One possibility is the same test used by Danley.
He runs the sound through the speakers, records, runs, records, does this N times... the variations are greatly increased as a result of each pass.
Presumably the recording media is "blameless. (Multiple loop throughs are easy enough to do here.) The mic of course is a factor, but measurement mics are pretty darn flat. And, in this case we are looking for a difference, not an absolute.
He runs the sound through the speakers, records, runs, records, does this N times... the variations are greatly increased as a result of each pass.
Presumably the recording media is "blameless. (Multiple loop throughs are easy enough to do here.) The mic of course is a factor, but measurement mics are pretty darn flat. And, in this case we are looking for a difference, not an absolute.
There is a slight RMS amplitude difference :
ch0 : +0.20% (+0.02 dB)
ch1 : +0.75% (+0.07 dB)
I was also trying to align both files, but it seems they were not recorded with synchronous DAC and ADC. WTF ? Was the source a CD player and not the PC soundcard with ADC and DAC synchronized ?
ch0 : +0.20% (+0.02 dB)
ch1 : +0.75% (+0.07 dB)
I was also trying to align both files, but it seems they were not recorded with synchronous DAC and ADC. WTF ? Was the source a CD player and not the PC soundcard with ADC and DAC synchronized ?
Last edited:
Indeed it is, but I would like to investigate where the effect comes from, and this is going to be pretty difficult if the wav files were not recorded with synchronous ADC-DAC, any processing which would compare both files becomes rather useless, since differences could as well come either from recording clock drift ...
Let's take 256K samples out of the file (around the part that I used for ABX testing) and plot the spectrums.
They look similar, but calculating the difference between file (second plot) shows one channel matches almost perfectly, while the other one has more treble by about 0.25dB. The file with more treble is the one which seems to have more "ambience" or "sound stage" in ABX.
The fact that one channel matches indicates clock drift is not an issue here. So, one of the amps (or both) could have a frequency response flatness problem in one channel but not the other. This could also come from the measurement chain since channels were originally in the wrong order...
They look similar, but calculating the difference between file (second plot) shows one channel matches almost perfectly, while the other one has more treble by about 0.25dB. The file with more treble is the one which seems to have more "ambience" or "sound stage" in ABX.
The fact that one channel matches indicates clock drift is not an issue here. So, one of the amps (or both) could have a frequency response flatness problem in one channel but not the other. This could also come from the measurement chain since channels were originally in the wrong order...
Attachments
Right, let me explain 
This all started when I had an unscheduled repair to make on my main amp... for those of you that saw the thread on LED's running into a current sink... that was the amp. One of the watt meter backlight panel LED's failed OC. In the amps place I slotted a 30 ? year old Rotel RA820, a, no make that the classic budget amp of the 1980's. This was the amp that got recommended pretty much no matter what your budget was and what speakers you wanted... it was rated that highly at the time.
That is RM1... the reversed channel is a total rookie error on my part through not being able to see round the back of the unit when its slotted in its rack. The speaker leads are custom length and wouldn't reach easily without unrouting everything.
RM2 is my main amp, the lateral FET design I've used for around 10 years now.
I didn't want to say anything on how I think they sound... I still won't yet. What I will do is repost the files with RM1 corrected.
This all started when I had an unscheduled repair to make on my main amp... for those of you that saw the thread on LED's running into a current sink... that was the amp. One of the watt meter backlight panel LED's failed OC. In the amps place I slotted a 30 ? year old Rotel RA820, a, no make that the classic budget amp of the 1980's. This was the amp that got recommended pretty much no matter what your budget was and what speakers you wanted... it was rated that highly at the time.
That is RM1... the reversed channel is a total rookie error on my part through not being able to see round the back of the unit when its slotted in its rack. The speaker leads are custom length and wouldn't reach easily without unrouting everything.
RM2 is my main amp, the lateral FET design I've used for around 10 years now.
I didn't want to say anything on how I think they sound... I still won't yet. What I will do is repost the files with RM1 corrected.
.... and please check that left and right channels have the same gain
Yes boss
The whole idea was to see what the signal was really like at the sharp end... and I've never tried anything like this before. It was just a grabbed opportunity while I had the chance.
