November 2006 Archive 

I've made some progress with the "CX" graphic LCD controller boards. The photo below shows the introduction screen as seen on a Crystalfontz CFAG320240CX-TFH-T# display.

This model display is a "transreflectance" version, showing black characters on a light background, with the advantage that the characters may be read whether the backlight is on (transmission mode) or off (reflectance mode). The disadvantage of this format, in my opinion, is that it is not particularly legible in either transmission or reflectance mode. A major plus of this particular display model is that it is available with a touch screen overlay, which, it it had been available a year ago when I started the Z90, I would have used instead of the six soft key hardware switches.

Fortunately, the CX controller board is fitted to several different display types, including a blue/white CCFL version identical to the current Z90 display. However, the touch screen overlay is only offered by Crystalfontz as part of the transreflectance display.
 

I might shift to touch screen control, which has certain advantages, although it would require reworking the PCB layout. (The LED backlighting also will require reworking the PCB, but there's plenty of room for those revisions in the CCFL inverter area.) My main concern is whether the transreflectance display has sufficient visibility. I'll build up one or two CX-Z90's and see how they work for me, and perhaps circulate them to a few Z90 customers to solicit their views as well. That's a project not likely to be underway until next year, however. And, actually carrying it through to another run of kits depends on the continued interest the Z90. Based on my experience with the current Z90 & Z91 run, a future kit build will only be for the Z90 version and will require a minimum order commitment  of 40 to 50 customers.

I inadvertently tested the polyfuse today. I've been working on revising the Z90 firmware to use the newest design LCD modules with a S1D13700 controller. To do this, I've been using a Z90 PCB laying open on my desktop, with a soft key board plugged into it, and the display propped against my K2.

This evening, the PCB stopped working. Checking the voltage levels, I found the +12V supply was dragged down to a fraction of a volt and the polyfuse was very hot to the touch. After checking a few simple things, I found that the coaxial jumper between W201/202 and W101/102 was shorting the +12V line. A short but sharp bypass capacitor lead pierced the coax jumper's jacket and applied a dead short circuit to the +12V line.

Don't let this happen to you when testing your Z90.

I've posted a new Z90/91 firmware release, version 606, at the Software page. It makes the following changes:

A good part of yesterday and today were consumed by cleaning out the garage. Northern Virginia is an area of active termites, and the initial ground treatment when our house was built 20 years ago has dissipated and a new application is to be made tomorrow. Part of the treatment is to inject the anti-termite liquid into the ground around the garage's foundation, via a series of holes drilled through the floor around the perimeter. This means moving all the shelving and their contents from around the walls to the center of the garage. Although I threw out quite a bit of "stuff" there was still a lot to move. Then, we get to move it back on Wednesday, after the application.

Glenn, VK4TZL, has provided a photo of his Z91 sitting next to one of his K2s. I've posted the photo and his comments on the Z91 at the User Views page. Glenn also provided photos of his Z10000-K2 buffer amplifier installation and I'll post those as well on the Construction Updates page.

I've had a question about the odd odor from the panels. I believe this is from the protective film, and it will dissipate after a few days use. Or, you could lay the panels outside to air out while you are building the kit.

Today's E-mail also brought a few questions about the Z90. I've added the questions, and answers, to the FAQ page. The questions are:

• Why does the BFO leakage signal shift when the K2 mode is changed from USB to LSB and vice versa?
• Why can't I generate a signal above 60 MHz from the Z90's auxiliary signal output port?

I also had a request to modify the way the signal generator, custom IF and custom span, soft keys work. In particular, in the current software release, it can be difficult to set a value using the automatic repeat function, as when you stop pressing the button, the value continues to advance one step. I've reworked the automatic repeat delay function and corrected this behavior. As soon as I receive feedback from the requestor on the test firmware, I will make it available for general use. To some extent, the vertical position soft key suffers from a similar problem when it advances into the 10 dB/step mode. Unfortunately, this behavior is not likely to be altered soon, as it would require major reworking of the interaction between the screen display and button reading.

