Clifton Laboratories 7236 Clifton Road  Clifton VA 20124 tel: (703) 830 0368 fax: (703) 830 0711



To search within the Clifton Laboratories site, enter your search term below.


Software Updates
Softrock Lite 6.2
Adventures in Electronics and Radio
Elecraft K2 and K3 Transceivers



October 2006 Archive

Question for my Elecraft K2 Readers

 I can now see the end of the Z90 kitting process and it's time to think about some follow-on projects. It seems that the main community of kit builders center around Elecraft, and mostly the K2.

What kits would you like to see that are not currently offered by Elecraft or other suppliers? I'm thinking of K2-related ideas, but am not necessarily limited to that. And, since I've grown rather fond of programming 18F PICs using Swordfish, the project should use a PIC someplace.

Drop me an E-mail with your thoughts. No promises, and I have some ideas of my own that may or may not pan out, but I would like to know what is on the K2 Elecrafter's wish list.

And, no I don't think that I can design and offer a version of the mythical "K3" general coverage receiver equipped transceiver, with 160-6 meter coverage and 100 watts output built in. It would be an interesting project to try, and I have some receiver ideas, but it's way, way beyond what I'm thinking of.

I'll leave this question at the top of the updates page for the rest of the month.

31 October 2006

Today was mostly a paperwork day, including more editing of the Z90 documentation and printing copies to be bound into the documents shipped with the kits.

Stan, W5EWA, has an interesting report on Stage 5 that can be read by clicking here and scrolling down. Based on his experience and the similar experience Bob, K7HBG, also reported today, I've decided to go back to the original design's over-current protection, a 1.1A polyfuse. I've discussed the issue in my comments on Stan's report, so I won't duplicate it here. However, this is exactly the reason to conduct a limited pre-release build. It's much better to make a change now, before the kits are shipped.

30 October 2006 PM

Today involved a bit of metalwork. I punched PCB mounting holes in 10 chassis pans. To quickly and accurately locate the six holes required in each chassis pan, I made a template that attaches to the chassis pan via four existing holes. I then use a transfer punch to mark the six holes to be punched. Then, I use a Roper-Whitney XX punch to make the 1/8th inch holes. The result is a clean, round hole with a minimum of burrs. While I was punching the holes, I knocked the sharp edge off the sheared end with a file. Still, use sensible caution when handling the chassis pan, as it will still have burrs and sharp edges.

I also did a bit of experimentation with various size boxes. The Z90 and Z91 kits are shipped in a 12x12x8 inch cardboard carton. Inside that are two (I think, based on today's work) boxes. One is 10x10x3 inches and holds all the enclosure-related parts. It also will hold the optional power supply "wall wart" unit. I think the Z91 may fit into a 10x10x2 inch sub-box, but all my experiments today were with Z90 enclosures.

The electronics will be in a separate 10x10x3 or 10x10x4 inch box. It's a tight fit into a 3" box, so I likely will use a 4" box with extra padding.

I've also been searching for a way to provide a serial number plate and have found a solution. The kit will come with three small tags printed on polyester base with the serial number and your callsign. These have an aggressive adhesive and once in place are hard to remove. Being polyester, they are not subject to tearing like a paper tag. The kits will ship with three serial number tags, one for the PCB, one for the temporary back panel or enclosure bottom, and one for the final back panel.

I also programmed some 18F4620's with the current Z90 firmware to ship with this week's kits. The firmware can be updated as new releases may become available, of course, via the Z90's serial port and bootloader software.  I also made a simple test fixture from a Z90 front panel and checked all the upper and lower struts I made for fit. All passed this test. Two struts are reserved for my use, as I made errors with them that resulted in extra drill holes. (It's hard to mess up when using a milling machine with a digital readout, but I misread 0.100" as 0.000" a couple times.) Those struts are fine, but the extra holes are not right, so they will stay here for my use.

I also revised the Assembly Manual to incorporate the three-page addendum that I sent to Stan and Bob for their pre-release kits.

Stan may provide a later construction update and if so, I'll post it tomorrow morning.

30 October 2006 AM

I posted Stan's Stage 4 results this morning and also added his Stage 2 comments that I had overlooked.

I've also added the grid reference document to the Documents page. Or, you may click here to view it.

29 October 2006

I finished enough struts to fill this week's orders. I will go through them and test fit to a Z90 panel to verify my work.

Stan, W5EWA, has finished Stage 3 of the pre-release build. I've added a page with his photos and comments. You may view the page by clicking here or via the navigation menu at the top left of this page.

28 October 2006

I made up the "Dropsies" bags this morning. The contents includes surface mount parts for the Z90 and the Z10000-K2 & U buffer amplifiers. Between the parts in the Dropsies bag and the extra gain setting resistors in the Z10000 kits, you should be covered for lost SM parts. It also includes hardware extras. As a courtesy, the hardware spares included with the Dropsies bag will be provided to all international purchasers as it can be difficult to find 4-40 hardware in a metric country.

I started making the Z90 struts. I cut up 60 some feet of 1/2" x 1/2" Delrin stock into 9 inch segments, trimmed to length and then milled a dozen or so to the correct thickness. At that point, I got a bit careless and nipped a finger. It's difficult to work (and to type) with a band aid on my fingertip, so I turned the mill off for the day. I'll get back to it tomorrow and at least finish enough struts to take care of this weeks shipments.

Photos of the struts under manufacture will be posted tomorrow when I can type with all 10 fingers.

27 October 2006

If you have not done so, please read the Update on Delivery Schedule at the top of this page.

Mr. Murphy, of Murphy's law fame, paid a visit this week. The parts with the longest lead time, the front and rear panels, turn out to have a problem--unacceptable paint quality in about 80-85% of the parts. In discussing the problem with Emachineshop today, it turns out that they had also done a quality check of the panels and knew of the problem, but shipped them to me without a comment.

The panels will be replaced without charge to me, but as mentioned in the Update on Delivery Schedule, not until mid-December or so. I will ship kits, if the customer agrees, with the panels I have and will supply a replacement as soon as I receive them and they pass my inspection. Or, I will wait and ship a complete kit with good panels later this year. If you are a customer and have not provided me with your option, please drop me an E-mail message with your choice.

This problem is more than a bit irksome as I was looking forward to providing a good looking product and getting the kits into the builder's hands within the next week or so.

So far, about half the responses have been to go with the existing panels. I appreciate that vote of confidence.

Yesterday and today have been consumed with this problem, including unpacking and repacking a bit over 100 individual panels. I will be back to a normal schedule tomorrow, I hope. I plan to start on machining the upper and lower support struts. It's more than a single day's job but I can at least make a good start on it tomorrow.

I also will try and post an update to the Swordfish DDS page over the weekend. I know that I'm behind on my postings there. When the DDS page is finished, I will try the inverse -- explain how one does analog-to-digital conversion with Swordfish and an 18F-series PIC.

25 October 2006

One open issue on meeting all the Z90 orders involved a few un-delivered BK-959 enclosures from TenTec. TenTec's enclosure sales manager, Jim Wharton, offered to repaint enough of their existing blue stock to meet my requirements and I accepted that offer today, as the earlier than expected front and rear panel delivery has accelerated my delivery schedule. Dealing with TenTec on the enclosures has been a positive experience and if you are in the market for standard or custom enclosures, I recommend them.

The BK959 (Z90) and BK929 (Z91) enclosures are expensive, but are very high quality, all metal enclosures. A significant fraction of the cost of production is represented by the enclosures and panels. Indeed, about as much money goes into the enclosure-related hardware, including contracting services such as shearing the chassis pans, as goes into the electronic parts of a Z90 or Z91, excluding, of course, the Z90's graphical LCD module.

I had several hardware parts bags short a few 4-40 screws and DB9 jackscrews. Those parts arrived yesterday and were added to the missing parts to the kit bags this morning. That should represent the absolute end of adding parts to the existing bags, absent some oversight on my part.

