From: "Terrell W. Fritz"
Date: Wed, 29 May 2002 18:49:57 -0600
Subject: Re: [TSSP] Re: Top V Probe Design
Hi Paul,
The commercial ones are like:
http://www.pearsonelectronics.com/Pages/Capacitive_Voltage_Divider.html
Reading between the lines... They seem to like a larger center conductor
probably to limit corona in oil discharge around the center pin. The edges of
the center conductor have to have large enough radiuses so that breakdown in
the oil does not occur. I also note the easy network Pearson uses to match a
50 ohm cable. But a 2500 volt scope probe is safer. I note they float the
output ground for isolation.
I checked Ebay but I only ended up bidding on three more darn current monitors:
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=1733460750
I should learn to control myself :o)) They look a little rough but for 1 cent
on the dollar, can't go wrong. Spray paint does wonders ;-) If anyone want's
any, I may have three more than I need tomorrow ;-))
A coaxial capacitor has a classic formula of:
C = 2 pi e / (ln (b/a)) F/m
e0 = 8.8542 E-12 F/m
er = 2.25
If I had 1/8 inch spacing between the shield and pickup cylinder I get about
1.76n F/m or about 450pF for the setup shown.
If the pin were 1/4 inch, I would get about 8pF. So only a 1:56 voltage
divider. However, there is a trick I "think" they play. They simply add a
fixed cap (in those little boxes). Like 7.5nF mica. The boxes are also where
the spark gaps and matching resistor goes (50 ohms to match the cable and a 50
ohm termination). Then we get a 1000:1 ratio. Simple calculation demonstrates
that they can't get like a 10,000:1 ratio just from the geometry. Pad caps
must be used.
BTW - For the voltage divider folks. Check out US patent 5107201
Cheers,
Terry
At 09:23 PM 5/29/2002 +0100, you wrote:
>Terry wrote:
>
>> http://hot-streamer.com/temp/HVProbe-1.gif
>
>At first glance I really like the look of that - it carries
>all the advantages of the shielded tube.
>
>> One may want to shorten the windings so that the affect of the
>> internal can is uniform along the secondary to make modeling
>> easier. Just a uniform capacitance to ground along the coil.
>
>Don't worry. We can model any old shape of innards, so long as
>it has cylindrical symmetry. As for calibration, just put a
>step signal into it and capture a trace of the output signal.
>A bit of signal processing will deliver the phase and amplitude
>response over the range that we need (which isn't very much to
>begin with, I think - say 20Mhz would do very nicely).
>
>> if one is willing to calibrate it with math rather than
>> precision hardware, it is easy.
>
>Naturally we would do both as a cross check. We'd probably have
>to guess the permittivity of the oil and fiddle it to match
>the observed probe response.
>
>Sounds like there are lots of ideas and techniques out there
>used by the professionals - but our main requirement goes beyond
>these in that we want to achieve minimal coupling to both the
>secondary and the space charge around the topload. IMO Terry's
>plan seems to fit the bill and we need to put some numbers to it.
>
>The only number that really needs a careful decision is the diameter
>of the grounded shield. Too wide and the secondary has too much
>extra C loading. Too narrow and the field around the input conductor
>may break down the oil dielectric. Beyond that, neither the
>frequency response nor the division ratio, nor the extra
>C loading added to Ctop, are particularly critical. To a large
>extent we can work with whatever it comes out to be.
>
>Whatever is made, we just need to be absolutely sure that it won't
>form any sort of breakdown within the oil, so that when we see Vtop
>departing from the idealised behaviour, we can confidently attribute
>it to the streamer loading. In other words the probe must be
>linear at TC operating voltages. That's probably our most important
>requirement.
>--
>Paul Nicholson,
>--
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.