Pump up the Volume (in various projects)

[hi30u]

Pump up or attenuate it less, actually!

I would like to understand various solutions and compare them with only one criterion: volume control.

For this comparison, I would like to consider one digital, optical "source".
And three different types of power amps: Solid State (say mono bloc class A/B), or mono bloc Class D, or FDA.

First question: how is volume controlled in a Full Digital Amplifier (Power DAC) such as Sabaj A30A or SMSL VMV A2?
Is it done before or after PWM control?
In other words, is it achieved using bit-trimming or is it controlled inside the power stage?

Will it results in less dynamic range at low volume?
If the source is a CD (16bit/44.1kHz) will I lose quality at very low volume?

In class D, there is no volume control in Purifi or nCore modules, correct?
Could it be possible to have one internally?
Could it be in the power stage module, something like a programmable setpoint?
(A current or voltage adjustable regulation in the power stage would avoid bit trimming and clipping).

For mono bloc class A/B solid state amplifiers, (and class D amps if you answered "no" to the previous question) some people advise to use the preamp function of the DAC.
Does the preamp function of a DAC consists of using bit-trimming in recent (not R2R) DAC?

Is it done inside the AKM or ESS chip?
Will it results in less dynamic range at low volume?
If the source is a CD (16bit/44.1kHz) will I lose quality at very low volume?

Last question: my DAC is SMSL D6 (two AKM AK4493S): let say I use it as a preamp to feed 2 class D monoblocs and, one day, something goes wrong (the DAC resets itself or worst) and the volume is then at its max because of the failure. When I start to play music, will I break the speakers, the amplifiers and my ears?

Many thanks!

 

[staticV3]

Does the preamp function of a DAC consists of using bit-trimming in recent (not R2R) DAC?

Yes.

Will it results in less dynamic range at low volume?

Yes.

If the source is a CD (16bit/44.1kHz) will I lose quality at very low volume?

Regardless of input format, you are highly unlikely to lose quality when using bit-trimming preamp functions in modern DACs, as environmental noise and amplifier noise is much more likely to be capping the effective playback dynamic range, not the DAC or its volume control.

Basically: if you can't hear the DAC's noise floor at your MLP, then you won't lose quality no matter how many bits you trim off your tracks using the DAC's preamp function.

my DAC is SMSL D6 (two AKM AK4493S): let say I use it as a preamp to feed 2 class D monoblocs and, one day, something goes wrong (the DAC resets itself or worst) and the volume is then at its max because of the failure. When I start to play music, will I break the speakers, the amplifiers and my ears?

The Amp won't break. Possibly speakers and ears, but it depends.
(Amp power, Amp gain, speaker sensitivity, speaker power handling, listening distance, content input into the D-6, duration, etc.)

 

[MaxwellsEq]

What is "bit trimming"? Are you referring to coarse reduction in bit-depth e.g. by reducing bit-depth from 16 bits to 14 bits in order to attenuate by 12dB but keeping the noise floor constant? In practice, most modern implementations convert to floating point in DSP before attenuation. This means real-world attenuation in a preamplifier is not necessarily better or worse than DSP-based attenuation in a DAC.

There is nothing stopping designers from adding volume controls to the buffer board of NCore and Purifi amplifier modules. Most vendors ship with multiple gain settings. I suspect that adding a volume control may impact the overall excellent measurements of these solutions and so designers are not motivated to do this.

 

[Vincent Kars]

In class D, there is no volume control in Purifi or nCore modules, correct?

This has nothing to do with class D. All power amps runs at a fixed gain (25 dB most of the time). Volume control is done by the pre-amp.

 

[MCH]

This has nothing to do with class D. All power amps runs at a fixed gain (25 dB most of the time). Volume control is done by the pre-amp.

Sorry for my ignorance, but out of curiosity: the same class D amp operating with different VDD, will It have the exact same gain or slightly different because of the difference voltage?
Thanks.

 

[ppataki]

Here is how my 'gain structure' looks like (I have three A30A amps hence I thought I would share my setup)

1. Volume control is digital 64-bit precision at the source (=PC) using Jriver's Internal Volume control and it is upstream, meaning it is the first DSP module in the signal chain
2. Volume is untouched (=set to Max) in all the downstream DSP modules
3. Volume is untouched in the DDC (RME Digiface USB)
4. Volume of the three Sabaj A30A is currently set to 65 (out of 99) - I guess I might lose some bits (dynamic range) there, not sure (any comments are welcome about this)

So this way, I ensure that there is minimum dynamic range loss at least up to the amps
(before the A30A amps I had Hypex NC250MP modules without volume control)

 

[hi30u]

This has nothing to do with class D. All power amps runs at a fixed gain (25 dB most of the time). Volume control is done by the pre-amp.

As a general habit / usage or a a design limitation of any Class D module?

Sorry for my ignorance, but out of curiosity: the same class D amp operating with different VDD, will It have the exact same gain or slightly different because of the difference voltage?
Thanks.

Yes, this is what I imagined.
Or maybe by changing the frequency of the output switching clock (faster should deliver less current, lower power/volume.

 

[hi30u]

Here is how my 'gain structure' looks like (I have three A30A amps hence I thought I would share my setup)

1. Volume control is digital 64-bit precision at the source (=PC) using Jriver's Internal Volume control and it is upstream, meaning it is the first DSP module in the signal chain
2. Volume is untouched (=set to Max) in all the downstream DSP modules
3. Volume is untouched in the DDC (RME Digiface USB)
4. Volume of the three Sabaj A30A is currently set to 65 (out of 99) - I guess I might lose some bits (dynamic range) there, not sure (any comments are welcome about this)

So this way, I ensure that there is minimum dynamic range loss at least up to the amps
(before the A30A amps I had Hypex NC250MP modules without volume control)

Yes, how would you do if you only had a CD player with only optical output?
Just trying to understand how your FDA amplifier works internally.

