8116A Code 42

[MarkL]

An additional thought on the hot chips A2U33 and A2U37... They are ECL and powered by a -5.2V rail.  You probably have gotten to adjust it already, but if the -5.2V rail is too negative due to "pot recentering", it could be causing extra heat dissipation.  The -5.2V rail is adjusted with A1R12, as shown in the table on page 9-3, along with the other rails.

ECL is fairly robust, but if presented with too extreme a supply voltage, one or more chips could have sustained damage.  Just a theory.

Please post back your progress on the adjustment procedures and the option 001 verification when you can.

[Mick B]

Greetings, well I got to the end Pg. 9-8. I changed 5 pots for 10 turn pots, yes, the averaging out (for lack a better term) a little here a little there and when there both right move on to the next. It's very time consuming. When I got to Overshoot & Transition Time adjustments, after setting it up correctly the overshoot was always less than<4% but fluctuating. At Steps 5 & 6 is where I noticed a voltage problem at 200mv it was off the screen. I unhooked everything and checked the voltages, and they were way off 2v was 3.19v and so on. Can you explain step 20 Pg. 9.7 I don't understand where to, or how to check the transition time. I'm just not sure what transition they are referring to or how to measure it. 
I kept on, trying to get things in the ballpark. But ended up starting over and decided that to get the first adjustment step5 Pg.9.4 the voltage at 16v is... well look at the square wave photo " The first Procedure." The purple trace is the tracking output. also, the crossover distortion Photo " crossover" seems only to happen at high voltage, I can pin it down. I am wondering if it's because I only changed Q501 Its Hfe was way higher than the one I took out perhaps it's a mismatch, I have a new Q502 in a TO-92 pkg, but I also have the heatsinks for it. also, I just changed Q506 I have to do some checking to find out if I changed Q505 earlier I think I did. will see. I will deal with the IC later unless I'm thinking one problem at a time. but I could be wrong. (it happened once before)   
     

[MarkL]

Unless the pots had issues, I would have left them alone.  Depending on what you put in, and where, some 10-turn pots are wirewound and the internal coil of wire adds inductance to the circuit.  In an amplifier that runs up to 50MHz, it could be significant.  I really have to emphasize "if aint't broke, don't fix it".  Especially in the middle of a repair where the thing you're replacing is known working.  It adds more unknowns to what might be wrong.

Transition time (aka rise time or fall time), is how long a signal takes to get from one plateau to another.  A square wave is used for the calculation here.  A change from one voltage level to another doesn't happen instantly, and can be seen as a fast ramp.  The measurement for transition time is the points on the ramp between 10% and 90% of the starting and ending voltage levels, where the difference between starting and ending voltage levels is defined as 100%.  I don't have a Rigol scope, but I'm sure it has a button to measure it automatically for you, probably under the "Measure" menu.  You need to have your horizontal sweep set fast enough so that you can see the ramp of the signal.  If you're curious what it's doing, search for "measure rise time on an oscilloscope" for the old way of manually doing it.

The square wave is not looking great, and it may be related to the crossover issue.  What does it look like at a lower frequency, say 10kHz?  Looks like you have it set to 100kHz in the screen shot.

For the unusually high output voltage, I would start with verifying the amplifier input at TP4 as described on page 10.4-1 1.  If the input isn't right, certainly the output won't be.

In the one scope capture you say the purple is "tracking output"?  Do you mean the trigger output?

The crossover distortion could be caused by very mismatched hFE pairs, or it could be something wrong with the bias in one of the stages.  The transition from the positive side of the amplifier driving to the negative side, and visa-versa, is discontinuous.  You could try changing the frequency of the waveform and observe what it does to the distortion.  If it's less of a problem at lower frequencies, and the distortion gets closer to 0V, it's a problem with the speed of the changeover, perhaps capacitive in nature.


Also, note that the adjustment procedure has you constantly putting in attenuators and taking them out and changing the multiplier on the scope.  This is because HP has chosen an insanely over-powered 20GHz scope for the adjustment procedure that can only do 1mV/div to 80mV/div.  So, all these attenuators are compensating for the scope's deficiencies in this application.  Instead, you can simply put a 50 ohm terminator on your scope (or use the internal 50 ohm termination if it has one) and set the scope according to the needed V/div for any particular step.

[MarkL]

Also, it may be worthwhile to re-visit the static operating points on your unit now that it's partially working.  I don't think you need to re-post all fresh readings from your unit since that's a lot of work to transcribe them, but you should compare again to a working unit.  You might find an area with incorrect biasing that is contributing to the distortion.