I'm going to knock out more revisions on the Operation Manual tonight. Today has been consumed with many small tasks that make it difficult to generate the sustained concentration required for writing. Today's tasks included establishing a Clifton Laboratories checking account, should anyone prefer to write their checks that way. It also included a visit from the termite inspection company, who confirmed that we have some infestation. It seems early enough that structural damage is not an issue, but the original ground treatment applied when our house was built 20 years ago has dissipated.

Based on my experience packing the two early build kits, I ordered more internal boxes, so that inside the main 12"x12"x8" cardboard box will be at least two smaller cardboard boxes; one containing the enclosure and the second the electronic parts. Within the electronic parts box will be sub-boxes for the more fragile parts, such as the GLCD, the GLCD inverter and the DDS module. Plus, of course, the ubiquitous foam packing peanuts and bubble-wrap. With the fragile parts thus triple boxed, I do not anticipate shipping damage.

24 October 2006

I received a surprise today�the UPS man delivered the Z90 and Z91 panels. They fit as intended and this means that the kits should be ready to start shipping early next week, assuming Bob, K7HBG and Stan, W5EWA, don't find major problems with their pre-release builds. Based on how well my pre-build went from random kit bags, I'm not anticipating anything more than some fixes in the documentation. I know the Assembly Manual needs new photographs in a couple places showing the final panels.

Today I also started revising the Operating Manual. It's another couple days work to finish that, and to work in the new photographs, print the documentation package and have it coil bound.

The panels are 0.062" steel, by the way, and were laser cut. The cuts are smooth and straight. Being steel, they are heavy. The color may not translate well through the Internet. It's a light gray powder coating, with a matte finish. You can see the texture when viewing the panels in person. The lettering is black and red.

Here are the photos. I'll make these a bit larger than usual and not add captions. Clicking on an image will give the full resolution version.


23 October 2006 PM

I packaged one complete (except for front/rear panels) Z90 this afternoon. I'll ship it off to Stan, W5EWA, in the morning, and I'll send Bob, K7HBG, his advance Z90 Wednesday, after the rest of the shipping and packing material arrives. Assuming things like work and the real world don't intervene, I hope to receive their comments on the build and documentation beginning this weekend.

The next couple weeks have some house-related activities, including new storm gutters and a visit to the landscape advisor at Betty's Azalea Ranch in Fairfax. The idea is to fill in the gaps caused by the tree trimmers cutting down the overgrown bushes and trees. I will continue working on various Z90 projects between these other activities.

The 1800 uF/25V electrolytic capacitors arrived today and I went through all of Stage 1 bags and replaced the 82uF/63V components with the larger value parts. Tomorrow should also see arrival of the 4-40 hardware and, fingers crossed, I should be able to complete the last bagging and tagging.

On the meteor scatter front, I hauled out my Icom R7100 VHF/UHF receiver and set it to LSB mode and dialed in the aural carrier frequency for Channel 2, no offset, 55.250 MHz, setting the receiver 1 KHz high to get an audible beat note. My R7100 is not nearly as stable as my Racal RA6790/GM which can be parked on a frequency and will stay within 1 Hz all day. I don't expect the R7100 to steady down until it runs for a day or two.

My reading of the FCC's channel assignment table says the nearest zero offset Channel 2 TV station is New York City. That should be close enough to see some aircraft trails as well as meteor pings. Baltimore's Channel 2 station is a + offset, so it is 10 KHz above 55.250 MHz. Channel 2- offset stations are found in Pittsburgh PA, Greensboro NC and Nashville TN. I can hear signals there but Channel 2 zero offset is stronger.

The offset of -10, 0 and +10 KHz is an interference control measure, by the way. The idea is that if your TV set should receive interference from an undesired co-channel station, the intentional frequency offset causes less objectionable closely-spaced herringbones, rather than broad black bars including potential synchronization loss. (Stations with identical channels and offsets are spaced much further apart than stations with identical channels but different offsets.) With so much TV being sent via CATV, there is probably a generation of folks that don't appreciate the problems of co-channel TV interference.


Here's a 55.250 MHz signal capture. As you can see, the R7100 is drifting, about 12 Hz over the space of the data capture (504 seconds). The main carriers are the bright white lines sloping downward due to receive drift. The weaker tracks are aircraft reflections. The ticks are 20 seconds, so knowing the carrier frequency it is possible to compute the aircraft speed from the Doppler shift. The curving tracks represent changed relative speed; as the geometry of the Channel 2 transmitter, my receiver site and the aircraft changes, the relative speed changes even if the aircraft maintains the same ground track and ground speed. And, of course, aircraft also change speed and course.



There's a serious use to which this type of data and analysis can be placed. What the image reflects is a primitive "bi-static radar" arrangement, where the radar transmitter (the TV broadcast station) and the receiving site (my QTH) are separated by large distances. With knowledge of the exact locations of the transmitter and receiver location, and better frequency measurements (and, preferably data from multiple transmitting and receiving sites) it is possible to determine the location of the aircraft generating these tracks, to a highly accurate degree.

This become important, according to the open literature, in detecting aircraft employing stealth technology. Stealth is apparently exceedingly effective at typical microwave radar frequencies. But, at lower frequencies, such as the TV low band and FM broadcast band, it becomes considerably more difficult to vanish from radar screens. Rather than small parts of the aircraft reflecting signals, the entire airframe radiates, or at least a significant part of the airframe radiates.

23 October 2006 AM

I ran the ARGO spectrogram program overnight, connected to my Racal RA6790/GM receiver, tuned to CHU at 14670 KHz. The antenna is a 25-1300 MHz discone at 85 ft above ground, normally used for VHF/UHF listening.

The Orionid meteor shower was still present last night and early this AM, I believe. I'll post a couple of the more interesting images.

An excellent tutorial on interpretation of spectrogram images of meteor scatter can be found at

The following two images are in time sequence. Note the "E" shaped artifact in the first image and  the intense bursts of strong signal in both. As I understand it, vertical lines are distinctive signatures of meteor trails--very high Doppler rate. The stranger artifacts are propagation from the trail of ionization left by the meteor and may last many minutes and will exhibit Doppler as the trail dissipates, and multi-path propagation, as seen on these images and the ones Ron, K8AQC, collected a couple days earlier. The reflective trails are at about 60 miles (100 km) elevation from what I've read.



Two more images in time sequence.



22 October 2006

I ordered the a few remaining hardware items today, as I ran out of 4-40 x 1/4" screws and Nylon washers. I also ordered the boxes and some packing material and finished the packing list document. The kits will ship in a 12"x12"x8" box. I packed a sample Z90 kit this morning and that size box appears ample.

Ron, K8AQC, sent along two more spectrogram images of CHU's 14670 KHz signal.

The image below represents normal "no signal" propagation mode. Traces of CHU's carrier can be seen intermittently.



The image below has some very unusual propagation paths. Note that the signal has shifted 2 Hz or so and there are two distinct strong traces. This is likely multiple anomalous paths or a single reflective path that is changing height, so that the result is a Doppler-shifted signal. I don't pretend to know how to interpret these images, but it seems clear that something unusual is going on. The unusual propagation conditions hung around for several minutes.


21 October 2006

I bagged and tagged most of the Z90 hardware today. I ran short on 4-40 x 1/4" machine screws and Nylon washers, so I won't complete that job until Tuesday when the additional parts arrive. I also have most of a check-list assembled, identifying all the various separate bags or assemblies that should be packed for a Z90 or a Z91. I also stopped at the UPS Package store and picked up a couple of 12" x 12" x 8" boxes to see if that size is suitable for the Z90 and Z91. I'm sure it is but the real test will be when I use them to send out the two pre-build kits next week.