I finished both Z90 PCBs this afternoon and ran through a checkout. One board checks out perfectly, but the second has a problem with the 1 KHz bandpass filter. I suspect that I exchanged two capacitors when stuffing the PCB as the response has a dip in the passband. Or, I might have damaged one of the crystals by overheating when soldering the third wire to it. I'll dig into it tomorrow.

I didn't keep track of the exact build time, but I think it was about 16 hours total for two PCBs, no cabinet installation. There's only a modest time savings from building two boards in parallel. I used K7HBG's suggestion for finishing the coaxial jumper shields, wrapping them with small diameter bare wire, and it produces a nicer looking finish.

Stan, W5EWA, in an E-mail message to me, reminded himself of the benefits of the Z10010 4915 KHz bandpass filter provided with Z90/91's to be used with K2 transceivers. Here's his message:

Jack, I was experiencing some baseband lifting this morning on 20m. It was up 20 to 25dB at times. I thought I had something wrong with the Z90. I had everything disconnected while upgrading the K2 and hooking up the XV units. I was sure I had everything hooked up properly, and was puzzled as to what was going on. I finally looked around at the back of the Z90 and when I THOUGHT that I had hooked up the IF filter (Z10010), I actually had not. I had taken the short jumper that you provided and had it connected across the two connectors on the Z10010. Bottom line is, that when I hooked up the Z10010 everything started looking normal. It took the baseband down from 20dB to around 5dB or less. So it DOES work. I'm glad that happened. It gave me a real chance to see what the filter is capable of.

The Z90's dynamic range is about 60 dB, which is reasonably for a broadband device with four inexpensive crystals as the only IF filtering. However, the K2, as with most receivers, has a much greater dynamic range. When very strong signals are present within the K2's passband, an displacement of the Z90's baseband line (shift upward) is seen.  The K2's passband is on the order of 1 MHz, so the strong broadcast band signals in the 13 MHz band will be passed along to the Z90's input when tuning the 20 meter band, as Stan was doing. The Z10010 filter knocks down these out-of-band signals by 20 to 30 dB and thus significantly improves the Z90's dynamic range. The Z10010 won't do anything for strong in-band signals, and judicious use of the Z90's attenuators and the K2's RF Preamp/Attenuator will be beneficial.

Matt, KC0UKK, reports his Z90 is finished and provided a screen shot of a test signal, along with a question. See the Construction Updates page for the image and my answer.

I started building two Z90s for my own use yesterday and have one completed through Stage 6 and the other through Stage 5. I thought that after building six Z90's in various designs, that I didn't need the Assembly Manual. Wrong. I started before lunch yesterday and wanted to knock out a quick module before breaking for lunch, so I built two soft key boards. Well, actually, I built three soft key boards because I put the 10-pin connector on the wrong side of one of the boards. It was easier to scrap that board and build a new one than to try and get the connector off without pulling the through-hole plating out, even with solder wick and a Weller vacuum de-soldering station.

I also followed Matt's suggestion on crystal grounding and I've posted a photograph of his "three pin" crystal modification on the Construction Update page. I like it; seven out of the eight crystals have a ground via centered and those work well with the modification. The eighth crystal has a nearby via that can be used for grounding. I doubt this makes any improvement in performance, but it's a cleaner looking installation.

Matt, KC0UKK reports his Z90 is completed and operating, and I've added his comments, along with those from Glenn, VK4TZL, to the User Views page.

I've also added the following message to the Delivery Updates part of this page:

I've added Matt's (KC0UKK) suggestion for crystal grounding. Matt followed the technique developed by Bob, K7HBG, but pre-wired the ground lead. More details at the Constructions Update page.

I've also added a caution on cleaning solder flux from the LCD module after soldering the 20-pin connector. In a word, don't. The light diffuser can be damaged and result in in a splotchy backlight.

I received a couple of construction suggestions and photographs from Doug, N6TQS, today and have posted them at the Construction Updates page.

Friday, I received a trouble report about stray characters being seen on the Z90 display. The characters appear at random and remain in place until the Z90 is rebooted. (By the way, it isn't necessary to cycle the power switch to reboot a Z90--you can also do it by holding down the top and bottom soft key switches for approximately one second.)

I've had Z90's running for quite a while. The oldest has run more or less continuously for five or six months, and I have two others that have run close to 24/7 since mid October, without any indication of the stray character problem.

