Digital 'scope & spectrum analyzer Re: [Open-graphics] PCIe know-how?

[James Richard Tyrer]

Timothy Normand Miller wrote:
> On 3/4/07, Dieter <netbsd at sopwith.solgatos.com> wrote:
> 
>> Will OGC be able to output arbitrary waveforms, or only video? If
>> OGC can generate sine waves, square waves, triangle, etc. it would
>> be very useful as a piece of test equipment.  It could be a
>> tracking generator for the spectrum analyzer?

> I thought it would be interesting to generate RF signals entirely in 
> the digital domain and use a video DAC to turn it into analog.  For 
> instance, you can have a sine-wave generator, that's completely 
> digital, and it has parameters that you can use to adjust the 
> frequency, within a band.  BOOM.  You have an FM transmitter.  Attach
>  an amplifier, and you can send audio to a nearby radio.

You can purchase such devices as lab equipment -- Arbitrary Wave
Generator -- but they were rather expensive, but it is possible that the
price of the needed technology going down has changed that.

> Say we could do this where we mix a complete TV signal, on channel 
> 3/4, video with color subcarrier, and the audio channel.  Voila.  No 
> more need for a TV chip, and we can get both s-video and encoded.

> If the granularity of our control of the signal is 330Mhz, can we 
> encode all of the information in the TV signal?

The color signal is very messy even with SD-NTSC.  You are going to need
a DSP to compute it in real time.

> Would the steps between digital levels (1024 of them) be too noisy?
> Could we fix that with a low-pass filter?

I don't know how much of an effect quantization noise would have on
color TV.  The output LP filter will not eliminate quantization noise,
it only removes the unwanted higher frequencies generated by using the
digital output.

> I do know, for instance, that IP-over-power-line decoders do 
> everything in the digital domain.

Most current modems are digital as well.

> If you were to transmit the signal on a single band, the amplitude
> would interfere with electronic devices, so they use multiple
> low-power channels instead.  The decoder doesn't use caps and
> inductors to decode each channel, however.  They use a band-pass
> filter that gets all of the channels at once.  Then they digitize it
> and use a taylor series approximation of a FFT to split out the
> channels entirely in the digital domain.  We would be doing this in
> reverse, which is even easier.
> 
Actually, I think that a digital demodulator is easier (then a 
modulator) for complex signals.  For example, with SD-NTSC color TV, you 
have two color signals which are used as x & y to vector modulate 
(amplitude and phase) the color subcarrier which is then added to the AM 
luminance signal.  Except for the quadrature, this is simpler in analog.

-- 
JRT