HP 8657B Signal Generator repair

[Squiddaddy]

I recently picked up an HP8657B Signal Generator with options 001 and 003 for a decent price, knowing that it probably had some issues.
I was going through some initial tests, and here is what I found so far.

The option 001 10MHz OCXO appears to be rock solid.
All buttons and leds on the front panel appear to work.
I was going through the frequency ranges and levels, and noticed that signal drops out at about 50mV. But I can get all the way up to 1V.
So that tells me that there is probably some kind of attenuator issue that I need to address. A missing range or something.
Those attenuator relays are loud! LOL Louder than my 8656B.
I was able to get up to close to 1500MHz before I reached the limits of my frequency counter.
But then, as it was on for a while, I lost that ability. It seems like the doubler board was working, but then stopped at some point. Now cannot go above 1029.999999 MHz. Signal drops out after that.
I notice that the -5V line is very low at -4.2V, so either that supply is faulty, or something is dragging it down. So I need to work on that too. Maybe that is what is hampering the doubler board now.

My questions are, is this a common failure with these devices, and is it repairable, or am I going down a rabbit hole of no return? The doubler is the one that worries me. I hope it didn't fail completely. But it was working for a while.
I think it'll be a great project, but I don't really want to come up empty at the end.

I downloaded the service manuals I could find, but they did not have my serial prefix on them. This is a newer 3520U prefix from about 92 or 93.

Thanks for any advice or suggestions.

[bd139]

Have a read through this - sounds like common problems for this line of gear:

http://www.azurelectronics.com/Repair%20HP%208657B%20Signal%20Generator.htm

I only know this existed because I know someone who just had to do exactly the same repairs on their 8657B as the above site and sent it to me.

Reckon it's repairable.

[cmjones01]

I can't help with a solution, other than to refer to the Azur electronics site that's already been mentioned, but I also have an 8657B (bought in an industrial auction) which works but seems to have low, and varying with frequency, output above 1030MHz. One day I'll get round to looking at it on the spectrum analyser and see if I can work out any pattern in the output level which might give a clue to what's wrong.

Chris

[Squiddaddy]

Thanks bd139!!
That looks like a good place to start. And I have exactly the same issue he had. Doubler was working and then stopped.
And my output levels are out of spec too.

[Squiddaddy]

I came home today to try some things with a fresh mind.
I powered it up, and the -5V was back to correct.
I set 1300MHz, since that's the limit of my counter, and got output. Am able to go down to 15mV before the counter drops out.
Go back to 100MHz, have to jump back up to 50mV to get output.
So that at least gets signal with and without the doubler, and a lot of the attenuator steps so far.
After going through ranges and amplitudes, the -5V led started dropping and flickering.
So, things appear to function at times, so this is going to get interesting.
Got to nail down the power issue first I guess.
Also, there is a sticker on the doubler that says Modified Doubler. Looks factory.
Any other info anyone has will only help more.

[Squiddaddy]

A little more work time on this today.
I found that this is actually a 1995 model. And a lot of the service manual doesn't match.
The battery is leaking, even though it appears to not be dead. Need to replace.
My -5V supply gets pulled way down to almost -3 when it acts up. Have not verified the supply yet.
U25 rom on the processor board is not populated. Not sure yet what that was for.
Biggest concern will be if the doubler is faulty. These don't appear to be available.
Attenuators are, but not certain yet if I need them.

