HV p. supply- Sensing / limiting current

[atferrari]

The output of my dual variable p. supply goes from 1 to 45 V with a max. current of appr. 2A. Same specs for each rail.
The PICmicro in charge has to mannage the setting of current limitation requiring sensing of load current for each rail somewhere around the pass transistor.

When considering an opamp (diff. amplifier), to get the drop across the sensing resistor, measured wrt ground, I've found that:

a) Common opamps can not sustain working with such high input voltages (appr. 45 V). The uncommon ones that I could use are EXPENSIVE /HARD TO OBTAIN.

b) Every time I tried a differencial amplifier fed by a supply also feeding the circuit under measure, Vout was meaningless. With separate supplies (that is, separate commons) everything was fine.

My questions:

1 - How to sense current reliably, with Vout= 45V?

2 - Whatever is suggested, I need the same cirucit to sense current as reliably as above when Vout= 1V.

3 - If I solve question 1, I wonder if sensing BEFORE the pass transistor is better, since the Vin (even quite high)would be more or less always the same allowing to neglect problem in question 2.

4 - Is it any chance of avoiding an auxiliary supply (even small) for any differencial configuration?

5 -I think I fairly understand op amps to a certain depth but the necessity of separate supplies in differential amplifiers escape to my comprehension.

I browsed, several manuals, National's is one, and found no warnings on that. It seems that pros take that for granted.

I recall that for the LM725, there was a circuit which I reproduced with the mentioned surprising finding. Same surprise with the same circuit implemented with two OP77.

Enlightening replies will be appreciated. (Even more if you can follow the order of my questions!)

Muchas gracias

 

[BrianG]

Well, you could always reduce the common-mode voltage into the op-amp by taking the voltage from each side of the sense resistor to a lower level with a pot divider network to 0V from each side. However as the differential voltage will be much reduced and you will need close tolerance resistors to make it work reliably.

However, you don't have to put the sense resistor up at 45V. You can put it in the OV side - provided that you move the ground reference of the output voltage sensing circuit to the load side of the resistor.

That way you get the current-sense voltage at around ground potential. In some respects you still have the same op-amp problem, i.e. both inputs close to a supply (-ve) rail but it is much easier to find inexpensive devices that can have both inputs operating near the 0V rail than at 45V! From memory I think various bi-mos op-amps, or the 324 op-amp series do this rather well.

 

[ScottyUK]

You could use one of the many 'instrumentation amplifiers' available, something from the excellent INA*** series from Burr-Brown perhaps. These give you a better performance/cost ratio than the equivalent discrete circuits, and are available with up to 200V common-mode input.

 

[nbucska]

Alternative: Use a +rail referenced OP.AMP with level xlator
(FET,BJT or optocoupler ). The V- of the OP.amp is R from
GND, zener to V+. Use low current R2R op.amp.

If unclear, give fax# OR E-mail ( 1 MB )

<nbucska@pcperipherals.com>