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

E-mail: Jack.Smith@cliftonlaboratories.com
 

 

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

 


Home
Up
Documents
Book
Software Updates
Softrock Lite 6.2
Adventures in Electronics and Radio
Elecraft K2 and K3 Transceivers

31 August 2006

I've completed measuring my K2's performance "before and after" installing the most recent IF buffer amplifier. I've posted a brief summary of the test methodology and a sample before and after photo at a new web page, K2 Measurements. I found no significant differences after installing the buffer amplifier.

In the next day or two, I'll also update the K2 Interface page to show the most recent board and cable installation, and to discuss BFO leakage issues.

 
29 August 2006

[Latest 29 August 2006. I've made some performance measurements of my K2, but all I've had a chance to post are the transmitter intermodulation distortion data. This shows a practical application of the Z90/Z91, by the way. The data can be seen by clicking on this link: K2/100 IMD Measurements.

My K2 has arrived at the local FedEx distribution center, so I should receive it sometime today.

I've made measurements of the typical output spectrum from the Z90's auxiliary signal generator output port, and also the Z90's signal input port. I also had an opportunity to examine the audio output versus signal level input for my Racal RA6790/GM receiver with AGC enabled.

 

Here is the typical stray leakage out of the Z90's input port, when set for a center frequency of 4915 KHz.

The strong pip at the left is 0 MHz (in other words, it's not a real signal; it's the spectrum analyzer's LO.). The strongest leakage signal is at approximately 26 MHz, at -78 dBm. This is the second harmonic of the Z90's local oscillator. -78 dBm corresponds to 28 uV in 50 ohms.

 
Here's a typical auxiliary signal generator output spectrum view. The Z90 is set to signal generator mode at 10.000 MHz.

The worst harmonic output is approximately -35 dB below the fundamental.

 
The purpose of automatic gain control (AGC) is to maintain the audio output at a relatively constant level as the signal input changes strength and to prevent overloading the receiver's gain stages.

In order to improve signal to noise ratio, it's common for AGC to be "delayed" or to allow the receiver to operate at maximum gain up to a certain, defined, input signal level. At that threshold point, AGC is applied and maintains the audio output more or less constant over a wide range of signal input.

AGC operation is often tested by plotting audio output versus RF input signal level. Rather than manually stepping a signal generator though a variety of levels and recording the audio output, I automated the process with a simple Liberty Basic program to control my Panasonic VP8191 signal generator via an IEEE-488 (GPIB) interface and to read the receiver's audio output with my Fluke 189 digital multimeter over an RS-232 interface. The computer program then saved the RF signal generator level and resulting audio output level to a disk file for later analysis and plotting.

I set the receiver to 14.025 MHz, CW mode, 300 Hz bandwidth, AGC = On, fast attack/decay, AGC threshold = default.

 
Setup for automated AGC test

 
Resulting data. AGC threshold is at -110 dBm.

Over the range -110 dBm - +10 dBm, output is essentially unchanged.

The Racal RA6790/GM allows the user to set the AGC threshold point via a front panel adjustment, and -110 dBm is the default setting, which I find adequate for normal use.
 
24 August 2006

My K2 has been shipped, and should be in my hands Tuesday the 29th. Frank, W6NEK, received the replacement LCD diffuser assembly from Elecraft and has fitted it in place. I'm looking forward to receiving the K2 for both testing with the Z90 but also to make some real contacts with. It will likely reside in the basement shop area, but my antenna feedlines are accessible from the shop area through a jumper panel.

I've also received the second revision to the K2 interface PCB and should be able to assemble one later today. Assuming no errors in the layout--or at least no errors that can't be fixed with an X-Acto knife and jumper wires!--I'll send a kit to Stan for an independent assembly and test in his K2.

I've also had time to add the call sign customization hooks in the Z90 firmware and in the Z90-Display software. This allows you to enter your callsign (maximum 10 characters) into the Z90 (or Z91) EEPROM memory. Thereafter, whenever you power up the Z90, your callsign will be displayed on the LCD, and for both the Z90 and Z91, when used with the Z90-Display software, your callsign will appear at the upper left of the graticule.

