I don't really have an opinion regarding the Curryman DAC. I haven't tried it myself. The sample rate conversion in the ESS DAC is supposed to remove the problems caused by jitter. But I don't how well it works.
Personally I prefer to use a clean master clock whenever possible. Because this is the best possible solution.
Personally I prefer to use a clean master clock whenever possible. Because this is the best possible solution.
Wow, Jens. Thank you kindly for going through this exercise.
While this level of jitter is not ideal from a purely technical perspective, it does fall into the limits of acceptable behavior especially if using an interface that can correct for it. The previous link I provided to RME SteadyClock info shows correction of MADI jitter as bad as 80ns to 2ns and correction of "bad" 40ns of ADAT jitter to 0.7ns.
BTW, 16-bit or better FFT analysis yields more accurate results. My main system converters have 20ps of jitter.
ADAT will always be worse than SPDIF, but as long as good transmitters and receivers are used with the USBStreamer, it looks like industry-typical results should be expected. However connecting to converters that don't perform any clock correction of their own is another matter.
Again, a big thanks for following up on this information. Based on this discussion I've decided to try the USBStreamer "B" box for myself.
Hi JensH
Thanks for your answer. So to get the recording right, the input signals to the USBStreamer must always correspond to the device settings on the Mac. No detection or whatsoever. I thought about an ASRC board to reclock and feed the USBStreamer with 192/24 signals. This way all sample rates from the source would be supported. Any downside in doing so? The Dual XO Clock board from Ian: http://www.diyaudio.com/forums/group-buys/207438-ian-asynchronous-i2s-s-pdif-fifo-kit-group-buy.html looks like a good fit. Could that work out? I have to use 4 of them in paralell with good oscillators. In the end the digital chain should be somthing like this:
8ch I2S -> Reclocking -> Isolator/USBStreamer | --- | Mac | --- |
Isolator/USBStreamer -> BIII | --- | Amps
Thanks for your answer. So to get the recording right, the input signals to the USBStreamer must always correspond to the device settings on the Mac. No detection or whatsoever. I thought about an ASRC board to reclock and feed the USBStreamer with 192/24 signals. This way all sample rates from the source would be supported. Any downside in doing so? The Dual XO Clock board from Ian: http://www.diyaudio.com/forums/group-buys/207438-ian-asynchronous-i2s-s-pdif-fifo-kit-group-buy.html looks like a good fit. Could that work out? I have to use 4 of them in paralell with good oscillators. In the end the digital chain should be somthing like this:
8ch I2S -> Reclocking -> Isolator/USBStreamer | --- | Mac | --- |
Isolator/USBStreamer -> BIII | --- | Amps
The USBStreamer will always be master on its I2S bus. And all the clock outputs will reflect the sample rate set up by the PC/MAC application.
Using my master clock and isolator design does not really change this, except that the MCLK is now generated by an external clock oscillator (but the oscillator is not connected to the MCLK output). This oscillator still has to deliver the 22.5 or 24.5 MHz clock needed by the USBStreamer for a particular sample frequency. The advantage is that it can now be a clock signal with very low jitter. If you are only going to use 192/24 on the USBStreamer, you only need to provide a 24.576 MHz clock. So you don't need the 22.5 MHz in this case.
You can of course connect a sample rate converter between the ADC and the USBStreamer (with or without the isolation board).
The only downside I can think of is the risk of reduced signal quality caused by the ASRC.
Using my master clock and isolator design does not really change this, except that the MCLK is now generated by an external clock oscillator (but the oscillator is not connected to the MCLK output). This oscillator still has to deliver the 22.5 or 24.5 MHz clock needed by the USBStreamer for a particular sample frequency. The advantage is that it can now be a clock signal with very low jitter. If you are only going to use 192/24 on the USBStreamer, you only need to provide a 24.576 MHz clock. So you don't need the 22.5 MHz in this case.
You can of course connect a sample rate converter between the ADC and the USBStreamer (with or without the isolation board).
The only downside I can think of is the risk of reduced signal quality caused by the ASRC.
Hi Jens,
I have a USBStreamer that is connected to a Buffalo3 dac (8-ch mode) in order to do PC-based digital crossover and dsp. I use the toslink input on the USBStreamer for connecting sound sources. However, I can't get a solid lock of the signal on the BIII, which I believe is due to jitter problems (others have experienced the same thing with the USBStreamer-BIII combination). While searching for a possible solution I came across this thread.
Do you think your board could solve this? If so, do you have any boards left for sale (bare boards or complete with parts).
Best regards,
Mattias
I have a USBStreamer that is connected to a Buffalo3 dac (8-ch mode) in order to do PC-based digital crossover and dsp. I use the toslink input on the USBStreamer for connecting sound sources. However, I can't get a solid lock of the signal on the BIII, which I believe is due to jitter problems (others have experienced the same thing with the USBStreamer-BIII combination). While searching for a possible solution I came across this thread.
Do you think your board could solve this? If so, do you have any boards left for sale (bare boards or complete with parts).
