Results-MonteCarlo simulation

[riccart07]

Hi all,

I'm trying to get the histogram of the offset of my comparator.
Does anyone could tell me why the plot in attachement just stay on the positive side ?

Thanks

Riccart

 

[keith1200rs]

Because you have a systematic offset? What offset do you have from a 'typical' simulation?

Keith.

 

[riccart07]

First of all, thanks Keith.

My typical offset is about 410 uV. In attachment, you will find out how I got it.
I don't have a systematic offset if I am referring on the definition of a systematic offset.

Hugues

 

[keith1200rs]

I think the scale of your typical results is too coarse to show if there is a systematic offset. The histogram suggests there is a systematic offset of 400uV but you would not be able to see that on the typical graph you have shown - less than the thickness of the graph line. Try expanding that graph to see if there is a small offset (I assume there is).

Also, I would suggest you use a DC analysis for checking the offset, not a transient analysis, so you can separate out dynamic behavior from static behavior.

Keith.

 

[riccart07]

The Graph in attachment is most clear.
I addition, the comparator works with two clocks so I can't do the DC sweep.
To compensate this situation, I am using a clock at 20 Mhz and the tr and tf of the input signal is slower.

Do you have any idea how I can measure this offset ?

 

[keith1200rs]

OK, I see why you are doing a transient analysis. Your systematic offset is clear from that attachment BUT it raises another issue. You are sweeping your input signal at 50mV/us (differential). With a 20MHz clock the input will have moved 2.5mV between clock pulses - nowhere near slow enough to identify an offset as low as 400uV. You need to slow down the input sweep - just sweep from +1mV to -1mV over 10us, for example. Then reduce it further to make sure the results don't change.

Keith

 

[riccart07]

Thanks, It's works now.

Hugues

 

[keith1200rs]

 

[riccart07]

Results-MonteCarlo simulation - Plots

Hi Keith,

In attached, there is a graph of my Monte Carlo simulation. Please, let me know if the result is normal because it's very different to the habitual form.
P.S. It's for 200 runs

Thanks

Hugues

 

[keith1200rs]

Hugues,

Part of the problem is you have an "outlier" around 100uV which has increased the bin size so you only have 5 useful bins. So, it could be a skewed distribution but I am not sure you can be certain with 5 bins.

I think you can do one of three things:

1. find out why you have the outlier and fix it or

2. do a run of maybe 1000 to try and get more data in the -20 to +40uV area (I am not certain that will actually have the desired effect though).

3. manually set the bins to have more bins in the -20 to +40uV range.

Keith.