Know of a good high voltage, very low load probe?

[zappedagain]

(If this isn't the right forum for this let me know)

I'm looking for a probe that can handle +/-100V with very low capacitve loading (<1 pF). Rise and fall time of about 1mS is acceptable.

I have an Agilent 10076A (66 Megohm // 3pF) that doesn't quite do it so I'm searching for something better. I'm probing a pixel on a TFT (thin-film-transistor) panel; did you ever try probing a 0.3pF cap that gets charged through a MOSFET with a RDS-on of about 1 megohm?!?

I have a proprietary, two of a kind on the planet, probe with a leakage current of about 2 pA and lots of temperature drift. I'd rather not put that one on the production line...

 

[Warpspeed]

How about something like a Fluke 40Kv EHT probe ?

They have a 1,000:1 voltage reduction and are basically a simple voltage divider with a 1,000Meg in series with a 1meg resistor. I have no idea what the input capacitance or rise-time would be.

Apart from that, a proper FET oscilloscope probe might do the job, but the safe maximum input voltage may be rather limited.

 

[elfgriper]

zappedagain,
take a look at a Tektronix P6015 freq compensated probe or there newer replacement model and see if their specs meet your needs.
-elf

 

[ScottyUK]

How about the Tektronix P5100 HV probe? 2.75pF, 10M[&Omega;], 250MHz, 100:1, 1.75ns risetime, 2500V max input?

The Tek 6015 is a physically big HV probe and has 1000:1 attenuation, and a slower rise time & bandwidth.







----------------------------------

If we learn from our mistakes,
I'm getting a great education!

 

[Skogsgurra]

The +/-100 V does not need a HV probe. Your problem is the capacitance. A naked scope input has around 30 pF, so you cannot use that one. That's for sure.

Your modest rise-time needs make a multi-megohms passive resistor a viable solution. The time constant of the input is about 1 Megohms times 30 pF - or around 30 microseconds. Adding a resistor will not increase that time constant, it will decrease it (or leave it unchanged) since the added resistor is in parallel to the internal resistance. So you can add as much resistor as you want without having a longer time constant.

If you put a 100 Megohms (probably need to use ten 10 Megohms) resistor as close to the TFT pixel as possible and make the cable to the scope input very short, you will have something that might work. The sensitivity of a normal scope can usually be set to 2 or 5 mV/div and with this "probe" you will have 200 or 500 mV/div.

Why shall the connection be short? Simply because a metre of wire will add something like 100 pF capacitance to your input and turn your time constant up to more than 0.1 ms. Still acceptable for your application - but unecessary.

Using a 1 Gohms resistor will reduce loading even more and you will still have a less than 0.1 ms time constant. Remember that you need a very good screen - right down to the probe tip's last millimeter - to avoid any stray voltage pick up.

Gunnar Englund

http://www.gke.org

 

[ScottyUK]

Hi skogs,

Totally agree that +/- 100V does not need an HV probe. Most of the commercial low capcitance probes are limited to about +/-20 to 30V, which makes them unsuitable.



----------------------------------

If we learn from our mistakes,
I'm getting a great education!

 

[Skogsgurra]

I see. Didn't know that (the low voltage capability, I mean).

Gunnar Englund

http://www.gke.org

 

[IRstuff]

Skogs' idea is still tractable, I think, if you use the 1G resistor as a divider to knock down the voltage and then run that directly into a buffer amp attached to the probe. Something like a simple FET buffer amp mounted directly on the prober with the divider tacked in front of that.

I've used something similar to poke sharp sticks into the guts of integrated circuits.

TTFN

 

[zappedagain]

Thanks everyone for your help. As I started working through the divider circuit I realized I left out an important point. Because of the raster-scan drive technique my signal only gets refreshed every 50mS or so. That means the time constant with a probe attached needs to be much greater than that and pushes my divider network over 100 Gigohm. That gets out of the range to get a decent signal to the scope so a DC divider won't work (Heizenberg wins again). I'm back to driving the gate of a high voltage FET (my two on the planet approach). I guess I'll keep pushing my sister-company to get that design out of the lab and into production.

BTW - Fluke 40Kv probe (80K-40) - While it is great for DC, Fluke says the frequency response (-3dB) is only about 200 Hz, so they don't recommend using with any scope.

 

[IRstuff]

I'm confused why you want your scope time constant to be longer. That will smear out your signal.

TTFN

 

[Skogsgurra]

I guess that the little capacitor (0.3 pF) gets discharged very quickly when a resistance <100 Gohms is connected. Since it gets updated every 50 ms, the time constant needs to be in the 500, or more, ms range. The time constant of the probe/input capacitor needs to be in the 1 ms range.

Gunnar Englund

http://www.gke.org

 

[zappedagain]

Right. I want the scope to respond quickly, but I don't want it to discharge the <1pF cap that I'm trying to measure. DC leakage current is a killer in this app.

 

[IRstuff]

How about AC coupling?

TTFN

 

[itsmoked]

Man this sounds more like you need one of the noncontact probes. Used for measuring values inside ICs. I believe they are called "Force Probes".

 

[IRstuff]

Force probes are a different animal altogether; they measure profiles, but not voltage.


There are, however, SEM's modified for detecting voltages on circuits.

TTFN

 

[itsmoked]

My buddy did work for a San Jose company that made non-contact voltage probes for oscilloscopes. I'll have to dig around.

 

[Warpspeed]

Another way to do it might be to combine a voltage divider with some sort of sampling technique.

Suppose you use a very high value series resistor right at the probe tip as suggested previously, and then switch the other end electronically. When it was "off" input loading and leakage would be very low. When ot was "on" the input voltage could be sampled at a very small duty cycle. That would go towards solving the constant circuit loading problem.

The output signal might be reconstructed with a sample and hold circuit.

This is not going to be easy to implement, but it is just suggested as a line of possible further thought.