Best regards,
Mattias
Hi Mattias,
Assuming that jitter is creating the problems that you see, the V1 may actually be able to solve the problem. But actually I never mounted the flip-flop device needed for re-clocking the I2S signals. So I don't know if the timing will be OK.
Another option could be the V2. On V2 I have used faster isolators and I removed the flip-flops for two reasons:
1. I wanted to be able to make a loopback connection. Having additional delay on the I2S signals makes it difficult, if not impossible, to achieve this without a complex re-framing of the I2S signals.
2. Most converters only need a jitter free MCLK. The SCLK, LRCK and data are non-critical.
I recently made a V2A, where I included a small modification, which makes it behave better when the isolated side (non-USB) is not powered. The V2A still has on-board oscillators. With some additional (off-board) flip-flops it will be possible to re-clock the I2S signals. So if you want to try that, it is definitely an option. You could make this on a small add-on PCB. I would suggest to have an option to select the polarity of the MCLK, to achieve the best timing. Provided there is enough interest I may make a derivative of the V2A, where the flip-flops were put back in.
As long as this is only used for playback it should be possible. But if it solves your problem, I don't know. I have no experience with the Buffalo3 (or the ES9018).
I have V2A bare boards, as well as other types with no on-board oscillators. The latest is V7. The functionality of the V7 is the same as the V5. I just changed one component and corrected a couple of connections.
Jens
Assuming that jitter is creating the problems that you see, the V1 may actually be able to solve the problem. But actually I never mounted the flip-flop device needed for re-clocking the I2S signals. So I don't know if the timing will be OK.
Another option could be the V2. On V2 I have used faster isolators and I removed the flip-flops for two reasons:
1. I wanted to be able to make a loopback connection. Having additional delay on the I2S signals makes it difficult, if not impossible, to achieve this without a complex re-framing of the I2S signals.
2. Most converters only need a jitter free MCLK. The SCLK, LRCK and data are non-critical.
I recently made a V2A, where I included a small modification, which makes it behave better when the isolated side (non-USB) is not powered. The V2A still has on-board oscillators. With some additional (off-board) flip-flops it will be possible to re-clock the I2S signals. So if you want to try that, it is definitely an option. You could make this on a small add-on PCB. I would suggest to have an option to select the polarity of the MCLK, to achieve the best timing. Provided there is enough interest I may make a derivative of the V2A, where the flip-flops were put back in.
As long as this is only used for playback it should be possible. But if it solves your problem, I don't know. I have no experience with the Buffalo3 (or the ES9018).
I have V2A bare boards, as well as other types with no on-board oscillators. The latest is V7. The functionality of the V7 is the same as the V5. I just changed one component and corrected a couple of connections.
Jens
Hi Jens,
a few days ago I discovered the minidsp usbstreamer. I was quite exited because of its multichannel capabilities, but after some online search I read that it has some jitter problems.
But I also encountered your solution for the problem. I therefore wanted to ask you if you still have a board with the isolators and master clock for the usbstreamer?
I briefly want to describe my setup:
At the moment I use two digital active speakers with a Hypex dlcp, but I would like to expand it to a multichannel setup and would need a sound interface that provides multiple digital outputs.
I think the usbstreamer together with 4 i2s to spdif converters would fullfill my needs.
What do you think about this setup?
Best regards
Alexander
a few days ago I discovered the minidsp usbstreamer. I was quite exited because of its multichannel capabilities, but after some online search I read that it has some jitter problems.
But I also encountered your solution for the problem. I therefore wanted to ask you if you still have a board with the isolators and master clock for the usbstreamer?
I briefly want to describe my setup:
At the moment I use two digital active speakers with a Hypex dlcp, but I would like to expand it to a multichannel setup and would need a sound interface that provides multiple digital outputs.
I think the usbstreamer together with 4 i2s to spdif converters would fullfill my needs.
What do you think about this setup?
Best regards
Alexander
Hi Jens,
a few days ago I discovered the minidsp usbstreamer. I was quite exited because of its multichannel capabilities, but after some online search I read that it has some jitter problems.
But I also encountered your solution for the problem. I therefore wanted to ask you if you still have a board with the isolators and master clock for the usbstreamer?
I briefly want to describe my setup:
At the moment I use two digital active speakers with a Hypex dlcp, but I would like to expand it to a multichannel setup and would need a sound interface that provides multiple digital outputs.
I think the usbstreamer together with 4 i2s to spdif converters would fullfill my needs.
What do you think about this setup?
Best regards
Alexander
a few days ago I discovered the minidsp usbstreamer. I was quite exited because of its multichannel capabilities, but after some online search I read that it has some jitter problems.
But I also encountered your solution for the problem. I therefore wanted to ask you if you still have a board with the isolators and master clock for the usbstreamer?
I briefly want to describe my setup:
At the moment I use two digital active speakers with a Hypex dlcp, but I would like to expand it to a multichannel setup and would need a sound interface that provides multiple digital outputs.
I think the usbstreamer together with 4 i2s to spdif converters would fullfill my needs.