I've included my readings again below, updated with the two nodes from a few posts ago.

[Mick B]

Hello again, just an update. The voltage for the -5.2 rail is set at -5.4 I could not get it to adjust, and then I found this in "UPDATEING/CHANGES/ERRATA Pg.11 this mod had been done. I set the voltages the day I got it, it had no -24v rail. I also did the option 001 verification procedure test it produces a square wave, but it is always on there is no timing function it starts as soon as the output is on and does not stop the counter will continue forever.
And yes, I meant "trigger output" my-bad. and I know what rise and fall times are (you presented a great explanation) DOH! I just couldn't put the two terms together.
I dissected one of my 10 turn pots they are carbon, the ones that were in there should have been 10 turn to start with to call them sensitive is an understatement.
I wondered why all the attenuation, I'm so glad you explained what that was all about. that will shave off a lot of time.
Without the attenuation nonsense I have not been able to reproduce the crossover distortion, YET. I'm going to re-due the calibration today. and take measurements after.
Thanks, Mick
 

[Mick B]

Where to start? this is a half-ass calibration. It went through the procedure up until Pg.9-11 step 7. any width parameter higher than 984 turns on the error light. I putzed around after that. I still have voltage issues and the wave forms start to lose their shape around 7MHz. apparently, there is settings after the Power Supply Preliminary because from Pg.9-3 to 9-7 all to voltages are good up to 16+ volts the and the rise times are < 6.6 but fluctuate a lot. Pg. 9-7 step has to be a misprint. these settings put the signal off the top and out of sight. And then the voltages turnout low?
These are the voltage readings from Amplitude & Offset Performance Test. Pg.8-9 Just posting my results. in The Updating Chapter Pg. 8. there is an update for Offset voltages, & DC Output.  Unfortunately, these won't help my readings.
8v =
RMS 
Sine 2.72v
Trangle 2.23v
Square 3.70v

3v =
Sine 1.036v
Trangle 0.862v
Square 1.425v

1v =
Sine 0.344v
Trangle 0.299v
Square 0.494v

100 mv =
Sine 034.26mv
Trangle 029.78mv
Square 049.14mv

OFFSET
@ 100mv
7.50v =7.46v
5.0v =4.97v
3.0v =2.98v
1.0v =0.957v
100mv =090.8

@10mv
795mv =0.719mv
500mv =0.453mv
100mv =092.4mv

Frequency Performance Test Pg.8-2
50MHz = 50.047MHz
10MHz = 10.086MHz
10kHz = 10.082
1kHz = 998.8Hz

DUTY %
1Hz @ 10% =9.28%. @50%. =47.68%. @90%. =89.02%.
1kHz @ 10% =10.104%. @50%. =49.97%. @90%. =89.87%.
9.99MHz @ 20%. =20.00%. @50%. =50.18%. @80%. =80.00%.
Mark where do you think I should measure first? to take static measurements with the heatsinks on??  if you can pinpoint what nodes we might want to look at I will take a picture of the backside of the board remove the heatsinks and use a bright light to find the nodes on the backside and mark them on the photo so I can find them how did you measure yours?
     

[MarkL]

On width >984us errors, the test frequency is supposed to be 900Hz.  Are you getting that on the output?  Check that you have the proper voltages on the width controller IC U300, page 10.3-17.  On my unit the error light turns on at 1.10ms.  I'm not very familiar with the operation of U300, but the manual says error is asserted when the next trigger happens before a pulse cycle is complete.  So, my guess is that something is off with the duration of the pulse high, low, or repetition rate (frequency) that is making the cycle last too long.

On the poor looking waveforms >7Mhz, keep in mind the transition time of the square wave is 6.6ns or a little less.  In this unit, transition time is a constant since it's a limitation of (at least) the output amplifier.  As the square wave gets higher in frequency, the transition time will consume more and more of a waveform cycle.  By the time you reach 50MHz, the transition time will dominate and the waveform will look almost like a sine wave.

In terms of measuring, I did a lot of mine from the back when the heatsinks were in the way, and I used a flashlight from the top side to cast a shadow to verify I was probing the right pad underneath.  On some of the probe points, sometimes you can find a resistor or diode lead that's the same node and you can measure from the top.

Also keep in mind this unit is essentially an analog function generator that has a microprocessor instead of knobs.  Many of the adjustments are just calibrating the digital display to analog voltages that control frequency, amplitude, pulses, etc.  Because the it's all analog, don't expect precise output for anything like you would on an arbitrary or synthesized waveform generator.

