The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> Wrong behavior of dcblock and dcfeed in Cadence Spectre
https://designers-guide.org/forum/YaBB.pl?num=1518774343

Message started by cheap_salary on Feb 16^th, 2018, 1:45am

Title: Wrong behavior of dcblock and dcfeed in Cadence Spectre
Post by cheap_salary on Feb 16^th, 2018, 1:45am

"dcblock" and "dcfeed" are very common in RF simulator such as Keysight ADS and Keysight Goldengate.

They are available even in Cadence Spectre after version 14 or 11.

"dcblock" has to be "perfect short" except for DC and Tran Analyses.
"dcfeed" has to be "perfect open" except for DC and Tran Analyses.

These are implemented correctly for AC and PSS Analyses of Cadence Spectre.

But they are wrong for PAC Analysis of Cadence Spectre.

"dcblock" is treated as "open" for sideband=0 in PAC Analysis.
"dcfeed" is treated as "short" for sideband=0 in PAC Analysis.

These are wrong.

These are true for all slave small signal analyses subjected to master large signal analyses.

Behavior of "dcblock" and "dcfeed" are completely wrong in PAC, PXF, PNoise, PSP, PSTB, QPAC, QPXF, QPNoise, QPSP, HBAC, HBXF, HBNoise and HBSP.

My Spectre's Version is "17.1.0.124".

Title: Re: Wrong behavior of dcblock and dcfeed in Cadence Spectre
Post by Andrew Beckett on Feb 18^th, 2018, 12:35am

I tried this (in 16.1.0.567.isr12 and 17.1.0.160.isr2) and I see what you're seeing. I used this circuit:

Code:

V1 (in1 0) vsource type=sine freq=1M ampl=1 pacmag=1 dc=0.5
Iblock (in1 opblock) dcblock
Ifeed (in1 opfeed) dcfeed
R1 (opblock 0) resistor r=1k
R2 (opfeed 0) resistor r=1k

pss pss fund=1M harmonicbalance=yes harms=5
pac pac start=1k stop=100k

The opblock and opfeed responses appear to be the wrong way around for the pac analysis results.

This won't get fixed by stating it in this forum - please report it to Cadence Support.

Regards,

Andrew

Title: Re: Wrong behavior of dcblock and dcfeed in Cadence Spectre
Post by Frank Wiedmann on Feb 19^th, 2018, 2:37am

By the way, all of this only applies to harmonic balance (like in Andrew's example); these components are explicitly not supported for shooting pss (see the manual at https://support.cadence.com/apex/techpubDocViewerPage?path=spectreRF/spectreRF16.1/Chap8.html#frequency_dependent_components). I don't really understand why, as they should be able to use the same behavior as for a tran analysis. A corresponding enhancement request (CCR 1089307) that I filed more than 5 years ago is still in "Inactive" state.

Title: Re: Wrong behavior of dcblock and dcfeed in Cadence Spectre
Post by Andrew Beckett on Feb 19^th, 2018, 3:50am

Hi Frank,

I agree, it would be sensible to just implement them as large inductors or capacitors as with transient. I think it's just a matter of priorities meaning that more urgent things got done instead. The CCR filed on your behalf was filed with a low severity because the workaround is of course just to use a large inductor or capacitor; really the only point of using dcblock/dcfeed is if you're going to be using harmonic balance. If you're not going to use hb, then there's not that much benefit in using them in the first place.

Regards,

Andrew