Would it make sense to mate a Duet rcvr w external I2S DAC?

[Pete Fowler]

Jeez, I don't know where this attitude is coming from - did you not read or see my last 4 posts?

Apparently you were posting so vigorously my reply got bumped.

I missed that in the datasheet - if you knew this already & had mentioned it, it would be in the spirit of DIY but I seem to be coping an attitude from you that I don't understand - what's your problem with me?

I believe you're trying to help on some level. But (at least for me) you're not really helping.

Let's take an example.

Somewhere you read that on some DAC chip it's possible to run output transformers from right/left out to the VMID pin? Good so far. Definitely worth pursuing.

A prudent man would have studied the Wolfson spec in depth and/or researched the web to be sure trafos would work on this exact Wolfson VMID pin - after all there is a real chance you could have fried your DAC.

(God forbid we hook a vref to it). ;-)

My guess is you scanned the spec then asked if any of us knew about the VMID pin. No one did, including me. I did offer you info on bypassing the pin (all I knew, mate), and also referred to the part of this thread where the discussion took place in hopes it might help.

At that point I could have said "go read the Wolfson spec very carefully and let's discuss it, I think this could fry your DAC", but its not my job mon. It's a DIY thread. Hans and I have figured that out - things die in DIY

From your posts it's clear you soldered trafos to the VMID pin even though you didn't fully understand how it works for this particular Wolfson DAC. Okay fine, you took a leap and got lucky.

To me it that shows you don't filter your ideas before offering them - 'I read this on forum X'. It also tells me you don't read the specs too closely. If you're doing that sort of thing I'm not going to trust your other advice and insight, esp since your hands-on experience with the Duet is limited. What else do you expect?

So...my cranky comments are pointing to the same thing - get some meaningful experience on the Duet and then tell us about it. The other stuff...not so much.

Pete

[jkeny]

Apparently you were posting so vigorously my reply got bumped.



I believe you're trying to help on some level. But (at least for me) you're not really helping.

Ok, that's fine

Let's take an example.

Somewhere you read that on some DAC chip it's possible to run output transformers from right/left out to the VMID pin? Good so far. Definitely worth pursuing.

Read my post again - I didn't say I read it somewhere - I said I already did this with a PCM1793 Vout DAC (which has 1.65V bias on it's outputs & a Vref pin at VDD/2 (called VCOM). I said that this proved very successful. So maybe a bit more care in reading is called for on your part!

A prudent man would have studied the Wolfson spec in depth and/or researched the web to be sure trafos would work on this exact Wolfson VMID pin - after all there is a real chance you could have fried your DAC.

My experience told me it wouldn't fry the DAC - the DAC's output stage are capable of driving 16 ohm headphones (I already said I was impressed by this - you must have missed this also?) - my Sescom transformers have a DCR of 70ohm! So prudence doesn't come into it.

(God forbid we hook a vref to it). ;-)

My guess is you scanned the spec then asked if any of us knew about the VMID pin. No one did, including me. I did offer you info on bypassing the pin (all I knew, mate), and also referred to the part of this thread where the discussion took place in hopes it might help.

At that point I could have said "go read the Wolfson spec very carefully and let's discuss it, I think this could fry your DAC", but its not my job mon. It's a DIY thread. Hans and I have figured that out - things die in DIY

I usually find the spirit in other DIY threads/forums that if people have information or opinions that somebody might be making a mistake - they voice them or advise - you seem to absolve yourself from this spirit - that's your prerogative - but then you seem to offer advice to others?

From your posts it's clear you soldered trafos to the VMID pin even though you didn't fully understand how it works for this particular Wolfson DAC. Okay fine, you took a leap and got lucky.

You are attributing a cavalier attitude to me without having read what I already posted about my experience with trafos! Please stop building this house of straw around me - as I said already, where is this coming from?

To me it that shows you don't filter your ideas before offering them - 'I read this on forum X'. It also tells me you don't read the specs too closely. If you're doing that sort of thing I'm not going to trust your other advice and insight, esp since your hands-on experience with the Duet is limited. What else do you expect?