I asked the builder to return the display to me and sent a replacement to him today. In order to verify that I was sending a known good display, I ran one of the new displays for 24 hours without a problem. I then ran another new display for 12 hours without a problem but when I fired up the third new display I was able to duplicate the stray character problem. At the moment, it looks as if perhaps one out of 10 displays show this problem, but the statistical basis for this percentage is limited.

As you may recall, I developed the Z90 with two revision "C" Crystalfontz displays, which became obsolete a couple of months ago, replaced by the "C0" board, which uses a different controller chip, supposedly 100% compatible with the Epson controller in the "C" board. I've had one "C0" revision display running almost continuously for more than a month without a problem and a second "C0" display with at least two weeks 24/7 operation without an error. At least two builders have completed their Z90 kits, all with C0 boards, without display problems.

I had a lengthy chat today with an engineer at Crystalfontz, the display module supplier, and received assurances that if there is a problem with the displays, they will stand behind them. We discussed possible problems that could be seen with 10 or 20% of displays when tested with the same test bed and same software, but without coming up with a better answer than tolerance stackup on data transfer timing. The engineer suggested that I try increasing the time the data is on the bus before firing the strobe to latch the data into the display, although he agreed that the timing in the current Z90 software release was already generous based on the display's data sheet.

I've tried increasing and decreasing the various delay periods in the software, using the one display that exhibits the problem. So far, I'm unable to see a definitive correlation between delays and stray characters. In fact, when  I reduced the delay period to 200 ns, the display seemed to work better (but not perfectly) than with 1.2us delay. The display driver module, by the way, is posted on this site at the GLCD page and may be seen by clicking here.

I am continuing to work on the problem with Crystalfontz. If you have built your Z90, I would appreciate a brief report if you see stray characters or not. I assume that if you are seeing stray characters, I would have received a question, so this request is aimed more at those who have built a Z90 and have not seen a problem.

I'll copy this note to the Construction Update page as well.

I've had most of the day to work on my new K2-related project and it's moving along nicely. I will provide a bit more detail and a photo or two in a few weeks after working out a few more details.

The photo below is the Z90 Running Spares kit, which consists of one each of all the semiconductors, spare switches and the parts most likely to fail, such as the input attenuator resistors (should one make a gross blunder such as transmitting into the Z90's signal input.) and the connectors, as they are exposed to wear as they are mated/unmated. Not included are parts related to the display or the DDS module.

The landscape crew showed up this morning and managed to finish their work by 4 PM. The only radio related part of the job was to have some egg-size river gravel spread around the base of my tower. Looking at it now, it's apparent that it needs more of the same, but that will have to wait until spring. I imagine at that time we may wish to have more house-related landscaping work done.

I shipped two more Z90's internationally today. I believe that there are two pending orders for immediate shipment for which I am awaiting checks to arrive. All remaining orders are for consolidated shipping, when the replacement panels arrive. If you are expecting to receive your kit with interim panels and have not told me that the check is in the mail, please contact me. And, of course, contact me if you wish to change your mind and wish to take delivery now instead of waiting.

At this time, I know three Z90 kits have been built and are working, and one where the builder is almost finished but has a problem. These numbers do not count the three Z90's in my workshop for development and general band watching.

I find that setting the Z90 for 50 KHz span, 1 KHz RBW, skip mode "low" provides an excellent combination of bandspread and response time. I see about 2.5 scans/second in this configuration.

The day was otherwise  consumed with supervising the landscape work, picking up the last batch of bound manuals and similar things. I have found an hour or two for my next project and it is progressing nicely.

I had set aside the day to deal with a visit from the landscapers, but with 1.5" (nearly 40mm) of rain today, that job was washed out. It's supposed to rescheduled for tomorrow, but walking outside is similar to treading on a huge sponge. Perhaps by tomorrow morning the worst of the water will have run off or soaked in and the job can be completed. Since Northern Virginia is known for its dense clay soil, far more runs off than soaks in.

