Although for best quality I'd definitely use an indipendent power supply with (much) higher voltage, doing that would be great for portable use. Problem is, available voltage is really too small.
Perhaps for portable use you should consider a dedicated DC/DC converter which is completely disabled when providing an external power supply? (I'd say at least 12V!).
Last edited:
AW USB/I2S with TP "Buffalo" instead of AB?
In the AW thread on our italian forum a friend asked about the possibility to use the AW digital board (USB/I2S module) with the TP "Buffalo" DAC that he already owns. See post #94655, post #p94671, etc.
I'm not sure about all the details, so I ask it here 'cause I bet that there are quite many owners of a Buffalo DAC which may be interested to the answer...
Q: is it possible to use a Buffalo (I, II or III) in place of the AB? how easy/hard would it be to do so?
Q: What modifications needs to be done to the Buffalo and/or to the AW?
Q: What extra circuitry needs to be added?
(perhaps we should start another thread dedicated to this subject?)
In the AW thread on our italian forum a friend asked about the possibility to use the AW digital board (USB/I2S module) with the TP "Buffalo" DAC that he already owns. See post #94655, post #p94671, etc.
I'm not sure about all the details, so I ask it here 'cause I bet that there are quite many owners of a Buffalo DAC which may be interested to the answer...
Q: is it possible to use a Buffalo (I, II or III) in place of the AB? how easy/hard would it be to do so?
Q: What modifications needs to be done to the Buffalo and/or to the AW?
Q: What extra circuitry needs to be added?
(perhaps we should start another thread dedicated to this subject?)
Last edited:
Could you point me to an updated schematic for the Buffalo?
The signals that pass throgh the USB-I2S module are I2S (11.2896 or 12.288MHz bit clock) out and MCLK (same two frequencies) in. The module can be VBUS powered or use an external power supply. The latter case might pose some challenges to bootup and enumeration. But that is only speculation from my side.
Børge
The signals that pass throgh the USB-I2S module are I2S (11.2896 or 12.288MHz bit clock) out and MCLK (same two frequencies) in. The module can be VBUS powered or use an external power supply. The latter case might pose some challenges to bootup and enumeration. But that is only speculation from my side.
Børge
no idea where to find it.
On the TP site there is the page about the new Buffalo III, but I can't find the schematics for it. At the end of the page there are a few links to old Buffalo's (I, II & 32s) documentation, but no schematic whatsoever (except perhaps for the very 1st one).
Hopefully this helps, wiring from Musiland to Buffalo II.
Musiland 03 I2S to Buffalo II DAC: Playing 352.8Khz Music H i F i D U I N O
And the Buffalo III manual,
http://www.twistedpearaudio.com/doc...-Buffalo_III_DAC_Integration_Guide_V1.0.1.pdf
Musiland 03 I2S to Buffalo II DAC: Playing 352.8Khz Music H i F i D U I N O
And the Buffalo III manual,
http://www.twistedpearaudio.com/doc...-Buffalo_III_DAC_Integration_Guide_V1.0.1.pdf
no idea where to find it.
On the TP site there is the page about the new Buffalo III, but I can't find the schematics for it. At the end of the page there are a few links to old Buffalo's (I, II & 32s) documentation, but no schematic whatsoever (except perhaps for the very 1st one).
Hi George,
yes, the MCLK to the DAC is 22/24. But the serial output of the MCU won't run that fast, so there's a /2 JK flip-flop between the XO and the module.
The ES9023 needs to run its sigma-delta modulator at 22/24 in order to have high enough oversampling rates for 176.4/192 input. We could use 44/48MHz MCLKs too. The AB-1.1 board allows for that by coupling two JKs in series to divide by 4. Higher frequencies mean more jitter but also higher oversampling rates. Might be a profitable and audible trade off.
BTW, what is your clock setup like? AB-1.1 MUXes two fixed-frequency XOs from Golledge.
11/12MHz is sufficient for 24 + 24 bits at 176.4 / 192. I use the same rate in my CD player for 24 bits at 352.8 / 384. That leaves 8 dead bits which is practical since running I2S bumper-to-bumper can be a little uncomfortable.