I may have been wrong in some of my comments/suggestions but so what - this is DIY - I will get many things wrong - I don't expect the inquisition!

So...my cranky comments are pointing to the same thing - get some meaningful experience on the Duet and then tell us about it. The other stuff...not so much.

Pete

So re-read my posts & lay off please - I believe you are mistaken in your view of me & I again wonder what's behind it?

PS I already wrote to you privately (a couple of hours ago)about this so as not to pollute the thread but you seem to choose to air this in public - so be it!

[Pete Fowler]

So re-read my posts & lay off please

Fair enough. Not another word from me. Done.

Pete

P.S. Your email arrived well after I posted. But my agreement still stands. Done.

[jkeny]

Fair enough. Not another word from me. Done.

Pete

P.S. Your email arrived well after I posted. But my agreement still stands. Done.

Cool, normal programming will now resume

I just disconnected the 5V reg by lifting the inductor that feeds it 9v & I've attached a battery pack with 5V & tapped at halfway point for 2.5V. I was just coming back to the computer to fire up squeezebox server to have a listen but I'm having great difficulty getting this whole networking thing operating reliably - just have a blue light on the box - trying to sort out whaz-up & will report back!

PS, This thing is beginning to annoy the tits off me, not helped by the fact that I have to keep going back to the router under the stairs to see what status I have, blah, blah And did I already say that at times it seems to cause all PCs connected to the WiFi to lose their connection even though their status says connected i.e. page not available messages for any page accessed

[jkeny]

Ok, I found another router that allowed me to connect the SBR & PC into it - that way I have it local to the SB server & can play it without all the too-ing & fro-ing.
Anyway, listened to the output through transformers connected to headphones - I don't like the sound very mid-fi. As per my last post I have a battery pack (actually it's connected to a transformer charging the batteries at the moment) feeding 5V to DAC with a centre tap for 2.5V. One of the problems with this arrangement is the low output so the SNR is not good. Connecting the clean 2.5V to my VMID wire made no difference to the sound but I'm not even sure how this works see below. I am not investigating this configuration any further.

I then tried with caps on the Vouts (6.8uF Wima MKS) & ground connected to my headphones - the increase in volume was noticeable & the clarity was a lot better. Again VMID connected to 2.5V made no difference to the sound but see below. I like this sound, could live with it & better caps would probably masage the sound a bit but it's a good bit off my reference HiFace USB through my ESS DAC.

I can't understand how the external VMID will work - with power connected there is 2.5V on VMID generated internally by a resistor divider - if we connect a 2.5V supply to VMID no current flows so am I not understanding something?

Next I tried the ESS DAC plugged into I2S - a bit smoother & more relaxed plus a lower volume output but nothing like the low level of the first option. I would probably prefer this set-up of all 3 but I may be biased & would need blind testing to decide. These are brief listening results - will need a bit more time to decide on one. But again, this doesn't match my reference - more closed in, less texture, less distinct bass & all frequencies really, no lifelike venue acoustics.

Again, one piece of new info from these tests is that I had to connect the ESS DAC without any ground connection to the SBR & it seems to work without complaint (no difference if ground is connected) - I will apply this to my USB HiFace & see if it makes much difference (USB ground can be dirty coming as it does from the PC & it's not possible to galvanically isolate hi-speed USB 2.0 @ 480mBits/s)

[Pete Fowler]

Ok, I found another router that allowed me to connect the SBR & PC into it - that way I have it local to the SB server & can play it without all the too-ing & fro-ing.
Anyway, listened to the output through transformers connected to headphones - I don't like the sound very mid-fi. As per my last post I have a battery pack (actually it's connected to a transformer charging the batteries at the moment) feeding 5V to DAC with a centre tap for 2.5V. One of the problems with this arrangement is the low output so the SNR is not good. Connecting the clean 2.5V to my VMID wire made no difference to the sound but I'm not even sure how this works see below. I am not investigating this configuration any further.

