Hello everyone, I'm new to the forum, but I know there are some very skilled technicians in this group.
I'm slowly building my small audio repair lab, and one of my projects is to build a tube audio amplifier.
Among the instruments I’ve painstakingly purchased, I have a beautiful HP-339A distortion analyzer… the only problem is that it doesn’t work!
I’m trying to repair it, but I’ve encountered several difficulties.
The oscillator works correctly; on the oscilloscope, I measure a perfect sine wave with the correct frequency and amplitude from 1 Hz to 100 kHz.
However, the notch filter does not track properly. The LEDs indicating frequency adjustment often stay lit, even when using the internal oscillator directly connected to the distortion analyzer input.
Occasionally, the notch null occurs (usually when the third frequency knob is set to 1x or 100x, almost never at 1k or 10k), but most of the time, it doesn’t.
The problem has worsened over the past few weeks. I’ve checked all the electrolytic capacitors in the instrument, including the tantalum ones, without finding any issues.
I also cleaned the frequency switch contacts with isopropyl alcohol, but it didn’t improve the situation.
Therefore, I’m seeking help from anyone who might have faced a similar problem.
Thank you!
I'm slowly building my small audio repair lab, and one of my projects is to build a tube audio amplifier.
Among the instruments I’ve painstakingly purchased, I have a beautiful HP-339A distortion analyzer… the only problem is that it doesn’t work!
I’m trying to repair it, but I’ve encountered several difficulties.
The oscillator works correctly; on the oscilloscope, I measure a perfect sine wave with the correct frequency and amplitude from 1 Hz to 100 kHz.
However, the notch filter does not track properly. The LEDs indicating frequency adjustment often stay lit, even when using the internal oscillator directly connected to the distortion analyzer input.
Occasionally, the notch null occurs (usually when the third frequency knob is set to 1x or 100x, almost never at 1k or 10k), but most of the time, it doesn’t.
The problem has worsened over the past few weeks. I’ve checked all the electrolytic capacitors in the instrument, including the tantalum ones, without finding any issues.
I also cleaned the frequency switch contacts with isopropyl alcohol, but it didn’t improve the situation.
Therefore, I’m seeking help from anyone who might have faced a similar problem.
Thank you!
Maybe I can help. I've assisted with several members, a few years ago.
I see you have a scope; do you have x10 probes? If you have a second sine wave generator, that can be very helpful when internal oscillator doesn't track the notch. You have service manuals?
Have you started through performance checks? Where do you run into trouble?
I see you have a scope; do you have x10 probes? If you have a second sine wave generator, that can be very helpful when internal oscillator doesn't track the notch. You have service manuals?
Have you started through performance checks? Where do you run into trouble?
Yes , i have a scope (HP-465B) and 10x probes , as sine wave generator i can use the HP 3325A, i have the original service manual, but it's not the "old" HP service with troubleshooting root..
As i noticed that only in 10x and 1x multiplier setting i have a good notch i used an external oscillator (3325A) and have almost the same problem. I have measured the internal oscillator and it works fine and is in spec for level and for frequency. Also the internal voltmeter works perfectly.
My assumption is that something happened in the notch circuit . I have also checked power supply for voltage and for ripple and is OK.
I checked also electrolitics in circuit and i cannot find any leakeage or open/short problems.
The main issue is that i don't' understand how the notch circuit works , my knowledge of electronics is just hobbyst.
As i noticed that only in 10x and 1x multiplier setting i have a good notch i used an external oscillator (3325A) and have almost the same problem. I have measured the internal oscillator and it works fine and is in spec for level and for frequency. Also the internal voltmeter works perfectly.
My assumption is that something happened in the notch circuit . I have also checked power supply for voltage and for ripple and is OK.
I checked also electrolitics in circuit and i cannot find any leakeage or open/short problems.
The main issue is that i don't' understand how the notch circuit works , my knowledge of electronics is just hobbyst.
Do you have good notch depth (i.e. good residual distortion at all frequencies in the x1 and x10 ranges? That would tend to absolve the switched resistors as culprits. If this is not the case, perhaps they are still suspect.
It's a good sign that auto-nulling appears to be functioning. Any concerning issues in the two low frequency ranges? Do the rotary switches "feel" consistent/reliable--- "clean"?
