| Electronics > Projects, Designs, and Technical Stuff |
| Uber low noise current source suggestions |
| (1/5) > >> |
| grouchobyte:
I have to design a laser driver for a very expensive (>$10k) high performance laser. It needs 3 volts of compliance and deliver around 650 mA CW. Long term stability and drift is not too important but current noise below 100 khz needs to be below 2 nA p-p. What I am concerned with for now is short term stability and current noise. The laser will be temperature stabilized and there are multiple APC loops for stability. Dont ask me what the application is......unless you want some men in black suits showing up at your door. Suggestions are appreciated and of course questions are welcome. I have limited info from my client at this point but I will attempt to answer what I can @grouchobyte |
| David Hess:
Without feedback and filtering, I do not think you can get even close to that specification but that is ok. Besides filtering, use a feedback loop to lower noise within the feedback loop's bandwidth. Then this problem becomes one of voltage noise at the sense resistance and the feedback amplifier. |
| grouchobyte:
I did a little more calculation and I can probably live with less than say 10 nA of p-p noise in the specified 100k BW i think what David Hess is suggesting is probably the most reasonable approach plus careful component selection @grouchobyte |
| David Hess:
If a bipolar transistor was used as the pass element, then just the shot noise of the output transistor would clobber that specification but like I said, filtering the output and feedback can control output noise so the characteristics of the pass device lose some importance. Someone who knows about noise in power MOSFETs can comment on if they produce less noise which would make meeting your requirements easier. Also consider building in a way to verify the current noise as part of your development circuit. That may include an amplifier with a specific bandwidth to drive your oscilloscope. |
| Marco:
--- Quote from: blueskull on July 04, 2019, 02:39:16 am ---2nApp is insane. Even with 1R current sense resistor (horribly inefficient for a 3V 650mA diode) --- End quote --- It seems inescapable, but these guys get close at 470 mA with 15 pA/rtHz (lots of 1/f because no auto-zero, but lets ignore that) and I see no massively heatsinked sense resistor. How do they do it? PS. or maybe what I saw as 1/f is a consequence of using inductance in the feedback loop? |
| Navigation |
| Message Index |
| Next page |