HP 8662A repair part 2

Replacement A4A7 board has arrived. Will it fix the generator?
Review of the remaining issues
Overview of the A4A7 board
Trying out the new board
Where do we go from here?
This channel is made possible by the generous support of viewers and patreons. If you'd like access to support the channel and gain access to bonus content, please consider becoming a Patreon here: / bandersentv

[@garp32]

Hey hey! You're getting there! Thanks for the update!

[@greggsvintageworkshop8974]

Awesome that you have gotten it going as well as it is. I have no doubts that you will get it where you want it to be. I am always impressed with your knowledge and troubleshooting skills with everything you do.

[@robharley9838]

Great to see the progress with the 8662A! These beasts deserve a comfortable retirement, the tech in them is just amazing. I recently revived an 8656A from slightly earlier and my experience may shed a bit of optimism: I had to swap out the faceplate assembly with another parts unit and the process was a pretty straight ahead module replacement. I suspect when you get to the dive down on the removal details, the front face detachment won't be too bad. Best of luck with it! Thanks for sharing the journey.

[@bandersentv]

You are correct. I've been reading the disassembly instructions and it doesn't look that bad.

[@WolfmanDude]

What a find, if I was rich I would spend all my money on old HP test equipment. Also its funny how all HP things have the loudest fan physically possible :D

[@Skyler_Hagen]

The fan in the 8640B is pretty amazing. In typical HP fashion it’s way over designed. Brushless motor (3-phase I believe) and there’s a dedicated card in the unit that is just a driver for the fan. Basically an inverter with power transistors and feedback windings in the motor provide commutation signals for the drive. Works great though and that one isn’t too loud yet.

[@bandersentv]

Wow, I wonder if that's what they use in the 3336A? It sounds amazing when it starts up. Like no other fan I've heard.

[@Skyler_Hagen]

@@bandersentv I have to say some of the most amazing fan sounds I’ve heard have been from my old Nixie tube frequency counters when they start up. The EIP 351B and Eldorado 1650 both sound like some sort of Star Wars death ray charging up when you first turn them on. It’s almost hilarious. I think they use direct 120v line operated fans and you hear the mixing of frequencies from the stator and rotor (slip frequency) as it slowly reaches its operating speed. Adds to the enjoyment of using those instruments for sure.

[@notimetolooz9655]

When I worked at a HP service center I remember that there was an upgrade kit we used to replace the early brushless fan in some generators. I don't remember whether the new type fan was any quieter or just more reliable. I don't think the new fans had the separate control module.

[@bandersentv]

@@notimetolooz9655 The fan runs directly off the ac line

[@rlgrlg-oh6cc]

Looks like there are three separate AGC amplifiers, for the different bands. Do all of them show the amplitude variation? I think you said that the error LED is off, right? The circuit for that is a window detector that is looking to see if the sum of the level detect and the DAC output is close enough to 0V. One alternative to extender cards is to temporarily tack solder one or two short wires (AWG 30 is best) to strategic points on the back of the pcb so that you can look at those voltages with your scope. Maybe look at the level detect input for starters. You could also monitor the DAC output (at the emitter of Q1) on your old board to see if it's that part or possibly the op amp or transistor right after the DAC. HP engineering is always very impressive!

[@vaualbus]

can be my eyes but it seems that also te offset is shift a little bit. Maybe some of the RF output transistor are getting leaky?

[@bandersentv]

Unfortunately, I only have a 100 MHz scope and can't accurately view the upper bands. Yes, there are no errors. I just discovered there is an advanced option to disable output leveling. It's possible mine is just turned off. I'll check later tonight. The boards are a tight fit and I don't think even 30 awg would fit. I'll spend the $100 for a set of extender boards if I decided to probe that deep. I'll need them anyway if I go through the calibration procedure.

[@vaualbus]

@@bandersentv you can use the tiny spectrum analyzer you bough recently I guess too? In general for looking at the sig gen output a spectrum analyzer give you more information like of phase noise and spurious harmonics!

[@bandersentv]

I did some more tests as best I can. From 10kHz to 1Mhz, the output level is rock solid level. Above 1Mhz the output level starts varying in a sinewave like pattern repeating about every 10Mhz.

[@bandersentv]

I think my scope was not perfect adjusted for 0 on the y-axis. I just repeated some tests and the DC offset seems good. There is also an adjustment for it if I do need to tweak it. The output amps are sealed modules and I'd hate to have to replace one.

[@bzuidgeest]

Why start with replacing the unobtainable dac? Your sine error is asymmetric, if the dac is faulty i would not expect that. Fwiw i would look at the transistors and opamps first. Cheaper to replace. To bad you don't have a curve tracer i think. Sometimes transistors only break down at higher voltages then the quick testers use. You could also place the dac in a breadboard. Should not be hard to test using a voltmeter and a few supply voltages. You could jumper the control word simple enough. That way you know if your actually chasing the right part.

[@bandersentv]

This was designed 45 years ago and all the semiconductors are obsolete and most are difficult to source. I'm going after the part that is the most easily damaged and easy to replace. I just bought two for $12. If that doesn't fix it, I doubt I'll go any further. Removing transistors and testing them would take far more time than I am willing to invest.

