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



April 2008 Archive


30 April 2008

I've updated the Z10000 Assembly and Operations manual to reflect the new component packaging and also updated and expanded performance measurements. I also fixed a couple of errors pointed out by builders.

It may be downloaded by clicking here or via  the Documents page. The new manual is version 3.0 and has the same file name as the older manual.


28 April 2008

I spent most of the weekend making performance measurements on a K3-IF Softrock Lite 6.2. I don't know if I will post the data to a new web page in the next few days or wait until my K3 arrives so that a complete analysis is possible.

Today was occupied in assembling kit packages for another batch of Z10000s. The photo below shows how I handle the surface mount resistors. The blue tray is a bead storage box with 32 round  bottom bins, purchased at the local Michaels Arts and Craft store. I cut each resistor section from the longer tape with the short scissors in the photo.

As a cross check on accurate packaging, I double count the resistors when I cut and place them in the bin. If all goes well, all bins empty simultaneously. I use a similar process for the rest of the components.

Z10000 kit Surface Mount Resistors to be Individually Packaged
25 April 2008

More maintenance this week. This time, a piece of quasi-consumer electronics with a throw-away design.

I bought a Magellan MAP 330 GPS around 1998, and have used it quite successfully since. However, the control buttons have gotten progressively less reliable and a few months ago the "enter" button ceased to function. Enter button failure made the MAP 330 unusable, as  the screen after power-up gives you five seconds to hit the enter button, by which action you acknowledge that the data provided is for entertainment value only, etc. The "lawyer screen" in other words, although I sincerely doubt this disclaimer would prove all that useful. After all, one does not purchase a GPS receiver for entertainment; nor does the manufacturer sell them as entertainment devices.

Regardless, without a functioning enter button, the GPS is unusable; it automatically turns off five seconds after turn on. I called Magellan's service center and found that the MAP 330 is no longer supported. The best deal they offered is a trade-in on a refurbished current production GPS, but the price after trade-in turned out to be about the same that a new model of the replacement current production model could be purchased from a discount supplier.

With the understanding that the value of my non-functioning MAP 330 is zero, I undertook a repair. The problem was reasonably obvious; the printed circuit board pads mating with tactile control buttons elastomer pins were likely were dirty or corroded.

The first problem is disassembling the case. My experience is that consumer plastic cases are either held together with screws or plastic melted rivets, often hidden under stickers. Or, if not screwed or riveted, tabs and mating slots are used.

After removing stickers, no screws or rivets were to be seen. Likewise, twisting and inserting a thin bladed screwdriver between the front and  rear case housings showed no signs of tabs and slots. Rather, it seems  that the case, made of polystyrene, had the two halves glued together. This makes for a waterproof case, but also makes a clean repair job impossible without a new case.

Still, since the defective MAP 330 was worth nothing, I used a cutoff blade in a Dremel tool to slice around the case parting line. I tried to cut as carefully as possible, as it was impossible to know how much clearance existed between the case and the PCB.

MAP 330 with the case cut in half. The white surface is an insulator and the PCB pads for the keys are in the insulator holes.

When in this state, I was able to activate the keys by pressing a metallic conducting rod through the holes to short the pad fingers.

Front of rubber key pad, showing 10 years of accumulated grunge.

Back of keypad. The  round black holes are actually molded pins of conductive rubber  that project through the holes when the corresponding key pad is placed. I measured about 600  to 1000 ohms across the pin.

Frequently used keys, such as the power on, the infamous enter key and the escape key have double rods for increased life.

I removed the plastic insulator from the main PCB and gently cleaned the mating PCB pads with GC's "no-oxit" contact cleaner using a cotton swab. The photo is after cleaning. There was an extensive black reside built up on the enter pad as well as several other pads.

Enlarged view of the switch pad.

In case you are curious about the rest of the PCB, here it is. The large black cylinder is a sounder. The antenna can just be seen at the bottom left of the photo.  It's a quadrifilar helix, I believe. At 1500 MHz, it isn't all that large.

At this point, I  reassembled the MAP 330, holding it together with a pair of rubber bands. It worked perfectly, but the rubber bands were not the long  term solution.

I  trimmed the worst of the melted polystyrene from the case halves, as I found the Dremel tool's cutoff wheel melted the soft plastic as much as it cut it, and ran a bead of neutral cure (non corrosive) Dow Corning silicon rubber around the gap.

It certainly isn't as pretty as before my repair, but the case halves hold together reasonably well. And, if another repair is required, the silicon rubber can be cut open with a sharp hobby knife and the bead peeled off.