[Squiddaddy]

So here's where things stand. And I don't think they look good.
I tested out the -5v supply, and it is working ok. Also tested the others.
I opened the A12 frequency doubler, and I noticed a resistor that had been getting hot. Color bands faded, and discoloration on the board under it.
There is a 723 regulator on that board with an external pass transistor, and the 22 Ohm resistor in the collector lead is the one getting hot. So something on the doubler board is dragging down the power supply, and when it acts up, a red led lights on the board. Sometimes, after a little time, the led will go out and the voltages will rise, and the doubler works again.
That makes me think that the led is maybe a current limit alert?
But, also, when I looked at similar auctions on Ebay, most describe some kind of issue with the doubler, or it's frequency range.
So, this must be a common fault, and it's odd that HP did not release any service information on this board.
What is so special about it that it can't be repaired, or a schematic produced?
I doubt I have the skill to try to reverse engineer that board to figure out where the current draw is coming from.
I do notice that I have a 1995 board, and it is newer than others that I saw pictures of.
It has 2 Actel FPGA's, and a bunch of DAC's, and the RF section, but not much else.
So unless I can get some information from a similar device, I guess I have a doorstop.
I have not been able to locate a board for sale to replace it with.

[bob91343]

I have one of these also.  It has two problems, one of which I caused.

First, I activated the reverse power protection and after reset, the generator output is around 20 dB below what it should be.  I don't know where to look.

Second, when I got it, and still, it has distorted output when the level is above 0 dBm or so.  This isn't too important but I'd still like to fix it.

Incidentally, my frequency doubler works fine.  I get leveled output (wrong level) over the entire frequency range.

I have some manuals but can't seem to find the circuits I want.  I looked online and only found manuals from the factory, not a complete set.

[Squiddaddy]

I got both parts of the service manual from the Keysight web site.
https://www.keysight.com/main/techSupport.jspx?searchT=8657b&id=1000002203:epsg:pro&pid=1000002203:epsg:pro&cc=US&lc=eng

Reverse power protection is a relay, so maybe the issue is there. Something in the contacts?
But it looks like a sealed unit also, and may need replaced as a unit.
You could temporarily bypass it with the SMA connectors as a test though. Check the signal level right out of the attenuators.
I get output across the whole range also, for a while, as I guess something overheats.

[bob91343]

Where is this relay and the SMAs to bypass?  I have the unit open on the bottom and have loosened the board in the center, below the doubler.  Under it is another board, containing the memory battery (which shows corrosion and needs replacing, I assume a CR2032 type).  I don't know what to do next.

I do see the tiny rigid coaxial cables.  I have removed the cables from the module called 'sensor' and bypassed it, with no change in performance.

Someone mentioned that there are some diodes for protection that may be shorted.  Dunno where those might be.

This is a very sophisticated unit and I don't want to screw it up any more than it already is.

[Squiddaddy]

Here is a sample picture. The greyish can beside the attenuators with the rigid coax coming out the top.

It should be marked as reverse power protection on the sticker.
Mine has the extra pulse modulation, so it's not exactly like this picture.

[bob91343]

Yes that's the module I bypassed with no change in performance.  So the trouble isn't there.

[bob91343]

That module isn't the reverse power unit; it's just the reverse power sensor.  I don't know where the unit itself is and perhaps that's where my problem lies.  It seems the sensor, being close to the output, can see reverse power and then sends a signal to the reverse power unit, which then opens a relay.  Somewhere there is a damaged part, probably within the reverse power protection unit, wherever it is.

I wouldn't mind taking it completely out if I could locate it.  Any ideas?

[Squiddaddy]

It is inside that module. Look at the repair manual. It is shown as a block assembly. If you come off of the attenuators with low signal, then it's a different problem. It could then be the attenuators, the doubler, or the output board. Reverse power protection just opens/closes the signal path to protect those other circuits. If you exceeded the 25w, then maybe something else got damaged.

[bob91343]

When I bypass that module the output remains attenuated.  So you may be right; can you suggest some way of tracking it down?  The attenuators seem to work properly and the doubler still doubles.

Someone once suggested that there are two diodes that could be shorted, across the signal line somewhere as leading edge protection until the relay actuates.  Of course they aren't in that module, since the bypassing doesn't help.  I would not mind eliminating reverse power protection if that will get me operational again.

[bob91343]

Oh, and when I raise the output above about 0 dBm, I get the same distortion that I used to get when the unit was working otherwise.  That tells me the problem is probably after the output stage.  I wish I had a level diagram.

