This week, I purchased an HP-8903B from a local gentleman. I went to his home to view it and he mentioned that he serviced the unit by replacing the few capacitors in it and it was working "as it should" and within specifications.
So far, I have two problems with the HP-8903B.
PROBLEM 1 - Frequency Inaccuracy
Is it normal for the unit to not "hold" on a frequency and to have, what I consider, an abnormally high THD when simply using a loopback cable?
For example, if I turn off all filters and then just turn on a 1 kHz sine wave, the display will show a frequency from ~998 to ~1001.5. No matter what frequency I put in to try to compensate for the +/- 3-ish Hz, I can rarely get it to stick exactly on 1 kHz. Also, the longer I leave the HP powered on, the higher the frequency drifts.
The oscillator in the HP8903B is not synthesised so you can't expect perfect frequency accuracy. However, it shouldn't drift appreciably either as the the frequency is adjusted by the microcontroller using a "count and tune" method. It might be a good idea to follow through frequency monitoring and switching path from the oscillator to the microcontroller to ensure that the signal hasn't been lost along the way.
An unstable oscillator frequency could be contributing to the high distortion readings as the notch filter in the distortion measurement circuit will not be able follow the frequency variations and will be unable to notch out all the fundamental frequency. Check the oscillator O/P with a scope to see if there is any jitter in either the frequency or amplitude of the oscillator O/P.
PROBLEM 2 - THD %
When I use a loopback cable (regardless of the cable I try), I'm getting a "high" THD... AND the THD seems to also be dramatically affected by the amplitude.
Example #1: All filters off. Frequency set to 1 kHz. Amplitude set to 150 mA. Distortion reads ~ 0.1240. If I change the amplitude to 250 mA, the distortion drops to 0.77. If I change the amplitude to 1V, the distortion drops all the way down to 0.022.
Example #2: If I turn on the 30 kHz filter (1000 kHz tone @ 150 mA), the distortion goes from ~0.1240 down about half to 0.065. Choosing the 80 kHz filter results in a 0.09 distortion reading.
Am I expecting too much out of this machine or is something wrong with it? The main reason I purchased this unit was to measure distortion so if that is going to be unreliable, it is not going to be the machine for me.
It should definitely do much better than what you are seeing. For a direct loop back connection you should be seeing distortion figures of 0.005% or better.
A useful diagnostic is to connect a scope to the monitor O/P on the rear panel to view the residual signal after the fundamental notch filter. The presence of any discernible fundamental frequency components indicates the notch circuit is not tuning correctly resulting in an inaccurate distortion reading.
The fact that introducing the 30kHz and 80kHz LP filters significantly reduces the distortion readings indicates that there is significant HF noise in the system. This could be due to a noisy front end on the HP8903B or interference from external devices. Turn off all computers, cell phones, WiFi, routers and LED lights in the vicinity to see if that makes a difference. Again, look at the HP8903B monitor O/P to determine the nature of the interference.
You should also try the 400Hz HP filter to see if there is any hum contaminating the measurements. This will also be visible on the monitor O/P.