This last set of waveforms you posted are not so bad for the 8116A, but the square wave could be flatter on the bottom, and the triangle is looking a little non-linear on the downward slope.  Again, maybe the crossover distortion is still getting you.

[Mick B]

Thanks Mark, your help has been invaluable. I was testing the WID settings at 900Hz and @ 100us I got 102 on the counter, but the next setting @ 900us. turned on the error light and like I said it goes out when lowering the setting to 984us turns off the error light I saw -5.3v on pin 9 of U300. I tried hooking up the HP 5005A just because, changing the bridge [W3] to test and changing the jumper to FreeRun turns on the error, STA, DTY, mV and lol lights. It seems locked up, no buttons do anything, having never done this before I don't know what it supposed to do, reset does nothing hooking, up the 5005A does nothing either. where do we want to precede from here?

[MarkL]

To answer your questions...

More explicitly, I was asking how your waveforms compared with the troubleshooting procedure on 10.3-17, not just the voltage on the error signal.  It's fairly certain that the error output on U300 was being asserted correctly.

Free-run mode puts the processor in a tight loop executing a single instruction while incrementing the memory address lines.  The unit does not boot into any operational mode, nor does it even get to any initialization code for the display.  The display is just showing whatever state the latches powered up in.

Because the address lines are incrementing, it allows you to use a signature analyzer to verify ROMs, decode logic, and sometimes other areas that operate independently of the firmware (like the keyboard scan in this unit).


What do I think from here?...  I would go back to the tenet of one thing at a time and finish with the output amplifier repair.  You still have some unexplained crossover distortion, as last reported.  Something is still not right.

I would go through the adjustment procedure and try to adjust everything having to do with output signal quality: voltage levels, linearity, distortion, overshoot, etc.  If you can't get the pulse adjustments right, or some of the other timings right, skip them for now.

After adjusting what you can, if there's still crossover distortion, compare your quiescent voltages with what I posted from a working unit.  I would look for voltages that are unbalanced between the top and bottom drivers in all stages.  If that doesn't turn up anything, output a sine wave and try to find a frequency where the distortion is the worst, and take a look at the waveform along the heavy lines (signal path) shown in the schematic.  Again look for a stage where distortion starts to be introduced, or possibly where it changes (like it's inverted).  Because this amplifier has feedback, keep in mind the feedback is trying to compensate for distortion problems, so you might see some nodes that look distorted but it's actually the correction from the feedback.

And double-check the input of the amplifier to make sure the signal going into it is distortion-free.

If you were able to adjust out the crossover distortion, then great, you can try to get the pulse output adjusted properly and pick up with the troubleshooting procedure on 10.3-17 (U300).

[Mick B]

MarkL, I'm back at it, I have been tied up working on the SAAB, Well the Chinese Tech2 Scanner mostly. To the 8116A, as it became apparent that the CAL was just not going well and seemed pointless so many things would not adjust to specs. and I knew it had mismatched transistors. I replaced Q502 with what is supposed to be a direct replacement A (2N5771) to complement Q501 A (2N2369A) Note: this is a TO-18 package and the 2N5771 is a TO-92. I found some heatsinks that slip on the TO-92 Package. Also, as the manual says if the waveform is distorted on the leading edge look at Q503 and Q505 and Q506 and Q505 have already been replaced so I replaced Q503 with a 2N3094 and Q504 with a 2N3906. It was a bitch doing this with the heatsinks, and the other things in the way.
The GOOD NEWS, every adjustment up until the last step (7) of OFFSET ADJUSTMENT Pg.918 dialed right in even fixed the error light in the width adjustment problem.
Step 7 says if any offset adjustments were required (a very small one) repeat the "Overshoot & Transition Time Adjustment" procedure before continuing. I went back to that chapter and the overshoot was wacked I decided to reboot the 8116A and when it came on E-51 greeted me. 
I Printed out MarkL's known good readings again. I'm thinking one of the new transistors / HEATSINK. I'm not ruling out a solder joint These were ridiculously hard to solder.
We will see. I was just so happy; it was adjusting out like it was new. Voltage reading forth coming. My memory cards from England showed up and I'm expecting a new VCI module for my scanner, the wife's car takes president.

[Mick B]

As my last post stated, something went wrong in the middle of calibrating this unit. I'm posting the results of my voltage readings after the failure. I had these ready to go but decided to check one last time so to be sure NONE of the readings were transposed (I'm dyslexic) and I will post those readings as soon as I can SOMETHING MORE is very wrong. Mark please what do you see here?
 

