The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> Question about 'clans' in DStoolbox
https://designers-guide.org/forum/YaBB.pl?num=1282824340

Message started by moore on Aug 26^th, 2010, 5:05am

Title: Question about 'clans' in DStoolbox
Post by moore on Aug 26^th, 2010, 5:05am

ntf = clans(order, OSR,Q,rmax,opt)
I don't know exactly what do Q and rmax mean?

It is explained that Q=|h|-1. For 1 bit DSM, |h|<1.5 is a rule of thumb. If I just set |h|=1.5, does it mean that Q here is filled with 0.5?

Additionally, rmax is explained as the max radius for NTF poles. What performance does it relate to? Any rule of thumb? Because in the DStoolbox user guide, it is set to be 0.95, I guess it is a pretty large one?

Many thanks!

Title: Re: Question about 'clans' in DStoolbox
Post by carlgrace on Aug 26^th, 2010, 4:20pm

rmax relates to the stability of the modulator. To be stable, all the poles of the modulator need to be inside the unit circle, but as you get closer to the unit circuit the phase margin of the loop degrades. So, the trade off is between stability and performance (by performance I mean how well the loop filter rejects out-of-band noise). It is set at 0.95 to give a lower bound on the phase margin.

Carl

Title: Re: Question about 'clans' in DStoolbox
Post by moore on Aug 27^th, 2010, 6:09am

Thanks. But how about Q?

Title: Re: Question about 'clans' in DStoolbox
Post by carlgrace on Aug 27^th, 2010, 11:37am

Hi Moore,

Q in this context is the max noise gain. You are confusing ||h||1 with h_inf in your question. ||h||1 is all the impulse response values added up, and h_inf is the max out-of-bound gain of the NTF. So, you shouldn't use Q=0.5 since that would not give enough LSBs for feeding back quantization noise. You should start with 5 (as the documentation suggests) and vary it a bit, noting what happens with the SNR. I imagine it will fail with Q=0.5, but I haven't got access to MATLAB right now to try it out.

Good luck.

Title: Re: Question about 'clans' in DStoolbox
Post by moore on Aug 27^th, 2010, 4:59pm

Hi Carl:

Thanks for your reply.
I am still confused about clans.
It is illustrated in the document that "Q is the max number of quantization levels".
Does it mean that if I use 9 level quantization, Q will be 9?

Additionally, are you familiar with CT DSM?
If I have got [a,g,b,c] for a certain structrue of DT DSM, how can I fit CT's corresponding coefficient based on DT coefficient?

Many thanks!