There was same 9038d6k, Cosmos ADC, no preamp.
You're right about ES9038Q2M 12.5MHz MCLK noise shaping but it's hump is about 20dB lower compared to CS43131. Ok, it's not the topic.
Due to very low noise floor of CS43131 I see no reason to measure without preamp.
Setup: REW Gen 1kHz Sine@-60dB - USB ISO - Meizu Hifi DAC - Cosmos Scaler +26dB - Cosmos ADC - REW
Unweighted DR = 129.1dB [20..20000Hz]
A-weighted DR = 131.4dB(A)
Setup: REW Gen 1kHz Sine@-60dB - USB ISO - Meizu Hifi DAC - Cosmos Scaler +26dB - Cosmos ADC - REW
Unweighted DR = 129.1dB [20..20000Hz]
A-weighted DR = 131.4dB(A)
Of course, if there will be some harmonic higher -30dB. Unfortunately all noise floor comparations including TD+N measurement are not exact and are very dependent on FFT length. That is why I use THD+N.
When you compare the noise floor of course you should keep the FFT length the same. The noise number given by REW is not dependent on FFT length.
The IMD plot at ASR review of Meizu Hifi Pro seems to indicate that there is a noise issue with CS43131.
https://www.audiosciencereview.com/...i-audio-pro-imd-audio-measurements-png.42761/
Can you show your measurment of e.g. CCIF IMD?
The IMD plot at ASR review of Meizu Hifi Pro seems to indicate that there is a noise issue with CS43131.
https://www.audiosciencereview.com/...i-audio-pro-imd-audio-measurements-png.42761/
Can you show your measurment of e.g. CCIF IMD?
The TD+N numbers on those indicate that there is a noise issue although the noise floors look quite good. Nothing like what you showed in your OP with 1k+30k.
The best TD+N here is -91.4dB. The mentioned my "real DR" measurement -110.3dB is very much better. Also, all IMD measurements contain many harmonics, so TD+N does not reflect the noise level well.
I don't see any reason to check this, but if I reduce the IMD signal by 20-30dB, the noise floor will change in the same way.
In most DACs I've measured TD is dominating in TD+N. In your measurements it is the opposite which indicates that there is a noise issue. But as said same thing can be seen in the ASR review so nothing new here.
I have never seen comments in reviews that devices based on the CS43131/CS43198 do not sound the way a dynamic range of 130dB should sound.
To my ears, the CS43131 (meizu hifi dac) sounds significantly worse in range than the es9038q2m (e1da 9038d6k). And on the other hand, I cannot distinguish this CS by ear from the es9281 (hiby fc3: THD+N -102dB, DR 118.5dB).
I haven't read any reviews on CS43131/CS43198 so cannot comment on that. But clearly the ASR review of Meizu is flawed as it did not report this behaviour.
My interest is purely research. The CS43131 isn't really that bad, but it's far from the best for hifi. Thanks.
Just for comparision:
This is the multitone measurement of the Meizu Hifi Pro Dac. This is the device nick_I tested:
and this is the multitone measurement of Tanchjim Space:
Contrary to the Topping D30PRo the iFi GO Bar dosen't use DSD noise shaping with PCM streams:
So some dacs use it and some not. Is this due firmware/register settings?
This is the multitone measurement of the Meizu Hifi Pro Dac. This is the device nick_I tested:
and this is the multitone measurement of Tanchjim Space:
Contrary to the Topping D30PRo the iFi GO Bar dosen't use DSD noise shaping with PCM streams:
So some dacs use it and some not. Is this due firmware/register settings?
The PCM noise shaping of CS43131 starts at 45kHz but the graph shows only up to 40kHz. DSD noise shaping happens already when DSD is generated, not at dac.
Not exactly. My device is not Pro but it means nothing without load.
ASR measurements of both Meizu is too noisy.
This FFT was taken for me by one guy from e1da discord channel:
I can't answer for sure. As CS43131/CS43198 have a LPF "by design" the answer "no" is more likely.
Sorry, I can only trust the measurements at ASR because Amirm tested hundreds of devices in the same way, at the same place, with the same equpment etc. It's not about to get absolute numbers but the ability to compare the devices among each other. And at ASR the E1DA with ES9038Q2M scored worse than the best dongles with CS43131. In general you can only compare devices and not the dac chips itself.
Sorry, but as I said before in this case ASR measurements are flawed as they failed to notice the noise behaviour of CS43131 with IMD. IMO this is a prime example for those who argue that ASR measurements have little to do with how the device sounds.
I understand your position, but these conditions are very far from ideal. For example, how could information about the behavior of a dongle on someone else’s noisy USB host be valuable to you? Therefore, comparison of results is sometimes not correct at all.
But my topic is not about comparison, but about the fact that standard direct measurements of devices on the CS43131 cannot be completely trusted.
By the way, similar technologies (DRE) appeared in ESS chips. This is alarming.
