Designer’s Guide Community

New Member

Offline

Posts: 8

power supply & substrate noise
Jun 03^rd, 2004, 11:28pm

Hi everyone,

Who know the scale of power supply & substrate noise in mixed signal IC? Assume the power supply is 5V and the circuit is MCU. I plan to simulate the CMRR capability of differential circuit, hence add some noise sources is necessary, I think.

I have heard that power supply & substrate noise belongs low frequency noise. I want to know the range of low frequency noise, and is 10Hz~1MHz resonable? How about the amplitude of the noise, maybe 50mv?

If I know the frequency range & amplitude of power supply noise, then I can generate the noise signal by Matlab, which can be as a PWL noise source added to the circuit. Finally, running transient analysis can observe the CMRR capability.

Thanks & best regards,
Vincent

Back to top

try my best...

IP Logged

ywguo

Community Fellow

Offline

Posts: 943
Shanghai, PRC

Re: power supply & substrate noise
Reply #1 - Jun 10^th, 2004, 10:02pm

Hello, Vincent,

I think the substrate and power noises depend on your circuit, package, and PCB. You'd better search a few papers about substrate noise on IEEExplore.

Generally, the issue is divided into two faces. One is the generation of switching noise, ther other is the shielding of the sensitive circuits.

If you want to simulate with substrate noise, please add the bonding wire model in your circuit, and put a small serial resistor, say 1 or 2 Ohm, in your voltage source. I though that is much easier and more accurate than PWL source generated using MATLAb. Obviously, the switching noise on substrate and power is not a PWL source, but often a damping sinusoidal source.

By the way, CMRR is a AC parameter, I don't think you can get it from a transient simulation.

I hope it is helpful to you.

Best regards,

Yawei Guo

Back to top

IP Logged

modern_analog

Junior Member

Offline

Posts: 12

Re: power supply & substrate noise
Reply #2 - Jun 16^th, 2004, 11:31am

Its not entirely true that CMRR is not material in this case.

I think Vincent wants to apply noise on the supplies and see its effect on the output.

Vincent, remember to skew the differential resistors. If you don't do that your output will not change when noise is added, ie mismatch is the only thing that translates a perfectly common-mode signal into a differential output, thus giving a finite CMRR.

Back to top

IP Logged