Re: [Discuss-gnuradio] Cheap ADC at 35MHz

[Marius Hauki]

Hi Adrian,

does the chip have a separate input for the
sync tip sample input ? If it has, might it be possible
to hardwire this input to a fixed DC value via a decoupled
potmeter ?


Another idea :
Perhaps it would it be possible to AC couple the baseband input thru a
high pass filter. Then on the ADC side of the low pass filter, you could
insert a DC voltage that emulates the sync values. The filter must be dc 
blocked
towards the outer circuit so it does not barbecue the outer circuits in 
the downconverter.
The penalty with this arrangement is that you would loose some bandwidth 
in the lower part of the baseband.
The band on the low side of the low pass filter will be a "dead band" 
dedicated to the DC input circuit. This DC
input circuit could be a decuopled potmeter with a cap over the center 
and an inductor in series.
The cap is to avoid PSU noise to enter the ADC.
Anyway video is approx 5 MHz wide and I assume the sample rate is at 
least 10Ms / sec or more
so there should be enough bw anyway. If the LPF is sharp it would limit 
the dead band. It should
not be too sharp, becuase it will have large phase deviations close to 
the cut off.
It should be possible to characterize the input complex frequency 
response by doing a
sweep on the input and measure the real and imaginary data. Then one could
implement a equalizer in software to linearize the response in amplitude 
and phase.
To avoid interesting parts of the signal to appear at around zero and in 
the dead band, and thus be lost, the local
oscillator could be set slightly offset.

This is basically the same idea that is used in injecting a DC voltage 
on the coax to feed
a low noise microwave downconverter for satellite TV.

PS : posted the reply to the list since it might have general interest.

Best regards
Marius Hauki
LA9EEA






Adrian Godwin wrote:

> > where did you get access to the driver ? Is this
> > available as open source ? Would it perhaps be possible to


> > bypass the DC restoration circuit if it is external to the
> > chips used  ?
>
> The Bt848 is one of the best supported cheap framegrabbers. There's
> a driver in the Linux kernel, and quite a bit of information around.
> See http://www.metzlerbros.org/bttv.html
>
> I didn't need the driver sources to capture those sample dumps,
> but it would be necessary to change it in order to get continuous
> samples as the vbi input is intended to grab only the data from
> the lines in the vertical blanking interval (i.e. teletext)
>
> The problem with the DC restoration circuit is that it's partially
> insice the chip. Although some parts of the circuit (the AGC, really)
> is routed outside, it appears that the 0 level is defined by sampling
> the value of the data during the sync pulse. I'm not sure if it's
> possible to avoid this.
>
> -adrian