[@bzuidgeest]

@@bandersentv most of those transistors will have modern replacements. And with a desoldering gun they are easy to get out. And you don't have to do all of them if your lucky, just start with the ones likely to be the problem. In the video you clearly expected problems sourcing the DAC part but if you already found those. Then you might as well try it

[@notimetolooz9655]

@@bandersentv The thing is that is probably a multilayer board (5?). The plated through holes can easily be damaged by soldering. In fact it is often the case that a choice has to be made, do you want the part in good condition or do you want the board in good condition? Having both in good condition sometimes is unlikely. If you want the board to be good, clipping the leads of an DIP and removing the pins one by one gives the best results.

[@bandersentv]

​@@notimetolooz9655it isn't. Just double sided

[@rlgrlg-oh6cc]

OK. Looking at the AGC schem a little more, I see from the table of switch control versus output freq that the H FET control signal changes above 1 MHz. This signal is created from the range inputs on the edge connector by U9 and then level shifted by U5. Since things seem good up to 1 MHz, then start to go bad, perhaps it's related to that signal, or the FET that is controlled by it (Q16)? Q15 and 16 as I understand it reduce the bandwidth of the error amplifier at lower frequencies, so that the AGC doesn't follow the dips of the output waveform. You want it to just react to the peaks. If Q16 is turned on at higher frequencies, maybe it would cause what you are seeing. Not sure.

[@bandersentv]

Excellent observation! I shall investigate further.

[@Sc16180]

Bob, was the scope terminated in 50 Ohms? The voltage changes with freq could be SWR if gen is not terminated properly.

[@bandersentv]

Yes, it is terminated.

[@martinbaines3719]

Is it possible the fan is noisy because it's detecting an error, and the default setting is to run the fan at maximum - like what happens on several higher end computer systems? Their default is to run the fans at full speed if the management system detects any type of error.

[@bandersentv]

Good idea, but the fan speed is not variable and it runs on full ac line voltage.

[@vaualbus]

I was wrong than maybe.. anyway where you bought the schematics? The version I have have the sheets splitted so they are hard to read! Your version seem better!

[@bandersentv]

I downloaded the service manual for free: archive.org/details/hp_8662a_operating_service_manual_vol2

[@Skyler_Hagen]

Seems like you’re very close on the repair! The “open loop” portion of the amplitude control is working now (you commanding the amplitude to change). It just seems like there is an AGC feedback loop (probably some sort of “detector” that probes and measures the output level and makes corrections to keep amplitude constant) that is still not there. On my 8640B, that same circuit drives an output level meter (analog in my case). Does this unit have a true measured output level display? Or does it just display the commanded amplitude assuming the AGC circuit is working properly?

[@Skyler_Hagen]

An HP 400 series true RMS RF voltmeter is a tool I’d like to add to my arsenal, and would be a great way to measure the actual output level precisely (although I’m sure the frequency range of those meters would be exceeded by the upper end of your signal generator)

[@bandersentv]

I have several but have never tried using them

[@bandersentv]

It displayed the commanded amplitude. If it does not match the actual output level, it generates a #10 error. That's what I was getting before

[@PracticallyFixed]

Good stuff. That Z marker output is why I need to get around to fixing my Tek 465B, it has a Z drive input which my other scopes to not. Did you ever experiment with the Z marker output on the HP3336A you have (on the back) while sweeping?

[@bandersentv]

Yes ages ago, but I recall it just has one marker that goes high and stays there when the marker is reached. so you end up with a l response curve split into a dim and a bright portion. I have an old video showing that.

[@PracticallyFixed]

@@bandersentv Right, ok. When you mentioned the 8662A had multiple dots that got me interested as I understood the marker on the 3336 to be a single gate you you indicated, but it got my hopes up it could do the multiples as well (I have a 3336B). Thanks for the information.

[@Tysman909]

Might be fan speed controller has an issue and it's stuck on maximum

[@bandersentv]

Good idea, but the fan speed is not variable and it runs on full ac line voltage.

[@Tysman909]

@bandersentv oh in that case fan must be worn out. I sometimes apply a very small bit of thin silicone grease to back of fan if there's a cover on back to access end of shaft. If not I would just replace it

[@69Dartman]

Sometimes if I don't want to replace the fan you can pull the label and just hit the bearing with some wd 40 or whatever you trust to see if it quiets down. Been able to get some computer 80mm etc fans etc running quiet for a while. I've had older computers with fans exactly that loud, especially the 80mm ones. That's why most pc's went to 120 and bigger fans, more air flow and much quieter but you have to use what fits and maybe the noise doesn't really bother you anyways.

[@LJJKD1947]

Just a little bit confused ,I'm sorry, do you think it's a board problem or an external trigger or input issue ( like the decoder you mentioned) or possibly still some power supply bypass cap acting up? At any rate your getting closer!🤔

[@bandersentv]

None of those. I think it's an AGC adjustment that needs to be made. There is a lengthy calibration procedure to go through to get it fully up to spec. It would cost a lot of money to get all the tools needed so I may just live with it as is.

[@andybonneau9209]

Maybe the new board needs some new caps as well?

[@SeekingTheLoveThatGodMeans7648]

Hope it will, with no fireworks (pun intended)