advertisement


How about amps, then?

T

Tenson

Trade: AudioSmile
Following the raging success of the DAC/ADC test thread I thought we might try this with a cheap power-amp :)

I've recorded the same music samples as the other test through my DAC -> Power Amp -> Kensai Speakers and tapped off the speaker terminals to an ADC. We can compare these recordings with original CD file extractions.

I think this is a good test because the amp was driving real speakers (volume set about 90dB@1m each) and also because the amp (Alesis RA500) is mighty cheap at £190 but has good tech specs much like Serge's infamous Behringer amps that look similar. Popular thought would be that the recordings which passed through the £190 power amp would sound worse.

So have some fun and see if you can identify which files are the originals and which have been through a DAC, amp and ADC. Even better is to use the ABX application linked to see if you can identify the files against themselves.

Here are all the bits : http://www.audiosmile.com/forum/showthread.php?p=296231#post296231

[Note from site TL: AudioSmile own an educational music license so this material is exempt from pfm's no needledrop / no copyright rules given it was posted by the owner of that site]
 
I have been planning something similar for ages, and it is still half-way on the to-do list.

I would feed the output of the DAC (the one which also drives the amp) straight into one ADC channel, and the output of the amp into the other ADC channel.

Then import into a DAW, match channel levels, and take the difference. This way test stimulus and test output are fully synced in time, and both channels have seen the same loading effects on the DAC output stage. Any differences then can be attributed to real signal modification by the amp.
 
That's a nice way to do it. My ADC has an 8K input Z so it would load each DAC channel differently if only one were connected to ADC. Still you could load the other DAC channel with an 8K resistor in parallel to compensate.

Audio Diff Maker is a program that can adjust for timing differences and sample rate drift, plus level when doing a null. Today it seems to keep locking up though! I'm not that interested in finding the difference file though, as it will certainly be doing something to the signal, the only thing of matter is whether we can hear it.
 
Just the one DAC channel goes to both one amp channel and one ADC channel. Then the amp output to the other ADC channel.

I don't trust Audio Diff Maker for synchronisation. It is tricky to do really well, and with ADCs and DACs involved there are bound to be errors close to Nyquist. Better avoid *when possible*.

Getting a clean difference file can be instructive, especially when the amplifier is (subjectively) said to struggle with the particular speaker.

I expect.
 
Just had a quick go - Christ on a bike, that's hard! I'm surprised, I thought it'd be easier than the DAC/ADC test!

Only just dipped me toe in, but early indications are that I can't ABX it.

Alex
 
had a listen by doing my own comparison as I can't be bothered to do the real thing. I'd be interested in the result though.
 
No problem. If you prefer, just listen to each of the file sets and PM me your thoughts on which might be the 'amp pass' in each case. Presumably if there is an audible difference the amp one would sound worse.
 
Brill! I've been waiting for this and it will be the first I try. I also await the spoilers with amusement.
 
I spent 10 mins or so on it earlier and failed miserably. I need to have another listen to see if I can find a killer sample, but I don't fancy my chances with this one. If there is an audible difference it's the square root of sod all.

Alex
 
Thanks for doing this. I'll give it a go.

One issue for me with the DAC test was that you swapped the A/Bs around over the few days to ensure no "cheating", but my result got mis-interpreted because of it. This has put a huge doubt in my mind as to the validity of the results.

For this one could you just leave the same A/Bs in place and trust the PFMers to do it honestly?
 
You must log in or register to reply here.


advertisement


Back
Top