If so, let's move to the next problematic range and try to find an issue, ideally one with glaring misbehavior.
It's a good sign that auto-nulling appears to be functioning. Any concerning issues in the two low frequency ranges? Do the rotary switches "feel" consistent/reliable--- "clean"?
If so, let's move to the next problematic range and try to find an issue, ideally one with glaring misbehavior.
in x1 and x10 i have about 90-95 db down, but i see the distortion fluctuates a little (i will try to make a video) before i got problems i remember a better null (about 100 db) . In 1x i see a curious behaviour: the 339 nulls then loose null then get it again is like it tries to null a couple of time before getting it right.
in 1k and 10k it does not null at all.
I have checked the alignment of the 1,10,1k and 10k shaft, i think it was a little off, i realigned in the way i see that the contacts of the notch part are perfectly aligned, but without any result.
i have read that sometimes the octocouplers can fail... i am a little worried about that unobtanium component.
in 1k and 10k it does not null at all.
I have checked the alignment of the 1,10,1k and 10k shaft, i think it was a little off, i realigned in the way i see that the contacts of the notch part are perfectly aligned, but without any result.
i have read that sometimes the octocouplers can fail... i am a little worried about that unobtanium component.
Your HP339A appears to be in beautiful condition! Is there any indication that it's been repaired in the past?
There's some evidence of problems in the frequency setting resistors/switch contacts. All sections of the Tens and Units switches have to track properly to have consistent performance--- and there's a bunch of them. In the oscillator, there are two switched elements that establish frequency (each having 100 positions). In the analyzer section there are two more switched R sections that set notch frequency and a fifth section that tunes a 90 degree phase-shift network. All five sections must track together for proper operation.
I suggest first confirming that the oscillator tunes smoothly so that it's not asking the notch to tune to an incorrect frequency. If you have the luxury of a frequency counter use it. Tune the oscillator from 1.0 x100 to 1.9 x100 and confirm smooth increments in frequency. Then repeat from 2.0, 3.0, 4.0, ---, 10.0 x 100. If you don't have a counter, use your scope.
If you find the switch contacts stlll need attention, I've had good success with Deoxit D5, but it seems to have become pricey since I last bought. Use it very sparingly, almost contact-by-contact to avoid a lot residue.
More to follow in the analyzer section shortly, but I wanted to check in.
There's some evidence of problems in the frequency setting resistors/switch contacts. All sections of the Tens and Units switches have to track properly to have consistent performance--- and there's a bunch of them. In the oscillator, there are two switched elements that establish frequency (each having 100 positions). In the analyzer section there are two more switched R sections that set notch frequency and a fifth section that tunes a 90 degree phase-shift network. All five sections must track together for proper operation.
I suggest first confirming that the oscillator tunes smoothly so that it's not asking the notch to tune to an incorrect frequency. If you have the luxury of a frequency counter use it. Tune the oscillator from 1.0 x100 to 1.9 x100 and confirm smooth increments in frequency. Then repeat from 2.0, 3.0, 4.0, ---, 10.0 x 100. If you don't have a counter, use your scope.
If you find the switch contacts stlll need attention, I've had good success with Deoxit D5, but it seems to have become pricey since I last bought. Use it very sparingly, almost contact-by-contact to avoid a lot residue.
More to follow in the analyzer section shortly, but I wanted to check in.
In the analyzer notch, I'll steer you to the "deep end of the pool," as it will provide the quickest insight to problems.
Find TP2 and TP1 in the A4 circuit board. These points correspond to drive currents delivered to the optocouplers and indicate how the auto-tuning section is attempting to null the notch filter for best notch depth. You'll note that TP2 controls activation of the frequency arrows on the front panel. In perfection, TP2 would be about 1.35V, centered between the trip points of U6B and U6C, but any TP2 voltage within this window keeps the arrows extinguished. Similar constraints apply to TP1. Both these test points adjust dynamically to achieve maximum notch depth.
Monitor both these test points for indication, but TP2 will likely be the insightful point, as you are seeing frequency arrows light. Within a given decade range, these TP's should settle at some nominal voltage and show only subtle variation as you tune the Tens and Units knobs. Any dramatic deviation after settling is indication of a defective switched resistor path. Use these deviations from nominal as clues to the defective settings. Each 1.0 to 1.9 and 1.0, 2.0, --- to 10.0 resistor path is a possible culprit. Remember the 90 degree phase shift and its resistors are also possible defects.