Still on my "to do" list is a button or check-box to reverse the scan sense. That is a bit more complicated, as the Z90 LCD display routines are only distantly related to the Z90-Control windows software. But, it will be done.

 

The new Enter Callsign Dialog box. It needs a bit more work to balance the vertical and horizontal space, of course.
 
I've also revised the Z90-Control software by deleting the callsign box I had added a few days ago as a temporary measure. I've kept the ability to enter a title.
 
23 August 2006

I ordered the longest lead time items today, the front and rear panels for the Z90 and Z91. Delivery time is estimated at 50 days, consistent with my planned October shipment date. My supplier quotes the delivery quantity as +/- 4 units. I've ordered enough panels to meet the order requirements assuming the shipped quantity is -4 units, so if the actual delivery is +4, I can fill a few more orders. Hence, if you missed the original cutoff date, there's still a fairly good chance that I can accommodate a few more purchasers.

The panel color is gray, with white lettering (red for cautions).

The Z90 front and rear panels in a 3-D view. (The color used in the image is not the actual panel color.)
 

I've also sent off a final prototype DDS board to my assembler for an independent test.

It looks as if I won't receive my K2 until early next week, as Frank, W6NEK, has not yet received the replacement part from Elecraft.

 
22 August 2006

The order window closed Friday, 18 August. I plan to place the cabinet orders tomorrow and will order enough front and rear panels to meet the orders in hand as of tomorrow morning, a few extras for spares plus ones I will use in my shack. My supplier quotes the volume as "+ or - 2" so to ensure that I have enough on hand, I will order front and rear panels to meet the current order list assuming the delivery is -2 of the ordered quantity., If it's on the +2 side, I may well have enough parts to provide a few extra kits. So, I'll accept orders for the next couple of weeks on a contingent basis, with the contingency being enough extra parts being delivered to me by the supplier and receiving the orders in the next two or three weeks, at which time I must order some expensive electronic components.

I've reconsidered the color scheme and am now leaning toward black cabinet covers with gray front/rear panels, white lettering (and red lettering on the rear for the caution warnings). I'll make a final color decision tomorrow morning but that's where I', heading now.

 
17 August 2006

Reminder - the Z90/Z91 order window closes Friday, 18 August 2006.

I've made a few changes to the Z90-Control software.

The revised software provides for a user-entered title line and also displays the user's callsign and the current date and time. This makes the image more self-documenting.
 

At the moment, the user's callsign must be entered every time the software is launched, but the next Z90 firmware revision will add the necessary hooks permitting the Windows software to read callsign information from the Z90's EEPROM and to write a changed callsign to its EEPROM.

 
I've also had an interesting discussion with a Scottish ham concerning signal levels on the 40 meter band. His measurements show that 7 MHz shortwave broadcast signal levels can easily reach 0 dBm or stronger with even a modest wire antenna. A contest-level station might see signal levels well over +15 dBm.

To obtain a sense of signal levels on the east coast of the US, near Washington DC, I connected my HP8558B spectrum analyzer to my main HF antenna. It's an M2 "skip band" log periodic, with coverage of 7.0 - 7.2 MHz and 10-30 MHz continuous. The antenna is at 100 feet above ground and was pointed at Europe (roughly 045 degrees) during the measurements. The measurements were made earlier this week, at around 8 PM, which is in medium twilight. The path over the North Atlantic, of course, has been in darkness for several hours at this hour.

 

6.5 - 7.5 MHz band. Strongest signals are about -46 dBm.   The signal levels in the amateur band, however, are lower.

 

   
0 - 20 MHz. The strong signal at the left border is the spectrum analyzer's local oscillator. The strongest signals are around 8 MHz and measure -32 dBm.

 

 
I don't consider my station and antenna as ranking near the top of the league, but I believe it ranks as "above average."

Later in the evening, signal levels may increase, of course. I also sense that with the demise of the Cold War, shortwave signal signal strength levels have significantly dropped in the eastern US. Radio Moscow, for example, used to consistently provide S-meter-pinning signals on several frequencies in the 7 MHz band. Radio Havana would offer similar signal levels.  It's been a long time since I've seen those stations push the S-meter against the peg.