At this point, the kitting process is almost 100% done. I have to pack the integrated circuits, and before I can do that I need to program the 18F4620's so I plan to do that after I make a couple small changes to the firmware next week.

On the normal ham radio front, I have an interesting image I received from Ron, K8AQC, an old friend from the Detroit area.

The image is a spectrogram of CHU at 14670 KHz, as received on a TenTec RX320D, early this morning during the Orionid meteor shower. The signal burst about one-third from the left edge is a remarkable signal reflection from a meteor trail, I believe. CHU is not audible at that hour on that frequency in Allen Park, and the only signals received are from various indirect propagation paths, such as aircraft reflections and meteor trails and scatter. Contact Ron directly at if you wish to know more about the image.


In addition to the Z90 work, I re-installed my 80 meter inverted vee after the tree trimmers put up the wire supports Thursday. That antenna has been up nearly 20 years and used plastic end insulators. When I climbed the tower today to attach the new Dacron support rope to the insulators, I found that the outer layers had turned to a chalk-like substance to a considerable depth. I imagine it is the effect of ultraviolet light, but I did not trust their mechanical integrity, so I replaced them with porcelain egg-type insulators. I know that it isn't considered good practice to use compression-type insulators on dipole ends, but I've done it successfully on many antennas. And, I used a good size compression insulator with a leakage path that's longer than the old plastic insulator.

I ran a series of sweeps with my HP8752B vector network analyzer after re-hanging the antenna. Here are a couple of the more interesting ones.


SWR from 3.5 to 4.0 MHz. When installed, I cut the antenna to be resonant around 3.6 MHz as my most common mode of operation is RTTY. That resonant point is still present. The 3:1 SWR point on the upper end is 3862 KHz. With the FCC's most recent frequency changes, it looks like it will be good for SSB as well.
Smith chart as displayed on the HP8752B.

I also ran a sweep on my M2 7-10-30LP8 "Skip" Log Periodic antenna. It's at 100 ft above ground and provides continuous coverage 10-30 MHz and also a segment of the 40 meter band.


SWR plot of the log periodic. The red reference line is 2:1 and the antenna is below 2:1 over the range 9.5 - 30 MHz and about a 150 KHz segment of the 40 meter band. The plot's frequency axis is logarithmatic, so the frequency lines are 7, 8, 9, 20, 20 and 30 MHz.

This is the second log periodic antenna I've used. The first was a KLM skip-band log periodic with similar frequency coverage. It lasted about 17 years but, as Matt Strelow, KC1XX, owner of XX Tower, said when he did the replacement job, "the aluminum got tired" and it was falling apart. The M2 antenna has higher quality construction and I expect it to last longer.

These compact log periodic antennas are adequate performers, and work as well as a typical three element trapped tri-band beam for most purposes (not as much F/B as a tri-bander) but permit full frequency coverage with a single antenna.

20 October 2006

I've started to bag and tag the hardware and miscellaneous items. I should complete most of it over the weekend, although machining the Z90's struts will take a bit longer. I cut the tin plate for the log amplifier shields today. I have a large guillotine-type paper cutter that I use for cutting PCB material. It did a nice job with the 0.008" tin plate as well. 

I've had a chance to work more with the power supply (wall-wart type) that I will provide as an option for purchasers. I purchased and evaluated two power supplies. One is a very nice, small, efficient switching supply, with universal voltage acceptance (100V to 240V, 50/60 Hz). The other is of standard analog design and works only with 120V, 60 Hz.

After listening tests and some spectrum analyzer work, I can only recommend the switching supply if your antennas are remote from your receiver. Like all switching supplies, it generates broadband trash. Some of it can be reduced with a common mode choke at the power supply, but there's still pickup on the antenna unless it is some distance away and connected via coaxial cable. Consequently, I'll provide the switching supply only if someone insists on it. I'm disappointed, as my early tests showed it to be relatively noise free.

The linear supply has no trash problem, but of course it is not as efficient and it is somewhat larger. Those are small prices to pay for reduced noise, of course.

I was supposed to receive the graphic LCD modules yesterday. When they did not arrive, I checked with Crystalfontz and found that they had not yet been shipped. That problem is now corrected, and the displays are en-route, to be delivered Monday.

After working with the production PCB for a week, I made a couple of small changes in components. The kits that ship will have the changed components. The main changes are increasing the value of the electrolytic capacitor in the power feed to permit improved operation where the power source has high ripple levels. My original design assumed a close to pure DC power source and it's apparent that isn't necessarily the case. I also increased the size of the heat sinks on the two 7805 regulators. The hottest temperature measurement with the old heat sink was 171°F and with the new, larger one, the highest reading is 134°F. I've run prototype units for the last six months with the smaller heat sink without a problem, but cooler is better. The DDS board runs about the same temperature, 130°F or so, a reasonable temperature for these components.


19 October 2006 PM

The tree service showed up a couple of hours late, but jumped into work immediately. The trees intruding into my tower and the guy anchors are now a memory. And, my inverted vee end attachments were installed about 45 feet in the air. One pine tree and one oak tree. That won't make much of a difference in the signal, I imagine, as the center is at 80 feet and the trees are quite some distance from the end of the wire. Still, it's a better installation than before. What remains is to climb the tower to the 20 ft level and attach the new support Dacron rope to the end insulators. I may have to snake one inverted vee end around the lowest guy wires as well.

On the Z90 front, the DDS modules arrived today from Mike at Kit Builders. I checked all the assembled units by plugging them into one of my Z90 units and found 100% functional.

I also prepared a hardware list for the Z90 and Z91 kits. This is an update of my general hardware spreadsheet to reflect things that I had overlooked, such as lockwashers.


An AD9851 installed on a Z90 DDS module by Mike at Kit Builders. This is a much nicer soldering job than I can do on AD9851 size chips.
19 October 2006 AM

I finished the Z90 Assembly Manual, version 2.0. It's quite a bit better than version 1.0, even if I do say so myself. It has almost all new photographs, showing the production PCB. It also includes the Z90 stages (the earlier manual was Z91-oriented.) It's available for download by clicking here, or from the Documents page.

In order to keep the document at a manageable length for download, I've selected 96 dpi graphic resolution. Purchasers will receive a paper copy of the manual and a higher resolution PDF version on CD-ROM.

Version 2.0 is a lengthy document, 96 pages, and in the original Microsoft Word format about 215 MB. While I like Microsoft Word, it is grossly inefficient working with larger files. It takes close to two minutes for an auto-save. And that is on a computer with 2 GB RAM.

I'm expecting the tree service company to arrive today to remove some trees that have grown to the point where they are interfering with other things. A couple are near my tower and guy wires and others to be removed are blue spruce trees that were cute little things when planted 20 years ago but are now 30 feet tall and obscure the front of our house.

While the tree service is here, I'm having them install antenna supports for my 80 meter inverted "V" antenna. The ends are presently supported at about 20 feet above ground, as high as I could reach with an extension ladder.

18 October 2006

Another long day, started at 0530 and it's nearly 2000 now. Other than a couple of short trips to the train station to drop off my wife and pick her up, I've worked on the Z90 project all day.

I made four sets of struts from Delrin square stock, experimenting with techniques that will speed up producing the number needed for the Z90 deliveries. The four sets will permit me to ship out two test assembly kits later this week and allow me to assemble a Z90 in a cabinet as well. None will have front or rear panels, of course.

The remainder of the day has been spent in further photographs and revisions to the Assembly Manual. The remaining part to write is the Z90 cabinet assembly. I hope to knock that out in a couple hours tomorrow.


The first real PCB assembled and in the Z90 cabinet, before the display and switch boards are mounted.
17 October 2006

I picked up the sheared chassis pans from the sheet metal shop this morning. They look good, nice clean cut with just a trace of a lip. I also received the "linear grommet" material and verified that it fits over the chassis pan edge.