[bob91343]

Problem solved!  It turned out to be the frequency doubler.  I took it out of the circuit and, while I no longer get output above about 1 GHz, it's otherwise back to normal.  In fact, the level is a bit more than it should be so maybe I can find an adjustment.  Perhaps there was loss through the doubler that no longer is a factor.

I still get distortion but only with output above 800 mV, which isn't a problem for my purposes.  Frequency is spot on, and AM and FM work fine.

Taking the doubler out of the circuit was not trivial.  I had to modify a rigid miniature coaxial cable.  So now, even if I got a new doubler, I couldn't use it unless it came with the hard line cables.

[Squiddaddy]

Did you take it completely out, or just bypass the coax?
When I remove mine, it says 'No Doubler' on the display and stops working. So it needs connected to the CPU.
I think the doubler is the death of the 8657B, and the number of used units seems to reflect that. Every one I checked out has a bad doubler.
And they can't seem to be fixed, since no data exists for them.
I guess I'll have to go back to my 8656B.
But at least I didn't pay a lot for my 8657B. It was more for a repair attempt to learn something, and hopefully get a nice piece of gear.
I was going to try to reverse engineer the board, but it's way over my head. Mine is newer, with DAC's and FPGA's.
I guess HP made a lemon in this product.

[bob91343]

I think you are spot on about HP and their doublers.

I byoassed it but left it in.  Everything is still connected except the IN and OUT.  I pulled the two rigid cables out and modified one to fit into the module, maybe it's the pulse modulator.  Modifying it required cutting it off, sawing off some of the outer conductor and cutting the inner conductor to the right length.  It was easier than reading my description.

As for reverse engineering the doubler, it seems to have several ASICs which of course will forever remain a mystery.  I was thinking of digging in and seeing if something obvious is wrong, like a bad relay but decided it wasn't worth my time.  I don't need the upper octave, and I was so thrilled to get my generator back in operation that I just put it together.

Where is the adjustment for normalizing the voltage reading?  It seems to be about 1.2 dB more than indicated.  Again, not a problem and that may even be within spec.  So if I set up -73 dBm I can expect an actual output of -71.8 dBm.  I can't think of a situation where that would cause trouble, unless I am trying to calibrate a voltmeter.  And since I know the error, I can allow for it.

The remaining problem is the distortion.  More than 800 mV output and it starts to clip, sort of.  If I am running AM I have to reduce that to allow for modulation peaks.  The distortion occurs at all frequencies, and always at the same level.  As I recall, the wave starts to clip on the negative peak and then the whole thing goes into a bad waveform, and the amplitude won't go as high as it is supposed to.  I suspected the output stage but it does have the correct supply voltages.

[bob91343]

It wasn't easy but I found enough info in the various manuals to ascertain how to calibrate the output reading.

This has to be one of the worst manual setups I have ever seen.  There is no place where all the modules are identified.  In order to set up the amplitude measurement you have to locate two modules and a variable resistor on each.  One sets the level and the other sets the step calibration.  Maybe.  But there are two adjustments on two different modules, A6 and A11.  They are both accesssible by removing the bottom cover.  The procedure was written by someone who doesn't like writing procedures.  I still don't know how to do it, but my intuition should take care of me.  You twist the two pots until the readings are correct.  After a few twists you realize what the two adjustments do.

At least, that's my plan for the next time I feel like opening it.  Meanwhile I just use my 1.2 dB factor which, when converted into voltage, is a lot sloppier than one might think.  About a 20% error actually, intolerable in any ordinary cheap meter but sort of normal for radio frequency.  I just have to make sure I do my adjusting at levels low enough to avoid the distortion mentioned in the previous message.

The positive side to all of this is my much greater familiarity with the innards of this beast.  I am far less intimidated than before.  I may even look into the doubler, although since I cannibalized one of the cables I would have to find a replacement in case I can repair it.  Nothing to lose, right?

