4
\$\begingroup\$

I asked this question a few days ago, but I don't think I was clear enough, so I'm going to try again.

Here is a rough diagram of my sputtering system:
Chamber electrical sketch

Basically, a DC voltage supply holds the substrate at a large negative voltage (~-1000V) relative to the chamber chassis (Vs) and a DC current supply pushes a large current (~120A) through a plasma into the chassis and substrate (It). Only a small fraction of the total plasma current actually goes through the substrate (~2A out of the total 120A). Most of the current passes directly into the chassis. So, to clarify, there are two power supplies. The voltage supply is providing -1000V to the substrate through which ~2A are passing. The current supply is providing 120A through the target/plasma (at about 20V).

The resistor between the chassis and ground indicates that the chassis is poorly grounded. Furthermore, the resistor should be taken to be variable as the current flow through the stainless steel chassis fluctuates, meaning that the voltage of the chassis relative to earth ground fluctuate significantly over time.

Here is the problem: I want to measure the voltage between the substrate and a particular spot on the chassis without measuring any of the voltage due to the large current flowing through the chassis. The measurement doesn't have to be very precise, it's only used as a check (i.e., +/- 5V is fine).

Right now, this is accomplished using a Fluke battery powered DMM. Since it's battery powered, it makes a real floating differential measurement. What I'd like to do is replace this handheld meter with a non-battery powered solution that could also be hooked to a computer for data-logging purposes. I thought maybe using a 110V AC to 9V DC wall adapter that could hook into the Fluke's battery terminals might be an idea, but I guess there is no electrical isolation and all the current in the chassis would get dumped through the Fluke to the mains ground.

Can anyone suggest an approach? I'm pretty ignorant about this stuff. I've tried reading up on it but can't figure out anything that might work. If I can clarify in any way, I'd be happy to. Any suggestions (including "you're stupid. this can never work") would be appreciated!

Thanks a lot in advance, Brian

\$\endgroup\$
  • \$\begingroup\$ Wait wait wait... 120A at 1000V?? That's a LOT of power... unless I'm misreading something. \$\endgroup\$ – Thomas O Jan 10 '11 at 23:51
  • \$\begingroup\$ There are two separate power supplies. The voltage supply applies (up to) -1000V DC on the substrate. The current through the substrate is ~2A when the plasma is on. The current supply provides 120A to the target/plasma. Typical voltage there is about 20V DC. I'll update the post to clarify. \$\endgroup\$ – BASnappl Jan 11 '11 at 0:06
  • \$\begingroup\$ Where do you work, @BASnappl? Love the question! \$\endgroup\$ – tyblu Jan 11 '11 at 0:44
  • \$\begingroup\$ Acree Technologies Inc. in Concord, CA. Lots of sputtering systems. I'm just an ignorant intern :/ \$\endgroup\$ – BASnappl Jan 11 '11 at 1:11
1
\$\begingroup\$

While a wall adapter is usually isolated, it likely isn't rated to 1kV. A simple solution would be to find a DMM with logging capability and CAT III/IV 1kV isolated external power source or long battery life. Here are some candidates:

Another option is to use a simple micro with an ADC, float the whole circuit at chassis voltage, then use proper isolation techniques (more than just optoisolation -- if you're not sure I suggest another question) to communicate with a PC over your preferred serial connection (UART -> opto -> RS232 -> USB -> opto -> PC, with cable shield voltage measurement and warning, would be my choice). Note that this means you can't touch anything on the floating (hot chassis) side of the widget. This way you can eliminate the power supply isolation worries by just using a battery, and still run it easily for 6 months to a year without replacing it, following thought on power consumption and sleep modes (ie: MSP430). Also note that a sputtering machine generates electrical noise, so you may need to use RS485 with error detection/correction algorithms.

\$\endgroup\$
  • \$\begingroup\$ Thanks for the detailed response. I'm reading up on the suggestions. I'll likely post a follow-up question since I'm unfamiliar a lot of this. However, I'm trying hard to learn! \$\endgroup\$ – BASnappl Jan 11 '11 at 1:14
  • \$\begingroup\$ What's your take on high voltage probes? Are they unsuitable due to the potential for large common-mode voltage? What about isolation transformers for bench meters? \$\endgroup\$ – W5VO Jan 11 '11 at 5:38
  • \$\begingroup\$ @W5VO, High voltage probes sound like a good idea to me, especially for only 5% accuracy specs. Bench meters are generally expensive and are for more accurate readings than handheld DMMs. I'm not familiar with 1kV+ isolation transformers. \$\endgroup\$ – tyblu Jan 13 '11 at 0:57
1
\$\begingroup\$

The current flowing through the signals your trying to meter doesn't matter in this instance at all, all that matters is the voltage your attempting to measure.

Any decent voltage metering device will be relatively high impedance (like 10M ohm) the only current that will flow through the meter is a function of the voltage differential being measured between signals and that internal resistance. The current flowing in the signals being metered doesn't matter at all (other than how it influences the signals voltage).

You are correct that you want to measure differentially, if your meter is earth grounded it most likely needs to withstand the common-mode voltage of the signals. However its probably easiest just to use a resistor network to divide the voltage down to something safer to plug into your meter. Again the design of this resistor network doesn't depend on the currents in the signals, just their voltages. Obviously as this high a voltage you want to use fairly high value resistors.

\$\endgroup\$
  • \$\begingroup\$ The common-mode voltage is the problem with this solution, as far as I understand. Since the chassis can be at a considerable voltage relative to earth ground, the concern is that breakdown through the meter might be preferable to going through the "resistor" (see image) to ground, thereby burning up the meter. \$\endgroup\$ – BASnappl Jan 11 '11 at 1:16
  • \$\begingroup\$ Thats the reason to just use a simple pair of resistors as a voltage divider. Given the minimal accuracy requirements a simple 1000:1 divider on each input terminal would allow him to use almost any type of earth grounded meter. You can even get o-scope probes with such dividers built it that will function to very high voltages 25kV+, I would imagine many bench DMM supplies would offer similar probes. \$\endgroup\$ – Mark Jan 11 '11 at 4:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.