The rest of the day was devoted to revising the Assembly Manual. I have now finished all of the electronic stages and the remaining sections relate to assembling the chassis, struts, displays and the like. I cannot finish that until the front panels are here, but I will wrap up the rest of the text tomorrow.

I also received a note that the DDS modules have been completed and were shipped to me today. I should have them before the end of the week and can test their functionality. My fingers are crossed, as these are VERY expensive modules, considering the parts and assembly charges. I used Mike, WA6OUW, at Kit Builders to do the DDS assembly, as he has extensive experience with the similar DDS-60 module.

16 October 2006

I've spent most of the day working on updating the Z90 Assembly Manual. I have 100 or more revised photographs to work into the document, and at least as many text changes, expansions, deletions and general clean up work to do. I hope to have that finished by Wednesday evening and will post it on the Documents page when completed.

In addition to new front and rear panels, the standard TenTec BK-series enclosures require a modified chassis pan in order to provide room for the Z90's display and, in the case of the Z91, to provide access for the front panel LED. This requires trimming 1 inch (25 mm) from one side of TenTec's stock chassis pan. (The chassis pan is the horizontal aluminum structure to which the PCB is attached with stand-offs. It's shown in the middle photograph of 15 October.) I've cut the 1" strip with my metalworking bandsaw in making the prototypes, but the result required a great deal of cleanup afterward to remove burrs. I opened all the TenTec boxes this morning and removed the chassis pans and delivered them to a large sheet metal shop near Dulles Airport today. They will trim the 1" strip using a power shear, and they did one sample for me while I was there. The edge is very nice, no burrs. The rest of the chassis pans should be finished tomorrow. This may seem like a small task, and it is, but by the time it's finished, it will have consumed about six hours, all considered.

I also asked the sheet metal shop to shear the 0.008" (0.2mm) tinplate stock for the log amplifier shield. This may not be possible with their equipment, as it is so thin and flexible that it bends rather than shears, or if it does shear, it will is with a bent edge. If it can't be done with machinery, I will cut the shield strips with hand shears. I received a short lesson on how to do that job without burs and bends while hanging around the shop waiting for the sample to be cut. The trick is to (a) use shears that do not have a serrated edge and (more importantly) (b) take small cuts, with the work near the hinge. Don't cut more than 25% of the jaw length in any single cut.

The 0.008" stock is thin enough to be cut with a stout pair of scissors. Part of the builder's job is to bend the 1" (25mm) strip of stock into an open-top box, solder the edge together and then solder the completed box onto the PCB. Depending on how good you are at making bends in the strip, you may need to trim a bit off the supplied length.

If you really botch the job, stop at the local hobby shop that sells model airplanes and look for the metal stock area. 0.008" Tinplate should be available in 4" x 10" lengths for a couple dollars. That's where I obtained the supply for the Z90 kits, via the Hobby Hanger in Chantilly VA. It happens to be located next door to Dancon electronics parts supply so I usually look around the Hobby Hanger when I visit Dancon. Hobby shops are a good source of some small tools and parts, such as short lengths of brass and aluminum stock, 0-80 hardware and the like.


15 October 2006

I finished the remaining stages shortly after lunch today. Total time between 10 and 11 hours, including photographs and annotating the assembly manual. I believe the extra time involved in documenting changes in the assembly manual offsets my experience in building Z90s, so your time will run about the same. The Z91 will be a bit shorter, by perhaps an hour or so.

I had no problems with the build. Everything worked as I hoped it would.

By the way, I found a source of 6-32 black oxide pan head screws at Fastener-Express. My usual sources, such as McMaster Carr or MSC do not carry this length.

The completed Z90 in the enclosure, without the front and rear panels and the top and bottom covers. It's running but no signal is connected. The top and bottom struts support the display and soft key board and allow the front panel to be made without screw holes, other than at the four corners.
Rear view looking at the display module. The front edge of the chassis pan will be covered with a plastic edge protector to prevent chafing the 20-pin ribbon cable. The cable is intentionally short to reduce RFI potential.
Completed PCB installed. The DDS module (on the left) is an earlier prototype. I'm expecting the assembled DDS modules to arrive around October 26th.

I've also increased the heat sink size on the 7805 regulator feeding the CCFL inverter. The original heat sink runs around 165 degrees F, which allows the 7805 to operate within its ratings, but that's still hotter than my preference. The new heat sink runs significantly cooler.


The next order of business is to revise the Assembly Instructions and dispatch the two pre-release builds as soon as the displays arrive. I have a good two days of work to revise the manual. And I also need to machine several sets of support struts to go with the pre-release builds.

14 October 2006 PM

I've completed Stages 1 through 4 in about six hours. That includes time to photograph and make notes, but since this is now the fifth Z90 I've built on a PCB, I'm further along the experience curve than most. I imagine that most Z90 builders will run about five to six hours to complete stages 1-4.

No significant problems so far; all the parts were in the correct bags. I did find an extra resistor that is necessary for the special features of a prototype but omitted in the production version. I should have deleted it from the parts list but failed to do so. Consequently, as a bonus Z90 kit builders will receive a 100 ohm, 1/4 resistor.

I'll provide a few photos of the build to date and comment on my new Hakko 936 soldering stations first real use.

The PCB completed through Stage 4. The DDS module is an early prototype, as the silk screened and solder masked DDS modules are still with the assembler.
Output from the Aux Output port. It's un-terminated, so the voltage is higher than will be delivered into a 50 ohm load. This establishes that the DDS is functioning and that the output Gali-74 amplifier is working.
Terminal output seen at the end of Stage 2. This tests the microcontroller and RS232 interface. Both are working.

And, yes, the firmware will ship with my callsign, but it will not take you more than a few seconds to program in your call via the Z90's serial data port. It's too much bookkeeping to program each PIC with a different callsign and then keep track of them. And, it's unnecessary since the firmware permits you to enter whatever call you wish, up to a maximum of 10 characters.

Hakko 936 soldering station with Macro handpiece and soldering iron rest.
I've now used the Hakko 936 soldering station for four or five hours. It's a better product than my Weller soldering station. The handpieces are smaller, lighter and easier to handle. It heats up much faster, about 45 seconds from cold start to ready to use versus about three minutes with the Weller. Small parts do not stick to the tip, as there is no magnetic field such as the Weller uses for temperature regulation.

I used both the Micro and Macro handpieces today. I started out with the Micro handpiece, but soon found that it does not have enough heat storage to handle soldering through-hole part ground pins. Although the Z90 PCB's ground pads employ thermal relief design on both the top and bottom, the Micro tips are too small to store enough heat to keep the solder molten. The PCB ground plane acts as a massive heat sink and there's not enough stored heat energy in the smaller Micro tips. I switched to the Macro handpiece and had no problem soldering these connections. In both cases, I maintained the adjustable temperature at 700 degrees F.

When building the surface mount sections, I switched back to the Micro handpiece and found it pretty close to perfect. You still have to be careful when soldering the ground connection on a surface mount part but it isn't all that difficult after you've done the first two or three. It's a matter of making a good thermal connection between the iron tip and the pad and part. This, of course, is an essential part of all soldering.

If I were doing this for a living, Hakko's dual soldering station would be the way to go. It has two independent controllers in one enclosure�a Macro handpiece on one controller and a Micro handpiece on the other.


14 October 2006 AM

I'll start the experimental build later this morning, after returning from the weekly excursion to the Burke VA farmer's market.

A few photos of the Z90 kit in its current stage:


Here is an almost complete Z90 kit. The missing components are the assembled DDS modules and the front/rear panels and a small bag with hardware.

The cabinet is at the top, along with the display module, installed in the support struts, along with the switch module (upside down, as the switches are on the left when installed). The layout includes a K2 interface kit, the K2 bandpass filter and all the Z90 parts bags.

Also not shown are the printed manuals.

