Lets have a sensor amp discussion!

[itsmoked]

I have a board I designed that I'd like to improve but I'm fresh out of ideas. I'm wondering if anyone has any suggestions for what I think is an in-frequent but probably not uncommon problem.

I have a customer probe in a chamber that monitors the plasma within.

Typically the signal is brought in and buffered via some relatively quick amplifier. The amplifiers have obscene input resistances like 10¹⁶ohms so the probe is completely unloaded.

This is good because the probes have horrible source impedances that vary all over the place but tend to be between 700k and 1.5M ohms. Nasty!

If this was all there was to it I'd be thrilled. I'd be getting the required unsoiled reading owning to the extremely high amp input impedences.

The problem...
Frequently the plasma is turned off. This results in infinite impedance! So now the amplifier has a floated input. It promptly drifts up to the rail as does its output.. The solution is to put in a resistor from the input to ground.

Bye bye 10¹⁶.

A 1M pull down resistor up against a 1M sensor probe results in a loss of 50% of the signal. They need to know the DC levels here so I can't capacitively couple to the probe.

Looking for ideas on how to prevent leakage current from railing the amp while minimizing any loading on the sensor.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Skogsgurra]

The only thing that I can think of "earla in the morning" is Keithley (suitable name) electrometers and their all-Teflon input stages. I have used a 602 with 10¹⁵ ohms input impedance. I am sure there are better ones available now. Your 10¹⁶ indicates that. The 602 dates back to around 1983.

http://web.mit.edu/8.13/8.13d/manuals/keithly-602-electrometer.pdf

has lots of information, including schematics and BOM. There are also a few tips on common things you need to Think about.

How fast do you need? At 1000 MΩ (1 GΩ) any piece of wire or cable will slow it down. One m (3 feet) has around 100 pF, so the time constant of the transducer and 1 m of cable will be around 100 milliseconds, which means BW around 1.5 seconds. Mentioned as a background, not a solution to the problem at hand.

Drift when turning the plasma off is the problem? I would cheat and use the information (plasma off) to freeze the signal and release it when plasma is turned on again. A 12-bit A/D and a corresponding D/A should be transparent to the system at these speeds and the OFF signal freezes the A/D and the ON signal releases it again. Doable?





Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[IRstuff]

Possibly some sort of chopper configuration?

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[itsmoked]

I meant to include this in the OP.

That's not a comma Gunnar it's a period so I'm struggling with one and a half Meg not one thousand five hundred Meg. Probably not electrometer land. That manual is amazing! Not like these days is it. I'm looking at about 500kHz stuff when not monitoring the DC level.

You may have it there with the cheat. Funny thing, I penned this question then I explained this problem to my wife while we were walking the dog last night and she said, "Why don't you turn off the output whenever it exceeds a maximum value". You two are clearly colluding..

IR; I keep thinking that but choppers are not very good at fast stuff they seem more for DC stuff like buffering references and what not.





Keith Cress
kcress -

http://www.flaminsystems.com

 

[IRstuff]

Back in the OLD days, IR sensors almost always had some sort of chopper stabilization, into the kHz. This one claims a GBW of 1.2 MHz

http://www.analog.com/media/en/technical-documentation/data-sheets/lt1052cs.pdf

Other similar possibility is the infamous correlated double sample:

http://www.analog.com/media/en/technical-documentation/lt-journal-article/LTC2411_0501_Mag.pdf

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm