That would be bad. CT7302 is not an ultra-low phase noise reclocker. It sounds awful in comparison to what LTK is doing. I know, since I have tried CT7302 and consulted with their factory to try to get it to sound decent.
Probably there will be when @Cestrian gets ready to release it. Maybe he will chime in to let his know his plans on that.
You probably tried their EVM board? Evm is bad sounding. For examplle this one https://www.audiodesignguide.com/DSC2/index.html got 1st place in Vnav forum where audio enthusiasts gathered with their audio devices, like a little competition, which tells that ct7302 is not bad at all.
Last edited:
It sure is. I asked them if improving the power supplies from what they are using might help. They said no it won't help, there is nothing you can do to make better. With that, I said I can't use it.
In any case, CT7302 is still not going to be a good reclocker. You might be able to reclock its output with Cestrian's or with my reclocker board. You still need to do other things right too though. IMHO its worth trying PCM2DSD and JL Sounds I2SoverUSB. Even if you decide later to try using something else, at least you will have some standard to compare to. If you can get some other architecture to sound as good or better that then you probably have something very good. However, if you don't have a good reference dac then there is a problem which is that standard measurements will not be fully predictive of SQ -- you should have a reference dac system to listen to as well. Remember, LTK said this is now the best diy dac he has heard, and there are still other things to try to make even better. I can tell you I already tried most of those other things, and IMHO LTK hasn't heard that much of what Marcel's dac can do yet. It may be able to sound similar to a Chord DAVE with Prescaler or something else in that class. We are talking about maybe a >=$15k dac.
Okay, I'm done. If you are interested in trying Marcel's dac with good reclocking then I would encourage you to go ahead and do it. If not, then that's okay too.
In any case, CT7302 is still not going to be a good reclocker. You might be able to reclock its output with Cestrian's or with my reclocker board. You still need to do other things right too though. IMHO its worth trying PCM2DSD and JL Sounds I2SoverUSB. Even if you decide later to try using something else, at least you will have some standard to compare to. If you can get some other architecture to sound as good or better that then you probably have something very good. However, if you don't have a good reference dac then there is a problem which is that standard measurements will not be fully predictive of SQ -- you should have a reference dac system to listen to as well. Remember, LTK said this is now the best diy dac he has heard, and there are still other things to try to make even better. I can tell you I already tried most of those other things, and IMHO LTK hasn't heard that much of what Marcel's dac can do yet. It may be able to sound similar to a Chord DAVE with Prescaler or something else in that class. We are talking about maybe a >=$15k dac.
Okay, I'm done. If you are interested in trying Marcel's dac with good reclocking then I would encourage you to go ahead and do it. If not, then that's okay too.
Last edited:
Thats all subjective. Ct7302 can't have 32 bits dynamicrange(192dB), THD+N (-174dB) in case it have an phase noise isue. And E1DA Cosmos ADC would not outpreform some $30k audio measurement devices in case their ct7601 audio usb out have some isues. I think Comtrue is a real!
ok, I accept a different opinion! I will do some measurements soon, booth dsc2 and dsc2 frontend with ina849
ok, I accept a different opinion! I will do some measurements soon, booth dsc2 and dsc2 frontend with ina849
I will probably try in case when I find the right post which contain schematic, bom, gerber. And probably i will share ina849 based analog frontend too as I want to move to DDPD development fully.
Two years ago I had isue with dc servo, this is part of the frontend for dsc2, might help you get high gain on AC and eliminate common mode isuues on firdac due to offset errors?
https://e2e.ti.com/support/amplifie...-simulation-bug-with-ina846-or-thp210-in-tina
https://e2e.ti.com/support/amplifie...-simulation-bug-with-ina846-or-thp210-in-tina
Last edited:
I admit to being a little sceptical at first regards the re-clocking , I'm now just grateful to have been convinced to try it . What amazes me is the potential Marcels dac has and theres still more to squeeze out of it , I'm just getting a taster .
Last edited:
Sorry for the delay in releasing the gerbers/BOM. I am having a small issue with one of the three boards that have been built (One v13 and 2 x v14) which I am sure will be a soldering fault but I wanted to be 100% certain that it is a build fault and not a design error before releasing. I just haven't had the time to devote yet to fault tracing. Hopefully I can get to it this weekend.
