EEVblog Electronics Community Forum
Electronics => Repair => Topic started by: gdnichols31 on February 07, 2022, 01:11:20 am
-
I've got an HP 8903A that has been working fine for a year or 2. I've got several threads here where I have repaired other problems. This is one of 2 units I have so i'm not sure which problems this one has had in the past but It' been working well.
One day I was checking an amplifier and the AC voltmeter seemed high. I checked and it was.
When first turned on it's about 16% high and then as it warms up it goes to about 25% high. It seems to be across all the metered values and all ranges. I have focused on the AC voltmeter.
I have been working on this instrument for about a year on and off and I don't know where to go next. I need help.
First I'll go thru what I have done.
The oscillator is perfect, the frequency is accurate and the output voltage is accurate.
As mentioned above I have 2 of these and the other one works fine. I swapped cards in and out until I found that the problem is on the A4 Output Amplifier/Voltmeter card. So the problem follows this card from unit to unit.
First thing I did was to check all the lytic caps and all the white axial Caps (Film?). They all tested good on 2 different capacitor checkers.
I went thru the troubleshooting procedure for the voltmeter (Service Sheet 7) and everything checked out. I couldn't find anything that looked suspect. All the waveform examples looked good.
See the attached schematic for the voltmeter section.
[attach=2]
[attach=3]
Since the meter is high on all functions I reasoned that the problem must be after the U18 Mux. It checks out, the output is always equal to whichever input is selected.
I checked all the voltages that are called out on the drawing. They are all good.
I don't understand totally how this circuit works, but the description given in the manual is;
All values are changed to a DC voltage and presented to the + input of a comparator U20
The computer generates a ramp gate signal, that starts a counter counting in the computer.
The gate also starts a ramp voltage (starting @ about -6V) at the - input to the comparator.
When the ramp voltage equals the DC voltage on the + input the comparator turns on.
The comparator output +5V is used to signal the computer to stop the counter.
The higher the measured voltage the longer the ramp runs and thus the higher the count and the measured value.
See the attached O'scope shot of the comparator output (top trace in yellow) and the ramp signal (bottom trace in magenta)
[attach=1]
They look good to me, but a 25% difference might be hard to detect.
I thought that the problem would most likely be in the ramp circuit, around U15.
I started checking all the components around it.
Since C70 controls/generates the ramp I thought it might be bad or going bad. I had already checked it and it was good. I checked again and it was still good. I replaced it with a new one and nothing changed.
I checked Q11 and Q12 and they are good.
I checked Q9 and it is good.
I checked VR5, the voltage on it was 5.85V instead of 6.2V. I installed a new one with a voltage of 6.2V, but no change in the meter reading.
This is where I'm stuck, I feel like I'm missing something simple.
I don't know what to check next.
Any ideas from a fresh pair of eyes would be great!
Thanks
Gary
-
Is there some sort of calibration that needs investigation? Since everything works fine other than calibration, that's where I would look first. I know nothing about this instrument so it's a shot in the dark. Is the readout digital or analog?
-
It's digital.
I don't think it's calibration because it just went bad all of a sudden.
It was accurate and then it wasn't.
Gary
-
The analogue signals look OK, so maybe it's a digital problem. Is the (DVM) counter clock frequency correct? Is the counter logic working properly?
-
The A4 board is all analog. So the problem is somewhere on that board.
Last night I decided to start at the beginning (input) again and start testing. This time I did find a problem with the overvoltage detector. I will trouble shoot it tonight.
-
Well I finally have found a problem!
The overload circuit is overloading at a voltage lower than it's supposed to.
I'm going thru that circuit now.
-
I went thru the overload circuit and it appears to be good.
I'm thinking somehow the voltage coming in to it is too high causing it to trip early. When it trips it signals the computer to change the amp gains and/or attenuator settings making it read high. I'm guessing.
-
A couple of things come to mind when investigating faults on these units. I can share my experiences. I had a problem the A5 oscillator board on my 8903B. As the unit warmed up the oscillator distortion rose, particularly at low frequencies. I initially suspected the 2N4392 FETs that were associated with the high value resistors that were switched into the state variable oscillator circuit. As I did not have an extender card I took the board out and tested the FETs individually by biasing the gate voltage and to measure the drain current (down to 10nA using a current source) to determine whether the FET was fully turning off. All passed ok but if you have not already extracted it and tested it for low levels of leakage then in your case I would suspect Q9 and the associated Vgs to bias Q9 fully off. Simple testers do not always spot everything and they often apply only low voltages that do not necessarily reveal a problem. An example is the real fault on my A5 oscillator board. I could see that the waveform from the output of U14 (V Peak) was not consistent. I checked surrounding components and began to suspect U14. However, I looked closer at C45 (TRW 0.1uF white polypropylene), C45 measured OK in value and with multi-meter (40M Ohms max) C45 had no measurable leakage. I applied a bias voltage (c.15v) and then measured the leakage. The capacitor showed leakage of to order of 20-90nA, the leakage varied randomly and increased with temperature. C45 (sample and hold capacitor) is a critical component in the function of the S/H and this may not be the case for similar white polypropylene capacitors that are used in the 8093B but this experience make me a little wary of them.