So, what significance does this have for you? If you are not a receiver designer, it may help you appreciate the job done by receiver designers in building a product that lets you extract a weak CW signal (maybe at -130 dBm) whilst simultaneously handling signals 100 dB stronger on nearby frequencies.

And, if a similar measurement were made in the eastern US with a super-contest antenna system, at peak propagation hours, the measured 7 MHz signal levels would probably be in the -10 dBm range, or a more.  That's a LOT of signal strength, and at that level designers have to worry about things like non-linearity in inductors used for filters, for example.

The ARRL measured Elecraft's K2's dynamic range, which is a measure of how well the receiver section does at separating a weak desired signal from a nearby strong signal, at roughly 126 dB at 20 KHz spacing, dropping to 115 dB at narrow spacing. The received 3rd order intermodulation dynamc range is -96 dB to -66 dB at close spacing. The measured signal level plots should help you interpret how these numbers apply to real world signal levels.

A signal level of 0.3 uV (pretty weak signal and probably below the noise level unless you have a good, low noise QTH) is -117 dBm. Hence, you should not see dynamic range problems with a K2 unless the interfering signal is +10 dBm (-2 dBm for close frequency spacing.) This is, of course, a simplistic analysis, and you must consider the effect of multiple strong signals (3rd order intercept is an important performance measure here) and other effects.

 
16 August 2006

Stan, W5EWA, is a beta tester and reports that his Keyspan model USA-19HS USB-to-serial adapter works well with the Z90 prototype. I've added this model to the FAQ. I also found Keyspan makes a USB adapter with four serial ports. This may be useful for controlling multiple devices with a single adapter unit. I've added a link to Keyspan's site at the FAQ.

 
 
15 August 2006

I've added a new page describing how to measure (a) carrier suppression, (b) unwanted sideband suppression and (c) transmitted intermodulation distortion using the Z90, a coupler (such as Telepostinc's LP100 wattmeter coupler) and a two-tone generator. Click here to go to the page. In addition, you should read the related page IMD Measurements.

I've been asked to explain why, when using an Elecraft K2, some bands result in an "inverted" display, where higher frequencies display to the left of the center line. I've added a lengthy FAQ to address this point. I've mentioned it in the Operations Manual, but additional information may be helpful in understanding why this occurs.

Incidentally, inversion is not limited to the K2; the high IF output (40.455 MHz and 45 MHz, for example) of many commercial and military grade receivers is also inverted.

I've considered adding a "normal/inverted" option to the Z90-Control software, and I may do so. However, since the K2's normal/inverted status changes band-to-band, a switch may be less useful.

I've also made two more illustrative screen captures from my HP8558B spectrum analyzer.

 

This is a normal analog TV signal. At the left is the visual carrier and at the right is the aural carrier. They are spaced 4.5 MHz. The color burst subcarrier (3.58 MHz above the visual carrier) can be seen at 1 division to the right of the center. The visual signal is amplitude modulated on the visual carrier while the audio signal is frequency modulated on the aural carrier.

Note that the signal energy is far from uniformly distributed over the 6 MHz TV channel.

   
The digital TV signal, in contrast, spreads its energy nearly equally across the entire 6 MHz TV channel. The spike at the left edge is the pilot carrier.
   
14 August 2006

I've added an FAQ discussion concerning adding your callsign to the Z90's display and to the Z90-Control software's display. The short answer is that you will be able to customize both with your callsign, but that part of the firmware and Z90-Control software isn't in place yet.

I've been working with Stan, W5EWA, to test the prototype K2 interface board. Stan completed building and installing  K2 Interfacethe modified prototype yesterday and I've expanded the K2 Interface page to reflect Stan's test installation and to discuss design changes in the interface board. Stan's testing confirms the basic design works well, but we continue to discuss the optimum physical location for the signal exit and the buffer amplifier board location. A minimal K2, of course, is no problem, as there are ample extra holes in the rear panel and plenty of room inside. As you add options to your K2, however, space becomes more of a premium and the extra holes on the rear panel fill up. Stan and I are looking for a single solution that works even if you have every possible K2 option installed and still meets the quality consistent with the original K2.

Incidentally, in the K2, the Z90/91 also display your transmitted signal, since the K2 uses the same filters for transmit and receive.

 

Here's a photo of the Z90's screen connected to Stan's K2, tuned to the 40 meter band. The signals are shortwave broadcast stations.
   
