I have an EMI problem with a high speed system. There is a pair of wires which are taken from a high speed driver to a high impedance electrostatic plate which resides in a vacumm. The plate is enclosed in a metal enclosure so does not emit, but the wires do. They are an untwisted pair about 4 inches of which is exposed, between 2 metal boxes. The signal on the wires is about 500 volts with a 1 nanosecond rise time and about 48 khz repetition rate. The rise time may not be compromised.
I have tried coax type shielding, but this slows down the rise time. I was thinking the next step is a rigid conduit about 2 inch diameter, which may have a higher impedance/lower capacitance than a coax. Also, twisted pair is worth trying, but again I cannot afford to slow down the signal.
That sure looks like a trip to the next big hardware store. If you are lucky they might even have 2" copper pipe. See if 1" would do the trick, it's much easier to obtain. Sometimes they carry decorative brass tubing of almost 2" but that stuff is expensive (it looks cool though).
You could put the wires in a pipe, like Joerg suggests. Electrical conduit, with conduit fittings on the boxes, would give pretty good shielding too. Use some plastic disks to make separators/supports for the wires if needed, to keep them spaced nicely relative to each other and to the pipe. If one wire is ground, you can probably delete it and use the conduit for the return, forming essentially a hi-Z coax... that might actually improve risetime.
What sort of pulser makes the 1 ns, 500 volt edge? How wide is the pulse? What's the physics?
A copper pipe should be good. Choose a diameter based on the separation of your wire pair, put the pipe in a press and squash it into an elliptical cross section and locate the wires at the foci forming a dpi-axial cable.
If it's symmetrical w/o ground another option would be to use two pipes. When sticking with a standard diameter there is usually a nice selection of plastics to be found in the PVC department where the irrigation pipes are. Like those white end plugs. They are about 3/4" long which prevents them from tipping inside the metal pipe. The flange can be ground down so the whole thing slides into the pipe. Drill a hole through the middle, or two if both wires go through a single pipe.
The plumbing department is also called Plomeria. At least in Caleefoaneeyah :-)
But one has to be careful with corona and stuff like that. PVC can produce hazardous fumes when it gets toasty.
Yes, that's an interesting question. Off the top of my head I'd only know how to do that with tubes. But I bet Win would have a nice FET solution and Jim would probably do it in a chip.
Avalanche transistor, maybe. It's borderline for a fet. Fet+shockline would be elegant. I've done 2.4 KV into a small capacitive load, 3 ns FWHM, 500 KHz with a drift step-recovery diode and water cooling.
The op's combination of voltage, risetime, and rep-rate is interesting.
John Larkin wrote in news: snipped-for-privacy@4ax.com:
I will try some sort of pipe I think, the termination is the tricky bit. The idea about using the ground wire as the shield might be worth looking at.
Sorry, the physics behind the high speed circuit is commercial-in- confidence and it is not a trivial design, though it is simple with our implementation. If you want to know how to do these sorts of high speed/high voltage pulsers with classical techniques, look up using BJT transistors in avalanche mode. The pulse width ain't important, ie 100 ns is OK, it just needs a fast edge. I see harmonics up to about 250 MHz.
You mean connecting the end? Just cut out a section large enough to reach in. Cut out a section a bit larger from another piece of pipe, "unbend" it a wee bit, slide it over and affix it with two hose clamps (preferably the stainless steel grade). At frequencies above 100MHz line of sight is what counts most in terms of EMI.
Did you have to use a stripline transformer or something like that to get to 500V? I haven't seen good avalanching on any BJT past 100V or so. Then again most of my stuff has stiff BOM cost limits so maybe I missed a BJT rocket.
The Zetex avalanche transistors are good for about 400 volts, over 20 amps peak, 8 kilowatts peak from a SOT-23! It's common to stack avalanche transistors up into the kilovolts. But if you're driving a capacitive load, all you need is a fast 250 volt pulse into a transmission line... it'll double at the end.
PSPL will put an avalanche transistor in a box and sell it to you for $10K or so.
This sounds a bit like the electron-beam blanking elecrodes I ued to have to drive at these sorts of speeds for Cambridge Instruments electron microscopes.
