Looper's Delight Archive Top (Search)
Date Index
Thread Index
Author Index
Looper's Delight Home
Mailing List Info

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Date Index][Thread Index][Author Index]

Re: Sound quality issues (was: RE: echoplex or jamman?




well, first of all, this is an odd statement:

At 8:30 AM -0800 10/30/98, Chuck Zwicky dramatically wrote:
>the EDP. The AD1848 converter is quite horrible by todays standards.

because the echoplex does not use that part. That is a lousy part,
extremely noisy, which is why it's not in there. If yours had this part,
then someone else modified it. It wasn't stock.


Secondly, I think you are rather dramatically overstating the case here,
for reasons I fail to understand. Your experience clearly does not match
mine or other peoples, nor does it match test measurements made in the lab.
Perhaps there was something wrong with the unit you were using?


>At 09:07 AM 10/30/98 -0500, you wrote:
>>You're right about the tambourine dynamic range, Kim.  What's more,
>>tamborine is RICH in frequencies above 20 Khz, all the way to the 50 KHz
>>range.  This will get past the anti-aliasing filters, rendering them less
>>effective.
>>This is one of those special cases where higher digital sampling 
>frequency
>>would be of great advantage.
>
>I disagree. The Jamman does not have these problems, and it has a slightly
>lower sampling rate. The problem in the EDP are the analog electronics and
>the poor layout.
>
>On the Analog side there is NO anti-alias filter on the input before the
>converter.

I would suggest that you consider reading datasheets before you make really
stupid statements like this. As with practically all modern digital audio
convertors, the convertors in the echoplex have built in anti-alias
filters. The datasheets from Crystal describe these at length. You might
want to consider doing a little research before making completely incorrect
assertions like this.

The convertor's built in filters are quite good. They have a very steep
rolloff, with stop band rejection of 74dB by .6Fs and passband ripple of
less than +/-.1dB. Designing a discrete filter like that would be absurdly
expensive, which is why nobody designs audio circuits that way anymore. In
addition to this, however, there is an RC filter at the input to the
converter adding a bit more anti-aliasing filtering, and additional
filtering around 3 of the opamps prior to the codec. Now, how does this
equal "NO anti-alias filter"?


>This, in addition to the pre-emphasis, which starts boosting the
>high frequencies around 200HZ  at a rate of 6dB per octave makes the input
>very sensitive to aliasing.

First of all, it's a shelving pre-emphasis/de-emphasis filter, so the boost
is not that dramatic. It's there because we wanted to improve upon the
performance of the audio codec used, by shaping the noise floor based on
typical music signals. In most cases, I think this works great, and gives
us A-weighted numbers exceeding the expected performance of the part.

Secondly, your beloved JamMan circuit also uses a shelving
preemphasis/de-emphasis filter, with similar characteristics to the one in
the Echoplex.

Thirdly, as noted before, the distortion you heard is not aliasing. You
could demonstrate that very easily to yourself with a signal generator and
a scope. (which I guess you did not do?)  I tested for this extensively
with good test gear, and all my measurements showed the circuit did not
have any significant aliasing problems. This is in line with Crystal's
datasheet, so their part performs correctly.

If you had used that generator and scope, you could have seen that the
distortion you heard happens prior to the audio codec, in the VCA. When
this part is overloaded, the distortion sounds bad. However, when the part
is not overloaded (ie the input level is set right), everything sounds fine
throughout the audio bandwidth.


>I experimented with modifying the pre-emphasis/de-emphasis filter, but
>found the noise floor so obnoxious that I left it stock.

If you found the noise floor so obnoxious, as you say, did you have gains
set otherwise correctly? An awful lot of people are using the echoplex in
all manor of situations without finding the noise floor obnoxious, how is
it that they are successful and you are not? Maybe you could explain the
specific setup and settings that you used, so we could see if there was a
problem.


>The noise I am
>describing is the result of poor layout and grounding on the PC board. 
>Lots
>of metallic clocking artifacts.

The layout used was reviewed by Crystal's application engineers. In fact,
by the guy who wrote the application note on layout for Crystal. He thought
it was fine. There are certainly ways to improve upon the layout, but they
would all make the cost go way up, and I doubt they would result in much
change in audio quality.

Also, yours is a remarkable assertion. Given that the test numbers show
performance reaching the limits of the audio part used, wouldn't it make
more sense to look to that particular part as the limiting factor, instead
of other parts of the circuit? Since you report no data, I'm assuming you
did no tests on the analog portions of the board with the codec bypassed.
Granted, the part used is not the best quality part available, but for the
cost I think it performs quite reasonably.


>The Jamman has a very nice LRC filter on both the input and the output. I
>have never heard it get nasty, even when overdriven.

Personally, I was rather shocked to see an LRC filter and a completely
discrete conversion scheme in the Jamman. Nobody has done that since the
mid-eighties. LRC filters are problematic, since they are very expensive
and their characteristics can change dramatically as the parts age. Also,
the Jamman uses only a 5-pole filter for anti-alias, which is generally not
considered enough by modern audio standards. I believe the filter in the
echoplex's codec is 12-pole.

Also, Lexicon's discrete A/D design adds some coloratation to the sound.
Most people find the result pleasing, so that's fine, and probably has a
lot to do with why people like Lexicon gear. Personally, I prefer to do
digital audio as transparently as possible. I would prefer to add my own
coloration.


>I'm sorry that all this has made Kim so defensive. I hope that he will sit
>down with an open mind and listen to the Jamman and the EDP side by side.

I have done that listening test, with sophisticated test gear. Feel free to
do the same, and report some numbers. The quality of the JamMan is fine,
but I found the echoplex to beat it. Defensive? sure. I don't take kindly
to people who see the internet as a place to spread misinformation and make
flaming, unwarrented attacks on other people, companies, or products. In my
opinion, that's very irresponsible, and should be defended against.

kim

______________________________________________________________________
Kim Flint                   | Looper's Delight
kflint@annihilist.com       | http://www.annihilist.com/loop/loop.html
http://www.annihilist.com/  | Loopers-Delight-request@annihilist.com