The display and soft key switch board, looking at the side towards the front panel. At the right of the display is the CCFL inverter module that supplies about 600 volts to the illumination tube.
Reverse side of the display and soft key module mounted on the support struts.

The image is of the old display unit I use for software development, but the new RoHS compliant displays are physically identical.

13 October 2006

I've almost finished sorting and bagging parts. I found myself short of a couple components needed for the last stage, so I bagged everything else, and will go back and insert the short components when they arrive mid-week. I also need to package the mechanical parts, e.g., the screws, nuts, etc. that are not part of the standard TenTec enclosure.

Speaking of enclosures, I need to find a source of black 6-32 machine screws 1" long, fully threaded, Philips pan head, if possible. I can find these in shorter lengths, but nothing beyond 1/2." Other types, such as cap screws are available in black oxide finish, but not with the desired head and/or not threaded for their full length. These are for the Z90's front panel and I prefer a black screw if possible. If not, I have the required hardware on hand in bright stainless steel and will supply that. If you know of a source that will sell small quantities (200) of these, drop me a message.

Excellent news on the graphic LCD display front. Crystalfontz shipped me two variants of their RoHS compliant displays for approval and the first one I connected plays perfectly. No code changes, no hardware changes. That's a major relief, as the last thing I need at this stage of the project is to fix hardware or firmware to adjust to a new controller.

The new display model is CFAG320240C0-FMI-T and is one of the two replacements for the CFAG320240C-FMI-T that I used during Z90 development. The new display uses an RA8835 controller chip that is RoHS compliant and, more importantly, said to be fully compatible, pin-for-pin and without software changes for Epson's SED1335 and  S1D13305 controllers. Crystalfontz's old "C" controller used Epson's S1D13305 controller. Apparently Epson allowed these chips to go obsolete, rather than develop an RoHS version. Other than the controller chip, the new display is physically and electrically identical with the older one.

I phoned my approval and the displays will be shipped Monday, 3rd day FedEx, so I will have them in my hands Thursday. I received the CCFL inverters with the samples today. This leaves the front and rear panels as the only open item of significance. I have about 60% of the enclosures and will call TenTec next week to confirm when the last enclosures will be shipped.

I plan to grab a set of kitted parts at random from the pile of completed kits and build a Z90 this weekend. I'll start tomorrow and photograph the build to revise the assembly manual. I should be finished Sunday. Of course, the cabinet will be without a front or rear panel, but otherwise it should be complete. I'll post a selection of photos. If this goes as expected, I'll send out two pre-release build kits Thursday for an independent test build.

I had a chance to use my new Hakko 936 soldering station today. At Bob, K7HBG's recommendation, I also purchased the "micro" handpiece. I like it a great deal. It's much handier than my old Weller station and it seems to be ready to use within a minute or less of powering up. I'll take a photo or two of the Weller and the Hakko and post them over the weekend. If you are in the market for a new soldering station, I recommend the 936. If you can get it with the micro handpiece, go for it. Otherwise, you will spend more for the micro handpiece than for the complete 936 soldering station with a macro iron.

A follow-up on DigiKey. As promised, the correct capacitors were shipped and showed up early the next day and I finished bagging and packing that stage. A tip of the hat to DigiKey for excellent customer service.

12 October 2006

Another long day (about 10 hours total) sorting and bagging parts. I've finished Stage 1 and almost finished Stage 2. (Both include sub-stages as the builds are different for the Z90 and Z91.)

My intention was to make it easier for kit builders by separately bagging the parts for each build stage. After thinking about it more, this approach is likely counterproductive, or at least close to it, as the kit can be built in about 8 or 9 hours and thus several stages are likely to be built in any particular session. And, most kit builders wish to inventory all the parts before starting work, hence they will open all the bags. (I do not recommend this, but if you do this, pay very close attention to the surface mount capacitors as they are unmarked and can be identified only via the envelopes they are in or with a capacitance meter.) Hence, it would be little additional work for the kit builder, for example, to receive a bag of marked capacitors and sort them into value and then find the appropriate parts for any particular build stage.

Packing the kit this way, instead of sorting and packaging by build stage would save a great deal of my time, and probably be in keeping with the way the majority of kits are packaged. For example, the Z90 must use 20 0.1uF bypass capacitors. Almost every stage uses three or four or five of these parts. Instead of cutting 20 capacitors from the reel and putting them into the "capacitors" bag, I cut four for this stage, three for another stage and so on. This multiplies the number of times I have to handle parts several fold and increases the overall error rate, as every time I count, cut and bag adds another opportunity for a mistake.

In any event, I'm not going to change the packaging approach on this project but probably will in the future, assuming I do additional kits. I have a few things in mind, but can't even think about starting a new project until this one is completed to my satisfaction. I also have Z90 firmware improvements in mind and some modifications to the Z90 control software that won't be started until after the current versions are released. Plus, it will be helpful to have wider user feedback before undertaking these firmware and software changes. One of the changes may be to de-couple the Z90 and Z91 software (presently identical) as there are some potentially worthwhile things that I could do to the Z91 firmware that are incompatible with the LCD display routines. This would, of course, increase the firmware maintenance and development effort as I would have to support two parallel firmware loads.

Any comments from the kit building fraternity?

11 October 2006

I did my best to finish bagging and tagging Stage 6 today, but my efforts were thwarted by DigiKey. Stage 6 is the crystal filter section. I finished characterizing and packing all the crystals yesterday and today I started on the other parts, a couple dozen capacitors and four inductors.

I had ordered all the parts a couple weeks ago and didn't anticipate problems. However, when I opened the bag from DigiKey labeled 680 pF I found the parts were marked "471" for 470 pF. (The bag had the correct DigiKey part number for 680 pF printed on it.) I called DigiKey and spoke with a very pleasant sales representative who said DigiKey would send the replacement parts by overnight delivery and that she would call back with more information after checking with the warehouse department. I received a call a couple hours later and learned that 470 pF capacitors are binned next to the 680 pF parts and apparently someone grabbed the wrong parts. Since the value of the parts was modest, I was told to keep them.

Anyone can make a mistake and the real measure is how one recovers from it, and I was very pleased with DigiKey.

I found 15 680 pF capacitors of the same part number left over from my prototype builds, so I was able to finish kitting Stage 6 for 15 bags and the remainder are complete, save for the 680 pF capacitor that I will add tomorrow when the replacements arrive.

I also had a conversation with one of Crystalfontz's engineers about my order for the Z90's LCD displays. The model display used in the development models recently became obsolete due to RoHS issues and is replaced by two variants, with slightly different controller boards. Both are supposed to be "compatible" with the older display, one more so than the other. Crystalfontz is sending me one of each type for delivery Friday, so I should be able to make a definitive choice on which variant I will order in quantity. Of course, almost none of this information is explained on their web page, and in fact, the obsolete display is still shown as an orderable part. Crystalfontz has enough of the displays in stock to meet my order, once I decide with model to use.

The UPS man also delivered  the 0.5" x 0.5" Delrin stock to make the Z90's LCD support struts. I don't want to start machining these until the front panels are here so I can double check my design against the delivered panels.

I'll update the Swordfish DDS page later tonight.
10 October 2006

Today was "crystal filter day" at Clifton Laboratories. I started at 0730 and finished a few minutes before 2000, a good 12 hour day with a break for lunch and not much else. All the crystals for the 200 and 1000 Hz filters are bagged, tagged and ready to go.

After making a couple trial measurements, I decided to measure the crystal resonant frequency and Q using the automatic 3 dB bandwidth measurement features of my HP87510A. If the crystal passed my minimum Q standards, I marked it with the deviation from 8 MHz and put it into one of two trays, each with subdivisions. One tray for ≥ 8 MHz and one for < 8 MHz. I marked the tray divisions in 10 Hz increments and placed each crystal in the appropriate tray bin. I was able to measure a crystal, write the deviation on it and place it in the correct bin in about 25 seconds.

