On Wed, 2010-02-17 at 08:46 -0500, Kyle Lil wrote: > > > > From: [email protected] > > To: [email protected] > > Date: Mon, 15 Feb 2010 19:14:00 -0500 > > Subject: Re: [ivtv-users] hvr-1600, frame CRC mismatch - incomplete > frame > > > > On Mon, 2010-02-15 at 16:22 -0500, Kyle Lil wrote: > > > > > Previously I'd had the tuner connected at the end of additional > > > splitters and devices, but I shortened the chain in > troubleshooting. > > > I'll try attenuating the signal again to see if that helps. I > hadn't > > > consider the possibility that the cable signal might exceed the > > > devices input limits.
> > I think you'll just need to attenuate your cable signal before it > hits > > the HVR-1600. > > > > > I built a -10dB attenuator as you described in your link (thanks for > that) and still saw the glitches. In fact, I see them whether I > connect the cable line straight from the wall into the HVR-1600 or > through two splitters and the attenuator (and every other > combination). In trying to get a better sense of what's happening to > the signal, My hypothesis is the HVR-1600 digital tuner and demodulator AGC circuits cannot attenuate the signal enough to avoid occasional clipping of the analog signal. The clips show up as uncorrectable blocks. > I went into the service menu of my cable box, a Motorola DCH-3200, > which lists various signal statistics. One of the parameters it lists > is "AGC", which I'm guessing is some sort of automatic gain control > that puts the signal into the optimal range for the tuner. Yes. > If the cable box is connected after the second splitter, AGC is 18%. > If it's connected to the first splitter, it's a few percent lower. 18% is on the low side for an AGC to be operating. That indicates that you have a strong signal. > If I add the attenuator, the AGC goes up another 5%. It makes sense > to me that the attenuator would cause the AGC value to go up, but I'm > sure why it only went up 5% with a -10dB attenuator. Well, without knowing 5% of "what" (AGC full scale voltage?), it doesn't really matter as long as your attenuator appears to be attenuating the signal. > > Does the HVR-1600 have similar AGC built in? Yes, there should be two: an AGC in the MXL5005s and an AGC in the CX24227. The MXL5005s will be used for AGC until the AGC Take Over Point (TOP) setting in the CX24227 is reached, and then the CX24227 will take over with it's AGC. The idea is to amplify very weak signals as best you can while getting the best possible Noise Figure with the MXL5005s; or if the signal is strong enough, use the CX24227 AGC to get maximum signal while avoiding clipping. In my hypothesis, the "glitches" are caused by occasional clipping of the analog signal in either the MXL5005s or the CX24227. > If the cable box cuts the signal to less than 1/5th of its amplitude > (even after two splitters), No, that's likely 5% of AGC full scale. You don't have enough details to know what AGC percentage level corresponds to what gain and if it is expressed as a percantage for linear units or logarithmic (dB) units. FYI: -14 dB is 1/5 of signal amplitude, -7 dB is 1/5 of signal power. A 1-to-2 splitter will drop your signal power by more than 3 dB. > maybe this is an indication that I need to build an even stronger > attenuator? Do you have any insight into what value attenuator I > should be shooting for? Attenuate by 20, 30, or 40 dB I guess. You'll find one or more of: a. The glitches stop. (Good attenuation value to use.) b. The signal is so weak you can't lock the stations anymore. (Too much attenuation) c. My hypothesis is wrong. :P As an experiment, I can look at trying to adjust the CX24227's AGC TOP to kick in earlier, at a much lower threshold, in an attempt to prevent the MXL5005s from trying to do any amplification at all. However, a series of tests with external (cheap home-made) attenuators would be more deterministic in terms of experimental results. The experimental trials are bounded by an attenuator value so large that the signal cannot be tuned anymore. Regards, Andy _______________________________________________ ivtv-users mailing list [email protected] http://ivtvdriver.org/mailman/listinfo/ivtv-users
