LTSpice Wrong result INA253A2

Thread Starter

waulu

Joined Dec 23, 2016
59
Hello,

I am doing a simple simulation with the INA253A2, a voltage-output, current sense amplifier,

https://www.ti.com/lit/gpn/ina253

I just want to check the gain, so I plot the output divided by the current measured,

Capture.PNG
I was expecting 200 mV/A, but with the simulation, I obtain 300 mV/A. I tried the same thing with PSpice TI and I got the right result.

Is it possible to get the right result on LTSpice?

Best regards
 

Attachments

Last edited:

ericgibbs

Joined Jan 29, 2010
12,258
hi w,
This is what I get at the Inp and Inn inputs for 1Amp shunt current, ie: 3mV, should be 2mV.
The internal OPA gain lifts this to 3Vout.

There seems to be a problem in the 'shunt' section of the .lib/net

E

Update:
Tried in LTSpice INA253A1, A2 A3, they all give unexpected results.???
 

Attachments

Last edited:

Thread Starter

waulu

Joined Dec 23, 2016
59
Hi ericgibbs,

Yes, all the results are unexpected with the three models. But all of them work on PSpice TI, thats why I am not considering that there's a problem with the model.

I know nothing about spice models but I may try and take a look at it during the weekend.

Regards
 

ericgibbs

Joined Jan 29, 2010
12,258
Hi waulu.
Over the weekend I will try to compare the three NET files, it may highlight the problem, let you know if I find the problem.
E
 

ericgibbs

Joined Jan 29, 2010
12,258
hi waulu,
This is an edited version of the INA253A2.net [ change the txt extension back to net, after downloading]
Lines #72 and #77 of the Net listing set the internal gain,

I have created a sim showing all 3 versions A1,A2 & A3, the A2 plot is the edited plot for 200mV/A.

E
Update:
Added plot of all versions A1,A2 & A3

AAA 781 10.41.gif
 

Attachments

Last edited:

Thread Starter

waulu

Joined Dec 23, 2016
59
Hi ericgibbs,

Thank you very much. Yes, I tested your model and it's good. Today I don't have the time but tomorrow night or during the weekend I will take a look at what you changed.

Regards,
 

Thread Starter

waulu

Joined Dec 23, 2016
59
Hi ericgibbs,

Yes, the step limit was part of the random experiments, I forgot to remove it. Thank you.

Just for curiosity, I tried the model on PSpice TI and your model changed the gain in PSpice Ti too.

But anyway, I am planning to use LTSpice. Can you explain to me how to debug spice models? This is one as an example. How did you realise that you had to change those two lines?

Regards,
 

ericgibbs

Joined Jan 29, 2010
12,258
hi waulu.
On this occasion we had three versions of the INA253 Net listings A1, A2 & A3, with their different gain settings.
The very simplistic approach was to open all three Nets, side by side in a text Editor.

Then visually flick from one net list to the next, the differences in the text was quickly visually observed as the different text 'flickered'

Using this method I worked down thru all the pages of the listing.

Noting value of the resistor value changes for a given known Gain of 100,200 & 400, it was easy to calculation the required resistor value change.

Obviously this method is limited.

E
BTW:
I ran sims on all three versions, using single and dual power supplies and pulsed/sine current loads in order to check their overall performance.
 

Thread Starter

waulu

Joined Dec 23, 2016
59
Hi ericgibbs,

Then visually flick from one net list to the next, the differences in the text was quickly visually observed as the different text 'flickered'
Excellent idea.

I ran sims on all three versions, using single and dual power supplies and pulsed/sine current loads in order to check their overall performance.
Incredible, you are amazing. Thank you very much for the effort.

hi w,
These a the test sims for three types.
E
Once again, I appreciate all the effort you had with this, thank you.

Regards,
 
Top