SPL Volume8 Review

[lecoyote]

I am just not sure when the SPL makes sense. If you are doing multichannel you are a probably sticking with your AVR for volume control.

SPL products are targeted to the pro sector. Since professional setups are likely to have more constraints like having analogue devices in the chain and requiring quick and precise access to volume control at all time, these expensive volume controls make sense. Or at least, they *made* sense for a while... with the newer digitally controlled volume IC, I'm not sure it does?

 

[lecoyote]

Ha, what a neat idea! How did you get audio to the 3700? Straight out of the Avid(?) on to HDMI?

Signal is coming out of a Blackmagic MiniMonitor 4K (which has both SDI and HMDI, HDMI being used to feed the AVR).

At first I had the idea of building a digitally controlled analog in/out volume control—but then though about including a HMDI de-embedder, a DAC and a DSP for audio delay compensation and speaker calibration. In the end, it made more sense to buy an AVR which had all those features and just control it via telnet. The AVR plus the diy network controller replace a lot of converters and cables that I had to use before.

 

[wisechoice]

I'm using it as an example because it is what I actually use (instead of a miniDSP) and it has decent but not SOTA analog performance. Tons of other pro audio devices have similar performance, ones that come to mind off the top of my head...MOTU Ultralite AVB es, MOTU 8A, MOTU 624, MOTU 828es, MOTU 624, RME UCX series, RME UFX series, RME Fireface 802. I've mentioned this already in this thread but a RME Fireface 800 has better noise performance on the lowest output setting (-10 dBV) than an Okto dac8 pro and has been available since 2004.

If you want to talk availability the miniDSP 4x10HD has been out of stock longer than the Mk5...

Michael

Yeah, you're right. Some of those are still available. I guess I had my list of favourites, in case I was going to replace my 4 DACs with one. And almost pulled the trigger on one or two of them, but decided to live with what I have for now. I didn't begin this journey to explore multi-channel, it sort of happened by accident and I kept adding DACs until I had a 5.1.2 setup. Now I need a hardware volume controller. Nobody should use my own system as a model for their own though, lol.

 

[pma]

By the looks of things this isn't a fully balanced design. It takes the differential inputs, converts them to single ended, does the volume control stuff, then sends the signal back through a single ended to differential line driver. What a disaster

That’s the only correct way to do it.

 

[5th element]

That’s the only correct way to do it.

Why?

 

[pma]

Why?

It has been replied and explained many times, sorry. I am not going to open another infinite circle of opinions.

 

[5th element]

It has been replied and explained many times, sorry. I am not going to open another infinite circle of opinions.

How educational.


I can understand not wanting to attenuate balanced with standard potentiometers. Given their channel mismatch you'd destroy the rejection ratio of the balanced line. But when using volume control chips with good inter-channel balance? This unit would have better performance and far less crosstalk that way.

 

[PeteL]

How educational.


I can understand not wanting to attenuate balanced with standard potentiometers. Given their channel mismatch you'd destroy the rejection ratio of the balanced line. But when using volume control chips with good inter-channel balance? This unit would have better performance and far less crosstalk that way.

I think way to much fuss is being made about maintaining balanced signal internally. I don't get where you take your theory about less crosstalk and better performance. aside higher voltage, if you need it. A balanced signal have very little meaning, what matter is balanced connections, balanced cables, balanced interface, that's inter-device that ground problems have a meaning, that interferences are picked up, that's what you want rejected. After that keeping a differential (not "balanced") signal from input to output is a design decision that can be valid to some designer but you can't make conclusions that it will perform better unless you are trying to solve a specific problem. A device with balanced inputs and balanced outputs is balanced, It will do what it's supposed to be doing and it will solve what balanced connections are made for.

 

[mdsimon2]

Interesting, I do have a 2x4 HD at the moment, but always listening if I can get better actual audible performance. Can you quickly let me know the signal flow, like where is the PI in the chain and connected how?
To be fully honest, I do have a PI somehere in a drawer, looks like it’´s a PI3, and also a 8 channel focusrite scarlett interface that does nothing at the moment, Had use for it in the past. Probably not SOTA. Did I purchase a 2x4 for no reason? My use case is typical, sub cross over and bass management. I have analog sources, and converted from digital sources, connected to a freya preamp.

A Focusrite Scarlett is going to be pretty similar to a miniDSP 2X4HD in terms of analog performance. My 18i20 2nd gen has just a bit more noise than my 2X4HD.

