a problemn of ADC simulation with Hspice.

[gdhp]

hspice adc snr

Hi all

I use the HSPICE to simulate the whole pipeline adc.

I want to check its SNR. So i run the tran. simulation with 2048 points, then transfer the .tr0 file to matlab to fft to get the snr value.

If i use 0.01ns transient step in tran simulation, the fft results is about more 60db (My adc is 10-bit)

If i use 0.1ns transient step in tran simulation, the fft results is about 52db.

Both simulations use the same setup, its options are also same.

So can someone give me some suggestions about the phenomenon? Which result is more reliable?

Thanks
GDHP

 

[JoannesPaulus]

adc hspice

You are experiencing some "numerical noise" due to the interpolation between two simulation steps, i.e. your 2048 points do not always fall on one of the points calculated by the simulator. So, the 0.01ns simulation is probably the correct one.

 

[gdhp]

Thanks JoannesPaulus for your reply.
Can you explain the "numerical noise"?
From our test results, the 0.1ns step simulation results are
more near to the test results.

Anyone has the experience to simulate the SNR of ppadc?

thanks

 

[JoannesPaulus]

Sorry for the delay.
The fact that your test results are closer to the 52dB might mean that you are experiencing thermal noise or maybe jitter in your measurements.

Numerical noise is due to the accuracy that your simulator uses to resolve the nodal voltage in your netlist and is not a real effect - you just have to be aware of it so that you can choose the correct accuracy.