While I had the case apart, I added a diode and series resistor to charge the batteries when powered from an external supply. I generally use the MAP 330 in my pickup truck, powering it through a Magellan accessory socket power supply. This puts out 3.9 volts, while the 2400 mAH NiMH AA cells I use in it are about 2.8 to 2.9 volts for the series pair when fully charged. I sized the  resistor to trickle charge the AA cells at about 15 mA.

What I find irksome about this repair is that Magellan designed the MAP 330 case so that it had to be destroyed to be opened. This design permitted the case to be spash-proof and certainly discouraged people from  trying to repair it with a jackknife. But, it's still annoying that a simple repair that took no more than 5 minutes once the case was opened turned into a two hour effort, almost all of which involved destroying the case and cobbling together a fix. That's not good engineering in my view. It may be excellent marketing, as it would have forced most MAP 330 owners to scrap their units and replace them, of course.

Post-repair, with a white bead of non-corrosive Dow Corning silicon rubber holding the case together.
23 April 2008

I've owned a Cubic R-3030 LF/MF/SW receiver for seven or eight years now. It's a dual receiver, two completely independent receivers sharing a common enclosure. The military version of the R-3030 is an R-2411 and Cubic makes the instruction manual for it (with schematics omitted) available at should you be interested in more information on the receiver.

A few days ago, the left side receiver failed to power up, and I spent some time troubleshooting it over the weekend. I thought it might be interesting to show how "the big boys" do their receivers´┐Żit's quite a comparison to even high end amateur radio gear.

R-3030 receiver

First, every section of the receiver, except for the front panel is in a separate shielded, plug-in module.

All but one of the modules has an LED "fault lamp" at the top, which illuminates if the module is defective. Of course, not all failures can be detected, but it's surprisingly helpful.

Each independent receiver has 13 shielded, replaceable modules.

When the modules are installed in the correct order, the black strip is straight.

In my case, the problem was the left side power module, which I confirmed by two methods. First, the power supply module fault lamp was illuminated. Second, I swapped the right side module with the left side module and the trouble followed the module. So far, so good.

By the way, a condensed troubleshooting guide is printed on the inside of the top cover, a small portion of which is illustrated below.

Power supply module. The switch selects 120/240 V AC power. There is no fuse in the module; the power switch is actually a combination power switch and circuit breaker.
Power supply with the top cover removed. The enclosure is a casting, drilled and  tapped for the shielding screws. AC arrives over the connector at the left; DC out through the right side connector, which is a DB25.

This is, somewhat surprisingly, a switching power supply. The AC section is to the left of the transformer and the DC at the right.
AC and Primary DC section of the switching power supply.

Working on switching power supplies is not my favorite recreation for two reasons. First, the AC and DC primary circuitry is directly connected to the AC power mains. Second, switchers are notorious for devilishly complicated feedback mechanisms, particularly during startup.

The direct AC mains connection imposes a major electrical shock risk, and also makes it difficult to use normal AC powered test equipment. Fortunately, I have an AC isolation transformer and was able to ground the primary side's floating negative, thereby allowing me  to use an oscilloscope and other normal test gear.

In this case, the power supply was dead; would not start. After some troubleshooting and head scratching, I decided that the most likely failure was  the UC2842 switching power supply controller IC, the large square at the schematic's bottom. The questionable part had a 1985 date code, so I didn't have high hopes that I could find a replacement without going to an obsolete parts house. Surprisingly, I found the UC2842 in current production by Texas Instruments and other manufacturers and I ordered several replacements Monday. While I had the module apart, I removed the old part and added an 8-pin DIP socket. Sockets are a mixed blessing for ICs, as a significant number of failures can be directly attributed to poor pin-to-socket connections. On the other hand, it is certainly much easier to  replace an IC in a socket than one soldered in place.

I was not 100% confident of my troubleshooting, however, so I also ordered replacements of the other possible bad parts. Here I had less luck finding exact matches. In particular, the switching MOSFET is a Siemens BUZ-57A which is no longer manufactured. It's in an odd case, resembling a plastic cube with a metal mounting flange. You can see the flange bottom in the photo, as I had to remove it for access to the UC2842. I found several other MOSFETs with similar ratings (1000V, 4.2A) but in conventional TO220 packages, and I ordered a couple figuring that in the worst case, there was enough room to make it fit.

My parts arrived today and installing the UC2842 in the socket cured the problem. After a few hours of on-time, all seems normal.