With such a high dynamic range and such low THD+N values, I'm sure they used a completely digital measurement that ignores analogue imperfections such as phase noise.
What matters for the shift register FIRDAC is that the bit clock is clean. A bit of jitter on the data inputs doesn't matter, as the data inputs sdinL and sdinR are immediately synchronized with the bit clock by flip-flops U20A and U20B.
If you want to use a CT7302 and no reclocker board and if what I'm about to suggest is at all possible, the best solution would be to run a crystal oscillator off a clean supply, connect its output to the bit clock input of the DAC, also provide it to the CT7302 and let the CT7302 produce a DSD signal with a bit rate equal to the crystal oscillator's frequency.
I wrote "if what I'm about to suggest is at all possible" because it's not clear to me if it would fit with the set-up and hold times. Besides, it isn't really clear to me what the CT7302 can and cannot do. In particular, it's not clear to me what it does with its ASRC_REF_CLK_IN and crystal inputs.
I only see an "SRC software mode table" on page 13 of https://www.audiodesignguide.com/DSC2/eichhornia2_CT7302.pdf
Mode 000 has an output sample rate equal to the frequency at ASRC_REF_CLK_IN, but I guess that applies to PCM.
It can work with a crystal of about 12 MHz and produce DSD256, so then the output bit clock frequency must be equal to what is on XTAL_ON.
Mode 000 has an output sample rate equal to the frequency at ASRC_REF_CLK_IN, but I guess that applies to PCM.
It can work with a crystal of about 12 MHz and produce DSD256, so then the output bit clock frequency must be equal to what is on XTAL_ON.
CT7302 supports an external clock for clocking the output. However, there is another clock there which is not an audio frequency clock and runs at somewhere around 11MHz, IIRC. Lots of digital noise going on inside that thing, and the upsampling and conversion to DSD may involve some ASRC using non-audio, high-jitter 11MHz crystal.
No, ct7302 reguire fine routing and forget any pins like 2.54! Only fine routing or coaxial is alowed otherwise you getting digital noise as like on their evm. I faced too using 2.54 pins until moved to u-fl. Take a look for E1DA Cosmos ADC, very popular ultra low noise audio measurement devices, inside E1DA is ct7601 with their 12MHz crystal only! Crystal is enought for booth ct7302 and ct7601 or ct7602 recently developed.
Last edited:
However, ASRC requires a low phase noise reference clock. Also it tends to be more accurate if fed low phase noise PCM at the input, so the PPLL (poly-phase locked loop) doesn't have too much jitter to try to attenuate in the first place. You really need synchronous SRC for best results. Even then upsampling filter design choices, etc., can make a big difference. That's my experience, anyway.
Last edited:
Ofcourse, I'm looking for an nice clock for booth CT + DSP, probably I will do Driscoll because LMK https://www.ti.com/lit/ds/symlink/lmk05318b.pdf looks not good for audio with their -70dBc@10Hz, but will be good for an phase locked oven control and nothing more. Curently I'm looking for muti clock, need 12MHz + 12MHz + 12.288MHz, thinked to do it with LMK but its not god for audio I see now : ) And it will be ultra hard to realise with Driscoll. Probably I will buy two ocxo for this.
Last edited:
The idea for omitting the Driscoll based oscillator which also requires a stable temperature if a stable frequency is desired besides phase linearity is to use the ct7302 directly as a driver for the crystal, at the same time the ideal thing is to find a 12MHz sc cut crystal because that way the CT would be regularly used for something at the same time e.g. for anything2pcm -> dsp -> seccond ct7302 for pcm2dsd -> firdac, dsc2, ddpd or whatever, that's my idea. The only thing I need to do is make an ultra stable oven and then a very good result can be expected. Frontend with two ct7302 + ct7601 + adau1462 is my need now for many diferent digital projects, expecialy for direct digital power dac ddpd. : )
If you mean that you want to use the crystal oscillator on board the CT7302, that is bound to result in more crosstalk from the digital signal processing to the crystal clock than a separate oscillator on a clean supply. I can't tell you how much worse it gets, though. Unlike Mark, I'm not worried about close-in phase noise, but more about the phase noise floor and spurious tones, particularly spurious tones around an odd multiple of half the bit clock frequency.