I've sent an email to miniDSP to see if they can clarify their manual instructions in terms of output levels and Dirac. I'll be the first to admit I'm wrong if so, and that would be great news for folks like me and
@Kachda who understood it the same way. The miniDSP forums also talk about this -10db volume setting as well, fwiw, so it's not just us.
You are correct in your earlier observations but the problem needs to be better understood to see whether a solution exists. It is more of a UI/UX problem than a technical one. And the responses to you have not been entirely correct.
There is some misunderstanding in this thread of the difference between the digital and analog stages. Between digital clipping and analog clipping. Between digital levels and analog gain from the pre-amp. Between analog volume control and digital volume control.
1. Digital signals in the way they are represented can only go up to 0db. If you process the stream digitally to boost above that, it will digitally clip (in the way a register that can hold a fixed magnitude will overflow and result in something different than what you wanted). This is unrelated to whether it leads to analog clipping downstream.
2. The incoming audio can go all the way to 0db. The device has no control over that.
3. Without any DSP processing the same digital levels go to the DAC. It cannot send more than 0db max peak. The DAC converts this to analog and the pre-amp stage converts this to the output voltage with some gain that is decided by the design as a trade-off between noise/distortion and ability to drive downstream amps that require some minimum voltage to provide their peak output. The 0db input to DAC corresponds to max design voltage output of the pre-amp.
4. Any digital processing you do before the DAC will not affect this DAC/pre-amp stage since it will only see a 0db max signal. So boosting in the DSP processing in digital has no impact on whether it will clip in analog or further downstream. It has nothing to do with whether the amp downstream can handle the DSP boost or not. The latter simply has to do with matching the output levels of the pre-amp with the input sensitivity of the power amp. That is independent of whether you are using a DSP or not.
5. If the volume control is in analog, then it is affecting the gain in the pre-amp stage and it can be calibrated to have its 0db correspond to the max peak output of the pre-amp which corresponds to 0db digital signal reaching the DAC. So, the 0db level here will be unaffected by any use of DSP upstream. The volume level will just go down when you engage Dirac as explained below.
6. If the volume control is in the digital stage, then it needs to ensure that it maintains the max 0db digital value because it cannot increase it without digital clipping. If the volume control is after the DSP stage and pre-DAC, it is simply a pass through for 0db and attenuation for below that.
7. When you do DSP stuff like Dirac, that can boost the signal to compensate for room eq, you run into the possibility of digital clipping. For example, if you create a filter to boost by ANY amount at a particular frequency, since the original signal can potentially have a 0db peak at that frequency, it can digitally clip distorting the signal. Note that the maximum boost you can apply here has nothing to do with whether the downstream analog systems pre or power can handle it because the maximum you can send out to DAC is still 0db numerically which will still be the max output voltage of the pre-amp and if the power amp can handle it, there is no clipping in the analog stage.
8. So room eq systems typically reduce the overall level of the incoming signal by an amount corresponding to the maximum boost they will apply to keep any processed digital output not potentially exceed (in computation) 0db for any signal level. Dirac is fairly aggressive and can apply 10db boosts if it feels necessary, so its implementations reduce the overall levels of input by default anywhere from -10db to -18db (DLP) as default. This causes the overall signal level to fall and so the volume you get out of it will be lower with the EQ engaged. But you can still reach 0db for signals that are boosted.
9. This causes an issue for the implementor for accommodating the use cases with both Dirac enabled and without. In particular, how to calibrate the volume control. If you calibrate to the 0db without processing (or past the DSP processing stage), then the volume drops as soon as the eq is engaged even with the volume control at 0db. If you calibrate to the reduced level used with eq on, then without eq engaged, the volume dial will need to go beyond 0db to get to the actual max output. This is not a technical problem but just a UI/UX problem.
10. If you have the volume control as an attenuator at the digital input before feeding to DSP (and I suspect the SHD may be doing this) so you do not have the DSP automatically reducing the incoming signal to prevent digital clipping, then the volume control will have to be calibrated to have its 0db correspond to the attenuated level when eq is engaged to prevent digital clipping but it will have the same problem as 9 above when eq is not engaged. So, I suspect this is where the -10db requirement for the SHD comes in because it is calibrated to 0db input signal. Increasing above that with Dirac engaged may potentially lead to digital clipping while it will be fine without eq engaged.
This situation is more of a UI/UX issue than a deficiency or a bug in the unit.
The reduction of overall level is a necessary one when you have room EQ engaged and it is dependent on how much headroom eq requires for its max boosts.
You cannot lose what you can't have by reducing the volume to -10db with Dirac engaged. It is important to note that the pre-output can still hit the max 2V/4V even when the
digital volume is at -10db because the processed digital level can still get up to 0db where it is boosted, so it is not like it is shortchanging your amp.
On the other hand, increasing it above -10db may lead to digital clipping artifacts (unrelated to and with no impact on any analog clipping in the pre-amp or the power amp downstream) and hence the recommendation. My guess.
Like I said it is UI/UX problem.
If you increase the design pre-amp output voltage beyond the standard 2V/4V, then all you are doing is transferring the problem downstream. You may always have to keep the pre-volume at some quantity less than 0db (whether you are using EQ or not) to prevent the potential for far more destructive clipping downstream. Or have to let the user do their own gain staging/matching settings between the pre and power amps.
One solution is to make the volume be much smarter and integrated with the DSP processing and the pre-amp stage, so it can automatically change its calibration depending on whether EQ is engaged or not so 0db always corresponds to the maximum you can get and so you can leave it on that setting whether you are using EQ or not and not have to manually compensate for eq switched on. But it may not be worth the additional complexity. This just solves the UI/UX issue and people thinking they are getting short-changed.