The link in my signature has a tutorial for setting up a CamillaDSP setup using Ubuntu Server. I've done it on a RPi3 but a RPi4 is a lot less stressed. Basic idea is you have a CamillaDSP capture device which receives a source signal, for a Scarlett 18i20 this can be analog, SPDIF or ADAT inputs. If you have the RPi setup as a streamer you can also setup the capture device as an ALSA loopback so output of your streamer software is routed to CamillaDSP. You then have a playback device which is the output channels of the Scarlett, this is all analog and digital output channels. DSP is applied in between the capture and playback devices with all of the stuff you expect (mixer, PEQ, FIR, x-over, delay, level adjustment, etc).

Michael

 

[5th element]

I think way to much fuss is being made about maintaining balanced signal internally. I don't get where you take your theory about less crosstalk and better performance. aside higher voltage, if you need it. A balanced signal have very little meaning, what matter is balanced connections, balanced cables, balanced interface, that's inter-device that ground problems have a meaning, that interferences are picked up, that's what you want rejected. After that keeping a differential (not "balanced") signal from input to output is a design decision that can be valid to some designer but you can't make conclusions that it will perform better unless you are trying to solve a specific problem. A device with balanced inputs and balanced outputs is balanced, It will do what it's supposed to be doing and it will solve what balanced connections are made for.

This device was measured to have poor crosstalk. This is signal contamination which a fully balanced internal signal path would reject. It's reasonable to assume that the channel to channel signal injection is occurring at the points on the PCB where the single ended lines run close to one another. In other words around the pot.

Crosstalk is just as valid a negative performance criterion as distortion if not worse. It's signal that shouldn't be there. And when distortion is at -120dB that's very much not a concern. But when you've got crosstalk at -80>-70dB. Anything you play in an adjacent channel will dominate over any of the distortion. What's worse is that distortion is correlated to the original signal but crosstalk isn't. You could have a special effect blasting off in a surround channel coming out of your mains.

 

[PeteL]

This device was measured to have poor crosstalk. This is signal contamination which a fully balanced internal signal path would reject. It's reasonable to assume that the channel to channel signal injection is occurring at the points on the PCB where the single ended lines run close to one another. In other words around the pot.

Crosstalk is just as valid a negative performance criterion as distortion if not worse. It's signal that shouldn't be there. And when distortion is at -120dB that's very much not a concern. But when you've got crosstalk at -80>-70dB. Anything you play in an adjacent channel will dominate over any of the distortion. What's worse is that distortion is correlated to the original signal but crosstalk isn't. You could have a special effect blasting off in a surround channel coming out of your mains.

Sorry but CMRR doesn't know what to do with signal correlated induced distortion and there is no way that this induced distortion is equal or related on the differential pair.

 

[PeteL]

A Focusrite Scarlett is going to be pretty similar to a miniDSP 2X4HD in terms of analog performance. My 18i20 2nd gen has just a bit more noise than my 2X4HD.

The link in my signature has a tutorial for setting up a CamillaDSP setup using Ubuntu Server. I've done it on a RPi3 but a RPi4 is a lot less stressed. Basic idea is you have a CamillaDSP capture device which receives a source signal, for a Scarlett 18i20 this can be analog, SPDIF or ADAT inputs. If you have the RPi setup as a streamer you can also setup the capture device as an ALSA loopback so output of your streamer software is routed to CamillaDSP. You then have a playback device which is the output channels of the Scarlett, this is all analog and digital output channels. DSP is applied in between the capture and playback devices with all of the stuff you expect (mixer, PEQ, FIR, x-over, delay, level adjustment, etc).

Michael

Got ya, so using The Pi as a streamer, yes but it won't help me because I have a few sources including analog. My interface is also 18i20 2nd gen so no benefit neither. I'll keep my "noisy" setup, and analog volume from the Freya, those noise levels are inaudible to me but thanks.

 

[mdsimon2]

Got ya, so using The Pi as a streamer, yes but it won't help me because I have a few sources including analog. My interface is also 18i20 2nd gen so no benefit neither. I'll keep my "noisy" setup, and analog volume from the Freya, those noise levels are inaudible to me but thanks.

Did you read my post lol? You can use any input on the Focusrite, you do not need to use the RPi as a streamer. It is this ability that allowed me to remove all miniDSPs from my systems. I avoided software DSP for the longest time because I thought it meant your source needed to be a computer but that was incorrect.