Since the landscapers did not show up, I used the time to finish preparing a third batch of printed manuals, and dropped them off at Kinko's for binding, to be picked up tomorrow or Saturday. My usual route back from Kinko's was closed due to localized flooding and the creek running through Clifton was as high as I've seen it in the 20 years I've lived here. Fortunately, our house is on a small ridge and flooding is not a concern.

According to the Post Office's on-line tracking system, my three international orders shipped this week are either in the hands of the destination country customs or in the air on the way to their destinations. Global Express Mail is pretty efficient; I mailed one package Tuesday and it at German customs today, Thursday. The packages to Australia left Wednesday morning from the Clifton Post Office are are in the air on the way to VK-land now, Thursday evening, and should be in Australian customs in another 10 or 12 hours.

The price of shipping a package the size of a Z90 kit via Global Express Mail is in the US$ 70-80 range, depending on destination country and the exact weight. I priced competing services from UPS and FedEx and found them significantly more expensive, on the order of US$ 125-150 for similar delivery speed. I also understand that customs clearance fees are lower when goods are shipped by post.

I believe that all domestic shipments for which I received payment have been delivered. The Post Office's system for tracking packages is slow to update, with delivery notification showing up electronically two or three days after actual delivery, and in a few cases, the electronic delivery notification has failed to show up at all. I've found the Priority Mail parcel service to be very efficient, but the Post Office's tracking system is not up to the standards set by UPS or FedEx.

At this moment, I believe that there are only three or four customers that wanted delivery with interim panels but from whom I have not received payment. If you intended to mail the check but have not gotten around to it yet, please do so. I would like to reclaim a bit more of the living room from its temporary shipping and receiving functions. Or, if you have recently mailed a check and have not received an E-mail from me acknowledging receipt, please let me know, as my practice is to let customers know payment has been received the day I have it.

Builder Alert--I assembled a Z90 today for a second development test bed and discovered that my supplier shipped a quantity of 2-56 x 0.5" spacers instead of 4-40 x 0.5" spacers. These are used only in the Z90, not the Z91. Some of these incorrect spacers have almost certainly been bagged with the Z90 kits. If you can't see the difference by eye, try hand threading a 4-40 screw into the spacers. It will be immediately obvious that the 2-56 threaded spacers are too small to admit the 4-40 screw.

Please check your spacers and let me know if you received 2-56 threaded spacers instead of the correct 4-40 spacers. Future kit shipments will be checked for the correct spacers, of course. I will also post this at the Construction Update page.

Doug, N6TQS, has provided another suggestion for Z10000 builders and also photographs of the stand he made for his Hakko 808 de-soldering gun. Doug's information appears at the Construction Update page.
 

I shipped the first Z90 internationally today, as well as two domestic orders. If you plan to take delivery with the interim panels, please send your payment and your kit will go out a day or two after receipt.

Doug, N6TQS, reports that his Z90 is assembled and working. That's the first "real" completion, after pre-release builds by K7HBG and W5EWA. Doug had decided to mount the CCFL Inverter using the socket option and found that the result is somewhat fragile. This is one of the trade-offs in socket mounting the inverter module versus soldering it to the PCB. I've built both ways and probably will opt for direct soldering for future builds, but different folks may reach different conclusions. I'll add Doug's remarks to the Construction Update page.

Stan, W5EWA, reports that re-flashing his Z90 with firmware version 602 went without trouble. He pointed out one discrepancy in my instructions on the Software updates page, now fixed--the mcloader program does not use the term EEPROM; rather the box that should not be checked is labeled Program Data.

I have added a revised Z90 firmware release, version 602, to the Software Updates page. The revision shifts one of the stock IF frequencies from 4914 KHz to 4915 KHz for a better match with the K2's IF frequency.

The firmware is installed into the Z90 with the mcloader.exe program provided on your CD-ROM.

This release applies to both the Z90 and Z91.

I don't intend to make further changes to either the Z90-Control software or the Z90 firmware until some operational experience is collected by Z90 users, barring, of course, bug fixes as may be necessary.

I've posted a note about measuring the Z90's current during checkout at the Construction Updates page. Read this if you use a digital multimeter to measure current.

I've added the revised Z90-Control software to the Software updates page. The revised program adds on-line help. If you are running an earlier version, you may wish to download and install this update.

