Just out of curiosity, how many PVR 500s will work in one box, with a regular KnoppMyth install?

I'm thinking of doing 3 or 4 of them in one box. Is this possible?

-Jeff

How many PCI slot does your motherboard have? How big is your power supply?

_________________
cesman

When the source is open, the possibilities are endless!

I theory, 6. One for each PCI slot on a standard motherboard.

Although, I've not heard of anyone trying it. Go on, be the first, you know you wanna.

_________________
Mephi
-----------
R5.5
Opteron 2.Ghz
1Gb Ram, 200Gb Maxtor PATA
Avermedia 771 DVB-T
Hauppauge Nova-S2 (waiting to be installed)
MS MCE2 Remote

Actually the ivtv driver have a limit of the number on video sources it can use. And each pvr-500 coulnt as two video sources. I beleave it is limited to 16 video sources (eight pvr-500's).

video0-video15 are video sources
video16-video31 is for the pvr-350's decompressor
video32-viode63 are for accessing the compression engine

Ok so it sounds like 3 cards, 6 inputs, in one box would be reasonable. Thanks guys.

-Jeff

Just watch the number of splits on your cable. If you're feeding anything else on that line you may have to invest in a amp to keep the signal quality acceptable. The boundary is usually around 5-6 devices using a single n-way splitter. Don't daisy chain splitters as the loss is then cummulative.

TJC is correct about the splitting. A normal 2-way splitter effectively splits the original signal in half, so a 15db signal split 2 ways becomes ~7.3db on each connector. Now, if you're chaining splitters, another 2-way splitter on the first one would take the incoming 7.3db and halve it to ~3.6db.

In this scenario, if you had used a 4-way splitter originally, you'd have 15db in and ~3.75db coming out each port. So regardless of whether or not you're chaining, you incur the same loss across the board *if* your splitters are being put to full usage. Never, ever use a 3 headed splitter where a 2 headed splitter will suffice.

Keep in mind that alot of signals are _very_ tolerant. The sync range for the average DOCSIS-compliant cable modem is anywhere from -15 to +15db. An average install, with a line coming straight from a 2-way splitter on the side of your house straight into the cable modem, yields about +4db on that one line. If it's good enough for your cable modem, it's good enough for your digital set-top tuner, so don't be too paranoid about the signal strength unless your low analog channels look bad.

Ummm... I don't think that is quite right.

If I Have +6 coming off my drop cable and I add a 2-way splitter that loses 3.5db on each leg, then I Have 2.5db on each leg. If I use a 3 way splitter that is balanced, normally I'd lose 6db on each leg, bringing my level down to 0. If I added a 2way and then a 3 way I'd lose 9.5db. I don't halve my source signal, but I do introduce the loss that the splitter is rated for (unless the splitter goes bad, which does happen).

-Jeff

Actually if you look at the (logarithmic) maths of dB, splitting signals (two ways) does half the power, but this is represented as a ~3db loss (plus whatever losses are caused by the box/connectors ~1db). So a 15db signal would be reduced to 11-12db from a two port splitter. dB are not an absolute measurement of power, it is relative to a standard. So negative values represent lower power levels than the standard (1mW for dBm, 1W for dBW). So a 1db signal passed into a splitter will come out as -2-3db.

Some points of reference:

3dB = x2
6db = x4
10db = x10
20db = x100
30db = x1000
0dBm = 1mW
0dBW = 1W

The reason dB are used instead of Watts is that with dB you add and subtract when in fact you are multiplying and dividing the absolute power.

Here endeth the lecture

_________________
HP VL400 (PIII 866), Skystar2 2.6D, PVR350, Nvidia FX5200, 384MB, 200GB, KnoppMyth R5.5

I wasn't going to go into the complications of asymetric splitters (taps), or the math involved in cummulative loss from cascading splitters, or how to figure the loss per port on an N-way splitter, or the complications introduced by the connector losses, or ... There are radio engineering handbooks for that kind of thing.

I figured simple rules like, "don't cascade", "use screw on connectors, and make sure they're on tight", and "anything over 5 or 6 devices start thinking about an amp" are easier to remember.

Yeah sorry, kind of went overboard on that one. Thats the trouble when I don't know much about linux in general then suddenly hit something I know Getting a bit OT too.

_________________
HP VL400 (PIII 866), Skystar2 2.6D, PVR350, Nvidia FX5200, 384MB, 200GB, KnoppMyth R5.5

Good reading. Also note that the real world measurement for a cable RF signal is microvolts, but dB is used to keep things simple. It was borrowed from early television engineers and used as a baseline for signal quality measurement.

One more thing to note that besides using splitters, long drops (lengths) of cable will also introduce signal loss, particularly rg50 cable (the thin spaghetti kind). RG60 quad-shielded is commonly used these days, and the kind they bury in your backyard or string across from the pole to your house has even more protection (namely a greasy substance inside).

Well, in the 'olden' days, it was refered to as dbm or db above a milliwatt!