The R-3030 is an interesting receiver, with many useful features. However, the designers had an off day when it came to the AGC section. Its attack time is too slow (around 15 ms, instead of  the necessary under 5 ms) and although the decay time is selectable at .25, 1.0 and 3.0 seconds, all three options seem to have the same fast decay, measured in tens of milliseconds. Both receivers act identically, so it's not a fault in one side. The elastomer keypad is another annoyance as it requires the fingers of a safecracker to enter data without error.

20 April 2008

I've tinkered with the Using Softrock as a Panadapter for the K2 page. More usefully, I've added a section showing what comes out of the K2 IF port and how the out-of-band signals may be knocked down with a Z10010-K2 bandpass filter.

20 April 2008

I've recently acquired two HP 8657A synthesized RF generators and made quick and dirty phase noise studies today. The data includes comparisons with HP8640B, Panasonic VP8191A and Boonton 102D signal generators. And, for curiosity, I also ran a similar check of my Electraft K2 transceiver's receiver phase noise. Data at Signal Generator Phase Noise and Elecraft K2.


19 April 2008

I've added a new page "Using Softrock as a Panadapter for the K2" with photos and test results. Based on this data, the Z10000-K2 buffer amplifier can be run as recommended in the assembly manual; to provide around 0 dB net gain.

This new page represents a week's work in building and testing the Softrock.

17 April 2008

I'm continuing to work on the two Softrock Lite IF frequency evaluations. Yesterday, I tried an EMU-0202 USB-connected sound card with my new Dell laptop computer. (See my Softrock page for more information on the EMU-0202.) I'm seeing some strange behavior with the Dell, whether connected via the EMU-0202 or the Dell's internal soundcard´┐Żan audio "reset" every few seconds. This may be a laptop issue, or it may be in the connecting cables or other hardware. I hope to find the problem source later today.

I've also expanded by test equipment collection by adding two HP 8657A synthesized signal generators. One of my projects is a web page comparing my various RF signal generators for stability, phase noise and the like. No question that my HP 8640B is the cleanest source, but the 8640B is a mechanical engineer's dream, chocked full of gears and cams, with custom switches. The 8640B's are coming up on 30 years old now and the plastic gears commonly fail, and the printed circuit board switches stop working as well. My main 8640B no longer has functioning FM modulation, which I believe is a defective FM modulation switch, and my military surplus 8640B (option 323, no phase lock) has a defective bandswitch cam.

12 April 2008

I have now assembled two "official" Softrock Lite 6.2 IF frequency kits. One for the K2 and one for the K3. I've made preliminary measurements on both but it will be a few more days until I can mount the two Softrock receivers/panadapters in enclosures and run full tests.

I can confirm that the K2 version is around 11 dB less sensitive than the K3. This is expected because the K2 uses 3rd harmonic sampling, which has a theoretical sensitivity penalty of 9.5 dB. I measured a bit over 11 dB sensitivity reduction compared to a 10.125 MHz Softrock Lite  6.2 I built last year. The K3 version uses normal sampling and has similar sensitivity to the 10.125 MHz Softrock.

I can also confirm that the Z10000 is quite effective at reducing local oscillator leakage out of the Softrock K2 and K3 receivers, but until I mount the two Softrocks in the die cast enclosures, with filtered power supply, etc., it's difficult to make repeatable leakage measurements.

I have a couple of die cast enclosures ordered and expect them to arrive in five or six days.

Whether  the 11 dB sensitivity loss in the K2 Softrock is important, and whether you will wish to make up that loss by increasing  the Z10000's gain remains to be seen. The Softrocks are quite sensitive receivers, with a minimum detectible signal in the 0.1 to 0.2 microvolt range. Even with an 11 db sensitivity loss, there's likely more gain at 4915 KHz than necessary, except perhaps under extraordinarily quiet conditions, or, perhaps, for weak signal VHF/UHF operation.   I'll make measurements with the K2 Softrock connected to my K2 as part of the complete test. I don't have a delivery date for my K3, although the second receiver problem seems well on the way to being fixed, which brings my K3 closer to arrival, as the second receiver is the delaying factor.

The photos below are two of last week's visitors to our front yard.


10 April 2008

I've added a section to the Z10000 page showing how I mounted two of the amplifiers in Hammond die-cast enclosures and presents swept frequency response data for the amplifiers. The result is a useful general purpose laboratory amplifier with 10 dB  gain from 3 KHz to 175 MHz.

