Gang,
The problem with Firewire is that you still have to write a driver. While there is an ASYNC mode like USB in the Firewire specs. The problem is no OS supports it. Well that and the fact that more and more computers are shipping without the interface.
At least with USB you can implement easy Audio interfaces and even really great ones like asynchronous and be done and work with any OS. Well at least to 24/96, if you want higher than that (i.e. 24/192) you either have to use a MAC or write a driver XP/Vista/Linux.
Peufeu, I tried the isolators... I have to tell you I am not a fan the added jitter was way to much. Also where are you going to put them? Are you going to run the FX2 and FPGA off the VBUS and then isolate the dac at the serial engine? If you do that you need to put the MCLK right at the dac and reclock all the incomming pins (i.e. SCLK, WCLK and DATA). You will also need to do a little clock snubbing at the FPGA as these isolators may look good in some photo shoot they do for the data sheets, but in my experience they aren't nearly as good.
Thanks
Gordon
The problem with Firewire is that you still have to write a driver. While there is an ASYNC mode like USB in the Firewire specs. The problem is no OS supports it. Well that and the fact that more and more computers are shipping without the interface.
At least with USB you can implement easy Audio interfaces and even really great ones like asynchronous and be done and work with any OS. Well at least to 24/96, if you want higher than that (i.e. 24/192) you either have to use a MAC or write a driver XP/Vista/Linux.
Peufeu, I tried the isolators... I have to tell you I am not a fan the added jitter was way to much. Also where are you going to put them? Are you going to run the FX2 and FPGA off the VBUS and then isolate the dac at the serial engine? If you do that you need to put the MCLK right at the dac and reclock all the incomming pins (i.e. SCLK, WCLK and DATA). You will also need to do a little clock snubbing at the FPGA as these isolators may look good in some photo shoot they do for the data sheets, but in my experience they aren't nearly as good.
Thanks
Gordon
LOL, firewire, no way, too complex really. USB is the way. And I got it working @ 45 megabytes/s using 8% of 1 cpu core with a userspace libusb driver, so don't tell me it's slow, lol.
REPEAT 1 : For FireWire you need several tens of thousand lines of code and a powerful CPU. For USB you need a 8051 and about 200 lines of C, and it works.
REPEAT 2 : FireWire is dead anyway. USB is cheaper and this is what matters. If you need to put an ARM7 @ 50 MHz in your peripheral to do what a 8051 could do if you used USB, you gotta have a really good justification for it, because you're getting fired by the bean counters.
> Peufeu, I tried the isolators... I have to tell you I am not a fan the added
> jitter was way to much.
Well, the figures from the datasheet are impressive (for isolators) but still pretty nasty wrt audio standards...
> Also where are you going to put them?
> Are you going to run the FX2 and FPGA off the VBUS
Either USB powered or wall wart if > 500 mA.
> and then isolate the dac at the serial engine?
Isolators closest to DAC.
> If you do that you need to put the MCLK right at the dac
Why would I go to the trouble of implementing an asynchronous protocol if I did not do this ? The clock path will be about 1 cm. Master clock to DAC, everything slaved to it. A picture says more than a thousand words
> and reclock all the incomming pins (i.e. SCLK, WCLK and DATA).
Yep. Well, depends on the DAC. The sigma delta ones have a MCK input so I'll use that. Dunno if I'll reclock the other lines, perhaps worth to try. Definitely needed for the old multibit DACs.
> You will also need to do a little clock snubbing at the FPGA as these
> isolators may look good in some photo shoot they do for the data sheets,
> but in my experience they aren't nearly as good.
lol, I don't know why I'm not even surprised... thanks for the info
what source termination did you use with those ?
as long as the bits get through it's OK though.
> The drivers limit of firewire is going to be a non-issue
> when a standard driver will be provided in windows7,
> as now happens for mac and usb.
Yeah, a driver that, as usual puts the clock in the PC. And there still is a little unsolved interrogation about all the DRM cr*p. And how do I change the oversampling filters and the digital crossover in my FPGA using this driver ?
REPEAT 1 : For FireWire you need several tens of thousand lines of code and a powerful CPU. For USB you need a 8051 and about 200 lines of C, and it works.
REPEAT 2 : FireWire is dead anyway. USB is cheaper and this is what matters. If you need to put an ARM7 @ 50 MHz in your peripheral to do what a 8051 could do if you used USB, you gotta have a really good justification for it, because you're getting fired by the bean counters.
> Peufeu, I tried the isolators... I have to tell you I am not a fan the added
> jitter was way to much.
Well, the figures from the datasheet are impressive (for isolators) but still pretty nasty wrt audio standards...
> Also where are you going to put them?
> Are you going to run the FX2 and FPGA off the VBUS
Either USB powered or wall wart if > 500 mA.
