Volume control - where in the chain?

[Ugg10]

I am starting to plan a project with a dsp based 3 way speaker and wondered where the best place in the signal chain to control the volume is.

Plan is as follows -

Digital only sources (Chromecast audio via optical or cd via optical or coax)
Into
Minidsp nanodigi (dsp crossover plus room eq)
Into
4x khadas tone board DACs (or possibly a second hand Lucid 88192)
Into
Rotel rmb1066 amp
Into
3 way speakers (traditional floor standing hifi, still to confirm design but likely to be a MLTL 8”+2-3”dome+0.75” tweeter or small ribbon).

So volume can be controlled at almost any point in the chain:

Spotify app on pc/tablet, if using pc then also win10 master volume (digital)
Minidsp has volume control via remote control or added pot (digital)
Khadas has a hack for digital volume control on the ES9038Q2M chip (digital)
I can add a ladder attenuator between the minidsp and the amp, 2xEIZZ 4 channel attenuator, I have done this on a Audio GD DAC to good effect (no sound change) (analogue)

Now, my brain is saying, keep all of the digital stuff at full volume as you don’t know what a digital volume is doing to the signal (dropping bits etc), so go with the passive attenuators.

Any comments, confirmation, disagreement or other options would be very welcome.

 

[scottjoplin]

Look into "gain structure" Typically you want to keep the level as high as possible, without clipping, through the chain and attenuate last of all in order to maintain a high signal-to-noise ratio

 

[Ugg10]

@scottjoplin, thanks for confirming my thoughts, keeping all of the digital signal maxed out and add a passive attenuator just before the power amps.

 

[TNT]

Not necessarily "maxed out". Keep the original at 0 dB gain (i.e. data unaltered) as long as possible - sometimes it's possible to add gain.... Probably the ES chip has the "best" attenuator.

//

 

[MarcelvdG]

Please keep in mind the intersample overshoot issue:

Audio That Goes to 11 - Benchmark Media Systems

Most digital signal processing and DAC chips can't handle peak sample normalized recordings without clipping, and many recordings are peak sample normalized.

 

[Ugg10]

Many thanks again, will look into those areas highlighted.

 

[NATDBERG]

@scottjoplin, thanks for confirming my thoughts, keeping all of the digital signal maxed out and add a passive attenuator just before the power amps.

Not necessarily passive either.. I think Bruno Putzey's BPPBP (a preamp/volume control) is a great thing for this. You'd need three boards. Member alexcp has created a circuit based on it too, available for about $12

For background:

BPPBP - Bruno Putzey's Purist Balanced Preamp (well a balanced volume control really)

For alexcp's board project:

Balanced Volume Controller / Line Stage

Depends on your power-amps and how best to interface them with line-level. If you're using any kind of differential topology like Hypex amp modules, a simple, single-ended passive attenuator might not be the most suitable. And given the cost of so many high-end attenuators, you might be better off cost-wise with a BPPBP-style volume control. Also a buffer like this is probably going to help avoid any impedance mis-matches which could change frequency response before the power-amps (not what you want for an active speaker, else you'll be adjusting DSP depending on attenuator position!).

 

[Ugg10]

Thanks again, power amps are the rotel rmb1066 which are a pretty standard class AB as far as I can tell.

Service manual - Rotel Rmb-1066 Technical Manual | Distortion | Electrical Components

Basic specs -
Input impedance 22kohm
Input sensitivity 1.5v ( so may need to attenuate down the khadas as these are 2v output)
6x60w into 8ohm or 3x150w bridged into 8ohm
500kva transformer and 6800uf per channel.

Any advice on a preamp/volume control to sit between the dac and the power amp would be welcome also where to attenuate down to 1.5v input into the power amp (minidsp or after dac).

Thanks in advance.

 

[soundcheck]

Now, my brain is saying, keep all of the digital stuff at full volume as you don’t know what a digital volume is doing to the signal (dropping bits etc), so go with the passive attenuators.

By using a DSP as crossover and roomEQ, you attentuate the signal and you filter the signal already quite heavily. If you really listen to your brain, you better skip the whole project. 😉

That's what I did exactly because of this long time ago.

I prefer a SW control (e.g. Sabre) feeding a properly gain shaped DAC->amp->speaker setup.

Good luck.