ESS 9018 Volume control?s
Hi all. I am wondering a little about ESS' volume control implementation. The white paper says all (digital) noise products are below -170 dB coming out of the DSP section of the chip, two questions:
1. does this include noise products caused by applying digital attenuation-ie quantisation noise?
2. does ESS apply dither when the volume control is active to smooth out the digital noise floor?
I am not at all concerend about resolution loss (bit loss), but I would like to know more about how the volume control is handled. Since is not forthcoming to the public with these details, perhaps someone here knows more which can be divulged (Russ)?
Thanks
Hi all. I am wondering a little about ESS' volume control implementation. The white paper says all (digital) noise products are below -170 dB coming out of the DSP section of the chip, two questions:
1. does this include noise products caused by applying digital attenuation-ie quantisation noise?
2. does ESS apply dither when the volume control is active to smooth out the digital noise floor?
I am not at all concerend about resolution loss (bit loss), but I would like to know more about how the volume control is handled. Since is not forthcoming to the public with these details, perhaps someone here knows more which can be divulged (Russ)?
Thanks
No:
That is an overly simplistic view and not entirely correct. Applying digital attenuation also results in quantisation noise, which can be spiky noise artifacts which can be audible unless the signal is properly dithered. Additionally, digital attenuation lowers the overall S/N of the DAC, as the attenuation (unlike most analog attenuation approaches) only attenuates the signal, and not the noise floor of the DAC (including that of the analog output stage).
The noise issue would only be a concern at high levels of attenuation, and/or with a system with poorly matched gain structures, but quantisation noise can be a real issue if the signal is not properly handled, hence my question above: if ESS' white paper statement that all noise products from the DSP section of the chip are below -170 dB applies to volume control as well, then there is no concern (and to me this would indicate both the use of dither and some rather tricky noise shaping applies to the volume control).
That is an overly simplistic view and not entirely correct. Applying digital attenuation also results in quantisation noise, which can be spiky noise artifacts which can be audible unless the signal is properly dithered. Additionally, digital attenuation lowers the overall S/N of the DAC, as the attenuation (unlike most analog attenuation approaches) only attenuates the signal, and not the noise floor of the DAC (including that of the analog output stage).
The noise issue would only be a concern at high levels of attenuation, and/or with a system with poorly matched gain structures, but quantisation noise can be a real issue if the signal is not properly handled, hence my question above: if ESS' white paper statement that all noise products from the DSP section of the chip are below -170 dB applies to volume control as well, then there is no concern (and to me this would indicate both the use of dither and some rather tricky noise shaping applies to the volume control).
The noise shaping needn't be that tricky if the attenuation happens at 2x or 4x the sampling rate.
Just received a Buffalo II DAC and using a USB to I2S interface. I've had trouble with hiccups. I removed the firmware chip and now it's working fine. I know there's a new firmware chip.
But here's my question and concern. I'm hearing that the cause for the dropouts is related to the I2S source having a high level of jitter. Since I bought this interface to get a connection with low levels of jitter, this is obviously a concern to me.
Can Twisted Pear Audio say for sure that it's the source's jitter causing the dropouts or is it simply a weakness with the firmware?
And if someone can recommend a solution (SPDIF or I2S) together with Buffalo II that gives very low level of jitter, that would be great.
But here's my question and concern. I'm hearing that the cause for the dropouts is related to the I2S source having a high level of jitter. Since I bought this interface to get a connection with low levels of jitter, this is obviously a concern to me.
Can Twisted Pear Audio say for sure that it's the source's jitter causing the dropouts or is it simply a weakness with the firmware?
And if someone can recommend a solution (SPDIF or I2S) together with Buffalo II that gives very low level of jitter, that would be great.
(based on experimental evidence and reports from other users) It is a combination of setting the DPLL bandwidth at its lowest level in the current firmware and the fact that almost no board can feed signal clean enough for that setting. You will observe that the hiccups will reduce as the chip warms up (say after 1/2 hour of power on) but will never go away with this setting. The only practical solution is to use spdif or wait for the new firmware that uses a higher setting for the dpll...
BTW, removing the firmware removes the inputs to 1/2 the dacs, so you loose 3 db of SNR (Can you tell?
)
BTW, removing the firmware removes the inputs to 1/2 the dacs, so you loose 3 db of SNR (Can you tell?