A long day in front of the milling machine. I finished making all my remaining 0.5" x 0..5" Delrin bar stock into upper and lower struts. I started early, around 0630 and finished around 1400 and I made 21 pairs of struts. That works out to about 10 minutes for each strut. Adding the time to cut the stock into rough lengths and then finish to the final length probably adds another three or four minutes to the production time, but I did that 10 days ago. And, yes, this time I managed to keep my fingers where they belong and away from the milling cutter.

I still have about six persons who indicated they were prepared to purchase a Z90 or Z91, but who have not responded to my follow-up messages. I'll send out one more round of messages to those persons, probably later today. If I do not receive an answer within a reasonable time, I will regard their order as canceled and will re-open the order window after meeting any requests in the queue.

One more Z90 kit went out the door this afternoon. It looks as if the Post Office Priority Mail is pretty consistently meeting second day delivery, but its Delivery Confirmation service has real problems. On Friday, 3 November, I mailed four packages, each with Delivery Confirmation, and as of a few minutes ago when I last checked none of the packages show anything other than Acceptance at the Clifton Post Office on the 3rd. In one case, a customer let me know that his package arrived but I don't know the status of the others. (That package arrived Monday.) One day earlier, on the 2nd, I mailed three packages and received Delivery Confirmations reasonably promptly. In one case, the confirmation must have been sent to me via E-mail within a few minutes of delivery, and the others were a bit slower, but still showed up within a day of the actual delivery.

I've added a comment from Matt, KC0UKK, at the User Views page, with his first reactions after opening the Z90 box.
 

I have posted a couple of suggestions in Construction Updates, one from K7HBG and one from N6TQS.

I've added a help file to the Z90 Control software. After I test it with a some friendly users, I will post the updated program file and help file at the Software updates page. The help file is a copy of the Z90 Operating Instructions.

I picked up the box of bound documents this morning and the one outstanding order shipped at noon. All the packages shipped Thursday last  week were delivered on Saturday, and that should mean Friday's packages will be delivered today.

I dropped off a box of 10 sets of documentation to be coil bound this morning and will pick them up tomorrow. I have one outstanding order that will ship Monday after I pick up the bound documentation.

A series of posts in the Elecraft forum has discussed whether the K2's RF output has overshoot on some bands in SSB mode. In other words, does the initial few milliseconds of audio input cause the output power to exceed the target. I've run a short test at 7 MHz with varying degrees of audio drive. The results may be viewed at the Documents page or by clicking here.

After this morning's measurements, I ran a few errands and returned to start producing the next batch of documentation. I now have 10 sets printed, ready for me to insert the tabs and run them over to Kinko's tomorrow. I should be able to pick them up Monday. I also went through the Stage 1 parts bags, removed the two picofuses and replaced them with the 1.1A polyfuse.

At least two of the kits shipped Thursday and Friday have arrived. One to California and one to Michigan. That's excellent service from the Post Office for those two packages. I'm not sure about the other four, as the Post Office's tracking system seems to have variable update speed; sometimes the delivery will be reported within minutes and other times the report does not arrive until the late night batch update.

I have posted my conclusions and a link to the supporting measurements concerning top versus bottom crystal grounding in the Z90. My bottom line conclusion: No real performance difference between top bus and bottom individual grounding. More information at Construction Updates and the supporting data may be found by clicking here or visiting the Documents page.

Three more kits went out the door today.

I have now sent E-mail messages to all Z90 customers asking about their desire to take immediate delivery or wait for the re-made panels. About one-third of my E-mails have not been answered so follow up messages will be required. If you have not responded to my messages, please do so as soon as you can.

 I've been working on a more efficient method of handling the paperwork and the mechanical aspects of packaging the kits and getting them out the door with a minimum of errors.

All individual stage parts bags have long been finished (except that I have to open Stage 1 and replace the picofuse with a polyfuse) and are individual boxes. When I'm ready to ship an order, I print a detailed packing list / check list (with about 50 items) and check off the box when each bag or part goes into the box. I pre-prepared 10 Z90 cabinet assemblies, punched the chassis pan, selected front and rear panels, made up the upper and lower struts and the like.  It seems that small to medium size batches work the best for me.