The bag of crystals waiting for measurement and sorting.
A crystal in my homebrew 12.5 ohm resistive pi test fixture.
After sorting all the crystals, I then grouped them by frequency and tested them in either a 200 or 1000 Hz filter test fixture.
A group of crystals, marked with their frequency offset from 8 MHz, ready to be grouped into fours to go into the test filter.
The filter test fixtures are on either side of the crystal parameter fixture. The filter fixtures have variable trimming capacitors, but these are not adjusted during the test.
A typical frequency response of a group of matched crystals in the 1 KHz test filter. I found the measured bandwidth ran from about 1020 Hz maximum to 915 Hz minimum. The bandwidth you will see in your Z90 will depend on the tolerance of the capacitors and inductors in your particular Z90, of course, but the prototypes I've built have all come in within this same range. The 200 Hz filters typically come in a bit narrower than the design value, usually around 170-180 Hz.

I counted the distribution of crystals in each 10 Hz bin, as an exercise to see if the expected Gaussian distribution would materialize. The flanks certainly appear to follow a Gaussian distribution, but the expected peak is not there -- in fact there's a dip, not a peak. I suppose this could be a bi-modal distribution. Or, it's just the luck of the draw.

I allowed 30% rejects for poor Q. In fact, the rejection rate I found today was almost exactly 10%, so I was a bit pessimistic.
For completeness, here's is a plot of one of the crystal sets for a 200 Hz filter, measured in the filter test fixture. This is very close to the response I've seen in the Z90 prototypes.

No Swordfish DDS update tonight. It's too late and I'm ready to knock off for the evening.


09 October 2006 (PM)

Stage 5 is bagged, tagged and ready to go. For some reason, this was the most difficult one to sort and bag so far. One of the cross-checks I use is to first count out the number of components that go into a particular stage and put them into trays. I then count out the number of bags or envelopes required for the stage and apply the labels. Then fill the bags with the appropriate parts. At the end of the process, the last part should go into the last bag and there will be no parts or bags left over.

It's easy enough to make a mistake when counting out the parts, of course. Stage 5 includes a Coilcraft WB1040 transformer and I opened the package today and subtracted from the total the number of transformers required for Stage 5 and determined that I had 10 more than I needed, exactly as planned. I removed 10 transformers and put the rest into a tray. After bag stuffing, I found one extra transformer in the tray. I went through all the completed bags (in clear plastic zip-lock bags, so no need to open them up) and observed a transformer in each bag. Counted the bags and that matched, not to mention the fact that the only component left over from the several hundred in trays was the sole transformer. I had my wife go through the bags and double check and she agrees all have a transformer. Coilcraft must have shipped one extra, or my wife and I both missed it. The real lesson, of course, is that I should not have trusted Coilcraft's count.

I need a break from sorting parts, so tomorrow I'll start grading crystals. Each Z90 or Z91 requires eight 8 MHz crystals,  four for the 200 Hz filter and four for the 1 KHz filter. Each crystal has to be graded for minimum Q and and matched for frequency. The normal rule is to match the crystals to within 10% of the filter bandwidth, 20 Hz and 100 Hz, respectively. I'll use my HP87510A and home made pi-style resistive crystal test fixture for the measurements.

Based on experience, I ordered about 30% extras. Extras are needed not for frequency matching, but rather to take care of low Q rejection. In the past, I've assigned each crystal a unique serial number and entered the Q and motional parameters, including frequency, in an Excel spread sheet, if the crystal meets the minimum Q standard. Then, sort for frequency and go down the list selecting 4 crystals in frequency order that are within the minimum spacing. That's a very labor intensive process. I could automate some of it via the 87510's Instrument Basic programming functions, or I could write a BASIC program to control the 87510A via its IEEE-488 interface. I'll give this more thought tomorrow, and see how long it takes to do an abbreviated test via the 87510A's crystal parameterization function. No need to record the crystal motional parameters, so all I need to is select for Q and note the resonant frequency and write an identifier on the crystal. Instead of a unique serial number, I could write the offset in Hz from 8 MHz, further simplifying things.

The final test I'll make after I group and select crystals is to plug them into a stand-alone test filter. These are duplicates of the 200 and 1000 Hz filter circuits in the Z90, except they are connectorized and the crystals plug into sockets. The performance in the test fixture is not necessarily equal to that in any particular Z90 due to capacitor and inductor tolerance, strays, socket effects and so forth. Still, it's a good final verification that the particular crystals shipped have a good chance of working as expected when you build your Z90. 

Of course, if I were building several hundred Z90's, I would call up International Crystal or another supplier and purchase crystals optimized for filters, matched for frequency and all meeting my Q requirements. That's scarcely practical at the Z90's current volume.

09 October 2006 (AM)

Bob, K7HBG, has alerted me to a couple of small discrepancies between the notations on the Z10000 buffer amplifier silk screening and the schematics relating to connector numbering. I've revised the schematics and updated the Z10000 Assembly and Operating Manual and posted the revised manual at the Documents page. The latest revision includes Bob's dimensioned drawing showing where he mounted the SMA connector in his fully loaded K2.

If you have a Z10000 board and have not yet assembled it, no need for concern as the discrepancies relate to schematic nomenclature. The schematics now identify not just the individual input, output and power pads, but also associates them with J90x connectors to track the printed circuit board's silk screening.


08 October 2006 (PM)

I managed to complete the tasks I wanted to today--Stage 3 parts are bagged, tagged and ready to go. Also, Stage 4 parts are bagged, tagged and ready to go. I also ordered the (I hope) last parts that I either missed or failed to order earlier in sufficient quantity. By mid-week, all the parts and packaging supplies I need to finish packing up the various stage parts bags will be here.

Bob, K7HBG, has also supplied a dimensioned drawing showing where he located the SMA bulkhead connector in his K2. Since his K2 has every possible option, he had no unused holes. I will add his drawing to the Z10000 documentation and post a revised copy later tonight or tomorrow.

A few photos of the work that went into preparing the Stage 3 parts.


Spacer chucked into my Myford Super 7 lathe, ready to have 1/16th inch removed from its length.
I used a parting tool to reduce the spacer's length. When you receiver your Z90 kit, you may notice some slight marking on the spacer where it was held in the 3-jaw chuck.
After each spacer was cut to length, I ran a 4-40 tap through it to ensure the threads were undamaged.

Holding the tap wrench perpendicular to the work is ensured by using a center in the lathe tailstock. The center's point fits into a recess in the tap wrench, preventing side thrust on the tap, which will, sooner or later, break the tap.

I like working with good tools, and have found that a cheap tool is no bargain if it spoils the work. I'm far from proficient at metalworking, but I have found a few things that help me get results far better than when I started out building electronic equipment. With respect to tapping holes, for example:


Don't depend on hardware store taps. For aluminum, I like (no, make it love) Greenfield's EM-AL series taps, designed specially for aluminum. This is a "gun flute" style tap and has a nitride surface finish.

These ain't cheap, but if you handle it right, it will last a lifetime. Actually, it will be an inter-generational tool.

A good tap deserves a good tap holder. I like my Starrett 93A and 93B tap wrenches. The 93A is smaller and better suited for 4-40 and 6-32 taps.
The single largest cause of broken taps is side pressure. If you do not start and maintain the tap exactly perpendicular to the work, you will break it. Maybe not in the first hole, but before your job is done, it will happen.

There are a variety of tools that hold the tap perpendicular to the work. I made the one at the right as a copy of a commercial product. It's a length of 1.25" steel bar stock that I turned down and knurled and then drilled holes through, sized to fit common tap shank sizes.

The bottom of the fixture is counterbored to give room for the chips thrown off during tapping to collect.

