- #1
0xDEADBEEF
- 816
- 1
I have a very small signal coming from a thermocouple [tex]\mu V \mathrm{to}\, nV[/tex] range, which gets drowned in noise when I transmit it via a 2m BNC cable, even the lock in runs into problems.
I am about to amplify it at the source, and while I am at it, I will try to do a few more tricks. Electronics always contained a fair amount of voodoo to me, so maybe someone here with some ee intuition can tell me what works.
Some info about the signal.
1) Anything above 500Hz is noise
2) A crappy passive R-C lowpass did not do enough to clean the signal
3) Voltage levels are ok up to 1V at the receiver.
4) The signal must go into a very high impedance, since it cannot provide any appreciable power.
5) We are working in fairly high magnetic fields
So the maximal setup I came up with is this (in order how the signals passes it):
1) active 500Hz lowpass
2) Amplifier 1000x ?
3) split the signal and invert it
4) three pin lemo plug for ground and +/- Signal
5) Cat 6 STP cable put +/- on one pair ground on another one
6) two active 500Hz Low passes
7) then into the differential input of the Lock in
But since I don't understand much about the tricks of the trade I don't know where the problems would be.
So my thoughts about the different components so far:
1) The initial low pass is just there because I see no point in amplifying something that I know is noise. I was told this could lead to a DC drift, which should not cause a problems if we are talking less then 1[tex]\mu V/\mathrm{day}[/tex]. I hope this element does not produce noise, distort, or draw current.
2) This sounds like a job for an op amp with a non inverting voltage amplification set up. Is there a reason to limit oneself to some amplification. Can I just do 10^6 or so.
3) Here I am really confused. I didn't see any setups for this. I would expect this to be a common task for differential signal transmission. But somehow they always seem to use two wires against ground, not caring about the symmetry. (So we know the difference between the wires, but we leave the voltage to ground floating) Voltage inversion seems to require some finite impedance at least with the inverting amplifiers that I saw.
4) Lemo is niiiice, I like, expensive though
5) I have little clue about cables. Is it better not to ground one side of the signal, and not use the shield as the return path. What are good cables? I was also thinking about triax, but there the wires are not balanced, so I don't think that they are suitable for differential transmission. Is that correct?
6) DC drift again, I hope I can get rid of it by pushing the "relative" button of the meters in the morning...
Questions:
1) Do you see problems with the above set up?
2) Do you have recommendation for parts?
3) Do you know a box that I can buy that does what I want?
4) What Do you think will give the greatest noise reduction? Early amplification, differential transmission, good cable, frequency filters, avoiding shield as return path?
5) If it's all build in house, how much work and time do I have to expect, for placing magical capacitors and tuning the resistances...
Any help appreciated
I am about to amplify it at the source, and while I am at it, I will try to do a few more tricks. Electronics always contained a fair amount of voodoo to me, so maybe someone here with some ee intuition can tell me what works.
Some info about the signal.
1) Anything above 500Hz is noise
2) A crappy passive R-C lowpass did not do enough to clean the signal
3) Voltage levels are ok up to 1V at the receiver.
4) The signal must go into a very high impedance, since it cannot provide any appreciable power.
5) We are working in fairly high magnetic fields
So the maximal setup I came up with is this (in order how the signals passes it):
1) active 500Hz lowpass
2) Amplifier 1000x ?
3) split the signal and invert it
4) three pin lemo plug for ground and +/- Signal
5) Cat 6 STP cable put +/- on one pair ground on another one
6) two active 500Hz Low passes
7) then into the differential input of the Lock in
But since I don't understand much about the tricks of the trade I don't know where the problems would be.
So my thoughts about the different components so far:
1) The initial low pass is just there because I see no point in amplifying something that I know is noise. I was told this could lead to a DC drift, which should not cause a problems if we are talking less then 1[tex]\mu V/\mathrm{day}[/tex]. I hope this element does not produce noise, distort, or draw current.
2) This sounds like a job for an op amp with a non inverting voltage amplification set up. Is there a reason to limit oneself to some amplification. Can I just do 10^6 or so.
3) Here I am really confused. I didn't see any setups for this. I would expect this to be a common task for differential signal transmission. But somehow they always seem to use two wires against ground, not caring about the symmetry. (So we know the difference between the wires, but we leave the voltage to ground floating) Voltage inversion seems to require some finite impedance at least with the inverting amplifiers that I saw.
4) Lemo is niiiice, I like, expensive though
5) I have little clue about cables. Is it better not to ground one side of the signal, and not use the shield as the return path. What are good cables? I was also thinking about triax, but there the wires are not balanced, so I don't think that they are suitable for differential transmission. Is that correct?
6) DC drift again, I hope I can get rid of it by pushing the "relative" button of the meters in the morning...
Questions:
1) Do you see problems with the above set up?
2) Do you have recommendation for parts?
3) Do you know a box that I can buy that does what I want?
4) What Do you think will give the greatest noise reduction? Early amplification, differential transmission, good cable, frequency filters, avoiding shield as return path?
5) If it's all build in house, how much work and time do I have to expect, for placing magical capacitors and tuning the resistances...
Any help appreciated