Auto-scaling an analog meter has accuracy problems

Thread Starter

splitDiode

Joined Jan 24, 2021
6
OK, I’m about ready to tear my hair out.

I started this project last year, first because I wanted a variable linear power supply to work on another upcoming project. I have lots of OLD parts from years ago, and wanted to use them as much as possible although after a lot of grief trying to implement this with just 741’s, I did break down and buy some quieter op-amps.

The supply itself wasn’t a problem. It is based on a LM723 with 2N3053/2N3055 pass transistors. Output is 0–30V, 0-2 A. Ground is depressed with a 7805 regulator to achieve the 0V part of the range.

As I said I have lots of old parts and wanted to use one of the analog meters for displaying the current.

This led to a Current sense and limit circuit that runs on separate supplies to avoid Common Mode limits. This system reads a 0.1Ω sense resistor on the board of the main supply.

To allow me to read currents to greater precision, I designed a system to auto scale, with ranges of
0–20mA, 0–200 mA, and 0–2 A.

This measurement system consists of

  • A differential sense amp, X10, Current Sense Out (CSO)
  • A low pass filter to remove 60 Hz hum
  • Voltage references
  • A X10 amplifier
  • A LM311 Comparator to detect when current sense is greater than 20 mV.
  • Another LM311 Comparator to detect when current sense is greater than 200 mV
  • Another X10 amplifier
  • TTL logic to generate control signals for switching
  • A CD4016 Quad bilateral switch to select the proper signal to the meter control/Full scale adjust. The three input signals are, CSO, CSOX10, CSOX100. The output is 0-2V.
  • A 741 buffer followed by a resistive divider to the meter.
The autoscaling works, but I am having 2 problem with the readings

  • The ranges are not congruent, a sense voltage 10 or 100 times another does not reliably produce the same reading as it should. Seems to be a linearity/accuracy problem.
  • Sometimes when I start the supply at or near 0 volts, the current sense system shows a negative current even though independent measurement shows a positive voltage. The voltage is appropriately 0 or positive coming out of the differential sense amp, but negative after the low pass filter, (about -150mV). This self-corrects after raising the Power supply voltage enough.
Constant Current Preset Schematic Page 1 of 4.jpgConstant Current Preset Schematic Page 2 of 4.jpgConstant Current Preset Schematic Page 3 of 4.JPG
 

Attachments

MrChips

Joined Oct 2, 2009
30,711
I see two problems.

1) Why are you still using LM741? Don't let AG catch you doing that!
2) I don't see any DC bias adjustment. All op-amps have DC offset that need to be nulled.
 

Thread Starter

splitDiode

Joined Jan 24, 2021
6
I see two problems.

1) Why are you still using LM741? Don't let AG catch you doing that!
2) I don't see any DC bias adjustment. All op-amps have DC offset that need to be nulled.
I am using 741 because I have a drawer full of them. Ditto the 4558. Who is AG and why does he care?
Their is DC bias adjustment of questionable usefulness on the initial OPA177, even though the data sheet for it and the OP07 says probably don't use. The OPA177 is nulled and there is less than 1 mV (undetectable by my meter) after the first OP07. Any other offset can be accommodated in the final full scale meter adjustment.
 

Thread Starter

splitDiode

Joined Jan 24, 2021
6
I will test again in table form for you. My last tests were kinda messy.
Attached are the test results. The biggest problem is obviously in the first decade. I had to restart the measurements several times because UV11-1, 4, 11 were showing negative values around 150mV when the load voltage was showing 0. This slowly returned to 0 as I raised the load voltage until around 1.25 V when it went to a reasonable positive value. As I lowered the load voltage, the results came down - SOMETIMES. Occasionally it would flip back negative. It was a struggle to get data for load values below 100 mV. TP1 and TP2 showed appropriate if inaccurate values throughout.Current measurement data 2.jpg
 

crutschow

Joined Mar 14, 2008
34,282
It would see likely that you are seeing the errors due to component (resistor) tolerances and op amp offsets.
What are the tolerances of the resistors you used?
 
Top