A Cursory Examination of the Glow Discharge Panel   -   15FEB00

Jean-Louis Naudin (JLN) recently constructed and tested a glow discharge panel (GDP) with surprising results, including a large over-unity power balance.  His efforts can be reviewed here.  We decided to construct our own panel to confirm his measurements.

This photo shows our setup.  The glow discharge panel (left) was made by wrapping Radio Shack speaker wire (zip cord) around a small slab of fiberglass reinforced epoxy laminate (G10).  The capacitance of our panel measures 140 pF.

The panel is driven by a spark coil whose primary is excited by a MOSFET switching circuit very similar to that used by JLN, except that we used an IRFZ40 MOSFET.  We also elected to drive the gate of our MOSFET with a square wave from a signal generator (not shown) to facilitate frequency adjustment.  We experimented with a "freewheeling" diode across the coil and, although it made the primary current waveform very nice looking and was surely good for the MOSFET, it also nearly killed the GDP effect.  We conducted the tests described below without the freewheeling diode.

Our spark coil was obtained at a local Pep Boys auto supply.  DC power for the device is supplied by the Power Designs regulated DC supply visible in the background.

We constructed a 2 watt non-inductive precision 10 ohm resistor out of 4 1/2 watt precision 10 ohm resistors.  We placed this resistor on the HV output side of the coil in series with the GDP and connected the 10X P6113B scope probe across it as shown.  We also connected a TEK P6015 HV scope probe across the GDP itself.

Our probe connections are the same as those used by JLN (see copy of his schematic to the right).  With the probes connected like this, a positive power reading indicates power flowing to the panel from the coil.

Here is another view of the probe connections.  The tip of the large HV probe is connected to the high side of the coil output with a short red clip lead.  The green clip leads run out to the GDP.

Since the current viewing resistor is 10 ohms, a current of 1 ampere will produce a voltage of 10 volts.  However, this voltage is being observed with a 10X probe which attenuates the signal by a factor of 10.  Therefore a current of 1 amp will produce a signal at the scope of 1 volt.  

We connected this probe to Ch2 of our THS720A scope and configured the input as a current input with sensitivity of 1000mV/A.

The HV probe has a signal attenuation of 1000X so we connected it to Ch1 of the scope and configured the input as a voltage input with a 1000X probe.

The diagram below shows the complete arrangement of our circuit.  Note the connections to the spark coil.  Because we did not want the ground side of the HV output to be jerked up and down by the MOSFET, we decided to connect the primary up "backwards".  The primary side that we have connected to the plus voltage is marked (-) on the coil.  That is the terminal that is internally connected to the ground side of the HV output (as shown).  The primary side that we have connected to the drain of the MOSFET is marked (+) on the coil. The signal generator drive the gate directly with a square wave that goes from ground to +5 volts as shown.  

With the DC power supply delivering 20 volts at 1.8 amps to the drive circuit, these waveforms were obtained with the probe configuration described above.  The appearance of the voltage and current waveforms, including magnitudes, are very similar to those obtained by JLN, with the notable exception that our measured current is 10 times smaller than his.  The mean power being delivered to our panel is about 21 watts, a reasonable value considering the 36 watt input power.  Also our resonant frequency was only about 3.75 kHz.

After 10 or 15 minutes of GDP operation, the spark coil is perceptibly warm and so is the GDP itself.  The MOSFET, which is mounted on a heat sink, is just a little warm.

We then attempted to observe the asymmetry in GDP current that JLN has observed.  We connected a Fluke 87 DVM across the 10 ohm resistor mentioned above.

As you can see in this photo, the F87 is reading 326.8 mV, which indicates a current that is about 70% higher than the scope probe showed!

What could possibly be causing such an error?

We then noticed, to our amazement, that simply turning the meter on its side was sufficient to triple the reading!

Further scanning of the meter around the circuit showed that any reading from about 200 mV up to nearly 2.0 volts could be obtained.

Obviously, there is a severe interference operating here.  A very similar effect was observed using a completely different DVM from Radio Shack.

We also tried moving the scope around and observed small (~10%) changes in its readings as well.  We are inclined to accept the scope's readings since they look "normal" but there are clearly some interesting measurement challenges in this experiment.

Finally we tried to trick our digital camera into taking a picture of the glow from the GDP.  Such a camera is not very suitable for special photography but, as you can see in this photo, we did manage to capture some indication of the glow by setting the camera on high sensitivity, using a tripod, and turning off the automatic flash.

At the suggestion of Jones Beene, we looked for x-rays coming from the GPD.  We tried two types of radiation detectors, (1) a G-M tube instrument called the RADALERT Inspector made by International Medcom. and (2) a 2" NaI scintillator system pieced together from stuff made ages ago by Texas Nuclear.

The G-M tube unit would produce a flood of counts if held close enough to the panel but, after a little experimentation, it soon became obvious that the signal was electrical interference instead of x-rays.  For one thing, the signal did not obey the inverse square law at all.  The signal was essentially absent until the unit was about 5 cm away from the panel and then it burst into activity.  Moving the unit out to 7 cm stopped the signal entirely.  Further, when the G-M unit was near enough to obtain a signal, the signal was greatly affected by the presence/absence of my hand touching the unit.   Misleadingly, the signal was affected by small metal shields placed between the panel and the G-M tube entrance but the other observations prove that it is not a real x-ray signal.

The NaI scintillator system was apparently immune to the electrical noise problem.  With the probe located about 5 cm away from the panel, we obtained identical counts (i.e. background) with the panel operating or not.

See subsequent investigations here.