So, if you can't compile them, how do they run? (noob question, I am sure)
I think you mean dynamic range is 85dB, bandwidth is 0 to 100kHz...
Yeah, with the fundamental nulled, and IF you can get the residuals UP in level above the noise floor of the instrument, sure you can see "deeper"...
...I am thinking the limitation on the 8903A is both the noise floor inherent in the electronics (mostly power supply + active devices) and the self distortion of the internal oscillator and the self distortion on the measurement side (especially the opamps). But I have yet to really read the schematic, service manual, and circuit description, so I am shooting from the hip...
I do have an ultra low distortion oscillator that could be used to test the results, and another distortion analyzer that could be used to test the oscillator, etc...
(so why am I dealing with the HP? because with PMillet's software I can export and do graphs and automate...)
_-_-bear
I think you mean dynamic range is 85dB, bandwidth is 0 to 100kHz...
Yeah, with the fundamental nulled, and IF you can get the residuals UP in level above the noise floor of the instrument, sure you can see "deeper"...
...I am thinking the limitation on the 8903A is both the noise floor inherent in the electronics (mostly power supply + active devices) and the self distortion of the internal oscillator and the self distortion on the measurement side (especially the opamps). But I have yet to really read the schematic, service manual, and circuit description, so I am shooting from the hip...
I do have an ultra low distortion oscillator that could be used to test the results, and another distortion analyzer that could be used to test the oscillator, etc...
(so why am I dealing with the HP? because with PMillet's software I can export and do graphs and automate...)
_-_-bear
Last edited:
When compiled, they run with the VEE freeware. I run them in the native VEE.
Always challenged by nomenclature...
I have found Pete's software and the 8903 helpfull for general characterization. Saving the graph outputs as screenshots is quite helpful.
I have had good success getting the fundamentals up above the noise of the analyzer noise floor of about -125db.
I have a Krohn Hite 4402 oscillator that runs at .0005% THD, (verified via twin-t notch /spectrum analyzer combo). The best I've seen when connected to and measured by 8903 is .0019% Thd+N. So, I agree with your assessment of the 8903.
The 8903 has a good RMS meter and also quite helpful in measuring oscillator frequency.
Ken
Eh.... Dude. You have bandwidth and dynamic range mixed up. The dynamic range of the HP3562 is specified to be 80 dB (most units will do 85~90 dB). The frequency range is 0~100 kHz.
The dynamic range of the HP8903A is specified to 80 dB as well. Hence, the lowest THD+N you can measure is 0.01 %. Most units perform better than this as evidenced above.
~Tom
Here is an example of the dynamic range (thanks) output from the HP3562 down to it's apparent noise floor. The input signal was generated by an Krohn Hite 4402 at a level of .998Vrms (as measured by the HP8903A and as close as I could adjust to 1Vrms), nulled to -51db by Dick's Twin _T; the THD+N of the HP8903A reported at .0018% with the 400Hz and 30kHz filters turned on. The spikes at approx. 48kHz and 63kHz are environmental noise, not from the KH4402, but, to my knowledge are a part of the HP8903A THD+N calculation.
Ken
Attachments
You may want to try disengaging the 400 Hz HP filter unless you have a lot of 60 Hz hum causing erroneous results. The 400 Hz filter adds 2 op-amps and many passives into the signal path prior to the notch filter, so any noise and distortion created by those parts becomes part of the measured THD+N. The 30/80 kHz low pass filters are after the notch so they can't cause nearly as much harm.
As you suggest, any noise such as the 48 kHz and 63 kHz spikes will become part of the measured THD+N. By observing the monitor output of the 8903, you can determine how much contribution they make. I use the FFT of my digital scope; your spectrum analyzer is much better.
I have a 8903B and I am very interested in improving the measurement floor. If anyone else is interested, we should start up a new thread and swap ideas. I think that noise is the main culprit; I see a fair bit of noise on the monitor output, but virtually no trace of fundamental or harmonics. Certainly the 5534 op-amps, though great 30 years ago, can be replaced with something modern. And I wonder about the many CMOS analog switches in the signal path, used to adjust amplifier gain, select filters, etc. Relays might do better (or not?).
Regarding the settings, I went back and looked (all the settings are built into my VEE script), the 400Hz high pass was turned off and the 80kHz low pass on.
This morning I ran a 10kHz 1Vrms output from the 8903 through the aforementioned setup and got the following results. I would tend to believe that the 50kHz spike is abnormally high due to environmentals.
