Problem with buffers and stepping down to low voltages

[george2525]

Hi

Ive been scratching my head for 2 days on this. I hope somebody can help as ive ran out of ideas

Below is an R2R ladder that outputs 0.1V - 3.1V in steps of 0.1V

I need these steps to be scaled down to 0.0015v - 0.465V in steps of 0.0015V (1.5mV steps)

The first buffer seems ok

I then use the inverting op amp with correct ratio but it does not scale properly!

I chose the LMC6001 because it states that is has ultra low current input so I thought this might help a bit

at 3.1V from the R2R it seems to work ok but not at lower values.

I have tried variations such as use of a potential divider from the first buffer to scale the voltage, then buffering that to the rest of the circuit but that didnt work well enough either.

The rest of the circuit is a BJT based exponential converter where the 0.0015V steps are applied to the base of a BJT

I can show this if necessary but doesnt seem relevant now.

Im hoping someone can help as ive hit a brick wall.

Thanks. I always appreciate the advice here

 

[Harald Kapp]

Why is one amp connected to Vcc, the other to Vss? Are you sure the supply voltages are correct and have a reference to ground (0 V)?

 

[george2525]

ignore that

i was trying different op amps that operate on higher voltages so needed to use a seperate supply. the voltages are all equivalent

 

[AnalogKid]

The first case is 31 steps. The second case is 310 steps.

One side of each switch goes to a wire that is not connected to anything.

1.5 mV probably is below the input offset voltage error range of the opamp, especially in a circuit with "infinitely" different inverting and non-inverting input source impedances. Low bit values might be swamped by offset voltage and bias current errors.

U16 circuit gain is -0.015. With a 3.1 V input, the output is -0.0465 V.

ak

 

[george2525]

The wire thats not connected to anything just goes to a voltage supply (reference) of 3.2V.

1.5 mV probably is below the input offset voltage error range of the opamp, especially in a circuit with "infinitely" different inverting and non-inverting input source impedances. Low bit values might be swamped by offset voltage and bias current errors.


ak

Yes i think it was something like that. I actually managed to get a decent result by changing the op amp for a TL072 and making a low resistance potential divider

ill leave this post up in case anyone stumbles on it with a similar problem. heres the new version that works.