AD620 - Current sensing

[DzAnej]

Hi!

I am currently working on a programmable power supply project and I am using AD620 to amplify the voltage drop on the shunt ressistor (0,1?).
Here is the schematic:


I have set Rg(gain resistors R3, R4, R5) to 12.35k?) so that the gain of the amplifier is 5 - the max voltage on the resistor should not exceed 1V (because the max current is 10A) and the output should be from 0 to 5V (0 to 10A). Now, when I measure the voltage across the shunt resistor it goes from 0 to 500mV (I'm only going to 5A) but the output of the AD620 stays 0V...
Here are some measurments:
Current: 0A
AD620 out: 16.5mV
+IN: 15,13V (CH1 - Yellow)
- IN: 15,13V (CH2 - Blue)

Current: 5,13A
AD620 out: 20mV
+IN: 14,25V (CH1 - Yellow)
- IN: 14,77V (CH2 - Blue)
Shunt resistor: 522mV

I've tried to filter out a little bit the signal but hadn't succeed, here is what i tired:

What could be the cause of the output staying at 0V?

[Yansi]

Do you know what input common mode voltage is? You are exceeding it ad might be glad the AD620 did not let the magic smoke out yet.

How can you sense +15V rail when the AD620  V+ voltage is +12 only? (The maximum permissible operating value is +Vs-1.2V according to the datasheet)

Btw, why don't you put the shunt in the mosfet source and measure the voltage against ground?


It seems you are trying to design something the wrong way most complicated way possible. A MCU controlled regulated power supply can be built around 2 jellybean opamps, no need for anything special like the instrumentation amps. Ans at 15V maximum output, the whole thing is a primitive task. If you want, I may draw you a "skeleton" schematic for it.

Bottomline: Using Rg1+Rg2+Rg3 in series is really a strange, as can be easily calibrated by a trimpot on a suitable place, or even calibrated by the DAC or from the ADC value.
Also please note that 15V 10A supply without any preregulation or tap changing (if it uses a mains transformer) is very bad idea. It is NOT as easy to get 150 or more W of heat out of some semiconductors, as someone might think.

[DzAnej]

Thanks for your reply.
I looked the datasheet and the Differential Input Voltage was 25V so I thought that that was ok... so that means that it won't work up there at 15V?
Also I though that the supply voltage could be in range of the output and not input... So in my case i wouldn't have a greater voltage of 5V on the output...

Next, I am using the shunt resistor at the high side because if I used it at the low side (at the mosfet) then the mosfets source wouldn't be connected to ground and i would need to offset the gate voltage (I think).

I have LM324 but i wanted to use AD620 because of it's precision... In the end I wan't to create a precision power supply with fine resolution...

Rg1, Rg2, Rg3 are all 1% because I want to get as close to the gain of 5 as possible so that I would not need to calibrate it with software that much later on...

And for the supply I am planning on using switch mode power supply but if I find out that it would be too noisy I am going to have a custom transformer wound with multiple taps. For the semiconductors I am planning to use multiple of them so that the power will be sheared between them.

[Yansi]

No.  The differential voltage is differential. That is the voltage BETWEEN the amplifier input pins, commonmode voltage is their OFFSET with regards to power supply pins. Also the differential input voltage is specified with 30 or 36V supply across the AD620.

Next, I am using the shunt resistor at the high side because if I used it at the low side (at the mosfet) then the mosfets source wouldn't be connected to ground and i would need to offset the gate voltage (I think).

You think right, but that's the job of the voltage regulator amplifier. So stick the shunt down there and leave the AD620 alone  ..or for something that deserves it.

1% of the first 10K resistor is 100ohm, so the last 150ohm resistor is kinda halfway pointless there. So is to trying to calibrate the current sense this way.

Also the transistor in the negative terminal topology is kind of not of the greatest. I'd leave it in the high side, as it will simplify a lot of things.

Look for example on this, to get an idea: http://www.electronics-lab.com/project/0-30-vdc-stabilized-power-supply-with-current-control-0-002-3-a/
U1 can be replaced by any suitable voltage reference IC. Potentiometers can be replaced with DACs. (minor circuit modification will be required around U3).

[DzAnej]

You think right, but that's the job of the voltage regulator amplifier. So stick the shunt down there and leave the AD620 alone  ..or for something that deserves it.

Well that makes sense now, thanks for that.

The resistor thing, I've measured a bunch of them and got pritty damn close, actually I got it spot on  the only thing now is the precision of the multimeter.

I was thinking of getting P-channel mosfets so that they would be between +SUPPLY and LOAD but I wasn't quite sure how I was gonna drive them and I already had a bunch of IRFP250N's laying around so I used them... I was wondering, would it be possible to use the N-channel  like that (between +SUPPLY and LOAD), would the amp. handle that kind of voltage offset?

Oh and thanks for the schematic, I will surely be helpful. But I would like to finnish my schematic because the voltage regulation is already working splendid, it's the current limit that is bothering me.

[Yansi]

I haven't seen your full circuit schematic, so I cannot judge if there are any other possible problems to be solved. The above schematic posted of your circuit seems not to be complete - as it would not work like that in reality.

You do not need a PMOS. Why would you? You can use either a NMOS or NPN in the + rail as the series pass element, as an source or emitter follower respectively. (NMOS having the requirement of higher driving voltage, as the Vgsth is about 3-4V, compared to Vbe of 0.6V for NPN).

IF you need your supply to output up to 5V, using a mosfet as the regulator, the gate to circuit ground voltage required is 5V+Vgsth (if the control circuitry will be referenced to the negative output terminal - which I recommend to do). So you need your voltage regulator opamp to output at least 9V on the gate (measured against circuit ground, negative output terminal). So, if you power the control circuitry from a 12V or 15V I'd rather go for 15V to be on the safe side, any jellybean opamp will do fine. If you need somewhat precision of the power supply, look for an opamp with low offset voltage and low drift. (OP07 comes to mind, as a very cheap precision opamp. Also, same applies for the current regulating opamp!).

If you need further help, please provide us with the full schematic, I (or anyone else) can look through.