The prototype buffer amplifier installed in Stan's K2. The input coax and power is at the bottom of  the photo, and the output coax is at the right center.

The prototype has been modified to introduce frequency shaping. The reasons for this and the builder's options are discussed on the K2 Interface page.

   
A couple years ago, I built a simple PIC circuit and associated firmware and Windows software to read the auxiliary output signals from my 1970's vintage HP 8558B and 8557A spectrum analyzers and display the results via a Windows program. I wrote the firmware in a combination of MBasic and assembler for a 16F876A running with a 20 MHz clock.  The Windows display software was in Microsoft Visual Basic for NET.

I wanted to update both the firmware and display software so that I can capture and publish spectrum analyzer sweeps when I work with the K2 buffer amplifier performance verification work I plan to undertake in September. Rather than revise my old work, I decided to start fresh with an 18F2420 PIC and Swordfish and write a new Windows display program based on the Z90-Control software.

The Swordfish compiler continues to impress me with its speed. With an 18F2420 running at 40 MHz (10 MHz resonator and 4X multiplier engaged), Swordfish will execute a 10-bit ADC read and store the result into a 256 element word array in about 27 microseconds. Swordfish's speed means that I did not need any assembler code as a 100% Swordfish version more than meets my speed requirements.

Here are a few images captured with my HP 8558B spectrum analyzer using the 18F2420 PIC and the new Windows display software. It's connected to a 25-1300 MHz discone antenna at 75 feet above ground. To improve the spectrum analyzer's noise figure, I've added a Gali-74 broadband amplifier between the discone and the HP8558B. The 18F2420 firmware captures 256 samples with 10-bit resolution and outputs the data via a serial port. The firmware automatically adjusts the sample interval as the user changes the span control and adjusts the analog video filter control.

 

Here's the narrowband FM weather station operated by NOAA in Manassas, VA. The signals below the main signal are other NOAA weather stations at 162.400 MHz (Baltimore MD) and 162.475 MHz (Haggerstown MD).
   
The FM broadcast band, 88.0 to 108.0 MHz. Lots of strong signals are seen here near Washington D.C.
   
WAMU-FM, an educational FM broadcast station operating with digital subcarriers. The signal to the left is an educational station at 88.1 MHz (probably WYPR-FM in Baltimore or possibly WYPF in Frederick MD) and  the one at the right is an educational station at 88.9 MHz (WVEP in Martinsburg WV).

WAMU divides its 96 kb/s digital subcarrier into a 64 kb/s channel and a 32 kb/s channel.

   
This is WETA-FM, another educational FM broadcast station. You can see the 19 KHz stereo pilot beacons and the analog subcarrier signals (talking books, I believe).
   
09 August 2006
I mentioned that I was working on a re-design of the Z90's front panel hardware to avoid screws or an appliqué. I've finished a pair of prototype struts that seem feasible. They are machined from 0.5" square Delrin or Nylon 6/6 material. (I made both Delrin and Nylon versions. Delrin is easier to machine.)

 

   
The struts run along the top and bottom and are held in place by the screws that retain the front panel to the side rails. I'll have to make a few changes in the front panel dimensions and also revise the soft key PCB to fit the new arrangement, but the outside appearance of the Z90 should be improved by removing the eight flat head screws shown in the prototype photographs.

The front panel installs normally, and is flush against the front face of the upper and lower struts.

   
I've been recently asked about using the Z90-Control software with computers without a hardware RS-232 port. I've added a discussion to the FAQ summarizing my experience with two different USB-to-serial adapters.
   
06 August 2006

I've recently had an extended E-mail discussion with one of the beta testers about what should be seen on a Z90, what the resolution bandwidth, span, dwell and skip factors are and how they interact. The "Quick Operating Instructions" provide some information on this subject, but it's worthy of more detailed examination. It's far to complex to make it a simple FAQ answer, so I've added a new page Display to address these topics. Take a look at it. 

