Standards Lab Measurement Dilemma #2

******** See EDIT at bottom *************

I don't think I understand anything here because it looks to me as though the basic spec of the Datron 4700 is no better than 160 ppm at 100 Hz

This is the only data sheet I could find:-

Compared to the Fluke 4 ppm + 400 µV at 1100V(I assume this spec. is across all frequencies).

How many frequencies have you tried (is it worse at 120 Hz than 100 Hz)?

*****EDIT***** I just looked at the Fluke spec. again and the extract I have copied is for DC. The AC spec. doesn't look dramatically better than the Datron. *******

Hi Randall,

It has nothing to do with Datron specs, the Datron is only a source driving the standard TVC and the DUT TVC with parallel inputs. Any error in the Datron AC output affects the standard's output and is mathematically corrected. The DC output could affect it, but the calibration of that is very close.

-S

Some test results are out of tolerance because of impedance mismatches and the 5700 instrument. We had to use a 5200 or 5500 (it's been 5 years ago) on some AC measurements because of it.

AC + unmatched cables + frequency can equal mismatches.

Some new information:

I tried calibrating an old Holt 11 from the primary standard instead of the one I was using (which was modified). The results were good with the old one even when using the Datron 4700! This adds credibility to the theory that an RF "spur" is responsible. The old unit only has wirewound resistors in series with the thermal element. The modified unit has different wirewound resistors and physical capacitors (silver mica) to adjust the frequency response, as well as some "stray" air capacitance. But I have checked the output up to 13.2 GHz and found nothing. it's very hard to believe it has a spur above that frequency.

I had tried a makeshift low pass filter on the 4700, but it had little effect. It was only a series inductor in the output high lead working with the lead capacitance, and may not have been effective. It isn't easy to build one that rejects signals of several GHz. I think you need feed-through capacitors in copper shields, etc. Anyone want to propose a good LPF design? The load is 60K ohms in parallel with 50pF. It must be able to handle 1000V RMS. It should pass 100kHz with 1dB loss or less. The total load capacitance can't be over about 70pF, or the 4700 will trip out.

It does all look like EMC to me. My usual method for diagnosing the situation would be to use a tunable resonant loop antenna coupled to a spectrum analyser. But you could try draping singly-grounded aluminium foil around the place...

But if you need to consult an EMC specialist I would caution you only to "pay by results". The reason for my caution is the prevalence of "recipe specialists" - and I think you have already looked at most of the recipes they use. (I imagine that the single biggest problem is still eliminating unintended connections that mean your signals and grounds only appear to follow the same routes).

"It does all look like EMC to me. My usual method for diagnosing the situation would be to use a tunable resonant loop antenna coupled to a spectrum analyser."

I had the room checked with a wideband antenna, and later used a loop (non-resonant) that I had build myself. There is a radio station 1/2 mile from here broadcasting on 1520kHz at 5000 watts (I think). Spectrum analyzers have found their signal to be lower than could account for the problem - both in the air and in the power line. I have been wondering about ground waves that could be in the walls, and flowing through buss cables, etc. that I can't measure. I may arrange to come in at night when they are off the air and do some measuring.

Dedication beyond the call of duty, I'd say. Or perhaps high-grade technical curiosity. (Coming from me, either of those is tantamount to a compliment).