Is it technically possible to record the output of the tuner, either as an mp3 or as a wav?

It seems to me it should be, as the audio from the tuner passes through the DSP anyway, so all one had to do is write a piece of software that takes the audio there and puts it onto the harddisk. What am I forgetting?

What am I forgetting?

That the DSP audio system is output only, ie the SA1100 has no way of reading digital audio data
that it did not originate. From the tuner, for example...

Patrick.

Opinions expressed in this email may contain up to 42% water by weight, and are mine. All mine.

Somewhere burried in the programming fourm, Hugo responded to a request that I had about recording radio. Here is what he said in response to
Is the inability to record from the tuner due to the hard drive being mounted ro at the time? Or is it due to a hardware limatition, or other factor?
Hugo:
Mostly due to incompatible sampling rates, which means the sample clock can't be used. The radio runs at 38kHz.

---

Reg 1640, 6gb Blue empeg-car mark 1. SN unknown

That the DSP audio system is output only, ie the SA1100 has no way of reading digital audio data
that it did not originate. From the tuner, for example...

Er, what? The SA1100 can send digital audio to the DSP, but nor receive digital audio from the DSP? I'm afraid I do not fully understand this. By the way, what DSP is used?

Is the inability to record from the tuner due to the hard drive being mounted ro at the time? Or is it due to a hardware limatition, or
other factor?
Hugo:
Mostly due to incompatible sampling rates, which means the sample clock can't be used. The radio runs at 38kHz.

And again: er, what? Where does this (completely nonstandard) 38kHz sampling rate come from? Is it perhaps possible that Hugo confused that with 32kHz, which would be understandable, as there's nothing usable above 16kHz on the radio. What ADC is used? Where does the ADC send it's digital audio, the SA1100 or the DSP? And, lastly, so what? I'm completely satisfied if I can store 32kHz, or even 38kHz digital audio, I can always do a sample rate conversion on my PC later.

Note: I did not ask for this feature to be included in the 'standard' version of the empeg, I was merely interested if it was at all possible to do it. I would gladly write the software myself, if it was possible. But if it's not, I'm not interested at all in the empeg.

PS: A flow-chart or something describing the signal flow etc. inside the unit would greatly help to clarify my confusion right now.

Er, what? The SA1100 can send digital audio to the DSP, but nor receive digital audio from the DSP? I'm afraid I do not fully
understand this. By the way, what DSP is used?

The DSP, a philips 7705 car audio processor, talks to the SA1100 via I2C for control and I2S for raw audio data. The data to the
DSP from the SA is output only, ie we can send data to it but not get it back. The tuner outputs a raw fm multiplex signal
directly to the DSP, which then processes it, extracts the RDS data stream and the audio L/R signals, and passes the audio
straight through to the onboard DACs. There is no way of intercepting it enroute. The only way of recording the radio would be
to sample the audio outputs with a separate ADC, such as the one the mk2 board uses...

Patrick.

Opinions expressed in this email may contain up to 42% water by weight, and are mine. All mine.

And again: er, what? Where does this (completely nonstandard) 38kHz sampling rate come from? Is it perhaps possible that Hugo
confused that with 32kHz, which would be understandable, as there's nothing usable above 16kHz on the radio. What ADC is used?
Where does the ADC send it's digital audio, the SA1100 or the DSP? And, lastly, so what? I'm completely satisfied if I can store
32kHz, or even 38kHz digital audio, I can always do a sample rate conversion on my PC later.

The ADC is built in to the DSP, and in radio mode does indeed run at 38kHz, I believe (without going to look it up) due to needing
this rate for RDS data recovery, amongst other things. Unfortunately the I2S bus from the SA, besides being output only, runs at
a fixed 44.1kHz rate, divided down from the master 11.2896MHz DSP clock.

Patrick.

Opinions expressed in this email may contain up to 42% water by weight, and are mine. All mine.

The DSP, a philips 7705 car audio processor,

Ah! Now this explains a lot. I didn't know that Philips was making DSPs...

talks to the SA1100 via I2C for control and I2S for raw audio data. The data to the
DSP from the SA is output only, ie we can send data to it but not get it back. The tuner outputs a raw fm multiplex signal
directly to the DSP, which then processes it, extracts the RDS data stream and the audio L/R signals, and passes the audio
straight through to the onboard DACs. There is no way of intercepting it enroute. The only way of recording the radio would be
to sample the audio outputs with a separate ADC, such as the one the mk2 board uses...

Does this mean recording radio might be possible with the mk2? Is there hope?

BTW is there a reason for using a car DSP, other than convenience? Is it cheaper than, say, a Motorola 56600?

When asking my original question, I assumed that the architecture resembled this: the tuner module outputs audio L/R and RDS data separately, the audio goes through a ADC, then through the SA1100, then through a DSP, then through an DAC. Would such an architecture be more expensive, or just more difficult to implement?

Thanks for your patience.

But if it's not, I'm not interested at all in the empeg.

Just curious, but what exactly is the point of recording the radio? The sound quality sucks, and not even because of the empeg's reception. Assuming you intend on recording music, you'd probably miss the beginning of the song anyway.

Perhaps you could shed some light on the subject?

jester
Sunnyvale, CA
MkI / 20GB / Blue / SN00309

whell to record Trafic Anaunsmens and the lastest news

Mark
wait for mk III with a USB Host/slave
(USB->Ethernet)(USB->COM 1-8)(USB->You name it)

No, I meant 38kHz. When in FM mode, the DSP runs at 38kHz which I believe is based on the way FM multiplex is transmitted (ie, 0->19kHz is L+R, 19->38kHz is L-R, allowing stereo separation to be performed).

The FM signal comes in as multiplex (which includes L, R and RDS data) into the DSP. We don't see it again until it comes out as sound - at 38kHz. We can't get it into the main CPU as the CPU is sample-rate locked at 44.1kHz. The ADC for the FM (and the DACs for audio out) are all inside the DSP.

The Mk2 allows 16-bit stereo sampling of the output at 22.05kHz, which is sufficient for most things.

Hugo

Philips make a lot of DSPs - this one is mask programmed. It's a much better bet than a general purpose DSP as it has specific car audio features - plus internal ADCs, DACs, etc. It's much cheaper than a general-purpose DSP, and smaller - a general purpose DSP plus 4 DACs, 4 ADCs, software for FM stereo decoding and RDS extraction, program/data memory, etc, would simply be too large.

A general purpose architecture would mean a much more expensive product that probably wouldn't fit into the box, and would require lots more software effort on our part.

Hugo

...so anyway to get the little squeeky rodent making the chewing noise out of the dsp?:(


...proud to have one of the first Mark I units

The DSP is actually very quiet - but the fact that in-car you use power amps means that even the slightest noise on the outputs can be audible, which means the install has to be good. The empeg, having a fast CPU, DRAM, etc in the box also has other noise sources, which is why we use differential outputs to remove digital noise.

The Mk2 has a redesigned PSU section, but the same audio output circuitry (basically) and we've managed to improve the performance plus go back to "normal" (non-differential) output, though it can be converted for differential if needed.

Hugo

I was under the impression that the reason you did a differential output was to manage noise, be compatible with high end amps, and it was the direction industry was moving in.

Why the backpedaling on the Mk2?

Kamau Wanguhu
[email protected]
http://www.BORGcube.com

There's no backpedalling - just that installers are happier with normal outputs, and we've managed to reduce the noise levels by a large amount through a new PSU and various other mods.

Differential output is still an option, we just remove one link per channel.

Hugo