[Squiddaddy]

I can actually see when my doubler acts up. If I monitor the -5.2v supply, it drops or gets loaded down when it acts up. Then, after a while, it comes back. Also, there is a red led inside the doubler near a 723 voltage regulator that lights up when the doubler fails, and goes out when I comes back up. Almost like it's a current limit or something. And a slightly charred resistor in the pass transistor.
So maybe mine has something to do with the rf amplifiers in the doubler. Whatever is under the milled aluminum sink.
Understanding what the -5.2v and +38v supplies do may help, as they only go to the doubler.
But you're correct that the A6 and A11 boards control the output. There are some DAC controls in the doubler that may be affecting your signal. Focus on getting A6 right, and I think you'll get there.
I can't find anyone to sell just a doubler to try, and I don't want to buy another parts unit just to get nowhere.
I have an 8656B that goes to 990MHz, that just needs calibrated. And it's OCXO isn't the greatest.
I bought the 8657B because I have a 1300MHz frequency counter I fixed and wanted to calibrate.
I still like to think that it's an analog issue on the doubler, and not the FPGAs, but don't know how far to take it.
I don't have access to a spectrum analyzer, so I can't go too far anyway. Even used, they cost a lot.
I just wish we had more repair stories to get information from.

[bob91343]

The pot on A11 didn't have any effect.  I couldn''t find A6.  Then it hit me that there is a front panel function called Offset.  So I used that to calibrate the amplitude and now it's spot on, within about 1%.

I still have the distortion issue.

While I had the doubler open I took off that heat sink and it's filled with mostly SMT parts.  Of course there is no schematic diagram.  So I closed it up and am living with a 1030 MHz generator, far in excess of my current needs.

I do have spectrum analyzers.  In fact, I have four or five of them.  Most of them cover the entire range.  But in my case I don't see how it will help me.  It does show when the distortion comes in, at about 700 mV output.  But frankly I don't need anywhere near that much output.  This distortion occurs at all frequencies at about the same level.  But it's such a pleasure to have the output amplitude agree with the display.  And so very close, at that.

One thing I meant to do that I forgot was to see if there is a short across the input or output of the doubler.  If so, maybe I could track it down to a bad part.  But I am not going to open it again, at least for now.

[macboy]

Distortion beginning at 700 mV isn't the end of the world; you are getting +10 dBm at that point, and it only goes to +13 dBm at maximum output (1.0 V). Of course I am assuming you mean 700 mV RMS not peak or peak-to-peak.

I guess I'm lucky that my 8657B works fine.

[Squiddaddy]

You need to check the signal from A6 before and after the het switch. Still think your issue is in A6.

[bob91343]

I don't need any more than a half Volt output so I am not worried about the distortion.  I am very pleased to have all the other functions operating perfectly (except the doubler).  I don't have a current use for signals above 1 GHz, either.  Of course it would be great to have all functions correct.

The way I obtained this unit was a 'custoner' of my very occasional repair service brought me some power supplies to get going.  When he got them back, he dropped off this unit that he said wasn't working well.  To make a long story short, I told him I would take his generator instead of money and he readily agreed.  The story he gave was that it didn't pass the flatness test.  Well, that was due to the distortion.  I am delighted to have such a fancy unit that performs so well.

It has the high stability oven master oscillator and the pulse modulation options.  To show how much I enjoy playing with this stuff, I set the output at 600 MHz and measure the frequency on my 5328A counter (calibrated against a rubidium oscillator) and watch the drift.  Since the generator's master oscillator only runs with power on, there is a period of drift for around 45 minutes.  The counter's mastter oscillator runs 24/7 so there is no warmup transient.  As currently set up, the frequency at 600 MHz gets within about 3 Hz, not too shabby.

All that having been said, I still have a desire to fix the doubler and the distortion.  I will give it some thought and perhaps, if I have an idea of where to look, I will get into it to see what might be done.   The doubler may just have one shorted component in the signal line.  The distortion may be caused by an open resistor in the bias of the output stage, or something similar.  Once I feel I understand it better I may look around.