Although using the Freya will give the best noise performance if you have a relatively high noise sources like the 2x4HD (or 18i20) so I agree that there is no benefit to switching to digital volume control.

Michael

 

[5th element]

Sorry but CMNR doesn't know what to do with signal correlated induced signal and there is no way that this induced distortion is equal or related on the differential pair.

Of course it does. If you get an identical induced signal on both the hot and cold it'll be rejected. Likewise if all your internal signals are running as differential pairs on tightly routed parallel trace sets then not only will each differential pair reject any common mode signal that's induced, but the parallel traces sets will cause cancellation in any EM fields they generate.

 

[PeteL]

Did you read my post lol? You can use any input on the Focusrite, you do not need to use the RPi as a streamer. It is this ability that allowed me to remove all miniDSPs from my systems. I avoided software DSP for the longest time because I thought it meant your source needed to be a computer but that was incorrect.

Although using the Freya will give the best noise performance if you have a relatively high noise sources like the 2x4HD (or 18i20) so I agree that there is no benefit to switching to digital volume control.

Michael

I guess I read diagonally, it was a long post, sorry for misreading.

 

[PeteL]

Of course it does. If you get an identical induced signal on both the hot and cold it'll be rejected. Likewise if all your internal signals are running as differential pairs on tightly routed parallel trace sets then not only will each differential pair reject any common mode signal that's induced, but the parallel traces sets will cause cancellation in any EM fields they generate.

Why would it be identical? crosstalk appears to adjacent traces, you will never be adjacent to both hot and cold on a different Channel!

 

[5th element]

It reaches further over than just one trace. In the single ended version it's just going to be a mess and that's clearly visible in the measurements here. All those channels routed in a tiny space with them all converging at the pot. No doubt channel 1 influences all the other 7 channels to varying degrees.

With properly routed balanced pairs you're going to first get cancellation of the fields generated by each pair thus significantly reducing the influence that any one channel can have on another. Of the fields that are left you're going to have coupling into the adjacent pairs. No it's not going to be identical because one trace of the pair will be in a different physical location to the other just like the cancellation in the trace pairs won't be perfect either. But it all provides a reduction in the coupling between the channels. If you really wanted, use a 4 layer board, with the hots on layer 2, cold on layer 3, 0.4mm layer separation and the traces running on top of one another.

 

[PeteL]

It reaches further over than just one trace. In the single ended version it's just going to be a mess and that's clearly visible in the measurements here. All those channels routed in a tiny space with them all converging at the pot. No doubt channel 1 influences all the other 7 channels to varying degrees.

With properly routed balanced pairs you're going to first get cancellation of the fields generated by each pair thus significantly reducing the influence that any one channel can have on another. Of the fields that are left you're going to have coupling into the adjacent pairs. No it's not going to be identical because one trace of the pair will be in a different physical location to the other just like the cancellation in the trace pairs won't be perfect either. But it all provides a reduction in the coupling between the channels. If you really wanted, use a 4 layer board, with the hots on layer 2, cold on layer 3, 0.4mm layer separation and the traces running on top of one another.

First time I read that running a differential pair on 2 separate layers being a good practice. I don't think it's even an option if you want to use the different PCB software differential pair tool. I mean It's in the name, common mode rejection, I don't know 5th, It's for common mode noise, not random crosstalk distortion, I don't doubt your experience but you did not convince me, we probably have to let that go and accept that our opinion differs. We are not going to solve this debate, there may be some benefits to fully differential designs, in some cases, but there is no consensus on that from experts with more credentials than me, with some white papers as well, fully differential designs means more electronics, more heat, more radiation, more real estate, there are probably some benefits as well but personally I'm interested by the CMRR of the differential amp at the input, YMMV but there are advocates for both config, I think you are quick to judge that it's "a disaster" and that putting an all trough differential design is a magic solution to cross talk, it's not that simple.

 

[EnceinteAcoustix]

Simply unplug the ground terminals inside the unit to eliminate mains noise, overall my measurements are similar. source SMSL SU-8S measures with Audiophile 192.

 

[wisechoice]

Simply unplug the ground terminals inside the unit to eliminate mains noise, overall my measurements are similar. source SMSL SU-8S measures with Audiophile 192.
View attachment 213465

Wow, very interesting. Can you tell me what possible negative effects that might have, if any?

Our wiring here is very old, so old that it crumbles in the fixtures if you touch it. No ground. We live on the third floor of a multi-unit building so there's also no hope of changing that. But all audio connections are balanced.