| Electronics > Projects, Designs, and Technical Stuff |
| variable gain current amp |
| << < (3/8) > >> |
| Simon:
at a 1mV input the AD error is high. this can't be done with gain adjustments. It has to be in the the sense resistor switching. |
| Kleinstein:
The proposed current sense amplifier is just not very sensitive - more like 2-3 digit class. So it is just the wrong chip for a large dynamic range. It may run with higher gain, but it does not make too much sense with an input drift of some 7 µV/K and high noise. For a large range one may have to switch ranges, by switching the shunt, at least in 2 ranges. 20 mOhms are OK for a 5 A range ( 100 mV full scale), but this means 1 mA gives only 20 µV. So it would need a pretty good amplifier to do this in one range. For shunt switching on can have the 2 shunts in series. For switching one could use MOSFETs in 2 ways: a) short out the larger shunt and switch the sense point b) direct the current with 2 MOSFETs and always sense the voltage with both shunts in series. If the burden voltage is critical it may need a beefy FET with on ON resistance in the <20 mOhms range. |
| Vovk_Z:
I would use some precision opamp like op37, for example, with a constant gain of 50 (and a 20 mOhm shunt). It has a trim circuit to adjust a voltage offset. The only drawback is that it needs a dual power but it is usual for precision and ultra-precision opamps. |
| Simon:
--- Quote from: Kleinstein on May 14, 2020, 07:14:29 pm ---The proposed current sense amplifier is just not very sensitive - more like 2-3 digit class. So it is just the wrong chip for a large dynamic range. It may run with higher gain, but it does not make too much sense with an input drift of some 7 µV/K and high noise. For a large range one may have to switch ranges, by switching the shunt, at least in 2 ranges. 20 mOhms are OK for a 5 A range ( 100 mV full scale), but this means 1 mA gives only 20 µV. So it would need a pretty good amplifier to do this in one range. For shunt switching on can have the 2 shunts in series. For switching one could use MOSFETs in 2 ways: a) short out the larger shunt and switch the sense point b) direct the current with 2 MOSFETs and always sense the voltage with both shunts in series. If the burden voltage is critical it may need a beefy FET with on ON resistance in the <20 mOhms range. --- End quote --- Using fets to switch the shunt resistors is not so simple. I need serious fets with µR of resistance to not become a significant and unpredictable part of the measurement. with a 20mR fet I want something like 2 orders of magnitude or better (200µR) lowest i can find are 340µR which really means 680µR when you actually use them, two in parallel will give me a fighting chance or getting something like 200µR. Given that the solution is in switching the shunts the amplifier becomes pointless as that was all that was for to switch the gains. I can use a 200mV panel meter and set the decimal point on change over using it at up to 50mV |
| Kleinstein:
With a 2 nd FET or an analog switch one can avoid having the switch as part of the shunt. So the higher current FET would only need to be so large to keep the overall burden low enough (e.g. 10 mOhms range). The 2 nd FET would only be for the maximum current in the lower range, so less critical. The switching part may still need an auxiliary voltage. The alternative would be using a really good amplifier for current sense and possibly a slightly larger shunt despite of more heat loss. This would be a little like designing for a high resolution current measurement in the 6 digit range. |
| Navigation |
| Message Index |
| Next page |
| Previous page |