I've also added a table of contents with jump links to the Z10000 page to help navigation.


08 April 2008

I've added 3rd order intermodulation measurements of a Z10000 amplifier to the Z10000 page.

The short story is the amplifier shows an IP3 intercept of +36 dBm, an impressive number.

07 April 2008

I've had a couple international customers pay via International Money Orders, denominated in US$ recently. It turns out that my bank (BB&T) assesses a $5.00 "service charge" for depositing International Money Orders. Hence, I have to ask international customers wishing to pay with an IMO to include the US$ 5.00 fee.

PayPal is much more efficient in this case, with a decent foreign exchange rate and no international surcharge.

06 April 2008

I've added a section to the Z10010 bandpass filter page describing how I tune a filter.

After looking in out-of-the-way storage, I found two 4915 KHz (K2 IF) frequency bandpass filters that I had partially completed in 2006 but didn't need at the time. I'll finish those filters over the next couple days. I also found one complete 4915 KHz filter from my 2006 production, so I'll be able to ship three K2 filters this coming week.

I also built three 8215 KHz (K3 IF) bandpass filters with the last three 2006 PCBs and they will also be shipping this coming week.

Additional filters will be available when the revised PCBs arrive, still scheduled for mid-April.

05 April 2008

I've been asked how one acquires a Softrock Lite kit to be used as a panadapter for the K2 or K3. Tony Parks, KB9YIG, "Mr. Softrock" replied to my query as follows:

Hi Jack,
Yes, I can supply both of the IF kits you list below at a price of $12 per kit.  The K2 IF kit has a 4.898 MHz center frequency and the K3 kit has a 8.191 MHz center frequency.

You can send Tony the appropriate funds to his E-mail address and you will soon receive your Softrock kit.

I'll add this information to the Z10000 page as well.

05 April 2008

I've received confirmation from Bob Friess, N6CM, that the Z10000's isolation is well above my earlier quoted 70 dB:

Hi Jack,

The amplifier seems to work very well.  I am limited here in the desert by an old analog network analyzer, but S12 is something greater than 80 dB.


S21 is, of course, reverse isolation, signal applied to port 2 (output), measured at port 1 (input).

Hence, the correct isolation is in 80-90 dB range, as it's now clear my lower quoted figure must have been made with the power inadvertently disconnected. I've added a correction note to the Z10000 page.

04 April 2008

Yesterday and  today have been occupied with getting Z10000 orders out. To show part of the process, I took a few photos during construction of the SMA bulkhead cable supplied with the internal installation option. I assembled 12 cables today in the space of 4 hours, or 20 minutes a cable. I include a couple of breaks to check my E-mail in this time, so  that actual per-cable time is closer to 15 minutes, which is an embarrassingly slow pace.

RG-178 Teflon coaxial cable cut into approximately 2 foot lengths.

Ratcheting crimp tool, with interchangeable jaws. I made the drawing and instructions as the connectors came with little to no documentation.

My SMA connector storage box

Tools needed to assemble the connectors. A decimal inch scale and a sharp pointed pick for combing the braid back are necessary.

Stripping the outer jacket with Hakko FT-800 thermal strippers. I've set the depth gauge to the correct depth. As a backup measure, I've also made a mark on the jacket with a black permanent marker.

The thermal strippers yield a clean cut, with no risk of nicking the shield. I also use the thermal strippers for the center conductor

The parts that go into the bulkhead connector. The small white plastic insulator has a habit of  getting lost.
To solder  the center conductor, I use a piece of Teflon rod, drilled  to hold the center receptacle. This arrangement frees up both my hands for soldering.
Crimping the outer conductor.
Applying the heat shrink tubing boot over the connector.

Before I crimp the connector in place, I check the cable for shorts and opens. I check it again after criming and heat shrink application. Between the initial run in 2006/7 and the second  run, I've had to scrap two connectors out of about 65 assemblies. In both cases, the problem was a short, which I believe was caused by one of the center conductor strands not making it into the solder hole. This has lead me to the pre-crimp check for shorts and opens.
02 April 2008

As usual, I've moved the March 2008 page to archives, reachable by clicking here, or via the navigation table at the top of this page.

I shipped nearly 20 Z10000 buffer amplifiers today, in various configurations. Interestingly, about half the Z10000 purchasers are international. That's quite a bit greater than my experience with the Z100, where international orders were around 20% of the total. Given the current US/Euro exchange rate, US products are a bargain compared with historical exchange rates.