Ken
This morning I ran a 10kHz 1Vrms output from the 8903 through the aforementioned setup and got the following results. I would tend to believe that the 50kHz spike is abnormally high due to environmentals.
Ken
Attachments
I was nosing around at the HP/agilent Yahoo group and I saw that someone posted a link to a 8903A operator/service manual. I haven't had a look at it, but there are some circuit differences so this might be useful for owners of that older model.
HP8903A AUDIO ANALYSER OPERATING AND SERVICE MANUAL Service Manual free download, schematics, eeprom, repair info for electronics
HP8903A AUDIO ANALYSER OPERATING AND SERVICE MANUAL Service Manual free download, schematics, eeprom, repair info for electronics
The E version does not have a low distortion oss. in it Pete Millet site has some good info HP 8903 Software
Differences between 8903A and 8903B, straight from Agilent:
8903A, 8903B, 8903E Comparison | Agilent
Ahhh, I see.
The main diff is lower noise floor in the analyzer, and balanced input... plus the onboard filter switches...
...which means different boards.
And, although I have not checked the manuals, I would be somewhat surprised if the E model used the same components as the B, but I suppose it is possible... older ICs get hard to source for new production.
_-_-bear
The main diff is lower noise floor in the analyzer, and balanced input... plus the onboard filter switches...
...which means different boards.
And, although I have not checked the manuals, I would be somewhat surprised if the E model used the same components as the B, but I suppose it is possible... older ICs get hard to source for new production.
_-_-bear
Fwiw, and fyi, and pending any time to *actually do anything* I am acquiring some additional *boards* for the 8903A presently. Assuming I can get those that are doing the right stuff functional, then it may be possible to socket the ICs, look at the noise sources, improve the local regulation (hey there may be a real use for SMT? 
) and try some newer ultra low distortion opamps in the holes...
yikes! another big project... phew.
(yeah, but I could be the 8903A hero?)
_-_-bear
) and try some newer ultra low distortion opamps in the holes...yikes! another big project... phew.
(yeah, but I could be the 8903A hero?)
_-_-bear
I just got mine from fedex this am and although it is an "A", i also got the K18 upgrade option from Tucker as well so it really is a "B" in an "A" chassis. Now I have a new input board, not that there was anything wrong with the original, but the new one upgrades the "A" to the "B" with differential inputs. So I got the "A" in great condition, the upgrade kit, and all three brand new HP bound manuals for 400.00 delivered.
Merry Christmas and Happy New Year to me!!!!
Cheers
Merry Christmas and Happy New Year to me!!!!
Cheers
Could you mention or link some examples? This is one of those things that people almost always talk about in generalities, and I am interested in actually getting something for the workbench -- I honestly have no idea what you guys are talking about.
So would an older distortion analyzer, like the HP 339A, be useful?
And does any or all of this stuff measure IM distortion?
There are some threads on soundcards here... I need a new one myself.
The better new ones no only do 24bit/192kHz. but they do it nicely with low residual noise and meet the max spec. But honestly even an average modern 24/192 card is probably going to be pretty good unless ur getting really into some cutting edge low distortion designs...
So you take ur signal, be it from the analyzer output (it has one) or direct from the DUT, and build a "signal conditioning" box so you don't fry the input of your soundcard (limited voltage input range) and then run a FFT type software to get a graph of the amplitude vs. frequency, that shows your fundamental and all the harmonics.
IM is two tone.
Some analyzers come with two tone testing.
I personally do not know if the 339A or similar is really terribly useful for this. However, with a Cordell Distortion Magnifier box, it well could be pretty darn good. (that is documented either here and/or online)
Crazy stuff...
The better new ones no only do 24bit/192kHz. but they do it nicely with low residual noise and meet the max spec. But honestly even an average modern 24/192 card is probably going to be pretty good unless ur getting really into some cutting edge low distortion designs...
So you take ur signal, be it from the analyzer output (it has one) or direct from the DUT, and build a "signal conditioning" box so you don't fry the input of your soundcard (limited voltage input range) and then run a FFT type software to get a graph of the amplitude vs. frequency, that shows your fundamental and all the harmonics.
IM is two tone.
Some analyzers come with two tone testing.
I personally do not know if the 339A or similar is really terribly useful for this. However, with a Cordell Distortion Magnifier box, it well could be pretty darn good. (that is documented either here and/or online)
Crazy stuff...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.