[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Discuss-gnuradio] Status of BTTV/CX88 capture

From: Hew How Chee
Subject: Re: [Discuss-gnuradio] Status of BTTV/CX88 capture
Date: Mon, 22 Aug 2005 08:44:44 -0700 (PDT)

Hi Martin,

> Did you copy the risc code from another driver or
> did you specially make it for this driver?

It's a new risc code I guess. I did not really look at
the risc code for cx88 driver to see whether it is the
same and I doubt it is the same. I just been through
the code, notice there are garbage/redundant old code
inside. The function that create the current risc code
is in 'make_risc_instructions'. Other function that
generate risc code are either experimental or does not
work yet. The tough part is setting the registers
needed to work with the risc code. That took most of
the time. Also PCI latency timer need to be adjusted
so that it would not overflow, if I remembered

What the risc code does it something like this.

Total DMA buffer is 32 Mbyte

b) write 2kbyte of data to DMA first page address
c) write 2kbyte of data to DMA first page address +
   increment counter (to keep track of circular
   generate IRQ (after a couple of write)
d) do b) and c) for for each DMA page
   if it is the last page of the DMA buffer,
   reset counter 

e) jmp to b) (so that it write again to beginnnig
   to DMA buffer)

Total memory needed to risc instruction is 128+ 
kbytes. This cannot fit in the internal RAM and
therefore put in PC memory.

I was trying to rewrite the risc code so that it 
does not occupy consecutive 128+ kbyte of memory since
memory allocation sometimes fail. After noticing noise
in the sample, I did no further testing. Probably
next, after listening to what you did and when I have
the time, I will try to take the IF from inside the
tuner. This wil be interesting.


--- Martin Dvh <address@hidden> wrote:

> Hi Hew How Chee,
> Hew How Chee wrote:
> > Hi Martin,
> > 
> > Pretty good hacking you have done there. Maybe
> I'll
> > try it one day, after the warranty expires :)  
> > 
> >>I think I am going to look for a nice cx2388x
> >>capture card.
> >>What card are you using ?
> >>What tuner is on it? (Is it easy to hack?) 
> > 
> > I am using Pixelview PlayTV Pro Ultra card. It has
> a
> > cx23881 chipset on it. It has also a different
> tuner
> > from my bt878 tv card (which is similiar to the
> > Philips FI1216 tuner module). Never try hacking it
> > since the main concern is to get continuous adc
> data.
> > Pinout is different also, 
> > 
> > I have uploaded the source code  and a few more
> > capture.
> Thanks,
> Did you copy the risc code from another driver or
> did you specially make it for this driver?
> I have been strugling with the risc code in the
> bt878 driver to get all samples.
> (It drops samples which it thinks are illegal
> because they are marked as sync)
> Or did you copy a driver and just add the RAW bit to
> the configuration?
> > You might want to look at the noise I talk
> > about. The noise might be cause by the PCI burst
> > during DMA. I have never tried capture more than
> 32
> > Mbyte of data (due to my PC setup), perhaps others
> > can try out the code first before you invest in a
> > card. Even so, you might want to look at other
> cards
> > since I feel that the ADC power section of my card
> is
> > not clean and cause all this noise.
> My bt878 also has some noise and signals which
> shouldn't be there.
> Your signal actually looks cleaner then mine.
> Some of the noise is just short sparks.
> They go away when you use a (lowpass) fir filter to
> reduce the bandwidth.
> (Even if you just reduce the bandwidth from 14 Mhz
> to 8 Mhz you get rid of a lot of noise)
> My experience is that a lot of pci-cards are not
> well designed for analog signal quality.
> (TV-capture cards, audio cards, graphic cards)
> Sometimes you can increase the signal quality by
> placing good capacitors at the Vcc right next to the
> (ADC/DAC) chips.
> Greetings,
> Martin

Start your day with Yahoo! - make it your home page 

reply via email to

[Prev in Thread] Current Thread [Next in Thread]