Børge
yes, the MCLK to the DAC is 22/24. But the serial output of the MCU won't run that fast, so there's a /2 JK flip-flop between the XO and the module.
The ES9023 needs to run its sigma-delta modulator at 22/24 in order to have high enough oversampling rates for 176.4/192 input. We could use 44/48MHz MCLKs too. The AB-1.1 board allows for that by coupling two JKs in series to divide by 4. Higher frequencies mean more jitter but also higher oversampling rates. Might be a profitable and audible trade off.
BTW, what is your clock setup like? AB-1.1 MUXes two fixed-frequency XOs from Golledge.
11/12MHz is sufficient for 24 + 24 bits at 176.4 / 192. I use the same rate in my CD player for 24 bits at 352.8 / 384. That leaves 8 dead bits which is practical since running I2S bumper-to-bumper can be a little uncomfortable.
Børge
The manual says:
"The Buffalo III DAC board uses a custom on-board ultra low jitter, low phase noise 100Mhz clock from
Crystek. The clock module lowers jitter significantly when compared to normal clocks. The clock is placed
on the board near the DAC chip to keep phase noise at low levels."
That is not immediately compatible with the USB-I2S module. But I'll have to read a bit more to tell.
Børge
"The Buffalo III DAC board uses a custom on-board ultra low jitter, low phase noise 100Mhz clock from
Crystek. The clock module lowers jitter significantly when compared to normal clocks. The clock is placed
on the board near the DAC chip to keep phase noise at low levels."
That is not immediately compatible with the USB-I2S module. But I'll have to read a bit more to tell.
Børge
Hi Børge
I use a Si532 dual output clock into a divider to feed the cpu. Output freq is selected by the cpu depending on the bit rate requested.
Although it makes for a compact board the clock is custom ordered at $20 a pop not including shipping. warning::: do not order small quantities as the courier costs at almost $90 are a killer.
George
I use a Si532 dual output clock into a divider to feed the cpu. Output freq is selected by the cpu depending on the bit rate requested.
Although it makes for a compact board the clock is custom ordered at $20 a pop not including shipping. warning::: do not order small quantities as the courier costs at almost $90 are a killer.
George
as does version two. IIRC Buffalo II by default had a slower clock, either 40 or 80 MHz.
Hi all,
The Sabre DAC's can run in two modes: asynchronous and synchronous (these modes are internal to the DAC and not to be confused with USB asynchronous ISOCHRONOUS transfer).
In the AW, we are running the ES9023 in "synchronous" mode as the MCLK supplied to the DAC is at a "correct" integer multiple (eg 192fs) of the incoming I2S data rate.
To run the ES9023 in "asynchronous" mode, the MCLK supplied must be > 192fs. If fs is 192khz, 192fs is 36.864Mhz.
AFAIK, the Buffalo DAC's has clocks at 40/80/100 Mhz, thus > 36.864Mhz, and therefore it is running the DAC in "asynchronous" mode.
The following is my conjecture based on reading the datasheets and "Technical Details of the Sabre Audio DAC" by Martin Mallinson and Dustin Forman, ESS Technology Technical Staff
So to interface the USB-I2S module with the Buffalo DAC, you might just have to add a small daughter board with the low jitter XO's (exactly like the AB1.1) divided down to the 11.2896/12.288Mhz to supply the USB-I2S.
So the USB-I2S module uses the 11.2896/12.288Mhz to throttle the USB data stream from the host. The I2s data is streamed to the Buffalo DAC's I2s input header. The Buffalo DAC then uses its internal DSP to filter, upsample, and jitter eliminate (not that there is any !!!) the samples to the internal high speed clock rate of 40/80/100Mhz.
Alex
The Sabre DAC's can run in two modes: asynchronous and synchronous (these modes are internal to the DAC and not to be confused with USB asynchronous ISOCHRONOUS transfer).
In the AW, we are running the ES9023 in "synchronous" mode as the MCLK supplied to the DAC is at a "correct" integer multiple (eg 192fs) of the incoming I2S data rate.