I then tried with caps on the Vouts (6.8uF Wima MKS) & ground connected to my headphones - the increase in volume was noticeable & the clarity was a lot better. Again VMID connected to 2.5V made no difference to the sound but see below. I like this sound, could live with it & better caps would probably masage the sound a bit but it's a good bit off my reference HiFace USB through my ESS DAC.

I can't understand how the external VMID will work - with power connected there is 2.5V on VMID generated internally by a resistor divider - if we connect a 2.5V supply to VMID no current flows so am I not understanding something?

Next I tried the ESS DAC plugged into I2S - a bit smoother & more relaxed plus a lower volume output but nothing like the low level of the first option. I would probably prefer this set-up of all 3 but I may be biased & would need blind testing to decide. These are brief listening results - will need a bit more time to decide on one. But again, this doesn't match my reference - more closed in, less texture, less distinct bass & all frequencies really, no lifelike venue acoustics.

Again, one piece of new info from these tests is that I had to connect the ESS DAC without any ground connection to the SBR & it seems to work without complaint (no difference if ground is connected) - I will apply this to my USB HiFace & see if it makes much difference (USB ground can be dirty coming as it does from the PC & it's not possible to galvanically isolate hi-speed USB 2.0 @ 480mBits/s)

Re the VMID pin, there is a discussion on what it does and how to deal with it (within my limited knowledge) starting here:

http://forums.slimdevices.com/showth...t=55044&page=3

Re not providing a ground reference to the I2S from the Duet, it may be working but in general the lack of ground reference can cause signal degradation. Generally not considered a good thing. Since the ground plane of the Duet isn't noisy like a PC can be, I would be in favor of grounding the IS2 lines even if the diff isn't audible in the macro sense.

Pete

[Bertel]

Re not providing a ground reference to the I2S from the Duet, it may be working but in general the lack of ground reference can cause signal degradation. Generally not considered a good thing. Since the ground plane of the Duet isn't noisy like a PC can be, I would be in favor of grounding the IS2 lines even if the diff isn't audible in the macro sense.

Pete, now this is interesting - I was just about to sit down and write the opposite!

I am using my SBR with i2s out to my Twisted Pear Buffalo32S which is based on the ESS9018 chip. The BUF has inputs for the three i2s lines and for GND as well, you have to 'base' these signals on some GND either from the source or from the DAC (that's the ground reference you mention), so it lets you choose which one you'd like to select, thus either coupling both units GND-wise or floating them.

A friend of mine who has a HF lab recommended to float them for better results (in order not to 'spill' noise on the GND planes from one unit to the other), I found no clear recommendation from Twisted Pear on this, so I tried both. I think I at least can say that the sound does not degrade in my setup when I float the units, I'd however have to say I even like it better when the BUF32S takes its GND reference from 'itself' (i.e. its digital part and the battery that feeds it) rather than from the SBR, but that can be perception. Maybe the reason for this is the way the ESS chip does its rigorous resampling, I don't know...

I will apply this to my USB HiFace & see if it makes much difference (USB ground can be dirty coming as it does from the PC & it's not possible to galvanically isolate hi-speed USB 2.0 @ 480mBits/s)

When heavily modding my PC system for cMP/cPlay earlier last year, I sourced power for the USB line externally (batteries+caps). You can see that as item #16 here: http://cplay.sourceforge.net/pmwiki....P.05Components Could this help?

Best
Robert

[jkeny]

Re the VMID pin, there is a discussion on what it does and how to deal with it (within my limited knowledge) starting here:

http://forums.slimdevices.com/showth...t=55044&page=3

Thanks, Pete, I read all those posts already - about bypassing, ESR, etc. Useful but I wanted to supply VMID externally. But my question is how can I do this as according to the WM datasheet VMID is connected to ground internally via a resistor voltage divider. In this case it doesn't make sense in trying to provide an external 2.5V directly to VMID? Am I missing something?

Re not providing a ground reference to the I2S from the Duet, it may be working but in general the lack of ground reference can cause signal degradation. Generally not considered a good thing. Since the ground plane of the Duet isn't noisy like a PC can be, I would be in favor of grounding the IS2 lines even if the diff isn't audible in the macro sense.

