F
Fred Bartoli
- Jan 1, 1970
- 0
I've some requirements for some challenging voltage measurement (I know,
subject said current, but see later). Namely:
- up to+/-4kV pk WRT ground
- % order accuracy up to 50kHz
- duty cycle max 10%
'til now, no pb.
Then the fun part:
- 80dB CMRR up to 100kHz
- 60dB CMRR up to 10MHz (yes!)
- the signal can go down to 4V and yet 1% accuracy is required (which is
an overall 16b accuracy)
No pb with power drawn by the probe but they don't want maintenance,
periodic cal and so on...
We are politely (well not so politely) asked to do this à la 2x
compensated dividers then diff amplifiers. Yeah...
OK. Now I have to backup that crap with a working solution, or as near
as possible to the specs.
My first thought was a load resistor (told you it was current
measurement and two current shielded xfrmrs (two to reject the CM
currents from load to shield) for the BW, and a servoed DC-LF
compensating current, all that almost working like the Tek probe.
Unfortunately, looking closer, the LF-DC part isn't going to be that
easy (mainly Hall sensor noise, offset and low sensibility...) :
- the load can't go much below 100K (10W on average) giving a 40mA-ish
pk current (plus the needed HF zero).
- Hall sensors seems to be somewhat noisy
- can't use multi turn primary, due to the then uncontrollable
parasitics (still CMRR)
Thought about fluxgate but the switching frequency will be well into the
signal BW, so it's probably not OK. Plus I can't think of a simple way
to implement a gate suitable to the design (100, maybe 200 units made).
Any clever idea welcomed.
subject said current, but see later). Namely:
- up to+/-4kV pk WRT ground
- % order accuracy up to 50kHz
- duty cycle max 10%
'til now, no pb.
Then the fun part:
- 80dB CMRR up to 100kHz
- 60dB CMRR up to 10MHz (yes!)
- the signal can go down to 4V and yet 1% accuracy is required (which is
an overall 16b accuracy)
No pb with power drawn by the probe but they don't want maintenance,
periodic cal and so on...
We are politely (well not so politely) asked to do this à la 2x
compensated dividers then diff amplifiers. Yeah...
OK. Now I have to backup that crap with a working solution, or as near
as possible to the specs.
My first thought was a load resistor (told you it was current
measurement and two current shielded xfrmrs (two to reject the CM
currents from load to shield) for the BW, and a servoed DC-LF
compensating current, all that almost working like the Tek probe.
Unfortunately, looking closer, the LF-DC part isn't going to be that
easy (mainly Hall sensor noise, offset and low sensibility...) :
- the load can't go much below 100K (10W on average) giving a 40mA-ish
pk current (plus the needed HF zero).
- Hall sensors seems to be somewhat noisy
- can't use multi turn primary, due to the then uncontrollable
parasitics (still CMRR)
Thought about fluxgate but the switching frequency will be well into the
signal BW, so it's probably not OK. Plus I can't think of a simple way
to implement a gate suitable to the design (100, maybe 200 units made).
Any clever idea welcomed.