HP 334A troubleshooting help

[dave the sound guy]

I'm attempting to align my 334A distortion analyzer and have hit a snag.  After setting the 400Hz CAL as described on page 5-10 of the manual it calls to reduce the input to 300uV and verify correct indication on the .0003V range.  The problem is that below around 400uV of input the meter accuracy drops off sharply and it begins to read much too high (full deflection on the .0003 range corresponds to a 144uV input).  The error is present on both .001 and .0003 ranges.

I'm suspecting the issue is in the meter amplifier, as I've compared the panel input with the signal present at input pin 6 of the meter amp board and they closely agree.  I replaced A2R30 with a trimmer pot but found adjusting it would sway the pointer only a tiny amount in either direction. 

When I tested voltages I found some that were a bit off, especially at the meter (see attachment), but I'm not entirely sure how to interpret the readings.  I also removed C18, C26, C27, and C28 and tested for leakage.  C28 exhibited 1.2uA @ 25VDC, the others were lower. 

How much leakage is permissible?  Any ideas what might be going on?

Thanks for any help.

[Jeff eelcr]

Dave, Do you mean HP 334A? On these units serial number and PCB numbers may help also.
Jeff 

[dave the sound guy]

Whoops!  yes, 334A.  Serial 743-02398...A2 board is 00333-66502 Rev D.  I believe the meter amp circuit is where the fault resides because the meter exhibits full deflection with only 144uV present on A2 pin 6 with the .0003V range selected.

I just tried bypassing the electrolytic capacitors in the meter amp one at a time with a 10 meg resistor to simulate leakage and see if I could get the reading to change but got nothing.  I'm going to bet the capacitors are not the problem here...

Q5 tests good on my DCA75.  So does CR6 and CR7, no current in reverse bias.

[dave the sound guy]

Not sure why I didn't do this before but I had a look at the signal path on my scope.  It looks there's some noise being generated upstream and it doesn't appear to be externally induced.

Input of meter circuit from impedance converter and post AT. (A2 pin 6):


Output to meter


No signal (pin 6)


Switching noise from the FET?

I also just noticed C4 has been changed to 3.9uF...

[wn1fju]

Not quite sure what is the problem, but I do know that these units will tend to generate internal noise at the lowest voltmeter settings.  Have you done some basic no-input checks?

I have both the 332A (manual) and 334A (automatic) versions.  On my 332A, when I terminate the input with 50 ohms, the meter will read 0.1
on the 3-scale at the 0.0003V setting (10uV).  Unterminated, the meter will go to about 80% of full scale if I bring my finger within 1/2 inch of the input terminals.

On my 334A, it is much worse.  Same 10uV with a terminated input, but unterminated, the meter will peg on the 0.0003V scale and nearly
peg on the 0.001V scale.

Yet both meters read perfectly fine with real sinusoidal inputs all the way down to 25uV and the distortion measurements are accurate.

Don't know exactly what all this means - just an observation and another data point for you.  Good luck.

[dave the sound guy]

My unit behaves much the same way in regards to sensitivity, except that it reads significantly high in the low range with an input.  If I ground the input on the .0003 range the meter reads about 2 minor ticks above zero...I believe my problem is one of scale and not offset. 

Could my setup be part of the problem?  I'm using a restored 200CD with wire leads and to make it usable in the microvolts I had to place a voltage divider at the input of the 334 consisting of a 10KR in series and a 1R across the input and ground.   To read the input voltage I have a 34401A clipped across the 1R. 


To clarify my previous post I was going through the troubleshooting guide figure 7-9b in the manual.  I think the positions of the negative voltages are flipped top-bottom.


(1) I verified the front input and the meter circuit input agree.  I can adjust amplitude and the readings stay within several uV of each other.  For the following checks and to read the meter the multimeter leads are removed from pin 6 and 15 (ground).  On .0003 range meter is off scale above a front input of 150uV. 


(2) Input signal returned to 300uV.  Q7 output looks about right...


(3) Amplitude is too high. 


(4) Also too high...


The bizarre thing is that on other ranges the waveform voltages are correct at full deflection and it reads accurately, and the error affects both .0003 and .001 ranges.   The periodic noise can't be normal.  It's worth noting that it looks like A2Q1 was desoldered for either testing or replacement.

It seems as though I cannot post images in-line...?

[wn1fju]

I tried some measurements with a 1 ohm and 10,000 ohm divider and they were spot on at 100uV.  So that shouldn't be an issue.

On my 332A, I get the waveforms shown in the troubleshooting diagram all the way down to the lowest voltage scale, even though HP says you can't make the measurement below 0.01V because the circuit will oscillate.  The peak-to-peak values are approximately correct, but the raw values differ by as much as 0.5 volt due to the biasing of the transistors. 

I'm not going to be of much further help probably.  But might I suggest that you remove the boards, scrub them real good with alcohol, let them dry for a few hours and reinstall.  Perhaps there is some contamination left over from the previous owner.  There are, of course, very high impedance circuits in the distortion analyzer that could be experiencing some leakage.

[dave the sound guy]

Thanks for taking the time to check.  Were your waveforms also "fuzzy" like mine?  I'm fairly certain something is noisy in my unit, unless it's coming from my DS1054z. 

I did clean the A2 board at one point with 99% iso, and I always make sure to completely mop up any flux residue that forms around the joints.

[wn1fju]

I disconnected pin #6 and supplied my own 400 Hz signal onto that pin (and pin 15 ground).  I used 300 uV input for the 0.0003 scale and 1 mV input for the 0.001V scale.  The two pictures below are the measurement from Q7-Collector.  I even dug out my Rigol scope, which I hardly ever use, just so my pictures would be a good comparison to yours.  How's that for service?  The noisier figure is the one at 300 uV.

As you can see, my unit is even worse than yours in terms of high frequency noise.  So perhaps you are chasing a red herring on that front.

[dave the sound guy]

I really appreciate the help and testing on your end, but for it to be a true comparison can you leave pin 6 connected and supply the signal to the front input? 

On mine, the noise that's being imposed over the signal is periodic, at about 80khz.  I've unplugged all switchmode power supplies in the area and turned off any LED lights but there's no change. 

Here is what I see at Q7 collector with no signal present, input to ground...top is with the .0003 range, bottom is .001.  Note my probe AT was 1x but my scope was set to 10x.  It could very well be a red herring but I'm running out of ideas...

What's interesting is that on .0003 and .001 ranges the amplitude of the "spikes" are about the same, but it decreases as you continue up the selector.

I'm tempted to just throw new capacitors into the board and see what that does, but that's basically giving up and I want to understand what's causing the error, and of course it may not even solve the problem.

[dave the sound guy]

I've determined the noise is externally induced.  I see it when the 334 is off!

[wn1fju]

OK, no need for me to do further measurements.  Sounds like you have a mystery noise generator somewhere.  I'm sure you will find it.

The Rigol scope does emit its own noise, particularly from the display circuitry.  You can see this if you hook the tip of your probe to the attached ground clip to make a loop, then hold it up to the screen.  Of course, it is only a few mV and appears to be around 30 kHz.  Nothing like what you are seeing.

[dave the sound guy]

Whelp, I wasn't able to pinpoint the fault...but after throwing new Vishay/Sprague TVA caps and a couple new tantalums onto the A2 board the 334A now reads accurately into the microvolts, and distortion measurement agrees with my Keithley 2015.