Thanks for sharing the photo. I do something similar right now. In my case the 1121 plugs in from below, the Amanero plugs in from above and the I2S signal is routed to the 1121 through an isolator. Furthermore, my board also contains a B1-fashioned volume control with DS1882 + Atmega 328P + IR and a handful of Salas shunt regs which feed all these modules. Integrating the USB-I2S converter on board is the last missing step towards maximum compactness.
As much as I remember the onboard volume control option, it shall rely on the reduction of some kind of a reference voltage. Without falsely claiming that I fully understand all of its details this hints to me that the SNR may suffer at low volume levels, where the last bits may get lost in the noise floor. If I do the analog attenuation of the 0dB output I attenuate the signal and the noise equally, preserving the best possible SNR until I hit the noise floor of the next stage. Does this sound very naive...?
Another argument is the much lower output impedance of the B1 compared to the bare 1121 which never harms. (This could of course also be achieved with an external opamp, as I see in your photo.)
In my case the Atmega 328P is responsible for IR reception, some face plate indicator LEDs, setting and storing the DS1882 levels as well as soft-start and speaker relay controls + related safety features for the power amp boards. (The full setup is actually a digital input integrated amp with Amanero USB + 1121 + B1 + exchangeable AB power amp boards.)
What is the overall sonic experience with the 3 x Salas + 1121 combination?
Another argument is the much lower output impedance of the B1 compared to the bare 1121 which never harms. (This could of course also be achieved with an external opamp, as I see in your photo.)
In my case the Atmega 328P is responsible for IR reception, some face plate indicator LEDs, setting and storing the DS1882 levels as well as soft-start and speaker relay controls + related safety features for the power amp boards. (The full setup is actually a digital input integrated amp with Amanero USB + 1121 + B1 + exchangeable AB power amp boards.)
What is the overall sonic experience with the 3 x Salas + 1121 combination?
1121 dont have the volume "by pot" option - 1021 has it. But both operate in the digital domain. So output noise is not lowered as volume is lower - correct. A pot on output however does lower noise. Now, noise it not a problem with the DAM DACs.... so...
I bet that a 1121 using the digital volume is much cleaner than using a preamp. The only hit the DAM takes from a low impedance is a voltage drop.... if the restoration ladder is used directly that is (i.e. no OP-amps)
//
I bet that a 1121 using the digital volume is much cleaner than using a preamp. The only hit the DAM takes from a low impedance is a voltage drop.... if the restoration ladder is used directly that is (i.e. no OP-amps)
//
There is no "circuit" for the volume control. The volume is changed by recalculating in floating point aretmetics the incoming PCM stream before D/A conversion. How much it should be attenuated is either given by a command (1121 and 1021: set vol -10) or for 1021, corresponds to a resistance between two HW pins. This resistance is "measured" and a the corresponding attenuation is calculated - from this point on, the same thing happens as if you where to set it by command - the PCM stream is recalculated to represent the new wanted level and send forward to the actual D/A conversion.
This is how it works.
//
This is how it works.
//
Thank you for your explanation. This may clarify why I do not hear any significant difference between changing the volume on the 1021 by a pot / command or changing the volume on the source device (I2S from an RPI via its GPIO header). Nevertheless, how does such an approach affect the SNR, then? Naively, it seems to me that like this the SNR is expected to degrade at lower volumes. If so, that would indeed justify the 0dB output level+ analog attenuation approach.
SNR do degrade but the noise level is so low its not like it gets audible noisy at say -30 dB... I think you need to do you own comparisons - myself I have settled on "digital volume" control - many pots have bad balance between channels and will degrade contact as time goes with possible distortion addition - the is not a problem with the digital version. I think it can be wise to cater for a good gain structure - i.e. not running the input to the DAC at max - 30/40 dB... but rather see to that you can play your favourite tracks at desired level with yet say 3-5 dB margin. It all depends on the sensitivity of your amps and speakers.
//
//
Good advice all round. For what it’s worth, I too prefer the sound of the built-in digital volume control - in a gain-matched system. The digital volume code isn’t difficult to implement with a microcontroller. And you already have one in situ.
I tried the 1021 with the pot once and couldn’t hear any difference - and didn’t expect to as the pot simply varies the ‘vol = xx’ commands - but the ‘feel’ of using a pot compared to a rotary encoder or IR remote wasn’t to my liking.
I may have dreamt this but I could swear I read awhile back in this looongg thread that actual 0dB output was at ‘vol = 2’ as against ‘vol = 0’ when using the DAM’s digital volume control.
Also, I believe the stated 1.4VRMS output stated in the specs is at the DAM’s +10 setting. A quite important thing to bear in mind when gain matching. Actual 0dB output was measured by a friend of mine (on an Audio Precision) at around 1.1VRMS.
The output opamp in my earlier photo is set up to give the slight amplification necessary to get 1.1VRMS to the consumer-standard 2V output single-ended. I don’t implement any ‘+’ volume settings in the uC code.