Tomorrow's batch job will be to print and organize for binding six or eight document sets. I print these on an HP 4000 laser printer to provide the best B&W graphic quality I can and also to keep the documentation reasonably up-to-date. I then take the printed document package, insert tabs and deliver them to the local Kinko's for coil binding.

As I've mentioned before, documentation is a moving target and it's important that builders check the Construction Updates page before plugging in the soldering iron. Bob, K7HBG, has provided two good suggestions, including a new alternate crystal grounding method today.

I shipped three kits today, and will ship another three between tomorrow and Monday. Also, the last of my delayed cabinet order arrived from TenTec. I have (fingers crossed) no outstanding parts order, save for the replacement panels from emachineshop.

I expect to receive comments and suggestions and recommendations from builders that will be of interest to all those building the Z90 or Z91 kits, including the Z10000 buffer amplifiers. And, the written documentation that I mailed today with the kits is already out of date because of comments I received this afternoon from Bob, K7HBG. Accordingly, I've created a new page Construction Updates to present this information. Before you start building your Z90 or Z91, you should check Construction Updates. I've also added a PDF of the most recent errata sheets that may be accessed by clicking here or via the Documents page.

I've also had a question about the difference between a picofuse and a polyfuse; why Stan experienced problems with the picofuse and why I went back to the Z90's original polyfuse design. Here's an edited version of my answer:

To summarize the fuse types:

• Picofuse -- a traditional one-time only fuse, made in a very small form factor resembling an 1/8 watt through-hole resistor. When it blows, you have to replace it. It is soldered in, so replacing the fuse requires removing the old one and soldering in a new one.
   
• Polyfuse -- automatic resetting, when the overload is removed. This is a permanent device and should not require replacement during the life of the Z90. You may think of a polyfuse as a miniature solid state circuit breaker with automatic reset.

When the polyfuse is installed, there are no problems with switching on the Z90 using a high current power supply, as polyfuses are not nearly as sensitive to very short duration current spikes as are the picofuses. And, that short, high value current spike at turn-on was popping the picofuses. The current spike lasts only a few milliseconds, but that turns out to be enough to blow a picofuse. My error for not testing the picofuse with anything other than either a wall wart or an HP laboratory supply that I use for development work.

I shipped Stan's kit with a picofuse, as I had decided it was a better choice than the polyfuse I used during development. That turned out to be an error on my part, as my original polyfuse choice works better with the Z90, particularly when powered from a supply capable of delivering many amperes.

All kits are shipped with a polyfuse, and builders will not experience any of the problems Stan did with blowing the picofuses. (The Stage 1 parts bag says it contains two picofuses, as it was printed before I made the change. The errata sheet contains an appropriate correction.)

Speaking of documentation, here is the package that accompanies the Z90 and Z91.

1. Z90 Operating Instructions (approx 55 pages)
2. Z90 Assembly Instructions (approx 92 pages)
3. Assembly Support Documents (approx 23 pages total)
   3.1  11x17 fold out schematics
   3.2  11x17 fold out schematics with voltage and waveforms for troubleshooting
   3.3  Parts location map with index
   3.4  Parts list
4. Z10000 Buffer Amplifier Assembly and Operating Manual (approx 45 pages)
5. Z10010 4915 KHz bandpass filter Operating Manual (approx 8 pages)

Later tonight, I will finish packing up several kits to ship this week, except for the printed documentation. They should be ready to go out tomorrow morning after I pick up the bound copies.

Stan has an update that I've posted to the W5EWA build page. He is now finished with the build and has installed the enclosure and panels. Stan's total build time was 12H 25M. The Z90 should be a good three or four day kit, although I built one prototype board in a single marathon session.

I mentioned a few days ago that I would post a couple of photographs showing the Z90's struts being machined. Catching the end of my finger in the end mill threw a damper on further photos, but here are two that I took before my error. (No real damage, but it reinforced the lesson I should have known before to keep my hands away from moving cutters. I'm lucky it was only a nick that a band aid fixed as it easily could have been a serious injury. A 1" diameter sharp milling cutter driven by a 1HP motor is nothing to fool around with.)