The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> filter tuning circuit won't start
https://designers-guide.org/forum/YaBB.pl?num=1182241910

Message started by aaron_do on Jun 19^th, 2007, 1:31am

Title: filter tuning circuit won't start
Post by aaron_do on Jun 19^th, 2007, 1:31am

Hi all,

i'm designing a tuning circuit for my bandpass filter. I have a reference frequency which i feed into a replica of the filter, Vref. I take the output, Vout, and use a PFD (phase frequency detector) to compare its phase to Vref. Then I have a charge pump and a loop filter. I'm simulated extensively in the open loop, and it seems the loop should work. Unfortunately i noticed that if i input a pulse into the control voltage of the filter, the filter takes about 2 us to start up. I figure the filter only works around the bias point of about 600 mV. i.e. from say 550 mV to 650 mV. So i reduced the loop bandwidth to around 10 kHz and it seemed to work. The frequency was offset by about 60 kHz for the 2 MHz filter. Does this result seem correct?

Any better ideas on a tuning circuit?

thanks,
aaron

Title: Re: filter tuning circuit won't start
Post by ACWWong on Jun 20^th, 2007, 3:50am

do you need continuous tuning ? you should only need this if your filter varies greatly with temperature and voltage. If your design is using say MIM cap and polyR/extR/bandgapI/constant_gm_bias_vs_temp then this should not be the case. Doing a one-off hit to calibrate/tune out process variation makes more sense when it comes to power. This one-off hit could be only at power up/maunfacture or every receive burst depending on your system needs.

anyway making your bandpass filter (whole thing or replica or a resonant component part) into an osillator is the usual way I have done it. then compare the frequency of oscillation to a known tone (from crystal), will allow you trim/tune your tuneable component of your filter transfer function, be it gm, R, or C. Anyway the higher Q of your osillation generation, the less residual error you will get.

your phase comparison approach yields 60kHz in 2MHz... which seems quite a bit of an error for tuning scheme which should correct to near zero.

Title: Re: filter tuning circuit won't start
Post by aaron_do on Jun 20^th, 2007, 4:25am

thanks. You method also sounds good. I've reduced the error through matching and i hope to improve it further by improving the charge pump. I've never worked with PLLs, but the PFD and charge pump that i'm using seems to have about 0.5% steady state phase error. Is this normal? BTW my frequency in controlled by a control voltage. How would I store this value if? Would it have to be digital?

thanks,
Aaron