More Work with the Glow Discharge Panel   16FEB00

As you clearly see in this photo, I have confirmed the apparent asymmetrical current that JLN is observing.

For this experiment, I connected identical 10 ohm precision resistors in each leg leading to the GDP.  I connected identical Fluke 87 meters across each resistor in a symmetrical fashion.  BTW, the Fluke 87 specifications indicate that it has 1% accuracy on AC volts over the frequency range 45-5000 Hz on all voltage ranges.

The panel drive circuit was operating as before at 20 VDC and 1.8 amps.  Under these conditions the scope measurements indicate an RMS current to the panel of about 15-20 mA.   With a 10 ohm current viewing resistor, the voltage drop should therefore be about 150-200 mV.

As you can see, the meter in the ground leg (lower) reads .435 volts while the meter in the HV leg reads 2.905 volts!

The cause for this asymmetry is presently unknown but there are strong indications that it is a measurement error.  For example, when the meter lead length was increased to 1 meter, the lower meter produced a reading that agrees closely with the scope value....about .150 volts...but only when the other meter was disconnected!  Simply connecting the other meter, even with 1 meter leads, caused the lower meter's reading to nearly double.  While connected with 1 meter leads, the other meter still read above 2 volts.  

The automotive ignition coil has one side of the secondary internally connected to one side of its primary.  Therefore there is no way to see if these peculiar effects are due to the grounding differences.

In view of this problem, I tried driving the GDP with a neon sign transformer, which has a grounded center-tapped secondary, which produces perfectly symmetrical HV outputs at the ends of the secondary.

I drove the primary of the neon sign xfrm with a Variac at 60 Hz.  Presumably because of the low frequency, I had to raise the voltage across the panel to about 11,000 volts RMS before it became completely covered with glow discharge.  Even then the discharge did not seem as bright at it is at the 4 kHz frequency used earlier.   Amazingly the insulation on the speaker wire withstood this voltage!

As you can see in the photo on the right, the two meters still do not read the same...but they are much closer.  The left meter reads 231.5 mV and the right one says 157.2 mV.

It should be noted that these two Fluke 87's are of quite different vintage.  One of them was purchased about 7 years ago when the F87 was a new product...the other is only 1 year old. There is significant evidence of redesign in the new one.

I tried switching the two meters.  The readings stayed with the meters, indicating a fundamental difference between the two meters, even at 60 Hz.

In the process of switching the two meters, I noticed that proximity of my hand affected the meter reading significantly.  In this photo you can see that, by simply holding my hand near the right meter, I can make the two readings agree (left: 219.3 mV, right: 217.3 mV).

As a result of all these tests I believe that I will be forevermore suspicious of AC voltage measurements made in the vicinity of AC high voltage apparatus.

Radiation Measurements:

After learning that the range of 10 keV electrons in air is only 2 mm, I have discarded the idea that the G-M tube measurements reported earlier could be due to electrons striking the detector.  I never dared to hold it any closer than about 50 mm from the panel.

Regarding the sudden cessation of the signal as I moved the detector from 5 cm to 7 cm away from the panel:  I'm sure the electrical noise didn't suddenly disappear with that minor distance change but it is not unreasonable to assume that the noise level did fall below the threshold in the G-M tube electronics.  That would explain the sudden cessation.