To run the ES9023 in "asynchronous" mode, the MCLK supplied must be > 192fs. If fs is 192khz, 192fs is 36.864Mhz.
AFAIK, the Buffalo DAC's has clocks at 40/80/100 Mhz, thus > 36.864Mhz, and therefore it is running the DAC in "asynchronous" mode.
The following is my conjecture based on reading the datasheets and "Technical Details of the Sabre Audio DAC" by Martin Mallinson and Dustin Forman, ESS Technology Technical Staff
So to interface the USB-I2S module with the Buffalo DAC, you might just have to add a small daughter board with the low jitter XO's (exactly like the AB1.1) divided down to the 11.2896/12.288Mhz to supply the USB-I2S.
So the USB-I2S module uses the 11.2896/12.288Mhz to throttle the USB data stream from the host. The I2s data is streamed to the Buffalo DAC's I2s input header. The Buffalo DAC then uses its internal DSP to filter, upsample, and jitter eliminate (not that there is any !!!) the samples to the internal high speed clock rate of 40/80/100Mhz.
Alex
yes, I think so. It does use only one and the same clock freq. for any stream...
ok, this sounds reasonably easy. Maybe he can even use some prebuilt "clock upgrade" module for that?
BTW: isn't asynchronous (ASRC) operation "wasting" quite some of the benefits of the AW?
Indeed it is a "waste" to operate the Sabre in async mode. In fact if I were to design something similar I would clock the ES9018 synchronously exactly like the AB1.1.
However, I was responding to the question about interfacing to existing buffalo DAC's, which are already fixed with async clocks
Alex
However, I was responding to the question about interfacing to existing buffalo DAC's, which are already fixed with async clocks
Alex
...I'd love to see an AB2.0 designed with two ES9018 wired in mono mode (one per channel), optocouplers on I2S, independent mains PSUs with schottky rectifiers, inductive input filter (gyrators...), dedicated discrete CCS+Salas shunt reg on each supply line of every load, ...
(that would be a huge and heavy thing, definitely NOT intended for portable use
).sure (and many tnx for that!).However, I was responding to the question about interfacing to existing buffalo DAC's, which are already fixed with async clocks
I was just wondering wether it would be possibile (how hard would it be) to modify the Buffalo so that it may be operated synchronously as in the AB1.x...
Last edited:
Oh, BTW:
with 100MHz clock, the Buffalo is able to handle DXD streams (PCM 384 or 352.8 KHz / 24 or 32 bit). Would it be possible for the AW USB/I2S interface to handle that too, somehow?
384khz cannot be done with the current uController - a limitation of the I2S bus. You will have to wait for the next version - a different uC, and/or FPGA, and/or custom interface chip . We are looking more at 7.1 channel than 384.
But let us not try to run before we can walk - we really need to get the AW to more testers and do more tweaking of the analog side to get the best stereo sound out first
Most high quality music are currently at 88.2/24, 96/24 and 192/24. You will need a top end system to be able to hear the difference between 192 and 384, I would think. (personally no experience with 384/24 music)
Alex
. We are looking more at 7.1 channel than 384.But let us not try to run before we can walk - we really need to get the AW to more testers and do more tweaking of the analog side to get the best stereo sound out first
Most high quality music are currently at 88.2/24, 96/24 and 192/24. You will need a top end system to be able to hear the difference between 192 and 384, I would think. (personally no experience with 384/24 music)
Alex
Me neither, just curiosity.
At the moment, I've seen some DXD files (including free samples) available only from 2L...
Completely agree about maxing out 2ch audio quality (see my previous post about my whishlist design for AB2.0...
). Couldn't care less about multichannel.
Pureplayer with WASAPI
Anyway, latest Pureplay(2.7.1) works with AW in WASAPI mode now(W7 with thesycon 1.18), the best player I've tried (prefer over stealthplayer now).
Anyway, latest Pureplay(2.7.1) works with AW in WASAPI mode now(W7 with thesycon 1.18), the best player I've tried (prefer over stealthplayer now).