 

[ppataki]

Just trying to understand how your FDA amplifier works internally.

Have you asked Sabaj Support team about this?

 

[hi30u]

Regardless of input format, you are highly unlikely to lose quality when using bit-trimming preamp functions in modern DACs, as environmental noise and amplifier noise is much more likely to be capping the effective playback dynamic range, not the DAC or its volume control.

Basically: if you can't hear the DAC's noise floor at your MLP, then you won't lose quality no matter how many bits you trim off your tracks using the DAC's preamp function.

Yes, good point. Better improve the room acoustics and close the windows, right?

Just one question: if I leave the window open but I still have 16 bits at low volume with everything else perfect, will I get better instruments separation, depth, etc than at 12 or 14 bits ?

 

[Killingbeans]

In other words, is it achieved using bit-trimming or is it controlled inside the power stage?

Bit-trimming is a thing of the past. Modern chips calculate the attenuation as floating-point to keep the noise floor consistently miniscule.

 

[Vincent Kars]

It have the exact same gain

In general the gain is fixed.
Headphone amps sometimes comes with a low/high gain. The low one to be used for IEM's as they often have a high sensitivity.

 

[MCH]

In general the gain is fixed.
Headphone amps sometimes comes with a low/high gain. The low one to be used for IEM's as they often have a high sensitivity.

Thanks. I am not sure if I explained myself correctly, in case: we have a class D amplifier, with no gain switch whatsoever. Say you operate it with 50VDC, the gain is x.
Now you change the power supply to one that gives 25VDC. Is the gain still x or x-3 or x-y...?

(my confusion comes from the very simplistic idea I have of these amps modulating a PWM signal that switches between ground and VDD, and I wonder if that means that higher VDD means that the amp can take an input signal with higher amplitude without clipping, or that it can amplify the same signal a bit more (higher gain) or none of the two because inside the amp there are regulators or the comparators work also at a different voltage or whatever other reason i am not smart enough to understand )
Thank you for your patience

 

[Vincent Kars]

Bit-trimming is a thing of the past.

Never heard of bit-trimming.

To the best of my knowledge:
16 / 24 bit audio is converted to float
DSP (like volume control) is performed
Dither added (necessary in case of 16 bit)
Converted back to integer as audio protocols like S/PDIF or UAC1/2 expect integers.
Likewise DAC's expect integers.
As it are integers, there always will be a quantization error.

If you half the volume using a 16 bit register, you loose 8 bits
MSB LSB
1111111111111111
000000001111111111111111

Yes, we do loose resolution using volume control

If we play 16 bits program material on a 24 bit DAC and lower with 8 bits (48 dB) we still have all 16 bits in the register of the DAC.

111111111111111100000000
000000001111111111111111

Even with 48 dB reduction, we don't loose information (technically spoken).
What about a modern 32 bit DAC?

As others already pointed out, in practice the noise (the SNR) of our playback chain is the limit. Modern amps do have a SNR of 100 so able to resolve CD quality (96 dB). Very silent amps like Purify can even resolve 20 bits (120 dB), about the max of musical information you can find in a 24 bit audio file made with a very silent recording chain.

 

[hi30u]

In general the gain is fixed.

Yes, and the question is why.
Is there a way to modify or completely redesign a Class D module to allow the user to control its gain?
The purpose being getting rid of the pre-amp.

 

[hi30u]

As others already pointed out, in practice the noise (the SNR) of our playback chain is the limit. Modern amps do have a SNR of 100 so able to resolve CD quality (96 dB). Very silent amps like Purify can even resolve 20 bits (120 dB), about the max of musical information you can find in a 24 bit audio file made with a very silent recording chain.

Thanks,
The first part of your answer is what I call bit trimming, I am not sure of a better name.
About the quote: I understand SNR. Would it be scientifically accurate to include in SNR what some audio addicts express as 3d feeling, placement of the instruments, real life tint, beat responsiveness..

 

[MaxwellsEq]

Yes, and the question is why.
Is there a way to modify or completely redesign a Class D module to allow the user to control its gain?
The purpose being getting rid of the pre-amp.

For decades power amplifiers always have fixed gain without a volume control. The gain is usually about 25 to 30dB. This has NOTHING whatsoever to do with Class D. My last amplifier I bought in the late 1980s. It was Class A/B. It did not have a volume control and its gain was 28dB.

 

[restorer-john]

For decades power amplifiers always have fixed gain without a volume control.

Ah no. Most power amplifiers in the last 40 years have had attenuators either on the back panel or the front panel.

Sure some didn't, but most did. The ones that didn't were cheap, stripped down poverty packs, or so-called audiophile designs...

 

[MaxwellsEq]

Ah no. Most power amplifiers in the last 40 years have had attenuators either on the back panel or the front panel.

Sure some didn't, but most did. The ones that didn't were cheap, stripped down poverty packs, or so-called audiophile designs...

Fair enough. You have seen thousands more domestic devices than I have. Generally, pro power amplifiers (which I've seen more of) always have attenuators.

But an attenuator is not an essential capability of a power amplifier. The OP is confused, thinking the lack of an attenuator is an amplifier class thing.

I also get the feeling they believe that amplifier variable gain is achieved through adjusting the rail voltages (perhaps they are thinking of Class H etc.) or through changing the PWM behaviour of Class D.

 

[Doodski]

poverty packs

Hey! Don't insult my term for a 6 pack of beer.... LoL.