Closed loop for accurate voltage measurement

[Sandyholmarc]

Am currently working to make a closed loop with LT1970 and LTC6915 .LTC 1970 is used for four quadrant supply and LTC 6915 for voltage measurement . . output is oscillating and i don't know how to correct
it.any advice and suggsetions is greatly appreciated.

 

[KlausST]

Hi,

Try to use local HF feedback at LTC6915.

In your circuit it has infinite gain....this is always critical.
Do you need the LTC6915 at all?

Klaus

 

[Sandyholmarc]

Thanks for your reply. LTC 6915 used because of it's low offset voltage for accurate voltage measurement .Now I changed the circuit as LTC 6915 is only used for voltage measurement and add an another opamp ,connect it's input across the load and connect it's output back to the inverting input of LT1970.

 

[FvM]

The PMOSFET is wrongly connected in your schematic. You should also consider that the MOSFET booster circuit is adding gain and reducing phase margin so that the amplifier needs additional frequency compensation to achieve stability.

 

[Sandyholmarc]

Hi FvM thanks for the reply
LT1970 circuits output varies according to the input ( getting output from -1 to +1 according to i/p-0.33 to +0.33) with 250R load, means PMOSFET is working fine .But when connecting a 1R resistor load there is 50mv difference between applied input and output when applying 0.33v input insteadof 1V am getting 950mv that's the reason why i add an external Opamp AD820 for making zero error beween applied voltage and output.The output of AD820 is connected back to -Vin of LT1970 for minimum error.Can you please tell me about adding an external frequency compensation network in my circuit .

 

[FvM]

Before designing the feedback amplifier, you would make a specification. Load impedance range, amplifier DC and AC characteristic (either gain versus frequency or characterized in time domain), intended accuracy.

Presently you have a relative low, frequency independent loop gain (P controller). The gain won't suffice to cancel DC errors but can already cause instability at higher frequencies. Something like a PI gain characteristic is suggested.