 

05 August 2006

I've joined the K2 community. Or, to be more precise, I will soon join the K2 community. Since 90% of prospective Z90 purchasers own Elecraft K2 transceivers, it make sense for me to have one for testing purposes. (And, I have a few other ideas for accessories that might be of interest to K2 owners.) I don't have time to both build a K2 and keep the Z90  schedule I've committed to, so my K2 will be built by Frank, W6NEK. I've seen photos of Frank's work on both K2s and other kits and his craftsmanship is first class. His charges for assembly are extremely modest, considering the time involved.

Along the K2 line, I've finished assembling a prototype K2 Buffer Amplifier board from the PCBs I received on 03 August. The board layout works and I'll send a test kit, along with the draft Assembly and Operating Manual, to  Stan, W5EWA, when the post office opens this morning. I'll incorporate his comments into the next revision. I also have some ideas that I would like to incorporate into the K2 Buffer Amplifier, but those tests will wait until late August when I receive the K2 from Frank.

I'm also contemplating making the Buffer Amplifier kits available early (say sometime in September) so that they may be built, installed and tested prior to receiving the Z90/91 kits. If this is of interest to anyone, please drop me an E-mail message. There's also some utility in having a buffered IF output available for other purposes, such as looking at the waveform envelope with an oscilloscope. However, there's a difference between the optimum IF tap-off point for a panadapter connection and for IF envelope viewing. The panadapter connection must be ahead of the receiver's selective stages, but for envelope viewing, the connection is better made after selectivity. Otherwise, the oscilloscope will display a mix of the desired signal, all the other signals coming down your antenna,  local oscillator feed through and a variety of mixing products. In some rare cases, signal levels may work out to make this usable for an envelope view of a particular signal, but don't count on it.  If you are interested in an envelope view, it is possible to install a second Buffer Amplifier, after the receiver's selective filtering.  This is the approach taken in most professional receivers; they offer both a wideband panadapter-compatible buffered 1st IF output and a second post-filter, post-AGC buffered output.

I've also signed up for the Elecraft Reflector (you can read the archives at http://www.elecraft.com/elist.html ) and look forward to my K2 experience. I've ordered my K2 with the 100 watt amplifier, SSB and noise blanker options, as I believe that should be representative of the most commonly owned options. 

I've added a new page to the site, Surface Mount Assembly, an extract from the draft Buffer Amplifier Assembly and Operations manual, that relates the techniques I've found to work for me in installing surface mount components. I make no claim that these are the only procedures that work, or that these are the best procedures. They work for me, but if you have a different approach that works for you, I wouldn't change it.

 

03 August 2006

In order to keep the Updates page manageable, I've decided to monthly archive the prior month's material. To read a prior month's Updates page, click on the corresponding button at the top of this page.

In the last couple weeks, I've received several E-mail messages with questions about the project. Since some of these questions and my answers may be of broader interest, I've reformatted the correspondence and placed it on the FAQ page.

I should receive the new design K2 Buffer Amplifier printed circuit boards today. I've completed a draft manual (28 pages, and I never thought there was that much to say about a rather simple circuit) . I'll post the manual after I complete a test build of the new layout and after Stan, W5EWA, has a chance to build a second prototype and critique the manual. Since Stan owns a K2 and I don't (I suppose that I should own a K2 and probably will in the not too distant future) I will rely upon his comments for what works and doesn't work in the K2 installation.

A few prospective purchasers are interested in an assembled unit. Dario, N5QVF, has said he is willing to assemble Z90's and Z91's and I'll post his price list and contact information on the Pricing and Ordering Page soon. One of the FAQ's discusses the "degree of difficulty" in assembling the kit. I'm a bit handicapped in accurately assessing this, as I've now assembled four complete PCBs in various prototype designs, which biases my ability to approach construction from the prospect of a first time builder. Hence, my answer to this blends my views and Dario's experience in the independent Z91 assembly.

I've not been happy with the Z90's prototype front panel, as it has eight 4-40 flat head black oxide machine screws to hold the switch board and LCD in place. I much prefer a clean front panel, with, at most, the four corner screws. There are several alternatives to achieve a clean front panel, such as spot welded 4-40 studs through covering the current design with a Lexan surfaced appliqué. With a volume measured in a couple dozen units, none of these are ideal solutions. I've decided to work on a different approach, with an internal support arrangement. If my current idea works, I'll post a photo or two. If it does not, then I'll revisit other alternatives.