The last page of that thread is hilarious, they apparently used two mics to do the comparison and thus ended up with a -3dB/+10dB mod response vs. stock
There is pretty much no way two preamps can be that different, unless you purposely build in filter circuits. Then you have built an EQ, not a mic preamp.
I can't abide modders that can't generate a proper technical spec. I also don't understand the difficulty with simple loop tests to gauge an amplifier. There is nothing magical about a resistive load IC input preamp that means it must be attached to a microphone* rather than any other amplifier output. So use the amplifier connected to a 100% repeatable source to generate the comparison.
That would also illustrate limitations of the chain (D/A/D). Do a loop test with the converter to determine the limit of its response. If neither version of the preamp can be observed as a degradation of that response, then the converter is the limiting factor in the chain, and therefore the change in preamp performance cannot be determined.
So instead of a few minutes of basic lab work, we get totally invalid comparisons with multiple variables and no calibration of the test environment . . .
Another issue with that comparison is it doesn't demonstrate one of their claimed improvements, that of noise performance. The DMP3 starts out as a quiet enough amp for the noise of the amp to be totally swamped by the noise of any condenser microphone. So one would expect a test with a ribbon microphone instead.
Of course, it's simple enough to measure noise performance: attach a 150 ohm resistor to stock and mod units, crank the unit to maximum gain, measure the output noise, subtract the gain (and the noise of the resistor if it's a quiet enough unit for that to matter). State the spec as follows: stock, -125dBV unweighted, -129dBA; mod, -127dBV unweighted, -130dBA (or whatever the actual figures are). Post a plot of the comparative noise spectra if you really want to illustrate the improvement.
In the time it took me to type this, I could have done these tests . . . I keep a test file with 20 second bursts of various sine waves, rectified sine, noise, combined HF sines (looking for IMD), square . . . in three minutes I get a pretty good idea of how an amplifier performs by watching the scope and FFT.
* the exception is the performance of the preamp's phantom power circuit. That concern can be addressed in a few ways: first, the DC performance and AC noise of the phantom supply can be directly measured. One useful test is to short one pin to ground. That will shunt one leg of phantom noise to ground, while the noise signal will still be present on the other pin (loaded down by the input impedance of the amp, but still observable). That defeats the usual common-mode rejection of the amp which will ordinarily cancel out any phantom noise.
The problem is that only occurs if the microphone has properly matched impedance to ground on both signal pins. Most do, but a few don't, and you will see reports of noise problems with certain microphones where the phantom supply is noisy. This primarily occurs with portable recorders--I have measured 500uV of noise (-66dBV) on one recorder's phantom rail, which is wildly unacceptable. 10uV (-100dBV) is a much more reasonable maximum, especially for a wall-powered amp. So if a stock phantom rail is noisy, and a mod improves that, it won't show up in a standard noise spec, but it can be easily measured and stated.
If one is concerned that the phantom load might adversely affect the other power rails in the box, the loop test can be performed with a dummy load attached to the mic inputs; 1mA, 3mA, 10mA, whatever you like.
