HP 8903B Audio Analyzer Source Noise Level

[rastro]

Hello All;

I was checking out an HP 8903B and found what I believe is a problem but would like a comparison from other units.
When I monitor the source output there is quite a bit of noise on the source output on the low range 5.0 mV amplitude @ 20Hz.
You can see in the picture.
Could someone try this setup to see what the signal looks like on their instrument?
Settings On 8903B
- key in 41.0 SPCL to initialize the instrument.
- Set the INPUT and OUTPUT switches to ground.
- Amplitude: 5.0 mV
- Frequency: 20Hz
- Reading from HI output (source)

1. All the verification checks in the manual pass which verifies 90% of the instrument.
2. I also ran many of the Performance verification tests which also look good - in particular the distortion and noise performance test. 
3. Off course the noise on the signal may fall outside of the instruments frequency measurement range which may explain passing distortion tests???
4. Changing frequency and level doesn't seem to affect noise artifacts.
5. Hooked scope up to A6 board at the Hi/Lo differential output and the noise was really not there!  This is really close to the source output in signal chain - very few components before the output BNC connector where the noise is present??
6. Ran a signal from a signal generator at the same level and did not get noise; this helps eliminate test setup as cause of problem.

[macboy]

put a short or a BNC terminator on the output low terminal. Perhaps the ground switch doesn't work, leaving it floating.

[wn1fju]

I have an HP 8903B hooked up to a Rigol DS1104Z scope, similar to your setup.  My trace is actually slightly worse than yours!

However, you can definitely clean up the waveform on the scope end with a combination of bandwidth limiting, anti-aliasing, or noise rejection.

I have noticed with many HP pieces, for example function generators, there is often very high frequency junk lurking around.  With a low
bandwidth scope, you might never see it.  Hook it up to a modern scope (say 500 MHz), turn down the source amplitude to a few mV, and
good luck even triggering a stable trace due to the high frequency noise.  Vendors seem to only care about the frequency band that their unit is
spec-ed for.  For instance, I have a 40 MHz Wavetek generator that puts out a bunch of garbage at 600 MHz.

All this plus the inevitable ground loop you probably have between the distortion analyzer and scope.

By the way, the 8903B is a great unit! 

[danmcb]

so you are outputting 20Hz at 5mV, and you see some HF crap on the waveform?

run an FFT on it. If it is out of band, say 30-40kHz or higher, forget about it. Not that unusual for (digitally generated) audio gear to have stuff like this.

Nice unit, but an AP is the shit for audio testing!

[rastro]

Thanks for all the feedback.

Looks like probably the noise at this level/frequency is a non issue - especially if other units have similar artifacts. 

I probably would not have noticed this issue except that decided to verify the 20Hz frequency with my HP 5351B frequency counter and it was getting crazy readings.  However it could read comparable setting from my SRS DS345 signal generator.

1. I also verified that the grounding switch was OK before posting.  Also a floating ground would probably caused some of the other Verification/Performance tests to fail.
2. As far as ground loops it's interesting that my SRS DS354 does not show these artifacts on similar settings with all else being the same.
3. As I understand the source is analog not DDS.  It is only set digitally by switching in various resistance and capacitor values.  If it where a DDS signal then I would expect some DAC noise.
4. I looked at the signal on an HP 8566B most of the noise appears above 350KHz.  I'll post a picture showing the source set at 50mV/100KHz the artifact are basically the same as at 5mV/20Hz but it is easier to see the 100KHz source signal at 1MHz bandwidth on the SA.

danmcb - It seems like AP is the gold standard for audio analyzers.  Would you be able to post a similar source signal from your instrument for a reference?

-rastro

[danmcb]

sure:

with the DSO


with the CRO (slow rate makes it hard to take a pic)


The larger spikes there are mains born stuff getting in through the leads.

[danmcb]

Mine is a Portable One Plus - these use an digitally controlled analogue oscillator, with a lot of switched attenuator stuff, so the output stays very clean. But what really makes this a great piece of test gear is the measurement options - things like dB relative to whatever, W into however many ohms. And also the fact that the outputs can be balanced or unbal with a variety of source impedances. You can tell that the designers really thought about the needs of people doing audio test, rather than "let's make another bit of lab gear"./

The only thing is that there is PC output with this one - but there is a parallel printer output. I have a home brew board which captures the output and writes it to USB stick.  I can then use a piece of commercial software to translate the printer language to JPG. One future project is to make that into a nice PCB and see if I can sell a few ... one of these days ...