Finally, I prefer working with the 1121 because of its smaller size and the flexibility it affords in that you choose your own power supply and whether or not to go with isolators. And, in the realm of subjectivity, the lizard part of my mind has always preferred the sound of the 1121 - don’t ask me to explain why, my toes just tap more! Both boards are great DACs and the SNR is not adversely affected by the built-in volume control according to my ears and as I understand the sign magnitude implementation. But I’m not an EE.
I tried the 1021 with the pot once and couldn’t hear any difference - and didn’t expect to as the pot simply varies the ‘vol = xx’ commands - but the ‘feel’ of using a pot compared to a rotary encoder or IR remote wasn’t to my liking.
I may have dreamt this but I could swear I read awhile back in this looongg thread that actual 0dB output was at ‘vol = 2’ as against ‘vol = 0’ when using the DAM’s digital volume control.
Also, I believe the stated 1.4VRMS output stated in the specs is at the DAM’s +10 setting. A quite important thing to bear in mind when gain matching. Actual 0dB output was measured by a friend of mine (on an Audio Precision) at around 1.1VRMS.
The output opamp in my earlier photo is set up to give the slight amplification necessary to get 1.1VRMS to the consumer-standard 2V output single-ended. I don’t implement any ‘+’ volume settings in the uC code.
Finally, I prefer working with the 1121 because of its smaller size and the flexibility it affords in that you choose your own power supply and whether or not to go with isolators. And, in the realm of subjectivity, the lizard part of my mind has always preferred the sound of the 1121 - don’t ask me to explain why, my toes just tap more! Both boards are great DACs and the SNR is not adversely affected by the built-in volume control according to my ears and as I understand the sign magnitude implementation. But I’m not an EE.
Thank you both for your insights. Just to make sure that I understand "gain-matching" correctly: am I supposed to set the gain and sensitivity of the power amp such that the typical volume level of the 1121 (or the one of the B1 in my current arrangement) is not too low (SNR argument) but of course still low enough to provide some headroom for louder listening? Or do I miss something here?
Yes, the idea is to have all the components in the system operating at or near their highest SNR. Also, max output should not drive the following component into clipping.
TNT, yes, I have wondered if that perceived uptick isn’t due to that post-FPGA clocking implementation. I’ve tried with and without isolators and it didn’t seem to make a difference. Still sounded great either way. But, who knows… once your efforts give you pleasure then trust your ears.
I built a sort of test-bed 1121/RPI2W/Roon device with an isolator on the I2S lines and using the opamp output. It’s powered by separate RECOM SMPS’s for the 1121’s digital and analog PS requirements and a separate SMPS for the Pi (clean side of isolator powered by the 1121’s 5VDD) and it runs on that board with the embedded Atmega - display, IR, rot enc etc. Most audible difference I heard of any changes/experiments was when I shoe-horned in three of Mark Johnson’s filters. Worthwhile mod.
TNT, yes, I have wondered if that perceived uptick isn’t due to that post-FPGA clocking implementation. I’ve tried with and without isolators and it didn’t seem to make a difference. Still sounded great either way. But, who knows… once your efforts give you pleasure then trust your ears.
I built a sort of test-bed 1121/RPI2W/Roon device with an isolator on the I2S lines and using the opamp output. It’s powered by separate RECOM SMPS’s for the 1121’s digital and analog PS requirements and a separate SMPS for the Pi (clean side of isolator powered by the 1121’s 5VDD) and it runs on that board with the embedded Atmega - display, IR, rot enc etc. Most audible difference I heard of any changes/experiments was when I shoe-horned in three of Mark Johnson’s filters. Worthwhile mod.
I will. Have to open one of the 1121’s but that’s no problem.
I’ve not paid a lot of attention to Toslink other than implementing the input on a few 1021s I built for friends. Most of my efforts have used USB-I2S boards or gone straight from the I2S out of various RPIs - usually via a custom PCB.
In your opinion is there any downside to the necessary conversions at either end?
What are you using as an optical source?
I’ve not paid a lot of attention to Toslink other than implementing the input on a few 1021s I built for friends. Most of my efforts have used USB-I2S boards or gone straight from the I2S out of various RPIs - usually via a custom PCB.
In your opinion is there any downside to the necessary conversions at either end?
What are you using as an optical source?
I don't like to draw in computer communication type HF into a DAC so I try to avoid USB. I don't mind the 192ksps limit of the toslink - it's ok for me. I have to system in different places - in my active 2way I use a RME digiface toslink and in an other I connect directly to a Mac Mini. I find that if one stop the clock continuous clock adjustment in the DAMs, the isolation from the source seems total and the DAC performs to its maximum extent given the built in clock. It seems really fine and I'm satisfied with this architecture. It has a calm to it that is very nice and dynamics is wonderful - if one hen the right stuff for the rest.
//
//