30MHz (now 40MHz) Probe Testing Update.
From: terryf-at-verinet-dot-com [SMTP:terryf-at-verinet-dot-com]
Sent: Tuesday, April 07, 1998 10:37 PM
Subject: 30MHz (now 40MHz) Probe Testing Update.
Tonight I used wire cutters, soldering iron, and pliers to "enhance"
my signal generator to test the frequency characteristics of my probe. I
found a small ringing in the square wave response, which I tuned out easily.
Retesting showed this did not have a significant affect on what I have seen
so far. It did clean the signals a little. Still get heavy ~50MHz rings
and all. The probe now gives a -3dB frequency of 40MHz with a clean square
wave response. :-)
I did more EMI testing looking for ground noise, radiated noise,
conducted noise........ I found nothing like that was going on. The
fiber-optic voltage transducer picks up the spike noise badly but I have not
been using it recently. I have some die-cast enclosures on order which I
hope will keep the spike noise out of the voltage divider section.
I have found that the frequency of the ring can vary from 30Mhz to
60Mhz depending on many factors, none of which I could pinpoint. Power
level, gap width, how far the variac was turned up, etc. all had an affect.
I cannot really put much faith in the following because the levels
are so far out of the current shunt's design parameters but here are my best
The initial current spike with a 2000 volt gap (17nF cap, 120uH coil) is
about 4500 amps or more. No wonder the voltage drops like a rock!
The follow up spikes range from 500 to 1500 amps.
The after rings in these burst are in the hundreds of amps.
Note: using the theoretical Xc and Xl calculations the peak current in this
LC network should be 23 amps max!! I really do believe all the current in
the primary is being conducted in these current bursts and the rest of the
time the current in the gap is near zero.
I think the voltage and current spikes in these bursts overlap indicating
power disipation. I will have to fix the voltage transducer to check and
This I am pretty sure of:
All the burst rings are either in the positive or negative direction
depending on if the voltage is at a positive or negative peak. Cleaning the
square wave response of the probe shows this much better now.
The current bursts are almost perfectly centered at the peaks of the voltage
I can find no time were they are not present but the level of the burst can
The speed of light is about 300,000 km/sec. :-)) It is easy to see the
~35nS delay in the 10m fiber compared to the antenna 1m away. 2nS is
probably the fiber optics delay. They really are that fast!! They make
them 10 times faster still. To bad they are so expensive.
I am now convinced all this is in fact real and that what I am seeing is
essentially correct, give or take some signal level accuracy.
At these levels the transmit LEDs are seeing about 100mA in the positive
direction. They shut down (reverse bias) in the negative direction and the
scope shows this clearly.
Of course, It must be remembered that the shunt may not be a stable 0.01
ohms at this frequency and power level - despite my wonderful design :-)
I.e. it may reporting more current than is really there. Any ideas on a
high frequency high power shunt design is very welcome! It is probably
somewhat in the ballpark but no way to be sure.
I also hooked up my cheap 35 MHz transistorized scope and confirmed
that the same observation could be made given the limits of this scope.
Definitely use a terminated antenna for this test. I had forgotten how
worthless this monster was. The triggering was definitely not up to this
but I still got it to work well enough. I think I'll feed it to my coil
some day :-) The Tek 210 is a thousand times better!
I tried replacing my wide copper strap with a thin loose wire to see
what a bad wire would do to the primary system. The ~50Mhz signal persisted
but the level was reduced and the burst rang for a much longer time (like 5
times longer). The whole signal became very sick looking and was not nearly
as stable. It really looked much worse! I suspect this high frequency
phenomena is the reason big thick straps give better performance in the
primary. This would also explain the high resistance the primary circuit
always seems to have despite our nice caps, heavy copper inductors and thick
straps. It would also explain why caps that should easily survive in this
service fail so regularly.
I tried to turn the voltage way up at one point but stopped in fear
I would kill the transmit LED from the giant current. I really need a
I have 42 scope pictures that were captured during the first test.
I have posted them all to:
They are just big JPG files called 1-8.jpg, 9-16.jpg, .... I have
explanations of each but I will have to type them out which will probably
take a few days. some are using the 2.5MHz port so all the effects are not
Many thanks to all the people who have give me advice! It is
wonderful to have so many very knowledgeable people to draw on. Please
send along futher suggestions for tests or experiments you think I should