To use this fixture, grab it by the knurled ring and place the counterbored side to the work. With one hand, hold the fixture in place and with the other hand slide the tap through the proper fixture hole and start tapping. This, of course, assumes the work is flat.

Lastly, it's important to properly lubricate the tap. For aluminum, I like Tap Magic Aluminum, and for other materials, regular Tap Magic.

The small plastic bottle is a needle oilier. I found this gadget at a gun store and bought a few for general shop use. I keep one filled with Tap Magic - Aluminum and a slight squeeze on the bottle will deliver a small drop at the end of the long spout. Much better at delivering precise quantities of a liquid than the normal drop-dispensing spout that comes with the Tap Magic can.

And, of course, keep your tap clean; use a brush to remove chips before you start tapping a hole and periodically clean it during the tapping process.


I'll update the Swordfish DDS page later tonight.


08 October 2006 (AM)

Bob, K7HBG, has built the first Z10000-K2 buffer amplifier "in the wild," i.e., that is not part of the test build program and has made a few suggestions on attaching the 3-pin header socket to the underside of the K2's Noise Blanker board. I've added Bob's recommendations to the Z10000 manual, which you may access by clicking here or via the Documents page. Bob's suggestions are in the section beginning on Page 31. The manual is a 3 Mb PDF file. I also reduced the main text font size to 10 points to improve the document's appearance and save a few trees if printed.

Today's plan includes making the 7/16" 4-40 threaded spacers that support the DDS daughter card (part of Stage 3 construction). I could not find a stock 7/16" spacer, so it's a matter of chucking the 1/2" spacers into the lathe and taking off 0.062" or so with a parting tool. I'll take a photo or two of the process and post later today.

After modifying the spacers, it's back to parts sorting and stage assembly. Stage 3 was assembling the DDS board, but is now a matter of adding an 8-pin header to the main PCB and plugging in the wired and tested DDS module, so I should be able to get the Stage 3 parts bagged, tagged and ready to go. Stage 3 parts are the header socket, the modified 7/16" spacer and two 4-40 x 1/4" stainless steel screws, so the limiting factor is modifying the spacers. I hope to get a start, if not finish bagging and tagging Stage 4 as well.

07 October 2006 (PM)

A good part of the day was spent in sorting parts, and assembling components into bead boxes. The parts for Stages 1 and 2, as described in the Z90 Assembly Manual, are in their bead boxes, ready for packaging as soon as a couple of parts arrive. Stage 8, which is not in the on-line Assembly Manual yet, is complete--sorted, bagged and tagged and ready to go. Stage 8 is the Z90's soft key board. It's a rather simple stage, consisting of the PCB, six momentary operation switches, the associated key caps, a 10-position shrouded header and a 10-position ribbon cable.


Switches before being packaged.
One switch, close up. The switches are 12 mm (about a half-inch) square.

I'm taking photographs of each type of component and adding it to the stage parts list. That's very likely unnecessary for the experience level of Z90 kit builders. If you don't know what a resistor looks like, building a Z90 is probably not the best way to start your kit experience.

By the way, if you wish to "accessorize" your Z90, key caps are available in a range of colors. I'm supplying basic black switch buttons, but if you wish to order your own, white, red, green, blue and gray are available. The correct part is a 9 x 9 mm key cap for a 12 mm switch. Omron's part number is B32-12xx where xx are two digits identifying the color.

Mouser gets 18 cents each for them, and carries five colors. DigiKey also gets 18 cents for a key cap and carries six colors. Of course, if all you order from DigiKey is six key caps, their $25 minimum order will make them a bit pricey.

I bought a dozen or so each of several colors and after looking at them decided black was the best choice. White is not a bad alternative.


Basic black key cap that comes with the Z90.
07 October 2006 (AM)

I've posted updated operating manuals for both the Z10000 buffer amplifier and the Z10010 filter on the Documents page.

I've also added a new page Swordfish GLCD Module with source code for the library I wrote to control the Z90's quarter-VGA display.

06 October 2006 (PM)

I've had an exchange of E-mail messages with Emachineshop, the company I'm using to make the front and rear panels. The good news is that they caught a mistake I made. The front panels will be light gray (if you want to see the exact color, go to . It's "signal gray" color ID RAL 7004.) I had earlier tried a darker color with white silk screen (and red silk screen for the cautions) but decided the lighter gray was better. Unfortunately, I forgot to change the white silkscreen labels to black, as white letters against light gray is not a good combination. In any event, Emachineshop caught this and asked if I really wanted white lettering on light gray, suggesting black would work better. I agree and the change has been made.

So much for the good news from Emachineshop. The not so good news is that the current delivery date is the first week in November, about three weeks behind the original "55 days from August 23rd." The extra time means that I can finish all the other work, including assembling one  test kit myself and having two "early assemblers" also put Z90's together. These will be without a front and rear panel, but otherwise operational. Thus, the kits will be ready to ship within a day or two of the panels arriving. (He said, assuming that the holes in the panel line up with the parts on the PCB. This part, at least, I feel reasonably confident about, as I've made paper templates from the CAD drawings and they match perfectly with the PCB parts.)

I also have a query to Crystalfontz for an update on the LCD backorder status, but have not yet received a reply.

I found a suitable fan at the MicroCenter and installed it this afternoon. It's a generic single ball bearing 60 mm fan that cost $9.95, but it seems to move enough air to keep the CPU happy, and that's the important thing.

The Geeks to Go repairman was here for a brief spell this morning, and I used the time to discuss a few other minor issues that I've been experiencing with my XP installation. I gave him a short look at Swordfish and he spotted an error in the demo program up at my Swordfish DDS page. I had started to modify the program but did not carry through and hence had changed two variables in one place but not dimensioned them or initialized their values the last time I saved it. I restored the program to its original form, reassembled the test board and ran the restored code. I'll have more code analysis later tonight.

I promised a photo of the living room and parts warehouse. Here it is.

Fortunately, we do not use the living room much, so it's not a great handicap to use it for temporary storage.

Here's the DAC output with the corrected DDS program. If this sort of thing interests you, check my Swordfish DDS page and also visit the Swordfish home page.

Speaking of storage, I highly recommend the plastic "bead organizer" boxes made by Darice and carried at the Michaels Craft Stores in the bead department. These have 17 storage divisions and are inexpensive. I bought five today at $1.98 each and they are occasionally on sale at $1.50. I use them for parts storage. Michaels carries similar storage boxes in other departments at two or three times the price, so look carefully if you go there.


Here's a Bead Organizer box with some of the K2 bandpass filter coils.
I'll update the Swordfish DDS page later tonight. I've fixed the code error already but will also add text.
06 October 2006 (AM)

Computer problem diagnosed, if not solved. After thinking about it more,  the only explanation that made sense was the the CPU was running much slower than it's rated speed, as the problem was independent of Operating System. And, that the most likely explanation for that was that the CPU cooling fan had stopped working. (Recent Intel chips slow the processor speed way down as a power-reduction mechanism to prevent overheating.) Woke up at 0400 thinking about it more and decided to get up and pop the cover off the machine. Sure enough, the CPU cooling fan was not operating.

I removed the fan from the shroud and took it to the basement workshop and applied +12V. Still would not run and rotating the blades by hand showed some points of binding. After removing the nameplate sticker for access to the bearings, I applied a couple drops of light oil and got it running again. The fan is clearly on its last legs but with some luck it will last long enough to get a replacement.

The fan is a standard part (NMB 2406GL-04W-B50) but it does not show up at Mouser or DigiKey. A Google search is not promising either. With some luck, COMPAQ will stock it, or the complete shroud/fan assembly. I have a number of fans in my junkbox, removed from various computers and power supplies but, of course, nothing remotely matching this part in size or voltage.

