Originally Posted by AustinJerry
I still do not know why this affect me--I am not using the LFE channel in any of the measurements. In case you can't tell, I am seriously upset by this development.
I can imagine the multimeter is currently this close to being buried into the wall atm so lets see if we can save your multimeter from that fate, honestly you're really v close to having this sorted out so no need to be vexed.
Take a step back then, what you're doing is trying to put the minidsp into your signal chain between the preamp and the sub(s). The minidsp has a fixed input sensitivity (of either 0.9Vrms or 2.0Vrms) and provides *no* mechanism to attenuate the input. Therefore you *have* to make sure the signal that reaches the minidsp cannot possibly be more than 0.9Vrms (or 2.0Vrms depending on the jumper position). This is your goal for this exercise.
Now take one step back up the signal chain, to the SW preout from your processor, and look at what is on that channel. In a bass managed system, this is a mono signal that is the sum of 2 components;
1) the output of a low pass of each main channel
2) the LFE channel itself
***The key point is that the peak signal strength on the LFE channel is 10dB higher than that on a main channel. ***
Now step back one more step to your signal generator. You are using REW to generate that "peak signal" by using a -3dBFS (rms) sine wave. You are connected as the LR channels so, as far as the processor is concerned, your signal is a main channel. If you had the capability to connect as the LFE channel (either into multichannel analogue inputs or, more commonly, via HDMI and channel 4 using ASIO4ALL) then it would see you as the LFE and hence 0dBFS would be 10dB higher.
In the final reckoning, it just means that the actual max voltage your processor can put down that wire is much higher than you have measured so far. Therefore, to avoid clipping, you can either; restrain the MV to a max of -10 OR change the gain structure accordingly.
AIUI the calculations I gave in my previous post should accurately (subject to rounding errors given that a DMM only goes to 0.1V resolution) describe what you really need to set trims to (though obviously best to verify this with measurements).
Does that make sense?