Hi
TLDR: Has anyone experience in PCM9211 in high performance application (<112dB THD+N) - can it do it?
I'm designing a new SPDIF to PCM board using a PCM9211 to replace my AK4115-based solution due to discontinued IC. However, I'm having issues with the performance of the PCM9211.
I've got an AK4493S based DAC PCB that can do <-116dB THD+N with I2S or with my old AK4115-based receiver. However, with PCM9211 I get 'only' -112dB (1kHz). Moreover, the result deteriorates quite significantly with higher frequency. Also, it is sensitive to SPDIF voltage level; going up or down from nominal 500mVpp increases performance, and also the digital signal level heavily impacts the THD+N - mostly via increased noise floor. Also, performance slightly increases with sample rate. The performance is slightly better with Optical input than coaxial. All this is something I haven't seen with any other SPDIF receivers I've worked with.
I've been modifying my circuit and doing dirty layout tricks but nothing has helped so far. I haven't tried to analyse or compare the clock quality coming out of the PCM9211 yet. This is a fairly complex chip but I don't see anything obviously wrong in configuration, also when comparing to EVM config scripts. For example, I use 24.576M MCLK in all cases PLL wide/narrow mode doesn't change anything. It almost feels like this is somehow related to the analog parts of the IC, mainly the PLL. But afaik there is nothing to do about it, it uses an external filter but component values are given in the DS. All caps are C0G and very close to the IC.
Before digging deeper, I'd like to ask if anyone has experience with this IC in high performance applications? Just thinking am I trying to fix unfixable or is there something wrong with my circuit. Any experience is welcome. TBH I'd be a bit surprised if the performance of the IC so 'bad' - but then again this is aimed at AVRs and cheaper end receivers where this wouldn't be a limitation.
Thanks!
TLDR: Has anyone experience in PCM9211 in high performance application (<112dB THD+N) - can it do it?
I'm designing a new SPDIF to PCM board using a PCM9211 to replace my AK4115-based solution due to discontinued IC. However, I'm having issues with the performance of the PCM9211.
I've got an AK4493S based DAC PCB that can do <-116dB THD+N with I2S or with my old AK4115-based receiver. However, with PCM9211 I get 'only' -112dB (1kHz). Moreover, the result deteriorates quite significantly with higher frequency. Also, it is sensitive to SPDIF voltage level; going up or down from nominal 500mVpp increases performance, and also the digital signal level heavily impacts the THD+N - mostly via increased noise floor. Also, performance slightly increases with sample rate. The performance is slightly better with Optical input than coaxial. All this is something I haven't seen with any other SPDIF receivers I've worked with.
I've been modifying my circuit and doing dirty layout tricks but nothing has helped so far. I haven't tried to analyse or compare the clock quality coming out of the PCM9211 yet. This is a fairly complex chip but I don't see anything obviously wrong in configuration, also when comparing to EVM config scripts. For example, I use 24.576M MCLK in all cases PLL wide/narrow mode doesn't change anything. It almost feels like this is somehow related to the analog parts of the IC, mainly the PLL. But afaik there is nothing to do about it, it uses an external filter but component values are given in the DS. All caps are C0G and very close to the IC.
Before digging deeper, I'd like to ask if anyone has experience with this IC in high performance applications? Just thinking am I trying to fix unfixable or is there something wrong with my circuit. Any experience is welcome. TBH I'd be a bit surprised if the performance of the IC so 'bad' - but then again this is aimed at AVRs and cheaper end receivers where this wouldn't be a limitation.
Thanks!
AK4118 is available and if you use only the DIR feature of the PCM9211 it could be much better choice.
https://www.digikey.fr/fr/products/...18AEQ/2333356?s=N4IgTCBcDaIIYGsAsBGFAOEBdAvkA
https://www.digikey.fr/fr/products/...18AEQ/2333356?s=N4IgTCBcDaIIYGsAsBGFAOEBdAvkA
Unfortunately Wolfson had issues with 192k sample rates in WM8804/5, or at least it was awkward to detect?
I may need to re-design (again..) using AK4118. I did use other features of the PCM9211 in this design as well, offering external PCM routing and optional ADC etc as well.. and as it's a fairly complicated design (for audio receiver) and working otherwise perfectly it's a bit annoying if I need to ditch the whole design because of this DIR. So I'd like to know if it's fixable, otherwise it will likely be an AK4118 based design.
AK4115 was great to use without MCU thanks to the automatic MCLK/WCLK multiplier - you could get 24.576M out for all 48k/96k/192k rates automatically. AK4118 seems to miss this? So in order to achieve the same, you would need an MCU as well, right?
I may need to re-design (again..) using AK4118. I did use other features of the PCM9211 in this design as well, offering external PCM routing and optional ADC etc as well.. and as it's a fairly complicated design (for audio receiver) and working otherwise perfectly it's a bit annoying if I need to ditch the whole design because of this DIR. So I'd like to know if it's fixable, otherwise it will likely be an AK4118 based design.
AK4115 was great to use without MCU thanks to the automatic MCLK/WCLK multiplier - you could get 24.576M out for all 48k/96k/192k rates automatically. AK4118 seems to miss this? So in order to achieve the same, you would need an MCU as well, right?
Yes, you're right, in order to keep MCKO at 24.576M over 44.1k to 192k range you need an UC to change register 0 content...
I have no experience with this PCM9211 chip. Regarding the weird issues you describe, I'll check the supply, DS describe at least 3 different ones, with independent grounds, but you certainly already checked theses...
I have no experience with this PCM9211 chip. Regarding the weird issues you describe, I'll check the supply, DS describe at least 3 different ones, with independent grounds, but you certainly already checked theses...
I have never used PCM9211, but many time used it's brother - DIX9211 in various projects.
1) With the coaxial input, I got 1-2 dB less THD+N than through the optical one.
2) I did not see the influence of the +/-50% coax spdif amplitude (cable was 1.2m).
3) THD+N was 1-2dB worse at 48x than at 44x, but for the some reason, I'm not sure if was because of the receiver.
1) With the coaxial input, I got 1-2 dB less THD+N than through the optical one.
2) I did not see the influence of the +/-50% coax spdif amplitude (cable was 1.2m).
3) THD+N was 1-2dB worse at 48x than at 44x, but for the some reason, I'm not sure if was because of the receiver.