-
Thank you for the reply.
I also had to replace C45, the S/H capacitor on one of my units. It had shorted (1 or 2 ohms if I remember correctly). It was also causing a high THD on the oscillator.
I'm thinking you meant Q7 in my case, instead of Q9. Q9 turns the SINAD meter on and off. Q7 generates the overload signal to the computer.
The overload circuit is tripping at 3.95V. The manual said to enter 4.0V and it should not trip. Then they have you enter 4.2V and then it should trip. So I guess it should trip between 4.0V and 4.2V somewhere. The voltage on the - comparator input is exactly where it should be and within the spec on the schematic.
I will pull Q7 and check it and also watch the + comparator input and see what it does.
Thanks again for your input! I need all the help I can get.
Gary
-
I checked both comparator inputs and the Q7 output. It trips perfectly based in the value of the - input, which is 3.79V. I think Q7 is good because comparator output is off, Q7 is off and providing a 4.7V 'logic high' to the computer. When the comparator turns on, Q7 turns on and grounds the input to the computer at 50mV, plenty low enough for a logic low.
The 3.79V on the - input is right at the bottom of the spec (3.79V to 4.01V). I think I am going to check the value of R60 (10k) and see if it is high. If so I may choose another to try to hit the spec range in the middle.
-
I checked the values of R60 & R61. They were very close. I chose another resistor for R60, I think it was 9.93K. it gave me a 'trip voltage' on the - comparator input of 3.81V which is better but it still tripped early.
One of the things I noticed is the 15V supply is low. It's 14.64V on the A4 card. The spec is 14.67V. This will also make the circuit trip early.
I think I am going to order a new voltage regulator and see if I can raise that voltage up. It's coming out of the regulator at 14.67V.
Maybe I'll get lucky
Gary
-
I just reviewed the block diagram for the 8903. I had one but that was years ago. The instrument uses a voltage to time ADC solution. Lets start with some basics. The voltage measured is off at 10 MV and at 10V? Is it off by about the same? Going from 1V to 10V for instance with no range changes is the error constant? The goal here is to make sure there are no attenuators that are not switching. The internals use their attenuator and ADC to determine the voltage and any failing attenuator will cause big errors. Next is to compare errors on averaging vs. RMS. Both should be the same with a good sine wave. If the errors are constant with range and level the ADC is not converting right. Start with the active rectifier (A4 U5). It could have a dead Schottky diode. Those are HP specific most likely. It seems the AC-DC conversion is there and the ADC is the voltage to time stage (around u15). That could also have a bad cap or current source.
The HP manual shows a number of variations on that section. I did not look at all of them.
Another possibility: the master clock is off. Check the frequency of a known source with it to see if the counter circuits are working right.
The failure should be a single point- active device, resistor, cap. The errors are too large to be a small power supply error. And the gain should be independent of the supply voltage across a pretty large range. The numbers changing as it warms up suggests either a cooked resistor or a bad active part.
I hope this helps. Those are not east to troubleshoot.
-
The 15V supply is off by .36V. I know that isn't much and it is most likely not the problem, but I wanted to get the voltage right before I continued so I would know it wasn't the voltage.
The measured voltage on the 8903 is high by a consistent 25-30% from 0 to 6V. That's AC volts, the THD measurement is also high. The error starts out at about 15% high and then goes to 30% high.
I ordered 3 new TO-3 15V regulators and they are ALL 4.65V or 4.65V! I guess in the 80's 15V really meant 14.65V!
How would I check the counter? The CPU clock is the only thing I can think of. I wouldn't think it would speed up would it? I will check it.
Thanks
Gary
-
Measure the frequency of an incoming known source (even power line frequency). If the internal counter is working right then probably all the clocking is working right.
It seems from your description that the meter error changes with level. Also its present in different modes so its probably in the metering circuit itself. The service manual on page 8B-13 (158 of the PDF of the service manual part 1) has a detailed troubleshooting process for the meter circuits. Probably better than anyone else could suggest. Its tedious but will help focus your efforts.