I share your hope a forum member has some useful link.
Just two or three days ago, I did stumble into this very fascinating bit from the engineer at HP who designed the notch filter in the 339A--- fellow by the name of Joel Nesheim. His post is in the reader-comment portion of the link below. I don't know if he could be tracked down and might respond to a direct appeal.
Measuring Distortion with a HP 339A and a Keithley 2015 THD | gerrysweeney.com
Good luck.
Just two or three days ago, I did stumble into this very fascinating bit from the engineer at HP who designed the notch filter in the 339A--- fellow by the name of Joel Nesheim. His post is in the reader-comment portion of the link below. I don't know if he could be tracked down and might respond to a direct appeal.
Measuring Distortion with a HP 339A and a Keithley 2015 THD | gerrysweeney.com
Good luck.
Measuring Distortion with a HP 339A and a Keithley 2015 THD | gerrysweeney.com
The link didn't past very well. This may be no better.
The link didn't past very well. This may be no better.
404 Not Found
Joel Nesheim on August 11, 2014 at 2:07 am said:
Thank you for the video of the 339a distortion analyzer. I watched the tear down video as well. It was a trip down memory lane for me since I designed the notch circuit (part of which is on that large circuit board in the back that you had some trouble removing.
That board kept me awake many nights. By the way, both the notch circuit and the oscillator ciercuit are based upon the “bridged T” notch filter which is tuned by the resistors and capacitors chosen by the wafer switches, then fine tuned by two photodiode/LED modules.
I spent weeks doing calculations comparing the responses of Wien bridge, twin T, phase shifter, and bridged notch circuits before I chose the bridged T configuration.
The biggest headache was the slower speed of the auto null circuit as the frequency to be measured approached 10 Hz.
Production started in 1977 I think and I followed the 339a into production which was a very busy and frustrating time for me. You end up reporting to several bosses at once and their objectives may differ.
The oscillator and input circuits were designed by Fred Kitson and the power supply was designed by the project manager, John Manning.
Design and production were done at HP inLoveland Colorado. The 339 was later produced at HP in Japan.
Best regards, Joel Nesheim
Reply ↓
Obviously, I still don't know how to generate a link. ;(
Joel Nesheim on August 11, 2014 at 2:07 am said:
Thank you for the video of the 339a distortion analyzer. I watched the tear down video as well. It was a trip down memory lane for me since I designed the notch circuit (part of which is on that large circuit board in the back that you had some trouble removing.
That board kept me awake many nights. By the way, both the notch circuit and the oscillator ciercuit are based upon the “bridged T” notch filter which is tuned by the resistors and capacitors chosen by the wafer switches, then fine tuned by two photodiode/LED modules.
I spent weeks doing calculations comparing the responses of Wien bridge, twin T, phase shifter, and bridged notch circuits before I chose the bridged T configuration.
The biggest headache was the slower speed of the auto null circuit as the frequency to be measured approached 10 Hz.
Production started in 1977 I think and I followed the 339a into production which was a very busy and frustrating time for me. You end up reporting to several bosses at once and their objectives may differ.
The oscillator and input circuits were designed by Fred Kitson and the power supply was designed by the project manager, John Manning.
Design and production were done at HP inLoveland Colorado. The 339 was later produced at HP in Japan.
Best regards, Joel Nesheim
Reply ↓
Obviously, I still don't know how to generate a link. ;(
progress--kinda
Fitted the two quadamps to A4 board and,
Before with an external osc I had a null when 339A was set to 1kHz and the osc was at 860 Hz, now the null comes at 1,160Hz.
on and off null
Using the internal osc I can get a null using the osc Vernier at max clockwise, frequency measured at 1,160Hz.
In CAL position the internal osc frequencies are very close to perfect.
A4 board-
TP5, TP3 equal amp sine waves
TP4, sine wave with greater amp, lower with null
TP1, on null 3.25vdc, off null 1.2vdc . slow rise and fall
TP2, on null 0.45vdc off null 0.65vdc
Fitted the two quadamps to A4 board and,
Before with an external osc I had a null when 339A was set to 1kHz and the osc was at 860 Hz, now the null comes at 1,160Hz.
on and off null
Using the internal osc I can get a null using the osc Vernier at max clockwise, frequency measured at 1,160Hz.
In CAL position the internal osc frequencies are very close to perfect.
A4 board-
TP5, TP3 equal amp sine waves
TP4, sine wave with greater amp, lower with null
TP1, on null 3.25vdc, off null 1.2vdc . slow rise and fall
TP2, on null 0.45vdc off null 0.65vdc
Hi Microx. I hoped no news was good news.
Using external oscillator, I'd like to study behavior around A4 TP2, and U5 pin 9. These two voltages should always be equal, irrespective of quality of the null.
Assuming this is found to be the case, would you advise the behavior of TP2 at below the null frequency, at null, and above the null frequency? Any associated frequency arrow activity?
You might explore U3A pin 1 versus U5A pin 1. I think they should roughly similar waveforms when the instrument is at null.
Using external oscillator, I'd like to study behavior around A4 TP2, and U5 pin 9. These two voltages should always be equal, irrespective of quality of the null.
Assuming this is found to be the case, would you advise the behavior of TP2 at below the null frequency, at null, and above the null frequency? Any associated frequency arrow activity?
