Inspired by BH1RBG, I cracked open the HP8640B to have a look at the counter circuitry. Above is the view that greeted me. That is the main counter board after I pulled it out of its socket. You can see the seven little red LED display modules.
It is not as bad as it looks. In fact, I found the construction and accessibility of the HP8640B to be quite impressive (much better than the Tek 465 with all its flaky plug-in transistors). The manual has good, detailed info on how to get into the various compartments, and even as you work, instructions on which screws to remove or loosen appear on the tops of each RF-tight compartment. Nice. This thing was obviously built with the needs of a future repairman in mind.
Above is that same board flipped over. Again, not as bad as it looks.
Above is board A8A3, the board that I suspect is causing me trouble. When I go through the troubleshooting routines in the manual, I get to the point where they check decimal point position. Everything is fine UNTIL I GET TO 16 MHZ. Then the decimal point is not where it should be and the frequency displayed is very wrong. (This is in the internal mode -- the counter works fine with an external signal source). The manual then sends me to Service Sheet 15 which points to possible problems on this A8A3 board, U3, U7 or U6D.
Troubleshooting this will be tough. I do not have the extender board that would allow me to test this A8A3 board with the other counter board raised up above it and operating. BH1RBG noted that getting the extender boards is almost as tough as getting the HP8640B itself.
While it really bugs me (!) to have a part of this device not working properly, I could just leave it as is. The signal generator is working fine, and I could use the external counter input to check the frequency. But this is a real kludge.
What do you folks think? Fix it or leave well enough alone? Discretion or valor? Anyone have an extender board? Any ideas on where the fault might be?
Go for it, Bill! Just take your time and enjoy the education. Ponder in excess. Begin by checking voltages. Attach the oscilloscope and carefully follow the logical paths. Seldom do chips or semiconductors fail. Usually capacitors. Sometimes resistors may change value but most often it is simply a connection that, due to lack of exercise, has gotten lonely. So be sure to frequently go back and see if it hasn't resumed operation.
ReplyDeleteTread gently with abundant mindfulness. 73, Don
Bill - if anyone can, you can. Go slow, consider all the signs and plan your moves. Sucess! Ed
ReplyDeleteWhat do you have to lose...
ReplyDeleteGo in and tack a wire (or two or 3)to points on the circuit board that you want to check. One that is long enough to be brought out where it can be accessed with your test probe. I know you're at 16 MHz, but what you are really interested in is whether or not there is any digital activity at the point(s) of interest.
Once in a while, even a blind hog roots out a few acorns... Best wishes. Bruce - KK0S
I've not looked at the schematic but I would look around the control signals going to the digital counters/dividers. Check the chip that drives the LED driver for the decimal point. The finger test may find a failed chip overheating.
ReplyDeleteGood luck. I'm sure we'll soon read of your success finding the problem.
I think you might be overthinking the failure mode here. I had a similar problem with my unit and it was one of my early Covid-fix-it projects. There is an assembly that controls the bands on the front. It has the famous gears that crack. On the back side are two sets of rotary switches that control a lot of stuff. The switches are of a very unique HP design and offer a lot of advantages over traditional switches EXCEPT they fail in an unusual manner. A PC board has all of the interesting wiring and very tiny double leaf springs short tracks together as needed. The springs tend to break away from the plastic posts on the rotating plate and not make the needed contact. Very carefully examine the insides of the 8640 and your bench top to see if any have fallen out. These are difficult to buy or fabricate. The disk is designed to be rotated 180 and use a new set of posts to locate the springs. Use a small dab of epoxy to set in place. The totally mechanical repair fixed my unit that also would not read above 16 MHz. There are a number of web pages that give great detail of this repair. Also it is worth while to replace the Delrin gears if they are cracked. Replacement brass ones are available on eBay and they will permanently fix the gear problem.
ReplyDeleteThanks -- That would explain it. I flipped it over and at first didn't see anything wrong. But when I moved to put it right side up, I saw something move. Sure enough, one of the little springs! I wish it had been a transistor or a TTL chip! Getting that thing back in there does NOT look easy. Discretion may override valor at this point. But at least I now know what is wrong. And wow, that HP8640B is quite a piece of work. The innards look like something out of the Apollo program. I will put up a blog post about all of this, with pictures. Thanks very much and 73. Bill
DeleteGlad to help out Bill. If you have not stumbled into it yet here is the blog of K6JCA that goes into amazing detail of exactly how to do this repair. I used it and it was so much easier to follow in the steps of someone that had already done it. Make note of the snap ring plier also, it makes it much easier. http://k6jca.blogspot.com/2018/09/repair-log-hp-8640b-rotary-switches-and.html
DeleteGood luck with the mission. 73 Dave VE3EAC