Finding absolute accuracy of the circuit

[razor6271]

I am trying to find the accuracy of my circuit which aims to give V(set) as V(reg)-5. To do this, I want to see the effects of both tolerance and temperature simultaneously on the circuit and see the change in output value on stepping these.
To do this, I tried doing worst-case analysis in isolation and temperature analysis in isolation but my issue is that the IC I am using which is ADR1399 has its own tolerance and temperature coefficient which does not seem to be modeled in the application.
So,
a) How can I do both simulations of temperature and tolerance together?
b) Is there any way to model ADR1399 by myself to put these parameters in it or some code that I can put in the schematic?

 

[aliarifat794]

Have you tried proteus to do this simulation?

 

[razor6271]

Have you tried proteus to do this simulation?

I haven't explored Proteus much, so I haven't tried it yet. Can you link me to any resources which can help me to understand how can I simulate this?
Also, I cannot see the ADR1399 IC in proteus

 

[FvM]

It's nothing but an ignorant guess that Proteus can help you in this regard. No surprize that it doesn't.

You have discussed the problem of ADR1399 model quality in your previous thread. It's not completely clear to me which new insights you are seeking now.

Most of heated buried zener reference tolerances is non-deterministic, it can be accessed e.g. in Monte-Carlo simulation, but by nature, a basic SPICE model doesn't cover these kind of tolerances. As far as I see, ADR1399 model represents typical redidual T.C of the thermostat and reference impedance. For worst case analysis, refer to datasheet values and do pencil and paper or Excel sheet calculation.

You should be also aware that long term drift and hysteresis are important parameters for absolute accuracy.

--- Updated Jul 28, 2024 ---

Are you specifically referring to electrical connection of heater and reference circuit with "isolation"? If so, you know that ADR1399 model has errors in this regard as discussed in your previous thread https://www.edaboard.com/threads/adr1399-datasheet.411640 Unfortunately ADI didn't correct it during two years.

Datasheet clarifies that both circuits are sharing a common substrate and has clear requirements regarding ground potentials. Datasheet doesn't specify leakage between substrate and reference circuit or even possible impact of ground shift on accuracy. If your circuit involves intentional ground shift, I would ask ADI.

 

[D.A.(Tony)Stewart]

What do you hope to achieve in analysis not specified in the datasheet? Do a tolerance stack-up. Define your long-term requirements and get your buyer to request ADI for Eng analysis. Then validate with random samples. If this is critical ask someone at NIST for advice. I have already told you how to reduce thermal sensitivity through insulation.

 

[7f6grenade]

It's nothing but an ignorant guess that Proteus can help you in this regard. No surprize that it doesn't.

You have discussed the problem of ADR1399 model quality in your previous thread. It's not completely clear to me which new insights you are seeking now.

Most of heated buried zener reference tolerances is non-deterministic, it can be accessed e.g. in Monte-Carlo simulation, but by nature, a basic SPICE model doesn't cover these kind of tolerances. As far as I see, ADR1399 model represents typical redidual T.C of the thermostat and reference impedance. For worst case analysis, refer to datasheet values and do pencil and paper or Excel sheet calculation.

You should be also aware that long term drift and hysteresis are important parameters for absolute accuracy.

--- Updated Jul 28, 2024 ---

Are you specifically referring to electrical connection of heater and reference circuit with "isolation"? If so, you know that ADR1399 model has errors in this regard as discussed in your previous thread https://www.edaboard.com/threads/adr1399-datasheet.411640 Unfortunately ADI didn't correct it during two years.

Datasheet clarifies that both circuits are sharing a common substrate and has clear requirements regarding ground potentials. Datasheet doesn't specify leakage between substrate and reference circuit or even possible impact of ground shift on accuracy. If your circuit involves intentional ground shift, I would ask ADI.

(This is my other account)
Sorry for being unclear, but what I meant was that I have been trying to compare ADR1399 and LT1236-5 in terms of accuracy as I am using both ICs to get a definite stable voltage output. To compare, I need to compare the accuracy of both systems and since such things are not modeled in LTSpice, I have been using a voltage source replacing these ICs and putting the required temp coefficient and tolerance from datasheet in this source and running my simulations.
The problem I faced is during mc simulation, i stepped both temperature and the run parameter to get effects of both tolerances and temperature on the circuit but the problem is the .meas graph doesn't show for what temperature is the particular colored graph is but also I want to find things like standard deviation, what component is the most sensitive etc. in the circuit which I am unsure of how to proceed to find those.

In the zip I have attached, I want to find the effects on V(reg) and V(set) which would further be helpful for me in the next circuits where I am going to put another voltage source that replicates them with the accuracy and temp coeff i get from this result

 

[D.A.(Tony)Stewart]

"
a) How can I do both simulations of temperature and tolerance together?

You can't.

But you can extract from the datasheets and make a total error budget to compare the IC's to include; temperature 5 ppm/'C vs 1 ppm/'C, or 0.2 ppm/'C tolerance, noise, aging and other sensitivities to error. Be sure to include all assumptions.

b) Is there any way to model ADR1399 by myself to put these parameters in it or some code that I can put in the schematic?"

No. If it were easy, they would have done it.

Then you test samples to validate your specs.