What do you think about this setup?
Best regards
Alexander
Yes, I still have boards available. The boards are bare boards, not assembled.
Do you plan to use 4 x Hypex dlcp with SPDIF connections?
What SPDIF receiver does the dlcp use? If it can suppress the jitter efficiently you may not need the master clock, which is part of my design. I didn't know the dlcp, but I have briefly looked into the manual. As far as I can see, it is not straight forward to get the local clock signal out of the dlcp. And since there is a sample rate converter in the dlcp the clock may not even have the needed frequencies available for 44.1 and 48 kHz based playback.
So, I am not sure that my interface is really applicable in this case. It is mainly targeted at systems with I2S interface to the DAC(s), no sample rate conversion and a clean clock, either on the converter board (with PCB V7) or on the isolated interface board (with PCB V2A). Going via SPDIF will add jitter to the signal, so the effort to make a very clean master clock could be wasted.
It may be possible to use my interface, but I would need more detailed information to be able to work out if it is possible and whether it makes sense.
Jens
Do you plan to use 4 x Hypex dlcp with SPDIF connections?
What SPDIF receiver does the dlcp use? If it can suppress the jitter efficiently you may not need the master clock, which is part of my design. I didn't know the dlcp, but I have briefly looked into the manual. As far as I can see, it is not straight forward to get the local clock signal out of the dlcp. And since there is a sample rate converter in the dlcp the clock may not even have the needed frequencies available for 44.1 and 48 kHz based playback.
So, I am not sure that my interface is really applicable in this case. It is mainly targeted at systems with I2S interface to the DAC(s), no sample rate conversion and a clean clock, either on the converter board (with PCB V7) or on the isolated interface board (with PCB V2A). Going via SPDIF will add jitter to the signal, so the effort to make a very clean master clock could be wasted.
It may be possible to use my interface, but I would need more detailed information to be able to work out if it is possible and whether it makes sense.
Jens
Hi,
as far as I know the dlcp converts the sample rate of all digital singnals to 93,75 kHz with an asynchron sample rate converter.
I thought I can build a setup as follows:
UsbStreamer->your I2S interface with clean clock -> something like this: https://www.hifiberry.com/hbdigi -> long spdif connections to 2 dlcps (maybe I will add more later) in separate enclosures.
Does this make sense? Or do you think there is lots of jitter added due spdif connection?
regards
Alexander
as far as I know the dlcp converts the sample rate of all digital singnals to 93,75 kHz with an asynchron sample rate converter.
I thought I can build a setup as follows:
UsbStreamer->your I2S interface with clean clock -> something like this: https://www.hifiberry.com/hbdigi -> long spdif connections to 2 dlcps (maybe I will add more later) in separate enclosures.
Does this make sense? Or do you think there is lots of jitter added due spdif connection?
regards
Alexander
Hi Alexander,
I think it will work. But whether cleaning up the clock of the USBStreamer makes sense when there are so many links in the chain, I don't know.
You are welcome to try it of course. For this you will need the USBStreamer Interface V2A, with on-board oscillators.
Send me a PM if you want to buy a bare PCB.
Best regards,
Jens
I think it will work. But whether cleaning up the clock of the USBStreamer makes sense when there are so many links in the chain, I don't know.
You are welcome to try it of course. For this you will need the USBStreamer Interface V2A, with on-board oscillators.
Send me a PM if you want to buy a bare PCB.
Best regards,
Jens
Hi svon,
By indicators, do you mean a sample rate indicator? If so, I have made one as part of a DAC design. It is based on a 74LVC138, which is used to decode the 48/44.1, SRATE0 and SRATE1 driving 6 LED's. But not as a stand-alone PCB. It would be easy to make though.
Do you have the USBStreamer Isolator up and running?
By indicators, do you mean a sample rate indicator? If so, I have made one as part of a DAC design. It is based on a 74LVC138, which is used to decode the 48/44.1, SRATE0 and SRATE1 driving 6 LED's. But not as a stand-alone PCB. It would be easy to make though.
Do you have the USBStreamer Isolator up and running?
It would be great if you could post a picture showing the location of this resistor to remove and to which pad the external MCLCK signal should be attached to.
Hello Jensh,
I think order an USBstreamer to replace my dead ESI juli@ and to allow higher recording sampling rate. As many others DIYers, i would be very interested to know how it is possible to inject external clean MCLK to USB streamer !
Is that possible to wake up the thread and give the trick ?
Regards.
Frex
I think order an USBstreamer to replace my dead ESI juli@ and to allow higher recording sampling rate. As many others DIYers, i would be very interested to know how it is possible to inject external clean MCLK to USB streamer !
Is that possible to wake up the thread and give the trick ?
Regards.
Frex
The first post in the thread mentions that the 3.3V supply is tapped from the usbStreamer board. For my own little project, I've been unsuccessful in trying to find out where to pull it from. J1 and J2 pins are the only ones documented in the manual and they don't show a 3.3V pin. Any help would be appreciated.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.