[MarkL]

One thing that looks really strange is the -13.172V at the junction of R543, R544, R545, and R546.  The other side of those resistors are at a much lower potential and they're all only 10 ohms, so either they're cooking or they're open if all those measurements in that area are right.

Can you double check that node?  And on the other side of R547 since it's also at approx. -13.2V?

I'm not sure where the extreme negative voltage might be from, except maybe there's something wrong with CR513?

[Mick B]

Hello all, while I was going to go over the voltages one last time, but after the error 51 showed up the voltages were so bad and so far off "negative voltages everywhere". it was just to ridiculous to post. I pulled Q512 and it testes good. put it back again, tweaked a few pots and NO error 51 its gone!   
I apologize for the erroneous voltages Mark noted, it was milli-volts at R547 & R545~R546 
Please use these voltage Number 4.PDF file.  Looking at TP4 & pins 18 & 19 of Q401 they test OK except the voltage on TP4 is low by about 1/2. Pg 10.4.12 fig. 10.4.7 on the 1v test, as the voltage goes up these stay the same. see Triangle.JPG.
I'm going to try to recalibrate up to the Shaper, just to see if I can get the basics going again, I don't know what happened, to cause error 51, but that needs to be found.
Mark as always, I will defer to you. Thanks
 

[MarkL]

The voltage at TP4 should be a constant +/-0.5V peak when the output is set to 999mV, 10.0V 9.99V, or 16.0V.  The voltage at TP4 is fed into a set of relay-controlled attenuators that set the final voltage going to the output amplifier.  See the explanation on page 10.4-3 for more detail.

As for the readings at TP4 being low by 50%, first verify your scope is reading ok by looking at a fixed voltage.  Measure something not in the 8116A that you know is right.

Check the reference voltage on the upper side of R411.  It should be about 6.1V.  The adjustment of R410 affects the amplitude, so it might be misadjusted.  I don't know if it has the range to result in 50% like you're seeing, but it's something you're going to have check/adjust in the calibration procedure.


Your A1 board has seen a lot of issues and more are occurring every time you make some progress.  You might want to consider buying a replacement A1, or even a scrap 8116A and take parts from it.  There are also a lot of parts in common with the 8112A, so it would also be a good parts mule.  You can probably get an 8116A for < $80; there's one for $50 being bid on right now with no errors showing.  I guess it depends on your goal in pursuing the unit you have.


EDIT: Fixed voltage threshold; should be 9.99V.

[Mick B]

I made it farther through the cal. this time, most of it. I skipped over some of the shaper adjustments mostly THD Harmonics I have a Siglent 1032X spectrum analyzer but at these low frequency's I am finding the FET in the Rigol MSO5354 to be most excellent. This is where I stopped tonight, at the beginning of Amplitude Modulation Pg. 918 so far so good. This cal. straightened out the strange triangle waveform. I'm thinking some of the shaper adjustments that call for readjusting all the pots that control the waveform balance NOT the balance between normal out, and COMPL. but pos and neg & amplitude that were already set maybe causing issues. This is my 3rd calibration attempt. so far so good.
I took your advice and bid on a $79 Working 8116A with option 001 and WON! It should be here before the end of the month.
Mark, this has been such a great learning experience, I just hope you stay on, I would like to tackle the smoking hot IC's and try out the HP 5005 Signature Analyzer at some point. This unit has had some issues, I just don't want to be defeated. These Function generators are spectacular, I would not like to end up in a landfill. I do not think we are here YET. After finishing up the basic cal. what next?     

[MarkL]

A working one - that's even better!  A working unit of anything makes a great comparison reference to fix a dead one.  Duplicate the settings and start comparing various nodes in the trouble area with either a DMM or oscilloscope, or even signature analyzer.

The scope FFT is definitely the right thing to adjust the harmonic distortion on these units.

It's always a great goal to prevent more landfill and it's why I like fixing this old equipment.  It's often a lot slower with less features than their modern counterparts, but a volt is still a volt, and great bargains can be found.  I think the only exception to that is oscilloscopes  They are such a valuable tool with ever increasing ease-of-use and analysis features, that it's worth the investment to refresh every 5 to 10 years, at least in my opinion.

I have a few things that are beyond repair, and in general I have a working version too.  I keep the dead one for parts, which also keeps it out of a landfill.  But I always make sure it's clearly labeled as broken, and I also mark the areas where I remove parts so when I look at it years later, probably for another part, I'll know what I did to it (hopefully).

[Mick B]

I'm am going to give this a rest. I will be back investigating the very hot IC's Thanks for the help.