The real trick to was to put the electrodes in the right place in the beam so you didn't need to drive them with more than a few volts. With the Cambridge Instruments EBMF 10.5 electron beam microfabricator, this was impossible - short of completely re-engineering the column - so we had to generate +/-60V across a couple of bent tin blanking plates in a place that we could more or less get at.
The problem we had is that in order to sustain the high edge speeds, we had to make the blanking plates part of a properly terminated transmission line - in principle if you put the driver within a couple of inches of the blanking electrodes you can treat it as a lumped RLC circuit, but this never turned out to be really practical.
Regular transmission lines have impedances of the order of 50R. Purpuse built coaxial structures can have impedances up to about 200R - check out
formatting link
for a tool to calculate the impedance of a transmission line of arbitrary cross-section.
formatting link
gives simple formulas for simple cases - note that the impedance is proportional to the log of a ratio of dimensions.
Even if you can get to 200R, 500V is going to dissipate 1250W in any terminating network.
A step-up trasnmission line transformer might be offer an alternative.- see see R.E.Matick Proceeding of the IEEE volume 56 pages 47-62 (1968) .
For non-integer ratios, see the following references, snipped from a posting here on Thurs, Dec 7 2000 4:47 pm
"Donald A. McClure, "Broadband transmission line transformer family matches a wide range of impedances", RF Design, February 1994, pp62-66 and May 1995, pp40-49.
Daniel Myer, "Equal delay networks match impedances over wide bandwidths", Microwaves & RF, April 1990, pp179-188.
Both articles treat a generalization of Guanella transformers, allowing ratios which are not limited to simple integers, but rather to the quotient of two integers.
Based on the first reference, I built a transformer with a voltage ratio of 1.4 and a -3dB bandwidth of 65kHz to 380MHz. Still, this wasn't good enough to quench my insatiable appetite for bandwidth :-).
Joerg wrote in news:eOLNf.17483$ snipped-for-privacy@newssvr27.news.prodigy.net:
Pretty close, a very low leakage transformer is used. The interesting thing with avalanche transistors at greater than 500 volt apps is that they only last a few months and wear out. When they were used, they were regarded as consumables. I was not very involved with the design at this stage but the transistors were a TO220 pack, and had a reputation as being good for avalanche. Our Russian engineer was very familiar through his work on particle accelerators.
You should have no problem with coax if your driver can drive the impedance of the cable. It is not capacitive or inductive, but resistive (assuming it is lossless).
I have regularly sent sub ns pulses down coax. (i.e. risetime around
50ps, FWHM around 300ps). I used a microwave step recovery diode to generate the pulses. That would not generate 500V though!! I did some of this during my PhD too.
I used a driver that had an output Z or 50 Ohms, a load of 50 Ohms and
50 Ohm coax. Of course, the output voltage drops a factor of 2 from the source, as half the voltage is dumped across the source impedance of the driver.
If you use large coax, it will support higher order modes, so behaves quite differently at high frequencies than low ones. Semi-rigid coax is regularly used to 10's of GHz.
PS
I am the author of
formatting link
which someone else mentioned.
--
Dave K
Minefield Consultant and Solitaire Expert (MCSE).
Please note my email address changes periodically to avoid spam.
It is always of the form: month-year@domain. Hitting reply will work
for a couple of months only. Later set it manually.
In article , John Larkin wrote: [... 1nS rise at 500V ....]
How about a spark gap?
Back the mists of time, someone was doing microwave RF with a spark gap. I can't remember the name right now but a search on the right keywords may pop up something.
Some WWII radars used triggered, or even rotating-spiked-wheel, spark gaps as the magnetron pulsers. The RadLab book "Pulse Generators" has details.
Hertz used spark gaps, as both transmitter and receiver, to prove that Maxwell radiation existed. And Marconi's first radio transmissions used spark gap transmitters.
And of course the first plutonium bomb used krytrons to fire the explosive lens detonators, something like 64 of them I think.
You can make a nice light-beam radar with a spark gap and a photomultiplier tube, and bounce off clouds and stuff.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.