Those two amps sound totally different in reality... and I'm still not going to say what my impressions of them are yet.
One thing I wondered, if you took a "perfect" or blameless amp and fed it with (lets say) the signal from across the load of a SE triode amp... what would you hear ? Would the unique triode amp sound come through or would we find that the blameless wasn't quite so blameless after all under real dynamic loading.
Mad thoughts... there could be the makings of a unique group test here. If we decided on a suitable original test track, then we could all record the sound of our amps under identical conditions (say 1 watt/0db or whatever) and then compare notes. How valid it all is I don't know... a bit of fun yes, for sure.
Seriously though, we sweat the fine detail on the small signal stuff and yet the power amp/speaker interface is really where it all happens and where the unique characters of many designs stand and fall. Could an ABX test of actual amp output voltage vs original file be a way forward for those seeking the "piece of wire with gain".
You'd really need to synchronize dac and adc so if a difference is heard, it would be possible to do some processing on the signal to find what it actually is... that means using your soundcard as the source...
Also if you use the soundcard to record the amp output, and you connect the soundcard ground to the amp output ground, some of the speaker return current will flow in the recording cable shield, a balanced connection would be needed here.
Also if you use the soundcard to record the amp output, and you connect the soundcard ground to the amp output ground, some of the speaker return current will flow in the recording cable shield, a balanced connection would be needed here.
Very, very first, quick impressions ...
Aahhhh ... ... takes me back to so many listening sessions in dealer showrooms, and people's systems ... really, really crappy hifi sound, the sort of thing that would put a person off the game for life,.
Sorry about that, Karl ... - now, let it settle down, and another listen later ...
Aahhhh ... ... takes me back to so many listening sessions in dealer showrooms, and people's systems ... really, really crappy hifi sound, the sort of thing that would put a person off the game for life,
.Sorry about that, Karl ...
- now, let it settle down, and another listen later ...
The grounding is something I was very aware of, for one because of the unequal and left and right speaker cables (3ft vs 10ft) and as you say, the fact the speaker returns would be joined together at the soundcard input.
Very, very first, quick impressions ...
Aahhhh ... ... takes me back to so many listening sessions in dealer showrooms, and people's systems ... really, really crappy hifi sound, the sort of thing that would put a person off the game for life,
Sorry about that, Karl ...
Nothing to be sorry about... I don't deny it might sound awful or strange or whatever
because as I mentioned at the start, how do actually listen to something like this. Its a true representation (if we exclude the common ground issue) of the voltage at the load, after being modulated by the speaker resistance, including the high output impedance of the amp etc. I think the inadequacies of the recording chain pale into insignificance compared to the mauling about the signal has endured at its final destination.
I think as a tool for testing, that it could have promise. You could test in mono with one amp and essentially zero lead length to the speakers. For those serious enough, some form of balanced set up could be constructed as a piece of test gear as peufeu hinted at. The technique of remote feedback sensing at the load could be investigated etc.
My comment was a bit strong, wasn't it ...
I hadn't read the thread before listening and posting, but the sound ticked all the boxes for me, reminding me of all the less than stellar audio I've heard over the years, where one is 'fighting' the unpleasant attributes in the reproduction - the treble was a mess, unfortunately, like so many systems out there typically manage to do ...
I hadn't read the thread before listening and posting, but the sound ticked all the boxes for me, reminding me of all the less than stellar audio I've heard over the years, where one is 'fighting' the unpleasant attributes in the reproduction - the treble was a mess, unfortunately, like so many systems out there typically manage to do ...
I do have a bit of experience with testing of complete audio chain systems with potential or existing ground loops caused by test equipment. Then, balanced or transformer coupling is necessary. As an absolute minimum, you may try to disconnect shield of the input signal cable, in case you would create ground loop without such disconnect.
My comment was a bit strong, wasn't it ...
I hadn't read the thread before listening and posting, but the sound ticked all the boxes for me, reminding me of all the less than stellar audio I've heard over the years, where one is 'fighting' the unpleasant attributes in the reproduction - the treble was a mess, unfortunately, like so many systems out there typically manage to do ...
Not at all
I think many would be surprised at the "sound" of their systems if they heard them played back on another set up... which is what these files are showing.- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.