[music-dsp] Video FIR filters
bogdan at rinda.com
Tue Jan 13 13:24:02 EST 2004
Have you looked at 3Dlabs Wildcat Pro series video cards?
----- Original Message -----
From: "David Olofson" <david at olofson.net>
To: <music-dsp at shoko.calarts.edu>
Sent: Tuesday, January 13, 2004 10:47 AM
Subject: [music-dsp] Video FIR filters
> (Guess I should be asking on a list named "video-dsp" or something,
> but maybe someone around here has some ideas...)
> I'm using an Eizo F980 21" monitor, and (as expected) I'm having
> serious trouble finding video cards that can drive it in it's higest
> resolutions. 400 MHz RAMDACs are getting common, but most of the
> cards have mediocre output stages with crappy filters and
> insufficient amplifiers.
> For example, the ATI FireGL 8800 card I'm using now can't drive a 1.5
> m cable without the signal bouncing back and forth once, causing a
> slight ghost image effect on the screen. The old Matrox G400 MAX had
> no such problems, but it doesn't have much of a 3D accelerator by
> today's measures, and it can't do 2048x1536 at a usable refresh rate.
> Still no Linux OpenGL drivers for the Matrox Parhelia, so that's out.
> I don't know if nVidia finally learned to build RAMDACs, or if the new
> 400 MHz spec on the latest GeForce FX chips is just marketting bogus
> as usual. Anyone tried them on a serious monitor?
> Anyway, I had this idea today:
> What I'm having trouble with is a totally stable echo phenomenon in
> the cable. (Sharpness is actually pretty good with the FireGL 8800,
> though that's part because of an adjustable filter in the monitor's
> DSP.) So I figured one could throw in a FIR filter that negates the
> response of the RAMDAC/cable/monitor system. One would have to tweak
> the filter manually until the result is acceptable.
> It's just that we're talking about 300-400 MHz, three channels and 10
> bit dynamic range. My first idea was to use a DSP, but where would I
> found one that's anywere near the kind of performance I need? It
> would have to be heavily SIMD and/or have a clock frequency in the
> multiple GHz range.
> How about FIR filter chips? Seen some, but they're usually all digital
> (not S&H based) and 3-5 coefficients or so.
> I need FIR filters with a sample rate of at least 400 MHz, >60 dB SNR
> and at least 8 coefficients. They should preferably be programmable
> during operation, for easy tweaking. If it can be done via I2C, USB,
> RS-232, RS-485 or something, even better! A ton of trim pots would be
> ok, though.
> Are there any suitable chips for something like this?
> Would it be reasonable to build filters for this frequency range using
> CMOS switches, op-amps etc? If I'm thinking straight, each filter
> would be a chain of 8+ "double" S&H stages, each with a buffered
> output with gain control, all connected to a mixer that produces the
> filtered output. The S&H stages would be driven in parallel with a
> 400 MHz oscillator. (Which is why they must be "double" - or you
> wouldn't have much control over what happens when you open all of
> them at once. Alternatively, I could sweep backwards, but then I'd
> need some control logic and another S&H on the output, I think.)
> Ok, I know I'm nuts :-) but this monitor has insane performance, and
> it annoys me that nothing on the market seems to be able to drive it
> properly - and this is about three years after I bought it!
> //David Olofson - Programmer, Composer, Open Source Advocate
> .- Audiality -----------------------------------------------.
> | Free/Open Source audio engine for games and multimedia. |
> | MIDI, modular synthesis, real time effects, scripting,... |
> `-----------------------------------> http://audiality.org -'
> --- http://olofson.net --- http://www.reologica.se ---
> dupswapdrop -- the music-dsp mailing list and website:
> subscription info, FAQ, source code archive, list archive, book reviews,
More information about the music-dsp