HP 8662A repair part 1 - recapping the power supply and next steps
Вставка
- Опубліковано 28 чер 2023
- New power supply caps have arrived. Some are from a new(!) line of high performance axial leaded caps from Vishay. Will this fix the #10 error?
Update!! I just discovered that FM is working! Chalk that up to operator error.
If you'd like access to support the channel and gain access to bonus content, please consider becoming a Patreon here: / bandersentv - Наука та технологія
i just snagged a unit here in the UK for £32.50 after seeing your video, it was not powering up at all but i have replaced 2 caps on the A4A7 motherboard and voila i now have a working unit with no errors at all, i cant believe it think i might go buy a lottery ticket, i did have to drive 250 miles to collect it in person but total cost all in to me has been about £140
Fantastic! You got an outstanding deal 🙂
The build quality on these units is insane!
My hp 8640B signal generator (also a fantastic machine but many more mechanical intricacies that you have to address to restore it) has a tiny wrench mounted inside just for making those mini coax connections! They had the service guy in mind for sure. The 8640B is cavity tuned (not synthesized), and uses a calibrated mechanical attenuator for output level, but it has a neat digital counter/display that doubles as a PLL to lock the frequency to the accuracy of the internal timebase to correct for thermal drift. Very stable once locked. Sort of an analog/digital hybrid and certainly a wonder of 1970’s engineering.
I got mine out of the junk pile when I was in grad school and it had MANY unrelated failures. Bad integrated circuit in the digital counter, broken plastic gears which fortunately someone sells pricey brass replacements for on eBay, dirty slide switches, you name it. It was extremely gratifying to finally get it working perfectly. I’m sure you’re on track for the same experience!
Have you visited the HP memory project website? They have a good section on the 8640B www.hpmemoryproject.org/wb_pages/wall_b_page_10b.htm
Fascinating to see the evolution of signal generators.
I was just recalling that HP wanted to make sure those mini-coax were tight but also not too tight wherever they were used. Finger tight could just induce some minor errors sources...over tight and a broken/over-stressed connector.
I spent a lot of time rebuilding an 8662A more or less from a box of bits I got from a guy. The first problem was a non-working PSU. Apart from recapping I found a not-dead-but-sickly transistor that was upsetting the PSU switching. The other thing I found was one of the two attenuators had a 40dB stuck pad. They were clicking but the contacts were not working (due to a well known problem with failed o-rings in the attenuator pins). Replacing that meant that the power output suddenly corrected itself - so it is worth putting the output in to a power meter as this can help to narrow problems quite quickly. I too had a problem with the AGC - at some frequency ranges the AGC circuit oscillates and I've not found a way to prevent it - however, by changing the power level down and then back up to the original setting I have found that this quenches the oscillations - so it is a work-around rather than a fix. The clipping issue is odd and I don't have a theory about that one apart from whther there is some GC bias on output? Using a spec analyzer will be very useful for troubleshooting. Good luck - and please post some updates - it is a super (but complex) instrument and well worth the effort of getting it working again.
There are two more parts in the series. I did solve the clipping and lack of AM modulation.
Glad to read you did get the FM working!
Helps to read the manual 😁
Hope you get it all fixed up.
Our computerized test stations had one 1 8662 and 2 8672's
11:16 Well... When you see that circular dimple on a tantalum, that's a wet tantalum. They have better specs, higher ripple current capabilities, higher resistance to surge voltages, BUT, unlike hermetically sealed dry tantalums, wet tantalums do age. They can be re-formed like ordinary electrolytics, but if they are not hermatically sealed and you can see a rubber end cap on them, it is highly recommended to replace them, because their electrolyte is *sulfuric acid*, so if they start to leak as the rubber seal degrades, they instantly destroy everything around their vicinity.
Answering the question what to replace them with needs some research of datasheets. I'm just guessing that a very high-spec low-ESR high ripple current electrolytic is an OK replacement, but those are probably not available in axial package. That might be a case when you need to replace them with new wet tantalums. But be sure you buy hermetically sealed ones. And if the original ones are also hermetically sealed, it's probably OK to leave them alone. Maybe re-form them, and check the leakage current before and after the re-form.
There’s a huge table 6-3 in volume 1 of the service manual that lists all the board level components with their HP part numbers alongside the manufacturer’s part numbers. For example, the DAC chip you pointed to is an AD7520LN. Pretty much every HP instrument I’ve encountered with house numbered parts has a table like this in the service manual.
Huh, I seem to be missing that section. I'll look for it
Ah-ha, just found the parts list. Thank you for bringing it to my attention.
Those multi ribbed electrolytic caps vaguely reminded me of some NOS Philips I purchased a few years ago on E-bay.And another spec that I had to pay attention to was ripple current, something I hadn’t thought of until I looked at current draw from supply in that area. Some cap models vary quite a bit in spec’s.Those two tantalum’s together in corner of board might be called “wet slug” tantalum
Yes, I think those are "wet slug" tantalums also. The case would be steel, the electrolyte is more acid than that in aluminum caps I think. That type was used in most high reliability military hardware (missiles, nuclear bombs, etc.) $$$
@@notimetolooz9655 If I'm not mistaken, some of them are built with a solid high purity silver body! The other design uses a solid tantalum can. The steel-canned ones are dry tantalums I think.
@@mrnmrn1 Those points are probably true. It's been a while since I thought much about it. I do recall the cases are not aluminum and about the inside of wet ones were acid.