You might explore U3A pin 1 versus U5A pin 1. I think they should roughly similar waveforms when the instrument is at null.
I'm concerned that U5 pin 9 doesn't appear to track TP2. Perhaps it changed during the duration of the measurement. If you connect voltmeter leads between these points, difference should be less than 10mV regardless of null quality. Failure of this check suggests a problem with U5D, Q2, photo-coupler LED, etc.
If you monitor TP2, does it vary smoothly with changes in oscillator frequency? Over what frequency range is nulling achieved? What is null depth? I.e. is nulling loop working at all? Waveform you describe at U3A pin 1 sounds plausible. When loop is nulled properly, its average voltage will be 0.
I'm on a long road trip today and tomorrow, but I'll try to check in when possible.
Good luck.
If you monitor TP2, does it vary smoothly with changes in oscillator frequency? Over what frequency range is nulling achieved? What is null depth? I.e. is nulling loop working at all? Waveform you describe at U3A pin 1 sounds plausible. When loop is nulled properly, its average voltage will be 0.
I'm on a long road trip today and tomorrow, but I'll try to check in when possible.
Good luck.
U5 pin 9 does not track TPS, several volts difference.
TP2. If the 339a is set up for a null with external osc. and I switch it off and then on again the voltage on TP2 starts at about 2vdc and then rises very slowly to 8 to 10 v .
U5 has been replaced, I will check Q2. The photo resistor I will remove and test.
TP2. If the 339a is set up for a null with external osc. and I switch it off and then on again the voltage on TP2 starts at about 2vdc and then rises very slowly to 8 to 10 v .
U5 has been replaced, I will check Q2. The photo resistor I will remove and test.
more progress
Flooded pump room sorted so back to the beast,
Board A4,
U3, U5 , U6 changed for TL074CN. Before the change I had a null with ext osc under the frequency set by the 339A, after the change the null is at higher ext osc frequency. Maybe the TL074 has gain difference, suggestions...
Board A2
Meter crashing to full scale, random readings etc. Traced to U8, changed for TL074CN, fixed.
Noise on meter when engaging 80kHz filter, changed U1, fixed.
I have now changed all the quadamps.
Rigged up a test of the two frequency leds, one poor the other open circuit. Jury rigged two leds to find that with a null from ext osc the leds work perfectly, high and low, with knife edge sensitivity.
Still on it when time permits.
Flooded pump room sorted so back to the beast,
Board A4,
U3, U5 , U6 changed for TL074CN. Before the change I had a null with ext osc under the frequency set by the 339A, after the change the null is at higher ext osc frequency. Maybe the TL074 has gain difference, suggestions...
Board A2
Meter crashing to full scale, random readings etc. Traced to U8, changed for TL074CN, fixed.
Noise on meter when engaging 80kHz filter, changed U1, fixed.
I have now changed all the quadamps.
Rigged up a test of the two frequency leds, one poor the other open circuit. Jury rigged two leds to find that with a null from ext osc the leds work perfectly, high and low, with knife edge sensitivity.
Still on it when time permits.
Sounds like great progress.
On board A4: TP2 and R55 voltage should always track each other within a few mV, despite misbehavior elsewhere in the instrument. Same applies to TP1 and R22 voltage. If this is the observed behavior, then the optocoupler current sources are working properly and trouble lies elsewhere. I'm going to assume this is the case until advised differently.
If the instrument were working properly, the analyser should be able to null over a range of a few percent, centered around the frequency specified by the knobs. I take "knife edge sensitivity" to mean null tracking is very restricted? And higher than knob frequency? These symptoms lead me to guess that the TP2 is "stuck" near ~ 8V, where the cell resistance is near minimum. Hence, null frequency is higher than nominal and tracking range is very restricted. In this scenario, the problem probably lies somewhere in the circuitry bounded by TP2, TP3, and T4.
With knobs set to 1.0 x 1kHz, perhaps you can tabulate voltages versus frequency: TP2, TP1, null depth, and anything other behavior that seems noteworthy.
Best.
On board A4: TP2 and R55 voltage should always track each other within a few mV, despite misbehavior elsewhere in the instrument. Same applies to TP1 and R22 voltage. If this is the observed behavior, then the optocoupler current sources are working properly and trouble lies elsewhere. I'm going to assume this is the case until advised differently.
If the instrument were working properly, the analyser should be able to null over a range of a few percent, centered around the frequency specified by the knobs. I take "knife edge sensitivity" to mean null tracking is very restricted? And higher than knob frequency? These symptoms lead me to guess that the TP2 is "stuck" near ~ 8V, where the cell resistance is near minimum. Hence, null frequency is higher than nominal and tracking range is very restricted. In this scenario, the problem probably lies somewhere in the circuitry bounded by TP2, TP3, and T4.
With knobs set to 1.0 x 1kHz, perhaps you can tabulate voltages versus frequency: TP2, TP1, null depth, and anything other behavior that seems noteworthy.
Best.
That sounds "better" but still weird. Just confirming guess work: are we referring to R23 (not R22) =348 ohm? If you are reconfirming numbers above, would you report voltages TP1, U3C pins 8 and 9? Probably good to inspect with scope since complex waveforms might be reported inappropriately on a mutli-meter.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.