Tektronix p6015 HV probe trouble

[smar03]

Hi guys,
recently I bought a second-hand Tektronix p6015 HV probe off of ebay for around $120, it said fully working and tested. However during testing it was not measuring signals correctly, so I decided to test it step by step and disconnected the compensation box and connected the BNC directly between the probe output and the oscilloscope. Even though the probe schematic (shown below) shows that there should be only a 100MOhm resistor inside, together with some parasitic capacitance to ground (probe body), when I measured a clear sinewave signal (~50 Vrms) on a 50 ohm resistive dummy load, the probe showed a larger attenuation than expected, together with a small phase shift compared to my standard oscilloscope probes (Rigol DS1054Z probes+scope). Since I don't have a device that can measure on the order of ~100Mohm directly, i measured a 10MOhm resistor and used it in a ~10x voltage divider with the 100MOhm probe resistor. I tried this both at DC and at ~100 kHz and in both cases the measured amplitude is approximately 2.5X lower, indicating that the probe resistor would be 250 MOhm instead of 100MOhm. My question is, am I doing the wrong tests or am I missing something? Because this way it seems that the HV resistor is broken. Cheers

[tggzzz]

However during testing it was not measuring signals correctly

Define your tests in sufficient detail that someone else could replicate them.

It would help readers if you used paragraphs.

[bdunham7]

Let's see the details of your test--photos maybe?  Equipment used?  If you are using a simple 10M resistor as a voltage divider with no other components, no termination, etc, then it would be quite unlikely that you would get the exact same attentuation at DC and 100KHz.  I'd be surprised anyway.  I would think that the parasitic capacitance of the large resistor should be huge compared to the relative capacitance of the 10M resistor--even accounting for a scope probe.  I could be wrong,but....

[David Hess]

The probe head will not work right without the compensation box.  While the compensation box attenuates the signal, it also lowers the impedance into the oscilloscope's 20 picofarad input raising the bandwidth which may explain what you observed.

You should be able to measure the 100 megohm probe head resistance using a standard digital voltmeter.  With a 10 megohm input resistance, it will make a x11 divider so 10 volts will measure 0.909 volts.

[smar03]

Apologies for the incomplete description of my tests, I will pay attention to that. Attached there are some images of the probe itself
and the compensation circuit.

I did a voltage divider test with the probe head and a 20M resistor, giving a 1:5 ratio in voltage drop, indicating the probe head is
indeed 100M.

I have also measured all the components in the compensation box with an LCR meter and the values overlap reasonably with the
specifications. I removed the compensation box-to-scope BNC connector because it was loose and could not be tightened
properly, which was affecting the reading if moved even minimally.

When I now use the probe, including the compensation box and the scope set at 1000x, to measure a 100 khz signal on the 50Ohm
dummy load, and compare it to the reading obtained using a standard Rigol probe at 10x, the result is still a bit off (image attached).
There seems to be a DC offset, as well as a phase shift, and the wave is not a clean sine.

When I try to adjust the compensation box, the effects are minimal. R5 does not change the amplitude at all, and tuning the other
components as described in the manual changes the waveshape in the right way but this effect is limited. The DC offset is not affected.

What other tests could I try?

[David Hess]

Two different probes will not have the same phase shift or frequency response.  But that signal distortion should not be happening.

I can only suggest studying the probe calibration instructions in the service manual.

[bdunham7]

Apologies for the incomplete description of my tests, I will pay attention to that. Attached there are some images of the probe itself
and the compensation circuit.

I did a voltage divider test with the probe head and a 20M resistor, giving a 1:5 ratio in voltage drop, indicating the probe head is
indeed 100M.

I have also measured all the components in the compensation box with an LCR meter and the values overlap reasonably with the
specifications. I removed the compensation box-to-scope BNC connector because it was loose and could not be tightened
properly, which was affecting the reading if moved even minimally.

When I now use the probe, including the compensation box and the scope set at 1000x, to measure a 100 khz signal on the 50Ohm
dummy load, and compare it to the reading obtained using a standard Rigol probe at 10x, the result is still a bit off (image attached).
There seems to be a DC offset, as well as a phase shift, and the wave is not a clean sine.

When I try to adjust the compensation box, the effects are minimal. R5 does not change the amplitude at all, and tuning the other
components as described in the manual changes the waveshape in the right way but this effect is limited. The DC offset is not affected.

What other tests could I try?

I was going to ask if you had figured in your DMM input impedance, but since you have concluded that the 100M resistor is accurate, I'll assume you have all that figured out.

I am unable to explain that waveform distortion.  Perhaps you could do the same comparison with a good, sharp square wave?  That might be easier for me (and probably others) to understand.

This might sound odd, but it might be a good idea to clean (bathe) the compensation box.  Cleaning can range from an IPA bath and compressed air dry all the way up to a cycle in a dishwasher (the Tektronix way, believe it or not...).  I'm not sure about those pots in a dishwasher though, so probably better to start with something safe like the IPA. 

[smar03]

Thanks for the suggestions. I have managed to get it working almost perfectly.

I replaced the second BNC connector with a new one and I also replaced the 150 ohm (R6) and 100K (R4) resistors. Then i put
everything back in the enclosure.

The phase shift is gone and the DC offset also, for the most part at least (image attached). I could adjust the amplitude with C3, but its range
was just not enough to get it right so I measured its minimum and it was at ~6 pF so then i decided to just disconnect it.

The only thing left to fix is that for some reason, the negative peak has a slightly lower value than the positive
peak, even when the traces of the two probes are almost identical. This suggests a small residual DC offset. Any thoughts?