Interesting that the error LED was off at first after the re-cap, and the waveform was not clipped. I wonder if that indicates something that works when it's cold, then stops working after it warms up.
Unfortunately I haven't been able to reproduce. Also the waveform had a lot of noise when the error led was off. I also noticed the error light on the front was flashing
@@bandersentv Probably you have already downloaded the Feb 1981 HP Journal that describes the 8662A design. If so, on page 23 Fig 1 shows a block diagram of the output section. It seems the output can come from three different sources. If you can select each one and see if the clipping is present, that should tell if it's after the switch. (In the wideband output amplifier) Or maybe it's before that switch.
Was that built in Avondale, PA??
How would I tell?
@@bandersentv QC stamps inside, and labeling on the back or bottom.
Hard to tell without the schematic, my first order of operation would be to check what caused the red led to glow; is it an internal self-check health for the card or it means the signal given to it is corrupted
There is an analog comparator that samples the output and compares to what is being asked for. They do not match.
@@bandersentv a self-check, very nice, you have a very good shot at it by changing it. Another thing you can maybe look out for if it fixes your issue is the fan(s), lot of improvement over the years in the computer field for the cfm vs noise, you might find something quieter either for static pressure or simple airflow. Thanks for the reply, it's a really nice purchase
I did not publish my comments I wrote here sigh.
Anyway I don't think that it just that board. The interesting part is that also FM mod is not working. If you look at the overall schematics on the HpJournal (Feb 81 edition) regarding this instrument () you can see that the instrument rove both AM and FM modulation from the same internal source. So what is the drive modulation signal is bad resulting tin this amplitude errors because it drove the output amplifier to much? The modulation is also taken care inside the AGC loop so yeah it can make the loop drive too much the output amplifier. Also normally in signal generator the output is fixed and the amplitude is regulated using attenuators and by driving a transistor to have better linearity. The attenuation is done both using step relè attenuator and using pin diodes that is a type of diode with interesting junction form. Pin diodes ultimately are also used for amplitude modulation. Also the generator has three separate path for the signal depending on the output frequency so you better should check the amplitude in all of those using the spectrum analyze. Another thing come to mind is that you should have the scope matched to 50 ohm otherwise the amplitude would be off. So I would follow the service documentation checks to isolate the fault but I am not convinced at all is that board.
FM is working and I do have a 47 ohm load on the output. Coarse attenuation is done using solenoid switches after the output amp. Fine attenuation is done on this board using a DAC. It is combined with the AM modulation and the resulting AGC sent to several boards. I did further testing and the output amplitude does vary with frequency so the leveling is also not working. The new board will arrive on Monday so we'll soon see.
@@bandersentv Oh I thought you had no modulation at all but only sweeping ahah. Yeah we will see Monday than! Hope it would be that board!
@@vaualbus I discovered FM is work shortly after I made this video. I hadn't noticed than when I turned it on it was defaulting to external mod. As soon as I chose internal FM mod, it sprang to life. There are numerous warnings in the service manual about this board containing a sensitive COMS device (this DAC) and care should be taken. Maybe it's as simple as this chip is toast?
@@bandersentvi think if the DAC would be dead you wouldn't have anything at the output. So I don't think it's a problem.
How so? AM modulation feeds the voltage reference on the DAC. Inputs on the DAC are the fine RF level control. Output feeds an op-amp then one side of the AGC comparator. If the DAC is fried, the input to the comparator would be pegged at ground or one of the voltage rails. The AGC doesn't have the range to choke off the output.
The DAC not working would account for all the issues I'm seeing. If I had an extender card, I could see if the DAC output level varies for the output attenuator setting. That will be my next step in the new card doesn't work.
I was just watching some videos and they were all in HD. For some reason this is playing in 360P. Little hard to watch. Yes, I can't change it. My choices are Auto or 360. Auto looks like 360. Hard to watch.
Still processing
@@bandersentv Ahhh....For some reason I did not think they were active until the processing was finished. I will watch it now. Looks great.
I get that sometimes on his vids. I see the vid listed in my recommended. All I can get is 360P. I have seen in on other channels and if I wait 20 min or so, I get full HD. For some reason, Bobs vids sometimes takes several hours to populate the full resolutions. Most of the time it is fine. Guess it is related to location; he posts to one server in the midwest and I'm viewing from a server in the South. BTW, this one did have the full HD list for me
@@ScottHenionWhen I upload I set them to private then wait a few hours for processing and then set to public so viewers don't have to experience low res.
@@ScottHenion Sometimes I publish before it's done processing. Like if I'm going to bed
You probably shouldn't get in the habit of twisting those coax connectors to loosen them. You are also twisting the center pin. A small open ended wrench is not that expensive, try a hobby shop. When I worked at a HP service center they recommended a torque wrench for some connections else it could change the calibration. I haven't looked at the service data but HP liked to use YIG oscillators in their stuff. en.wikipedia.org/wiki/YIG_sphere
They seem like something out of a science fiction story. There is a main tuning coil (for main frequency control and sweep) and a smaller FM modulation coil. I bought one of those kits off ebay for my spectrum analyzer. The boards do not look etched, I think he uses a mill to separate the traces. Attaching the edge connector is easy.
Thanks for the info. I've been looking at SMA torque wrench on ebay. Next installment will be up in a few minutes :)