Recognize that the lower 5 positions on the frequency knobs select a singe resistor, but the upper 5 positions have two resistors selected in parallel.
When you find a problematic notch frequency, using your separate generator as a trouble shooting source to search for notch nulling may be helpful.
I hope my descriptions are adequate, but don't hesitate to ask for clarification.
Keep us posted. Good luck!
Find TP2 and TP1 in the A4 circuit board. These points correspond to drive currents delivered to the optocouplers and indicate how the auto-tuning section is attempting to null the notch filter for best notch depth. You'll note that TP2 controls activation of the frequency arrows on the front panel. In perfection, TP2 would be about 1.35V, centered between the trip points of U6B and U6C, but any TP2 voltage within this window keeps the arrows extinguished. Similar constraints apply to TP1. Both these test points adjust dynamically to achieve maximum notch depth.
Monitor both these test points for indication, but TP2 will likely be the insightful point, as you are seeing frequency arrows light. Within a given decade range, these TP's should settle at some nominal voltage and show only subtle variation as you tune the Tens and Units knobs. Any dramatic deviation after settling is indication of a defective switched resistor path. Use these deviations from nominal as clues to the defective settings. Each 1.0 to 1.9 and 1.0, 2.0, --- to 10.0 resistor path is a possible culprit. Remember the 90 degree phase shift and its resistors are also possible defects.
Recognize that the lower 5 positions on the frequency knobs select a singe resistor, but the upper 5 positions have two resistors selected in parallel.
When you find a problematic notch frequency, using your separate generator as a trouble shooting source to search for notch nulling may be helpful.
I hope my descriptions are adequate, but don't hesitate to ask for clarification.
Keep us posted. Good luck!
Good, it’s clear to me how to take the measurements on TP1 and TP2, and I’ll try in the coming days.
In the meantime, I can tell you that the oscillator works correctly across its entire operational range, as checked with a good frequency counter.
I must therefore assume that the issue lies either in the analyzer or in the synchronization between the two.
I’ll try setting the analyzer to 5 kHz and see if, using an external signal, there’s a frequency that turns off the LEDs. This is to determine whether it’s a misalignment issue.
The instrument was intact when purchased, with no modifications or component replacements, and it worked correctly until about a month ago.
Thank you very much for your help.
In the meantime, I can tell you that the oscillator works correctly across its entire operational range, as checked with a good frequency counter.
I must therefore assume that the issue lies either in the analyzer or in the synchronization between the two.
I’ll try setting the analyzer to 5 kHz and see if, using an external signal, there’s a frequency that turns off the LEDs. This is to determine whether it’s a misalignment issue.
The instrument was intact when purchased, with no modifications or component replacements, and it worked correctly until about a month ago.
Thank you very much for your help.
this morning a fast test: external source set at 5khz the distortion analyzer set at 5 khz told me to low the settings. I have the best notch setting the frequency of the notch circuit at 3,7 khz it's definitely a mild notch (about 40 db) but in this position and near this the low led turns off.
This morning while i was doing some tests, the 339 started to work again as usual. Null in all frequencies and more that 90db measure on the internal oscillator ..... it worked for 5 minutes flawlessy, then i turned off, again turned on and it failed exactly as before, with 1k and 10k bands not nulling anymore. At this point i am a little disappointed. it's not a mechanical issue it's for shure an electronic problem maybe of a marginal component. It's not to simple to trace!
That does sound like some latch-up phenomenon. But in the failed state, you're seeing consistent, repeatable misbehavior when alternating between working and non-working frequencies--- right? But if you can catch the analyzer in it's failed mode, you can still troubleshoot. If you have covers open, I'd double check supply voltages, especially in the failed state.
The symptoms sure seem like defects in the frequency control switches and/or their R's and C's, but there could be phenomena I don't foresee. I suggest a systematic survey of working/non-working frequencies, as follows:
Leave the Units knob set to 0. Form a table/spreadsheet/matrix so that you can more easily spot patterns of misbehavior. Maybe use Excel if you're familiar with it. I suggest ten rows of frequency, 1.0, 2.0, 3.0,---, 10.0 and four columns x10, x100, x1k, x10k. This will give you 40 entries to explore and will exercise all resistors, caps, and switch contacts (excluding Units). Mark offending frequencies in the matrix with an X or similar. You can acquire this data without even removing the covers, but monitoring TP1 and TP2 might be insightful.