> and then isolate the dac at the serial engine?
Isolators closest to DAC.
> If you do that you need to put the MCLK right at the dac
Why would I go to the trouble of implementing an asynchronous protocol if I did not do this ? The clock path will be about 1 cm. Master clock to DAC, everything slaved to it. A picture says more than a thousand words
An externally hosted image should be here but it was not working when we last tested it.
> and reclock all the incomming pins (i.e. SCLK, WCLK and DATA).
Yep. Well, depends on the DAC. The sigma delta ones have a MCK input so I'll use that. Dunno if I'll reclock the other lines, perhaps worth to try. Definitely needed for the old multibit DACs.
> You will also need to do a little clock snubbing at the FPGA as these
> isolators may look good in some photo shoot they do for the data sheets,
> but in my experience they aren't nearly as good.
lol, I don't know why I'm not even surprised... thanks for the info
what source termination did you use with those ?
as long as the bits get through it's OK though.
> The drivers limit of firewire is going to be a non-issue
> when a standard driver will be provided in windows7,
> as now happens for mac and usb.
Yeah, a driver that, as usual puts the clock in the PC. And there still is a little unsolved interrogation about all the DRM cr*p. And how do I change the oversampling filters and the digital crossover in my FPGA using this driver ?
Ethernet!
Why not use something like this: http://www.ethernut.de/ ? In particular the EIR-1. Just need to replace the decoder chip. Everything is open source and the whole ethernet interface is a done deal. For distribution around the house I don't see how you want to beat ethernet.
But in the end the real issue is software and for that I still personally prefer the Roku (or similarly Squeeze).
Anyway, we're getting off topic.
Peter
Why not use something like this: http://www.ethernut.de/ ? In particular the EIR-1. Just need to replace the decoder chip. Everything is open source and the whole ethernet interface is a done deal. For distribution around the house I don't see how you want to beat ethernet.
But in the end the real issue is software and for that I still personally prefer the Roku (or similarly Squeeze).
Anyway, we're getting off topic.
Peter
Let's stop polluting this thread, and go there :
http://www.diyaudio.com/forums/showthread.php?s=&threadid=135313
http://www.diyaudio.com/forums/showthread.php?s=&threadid=135313
"Oh and yea... ESS released yesterday 32 bit version of the Sabre AND you heard it here first. The first ever 32bit ADC....
Thanks
Gordon"
no standalone asrc part? im sold on the BB anyway though.
Looking forward for the ADC . Pleasant suprise (especially this time around) , same multiple channel type?
Thanks
Gordon"
no standalone asrc part? im sold on the BB anyway though.
Looking forward for the ADC . Pleasant suprise (especially this time around) , same multiple channel type?
well , i rephrased Gordon's msg (without his approval) , i definately think he is talkin bout an ADC:
- ESS released yesterday the 32 bit version of the Sabre, AND, you heard it here first: THE first ever 32bit ADC...
32 bit wordlength is more common than the 6bit 5mhz comnig from a PCM4222 so it makes enough sense to me... I also remember Dustin said it'd be fun to start out with an ADC project, esp since even audio precision 2 uses bog standard cirrus converters.
- ESS released yesterday the 32 bit version of the Sabre, AND, you heard it here first: THE first ever 32bit ADC...
32 bit wordlength is more common than the 6bit 5mhz comnig from a PCM4222 so it makes enough sense to me... I also remember Dustin said it'd be fun to start out with an ADC project, esp since even audio precision 2 uses bog standard cirrus converters.
BrianDonegan said:
This is very interesting, i must have missed something then
So it coudl be done in hardware in the dac without loss.
But how could it be implemented in software without the obvious resolution problems? Would that be possible?
AFAIK connected to a computer, the Sabre is seen as a 24 bit sound device.
About ES9008S in hardware mode.
It is stated in the datasheet that in this mode, the chip would only accept I2S data. And the default register for bit width is 24bit. Does this mean that hooking it up with USB receivers like PCM2706 is not possible? PCM2706 has 16bit I2S output.
Also for applications using USB receiver and PC as a digital source, I do not know if the output would be 44.1kHz or 48kHz, the choice of system clock seems difficult to decide. Some previous posts suggested the system clock needs not to be a multiple of the digital source, but the clock should be as fast as possible.
It is stated in the datasheet that in this mode, the chip would only accept I2S data. And the default register for bit width is 24bit. Does this mean that hooking it up with USB receivers like PCM2706 is not possible? PCM2706 has 16bit I2S output.
Also for applications using USB receiver and PC as a digital source, I do not know if the output would be 44.1kHz or 48kHz, the choice of system clock seems difficult to decide. Some previous posts suggested the system clock needs not to be a multiple of the digital source, but the clock should be as fast as possible.