05 October 2006

Not a good day at all. The 1.7 GHz COMPAQ desktop machine I use for E-mail and web browsing and as a general file and print server for the rest of the computers in the house acted strange this morning--like the CPU was immersed in molasses. CPU usage was pegged at 100% and moving the mouse caused the pointer to lag by 5 seconds or more. Powering down and rebooting was even more eventful. Booting took two hours and came up with a flag that the hardware had been changed and Windows XP was looking to be re-validated with three days.

I spent the day trying various things, all of which were almost impossible to accomplish as when it takes 20 minutes to launch a program and two or three minutes to respond to a keystroke or mouse click, trouble shooting is not easy. Windows Safe Mode has the same slowness problem.

I got out the Knoppix disk and managed to get the COMPAQ to boot with Knoppix and it shows the same slowness. Perhaps not quite as bad as Windows, but clearly not right. The Knoppix boot took nearly 30 minutes and although I've only used Knoppix a handful of times, I don't remember a half-hour bootup.

At this point, I'm at a loss as to what the problem is--possibly something BIOS related or a hardware problem seems most likely. The files are intact and can be accessed over the local area network, albeit at a glacially slow pace. All the Z90-related files are backed up on a separate computer (the one I use to update this web site) so this is much more of a problem for my normal computer usage (and my consulting activities) than it is for progressing the Z90 project.

I finally threw my hands up and called "Geeks on Call." Not cheap, but it's beyond my ability to diagnose and repair. I'll buy a new computer tomorrow or Saturday, but I need to get the COMPAQ running well enough to extract the data, or pull the two hard drives out and install them in the replacement machine. My main backups are on a second hard drive internal to the COMPAQ, which is probably not the optimum strategy, but I had not been happy with the external USB drive I also have as a backup system.

As far as the Z90 project goes, I had little time to work with it today. However, the DigiKey box arrived and that completes the major parts order. I suspect tomorrow is going to be devoured by the COMPAQ problem as well.

04 October 2006

Quick update tonight as my consulting project of September has returned for some additional work.

I've finished a few Z90 tasks before the message came in to jump back into the consulting project for a day or so.

  • I've packaged all the buffer amplifier kits into their distribution bags, which speeds up the final shipping and, more importantly, clears space to start on the main Z90 packaging.
  • My large Mouser order arrived, which enabled me to finish tuning and installing in their enclosures  the Z10010 bandpass filters. I was able to bring the one un-tunable filter into alignment after reducing four of the resonating capacitors by 5%, so the filters are packed in a box, ready to ship with the Z90/91s.
  • The initial TenTec cabinet order arrived. It's two very large boxes and one smaller box. In addition to the upper hallway being the assembly area, our living room is now the warehouse.
  • Shipped one more buffer amplifier, representing the last of the advance shipment requests.

No time to do a Swordfish DDS update today or to take and post any photos. Tomorrow should be back to Z90 work.

03 October 2006

Another vote for "Question Mark" has been received for the "name this insect" photograph, so I see a consensus building.

Yesterday and today have been busy.

The final main PCBs arrived late this afternoon. I have not yet carefully examined the boards, but there were no changes in part location or placement from the last successful prototype, so I'm not anticipating problems with the board.

The large rectangle is a the bare PCB ground plane where the log amplifier shield is placed.

I've spent a good part of the day tuning and assembling filters in their enclosures. I have 80% of the filters finished, and can wrap up the rest as soon as I receive the bag of 1/2" x 28 hex nuts I ordered Sunday. Each filter takes four hex nuts, as I use two as a spacer to control the distance between the filter board and the enclosure top. I originally thought the extra space was unnecessary and didn't have enough hex nuts on hand.  Each filter has a serial number and I'm capturing the filter response to be provided as part of the user's documentation. So far only one filter has failed to meet specification, although tolerance stack up means that some filters have a cleaner passband response than others.
The filter on the right is the original prototype, done with Manhattan-style construction using a PCB holder I made. I've discussed prototyping techniques at my Prototyping page. The filter on the left is one of the production units. The diagonal cuts on the production PCB are to permit the board to fit into the Hammond 1590A die-cast enclosure.
Yesterday, I received checks for two advance buffer amplifier orders and shipped them out this morning.

Tomorrow's task is to kit up the rest of the buffer amplifiers -- place the individually packaged bags into a larger plastic ziplock bag comprising a complete Z10000-K2 or Z10000-U kit. This will let me clear the decks at my kit packaging area for the avalanche of Z90 parts that will arrive this week. If all goes according to schedule, I can start packaging the individual Z90 build stages later this week. If the hex nuts arrive, I'll also finish the remaining filters.

Yesterday's Z90 work started at 0700 and finished at 2300, with a two hour break for errands and I hope to wrap up tonight a bit earlier as I'm not used to these long hours any more.

If you have tried to access the Swordfish page in the last few days, you have likely experienced problems due to a nearly week-long, system-wide failure by its web hosting service. The Swordfish site is mostly up now, but it may be a day or two longer until the last bits are restored.

The second Swordfish news item is that its official launch is scheduled for November 6th.

I'll add more to the Swordfish DDS page tonight.



02 October 2006

I have two entries in 01 October's "name this insect" mystery photo:

I'm going with Larry on this one, mostly because the Question Mark is found in this area according to some other Internet information I found by searching under its Latin name, Polygonia interrogationis, while Steve's suggestion, the Argynnis Paphia, seems to be a European insect. Still, the silver washed fritillary has a strong resemblance to the Question Mark. Perhaps they are related or it could be a case of parallel evolution. In any event, it was an unusual insect to see land outside our window.

01 October 2006

As usual at the first of the month, September's Updates have been moved to an archive page, reachable at the link at the top of the page. I'll also stop placing all the text on this page in bold.

I'm mentioned a few days ago that an important step was verifying the DDS Daughter board was correct, and that I was not looking forward to working with the AD9851 DDS chip, as I find it a bit on the small side for easy work.

One of the prospective Z90 purchasers, Bob, K7HBG, had earlier asked if he could receive his DDS Daughter card as a kit, since he enjoyed building circuits that use the AD9851. Bob recently agreed to build the first DDS Daughter board with the new PCB layout and take burden that off my hands. I awoke this morning to find a series of E-mails from Bob, reporting that he had built the DDS Daughter board the previous evening and that it worked exactly as it was supposed to. I was more than pleased to hear that. Now that the PCB layout is validated, I plan to pack up the DDS boards and associated parts and ship them off to the assembler in the next few days.

I've finished drilling the cases for the auxiliary K2 IF filter, and have assembled more filter boards. I should complete filter assembly by mid-week.

I also placed a major order with Mouser and DigiKey today, completing the last parts buy. I'm sure I've missed some small items but those can be obtained with a later order if necessary. The main Z90 PCBs have been shipped and will arrive Tuesday, 03 October. I will built one and photograph it to update the assembly manual, but I don't want to start building with the leftover parts I've used to for the prototypes, as a few parts have changed, mostly due to manufacturer discontinuation as RoHS parts replace those of older design. The RoHS parts are, of course, supposed to be identical with the ones they replace, but I still feel safer waiting until I can construct a test assembly with the identical parts that I will ship. And, of course, the front and rear panels are not yet shipped, the LCD modules are backordered until mid-October and about half the enclosure order from TenTec will ship this week, with the rest backordered until later in October. So, it still looks like the second half of October for Z90 and Z91 shipping. It's possible that the Z91's could go before the Z90's if the LCD modules are late.

I have a couple photographs of something totally unrelated to the Z90, or, for that matter, radio at all.

I have no idea what it is, but whatever it is, it landed on the railing along the back door.  For a scale reference, the vertical square post is 2" x 2". If you know what it is, drop me an E-mail. It resembles an autumn leaf when its wings are folded. It very late in the season to see a butterfly (or whatever this thing is) in Northern Virginia, as the flowering plants have long since dropped their petals.
I'll also add a new installment to the Swordfish DDS page tonight.