If you study the rotary switch wiring on the A3 board schematic, you'll see that this exercise should generate a fairly thorough exploration of the frequency setting components.
Good luck.
The symptoms sure seem like defects in the frequency control switches and/or their R's and C's, but there could be phenomena I don't foresee. I suggest a systematic survey of working/non-working frequencies, as follows:
Leave the Units knob set to 0. Form a table/spreadsheet/matrix so that you can more easily spot patterns of misbehavior. Maybe use Excel if you're familiar with it. I suggest ten rows of frequency, 1.0, 2.0, 3.0,---, 10.0 and four columns x10, x100, x1k, x10k. This will give you 40 entries to explore and will exercise all resistors, caps, and switch contacts (excluding Units). Mark offending frequencies in the matrix with an X or similar. You can acquire this data without even removing the covers, but monitoring TP1 and TP2 might be insightful.
If you study the rotary switch wiring on the A3 board schematic, you'll see that this exercise should generate a fairly thorough exploration of the frequency setting components.
Good luck.
there is a news..... i tried to warm the board and components near the dist. analyzer area. i used an hair phon. At one point the instruments started to work normally. After a few minutes it begun to malfunction as usual... now i have to find the cause. i tought it was in the c1 c2 c3 zone but i tested the components and they are good.
Hi,
I've pasted a table I generated in Excel. It details how the knobs' frequency determining resistors from amongst the array of R's and C's. Refer to Figure 8-3 in the service manual for a simplified view of how these components are defined, i.e. R, Ra, Ca and Cb. "R phase" and "C phase" are my own designators for R/C components that tune the 90 degrees phase shift circuit.
Note that in the lowest knob positions, there is only a single resistor selected in each of the three switch decks for the analyzer frequency, eg. R20, R31, R1 in the "1" position. So it would provide a lot of insight if you can a single x decade range in which all "Tens" positions perform properly, or alternately a singe Tens position ( any position 1 through 6) that works in all four decade ranges. Said differently, can you find one row or one column in bottom matrix in which the analyzer nulls? I entered two cells in the matrix (in dB) depicting the two 1kHz signals you highlighted in your video.
In your last post you mention suspicion of the C1 C2 C3 C4 caps in the phase shift circuit. You confirmed good capacitors, but of course the rotary switches are possible culprits. This section of the analyzer is relatively easy to test. Use your scope's channel 1 to monitor and sync TP4 in the A3 board; with the scope's channel 2, observe TP1. It should have the same amplitude as channel 1, but be shifted in phase by 90 degrees re channel 1. It should present this behavior at all operating frequencies. Confirming that the phase shift circuit operates properly lets you remove it from suspicion.
I've pasted a table I generated in Excel. It details how the knobs' frequency determining resistors from amongst the array of R's and C's. Refer to Figure 8-3 in the service manual for a simplified view of how these components are defined, i.e. R, Ra, Ca and Cb. "R phase" and "C phase" are my own designators for R/C components that tune the 90 degrees phase shift circuit.
Note that in the lowest knob positions, there is only a single resistor selected in each of the three switch decks for the analyzer frequency, eg. R20, R31, R1 in the "1" position. So it would provide a lot of insight if you can a single x decade range in which all "Tens" positions perform properly, or alternately a singe Tens position ( any position 1 through 6) that works in all four decade ranges. Said differently, can you find one row or one column in bottom matrix in which the analyzer nulls? I entered two cells in the matrix (in dB) depicting the two 1kHz signals you highlighted in your video.
In your last post you mention suspicion of the C1 C2 C3 C4 caps in the phase shift circuit. You confirmed good capacitors, but of course the rotary switches are possible culprits. This section of the analyzer is relatively easy to test. Use your scope's channel 1 to monitor and sync TP4 in the A3 board; with the scope's channel 2, observe TP1. It should have the same amplitude as channel 1, but be shifted in phase by 90 degrees re channel 1. It should present this behavior at all operating frequencies. Confirming that the phase shift circuit operates properly lets you remove it from suspicion.