Pete

Well, I would agree that the ground is used as the reference point for signals so I was surprised that it worked - but this is what DIY is about. I think care has to be used here though as to whether this can be applied to other situations that's why I gave the parameters of what I'm using. For instance, leaving the ground connection off a USB cable could likely be disastrous see here: http://db.audioasylum.com/cgi/m.mpl?...nection+usb&r=

I wonder if the I2S voltage requirements http://www.sparkfun.com/datasheets/B...rds/I2SBUS.pdf are more relaxed then USB http://www.beyondlogic.org/usbnutshell/usb2.htm. From these links it can be seen that the signal levels are somewhat the same: low is <0.3V USB or <0.8V I2S; High is > 2.8V USB or >2V I2S. But USB operates via a differential voltage level between D+ & D- of 200mV whereas the the signal levels in I2S are as above. Does this allow it to work? More testing required.

[jkeny]

Pete, now this is interesting - I was just about to sit down and write the opposite!

I am using my SBR with i2s out to my Twisted Pear Buffalo32S which is based on the ESS9018 chip. The BUF has inputs for the three i2s lines and for GND as well, you have to 'base' these signals on some GND either from the source or from the DAC (that's the ground reference you mention), so it lets you choose which one you'd like to select, thus either coupling both units GND-wise or floating them.

BUT I would have imagined that the ground ref would need to be the same as is used for generating the I2S signal? That's what has me confused. Floating grounds only work if all signal voltages are within the I2S spec & taking the ground from somewhere else does not guarantee this unless you have measured the V potential.

A friend of mine who has a HF lab recommended to float them for better results (in order not to 'spill' noise on the GND planes from one unit to the other), I found no clear recommendation from Twisted Pear on this, so I tried both. I think I at least can say that the sound does not degrade in my setup when I float the units, I'd however have to say I even like it better when the BUF32S takes its GND reference from 'itself' (i.e. its digital part and the battery that feeds it) rather than from the SBR, but that can be perception. Maybe the reason for this is the way the ESS chip does its rigorous resampling, I don't know...

I would agree with your friend that large ground loops should be avoided & clean grounds are essential to lower noise (& hence lower jitter in digital). Maybe it is just a phenomena of the ESS chips?

When heavily modding my PC system for cMP/cPlay earlier last year, I sourced power for the USB line externally (batteries+caps). You can see that as item #16 here: http://cplay.sourceforge.net/pmwiki....P.05Components Could this help?

Best
Robert

That USB cable is dangerous, I think, if it has disconnected the USB ground from the PC - see the Audioasylum link above - it could blow the USB receiver chip BUT I guess the same could be said about my I2S experiment ?
I'm trying to tease out a cogent reason for it - maybe if the voltage potential between the intended grounds is measured beforehand & it's close then it's OK (& it won't change based on a fluctuating grid !) I don't know the answer.

Edit: BTW, I just noticed I have moved from junior member to member - whoa! - the parties at my place

[Bertel]

What you say is very right, and taps right into a thought/idea I am carrying around with me for a while now:

Of course I measured the difference in potential, and since it is 'only' 175mV and not changing or varying at all, I concluded that this would just add to the i2s signals and shouldn't damage anything, so I gave it a try. I liked the result and so thought that possible downsides by the added voltage likely were compensated by the improvent in having GNDs and system noises riding on them separated.

Yet I still don't like the thought at all that (i) there is no 'safety mechanism' in place that can guarantee that this difference in potential doesn't get higher than these 175mV and damage anything, and even more (ii) that theres is an add-on to the i2s signals that is simply 'unnatural' and 'not original' and I cannot guarantee that it doesn't tamper with these tiny fragile i2s signals

So I thought maybe transformers would be a feasible solution to deal with that difference in potential while at the same time isolating both modules and leaving the signals untouched. I haven't yet found suitable ones that have been proven to work well at 11.289MHz (ok, I haven't looked too hard either...) but I can't see a reason why e.g. a good toroidal or air core transformer designed for HF/RF shouldn't work...