Noise reported

That's odd - which IC and Spectre sub-versions are you using? 

For me (and I don't recall this being wrong in living memory) it plots in V/sqrt(Hz) (a perfectly reasonable unit for noise as that's often used by analog designers). It looks as if the axes are just labelled wrongly in your case (although the magnitude is slightly off too - in the calculator if you enter sqrt(boltzmann*4*300.152*1k) it will give 4.071E-9 which matches what I see - maybe your temperature is not 27C?)

I'm using IC6.1.8 ISR18 and SPECTRE19.1 ISR18 in this case (I probably should have switched to SPECTRE20.1 but I forgot to change it). I get the same whether I use the obsolete ADE L, the newer ADE Explorer, or even if I ran standalone from a handwritten spectre netlist and used ViVA to plot the results from the results browser.

Andrew

The IC version

And the spectre version

The temperature is set to 300K/27C

And regarding Shawn's question,

I just plot it from the results browser in VIVA.

I've just repeated my tests using:

Virtuoso sub-version IC6.1.8-64b.500.9

Spectre sub-version 18.1.0.464.isr11 (@(#)$CDS: spectre version 18.1.0 32bit 09/21/2019 21:29 (csvcm20c-2) $)

so identical versions to you. I get the plot exactly as I shared it above - i.e. working fine.

Please can you post the contents of your input.scs? (so I can see the netlist, temperature, analysis setup and so on). Please post it as text rather than as a screen shot (makes it easier to search, copy and paste and so on).

I just noticed that in the results browser you have a transistor and a resistor - so that circuit is not the same as you described?

Also, I see you have "outsrc" showing, which suggests to me that you've enabled the "noise separation" option. Ah, if I turn that on, the values are still correct for me (so 4.071nV/sqrt(Hz) but the y-axis units are wrong...). This is a bug. I'll check back with the latest version to see if that bug has been resolved.

Andrew.

It's broken still (with noiseseparation turned on) in the latest Spectre version (20.1 ISR5); the iC version isn't relevant here.

There's an existing CCR, 2074763, with a couple of duplicates reporting this.

For now, if you want the units to be correct, run with noise separation turned off.

Regards,

Andrew

Cool, thanks for helping!

Cool, thanks for helping!

Dear Andrew,

Andrew Beckett said:

There's an existing CCR, 2074763, with a couple of duplicates reporting this.

For now, if you want the units to be correct, run with noise separation turned

Thank you - again for both your time and your diligence!! You satisfied my curiousity too as I was unable to duplicate greywander's anomalous result.

Shawn

Good news: having pushed the CCR a little over the last couple of days, R&D have now fixed the issue and the correct units should be on the y-axis in SPECTRE 20.1 ISR8 (due out before mid-July). Note, I've not tested the fix yet, but I will do as soon as the build is ready.

Regards,

Andrew.