I recently added some Minidsp Curryman DACs to my system. I'm wondering if anyone has swapped out the clock oscillator on this DAC or other ES9023 designs and found some improvement.
I went ahead and changed oscillators based on my past experience with the XPresso oscillator without doing any comparison. Anyone else try this?
The attached plot shows phase noise measurement of the clock from the Curryman and a standard clock oscillator.
I went ahead and changed oscillators based on my past experience with the XPresso oscillator without doing any comparison. Anyone else try this?
The attached plot shows phase noise measurement of the clock from the Curryman and a standard clock oscillator.
Attachments
Can you please explain a bit further what we see on the attached graph (yes, phase noise plots, but what XOs are those two curves?). Wondering a bit about the carrier frequency specified in the graph (50MHz) since AFAIK the curryman DAC from miniDSP uses 48MHz (given they did not change the production which I have no info about). Is it a measurement of the curryman DAC?
While there are a few XOs having lower phase noise in the audio band they are not available at a suitable price (given the overall price point of the curryman DAC). The FOX HC73 is doing comparably well.
Looking forward to hearing your results.
kind regards, Daniel
While there are a few XOs having lower phase noise in the audio band they are not available at a suitable price (given the overall price point of the curryman DAC). The FOX HC73 is doing comparably well.
Looking forward to hearing your results.
kind regards, Daniel
Clocks
Yes sorry some explanation would be helpful. The Fox clock was 48MHz, it's the upper plot in the figure. The part I had on hand was 50MHz. I had thought
I had seen 50MHz on your DAC development thread, but maybe not.
Would there be an issue using 50MHz?
I'm not sure about the cost issue, I deal with these things in high volume where the prices are very low. I do see that the lower noise part Vectron VCC1-B3D-50M0000 is $1.10 on Mouser, the Fox part is $1.35.
But my original question was about the sound. In theory a lower noise clock would be better, but has anyone tried it out.
Yes sorry some explanation would be helpful. The Fox clock was 48MHz, it's the upper plot in the figure. The part I had on hand was 50MHz. I had thought
I had seen 50MHz on your DAC development thread, but maybe not.
Would there be an issue using 50MHz?
I'm not sure about the cost issue, I deal with these things in high volume where the prices are very low. I do see that the lower noise part Vectron VCC1-B3D-50M0000 is $1.10 on Mouser, the Fox part is $1.35.
But my original question was about the sound. In theory a lower noise clock would be better, but has anyone tried it out.
Hi,
so the lower curve is the Vectron VCC1 XO? That would be real impressive. I only knew some Crystek and NDK parts playing in that league and they are indeed much more expensive or hard to get (NDK is only available in 2x3mm which is hard to solder for DIY).
You can use both 48MHz or 50MHz when you are using the asynchonous mode of the ES9023 which is the normal use case. 48MHz was chosen because there is the option for an onboard 4:1 divider using high speed low jitter flipflops to provide 12MHz clock for an WM8804 SPDIF receiver, which will run in synchronous mode then. However this option is not used at the curryman DAC as produced by miniDSP.
Not sure but maybe you can find some more info regarding different clocks for the ES9023 in the SUBBU DAC development thread.
kind regards, Daniel
so the lower curve is the Vectron VCC1 XO? That would be real impressive. I only knew some Crystek and NDK parts playing in that league and they are indeed much more expensive or hard to get (NDK is only available in 2x3mm which is hard to solder for DIY).
You can use both 48MHz or 50MHz when you are using the asynchonous mode of the ES9023 which is the normal use case. 48MHz was chosen because there is the option for an onboard 4:1 divider using high speed low jitter flipflops to provide 12MHz clock for an WM8804 SPDIF receiver, which will run in synchronous mode then. However this option is not used at the curryman DAC as produced by miniDSP.
Not sure but maybe you can find some more info regarding different clocks for the ES9023 in the SUBBU DAC development thread.
kind regards, Daniel
Hi we used a 50 MHz XO for the ES9023 in asynchronous mode and we used simple 12 MHz crystals for the WM8804 (first we wanted to use a 12 MHz XO for the receiver as well). I tried the 50 MHz Fox Xpresso myself with good results but I liked EuroQuartz XO91 slightly better (by ear, oh oh) and it consumes less power. Also TXC XO's were used but an extensive comparison was not done. As the Subbu DAC is low cost using high priced XO's was not considered so we picked the ones I mentioned by looking at the specs and tried to find the best specified ones that were also well priced. Truth be told we also don't have access to equipment to measure phase noise.
At first I thought the anti jitter mechanism in ES9023 would not care about XO quality but this turned out to be different as differences between XO's can be heard. It was also clear that using a good regulator and decoupling were mandatory. I think experimenting is certainly worthwhile. The Vectron VCC1-B3D-50M0000 seems a nice one. If it would have been available here I would certainly have tried it.
At first I thought the anti jitter mechanism in ES9023 would not care about XO quality but this turned out to be different as differences between XO's can be heard. It was also clear that using a good regulator and decoupling were mandatory. I think experimenting is certainly worthwhile. The Vectron VCC1-B3D-50M0000 seems a nice one. If it would have been available here I would certainly have tried it.
Last edited:
I swapped out the oscillator in a MiniDSP OpenDRC-DA (which uses an ES9023) with an NDK oscillator. It was indeed a booger to solder in (don't even try without a microscope!), but I think it made a nice difference.
- Status
- Not open for further replies.