)
There is already a new firmware chip in sale that takes care of this. But I don't want to take that route if I get high levels of jitter with this USB to I2s interface. That's what I'm trying to figure out. It would also be quite interesting considering that the whole idea behind this interface was to have super low jitter..... And yes, I'm also wondering what the threshold of audible jitter is.
Hehe. I did think I noticed something when I removed it, but I would have to do a AB testing to say something for certain.BTW, removing the firmware removes the inputs to 1/2 the dacs, so you loose 3 db of SNR (Can you tell?
I meant to say "wait in the mail" for the new firmware
You can try implemented an outboard uP (Arduino) and select the dpll bandwidth that works for you. I believe the dpll setting in the alternative firmware is "best". The scale is lowest, low, med-low, medium, med-high, high and highest. And then there is "best". I can't tell you exactly where "best" lies in the scale; can only speculate.
For me, after the chip warms up, I can set it to med-low (for 44.1K SR) with no hiccups. On a lucky day I can set it to "low", but that setting doesn't work all the time...
You can try implemented an outboard uP (Arduino) and select the dpll bandwidth that works for you. I believe the dpll setting in the alternative firmware is "best". The scale is lowest, low, med-low, medium, med-high, high and highest. And then there is "best". I can't tell you exactly where "best" lies in the scale; can only speculate.
For me, after the chip warms up, I can set it to med-low (for 44.1K SR) with no hiccups. On a lucky day I can set it to "low", but that setting doesn't work all the time...
Jitter??
I can tell you this much-TPA normally sets the DPLL bandwidth to the lowest setting (this should give best performance with low jitter sources). I use a Wavelength Audio, Wavelink USB-SPDIF converter, with a Stereovox XV2 BNC SPDIF cable to the B-II, and I have no issues with any glitches or dropouts at data rates up to 24/192. I suspect that if you are having dropouts, your interface may not have an output which is good enough (low jitter) for the low DPLL setting of the ESS 9018. My Wavelink appears to have no problem with jitter on its output.
I can tell you this much-TPA normally sets the DPLL bandwidth to the lowest setting (this should give best performance with low jitter sources). I use a Wavelength Audio, Wavelink USB-SPDIF converter, with a Stereovox XV2 BNC SPDIF cable to the B-II, and I have no issues with any glitches or dropouts at data rates up to 24/192. I suspect that if you are having dropouts, your interface may not have an output which is good enough (low jitter) for the low DPLL setting of the ESS 9018. My Wavelink appears to have no problem with jitter on its output.
I hope Twisted Pear Audio can give an answer here. I've already received an indication from them that the cause of the dropuouts is high level of jitter of from the I2S source. But I would like a confirmation. The I2S source we're talking about here had the goal in their design of absolutely lowest possible jitter and that's why I bought it. If the oppostite is the case, I would like to know about it. People should know about these things.
The goal and the realization are two completely different things.
There are literally hundreds of Buf II users out there, and quite a few use high speed I2S (>192khz).
the simple fact is that some sources will require higher bandwidth to retain a lock and some will not. The compromise is that opening up the DPLL bandwidth also allows more jitter (or more accurately phase noise) through the DAC. So pick your poison. If your source requires higher DPLL bandwidth you have the option to use that firmware, but most people would rather use a source that does not require that and have better phase noise rejection.
The fact that some source has a hard time locking does not necessarily mean that source is "bad" BTW it just means you may need to tweek things a bit.
There are literally hundreds of Buf II users out there, and quite a few use high speed I2S (>192khz).
the simple fact is that some sources will require higher bandwidth to retain a lock and some will not. The compromise is that opening up the DPLL bandwidth also allows more jitter (or more accurately phase noise) through the DAC. So pick your poison. If your source requires higher DPLL bandwidth you have the option to use that firmware, but most people would rather use a source that does not require that and have better phase noise rejection.
The fact that some source has a hard time locking does not necessarily mean that source is "bad" BTW it just means you may need to tweek things a bit.
Just because I like threads with an occasional picture.
Thanks Twisted Pear, the B2 sounds GOOD and reveals more every day !
Nice box, where did you bought? excellent & neat work, congrats
Just because I like threads with an occasional picture.
Thanks Twisted Pear, the B2 sounds GOOD and reveals more every day !
Wow that looks awesome. Nice use of the dcb1 buffers too.
What else is in that box?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.