living sounds
Well-known member
Don't use poly caps for PSU decoupling, they are inductive and thus useless at higher frequencies (where oscillation can easily occur). Ceramics are what you want for this application.
Behringer ADA8000 Op Amp Replacement - RMAA Measurements
- Thread starter usekgb
- Start date
Help Support GroupDIY Audio Forum:
usekgb
Well-known member
Noted. The Wavefront app note suggested using poly's or high quality ceramics, so I used the poly's that I had on hand. I'll see if I have any ceramics around. If not, I'll pick some up on Tuesday.
usekgb
Well-known member
I had a quick listen to the modified vs unmodified channels this evening. This is really only a comparison between the TL072 and the OPA4134 after the DA conversion. The OPA's definitely sounded cleaner and more open, but they also seemed to have an exaggerated, almost sibilant top end to them. The soundstage improved with the OPA's and familiar recordings seemed to open up a little bit. I'm just not sure if I like it or not.
No idea where your decoupling caps go but in any analogue circuit, the important caps are electrolytics.usekgb said:
Make sure you've got some AT important OPAs and the CODEC chips. They need to well be within 1". You decouple to a DIRTY ground which should be separate from your CLEAN feedback signal grounds.
Samuel Groner
Well-known member
usekgb said:
Yes, but a very poor one! Running at higher sampling frequencies might help.
Samuel
usekgb
Well-known member
Ricardo -
I added 10uF caps to the supply pins of the input OPA, but didn't have time to add them to the output. Are you suggesting adding a 'lectro to the V+ of the converter chip as well? As for grounds, the ADA8000 appears to have a big ass ground plane in stead of separate grounds for digital and analog. The caps I added to the supply pins of the OPA are connected to a nearby analog ground.
Samuel -
I am limited to the 48kHz sampling frequency of the ADA8000. I wonder if adding a pre-filter at 48kHz before the converter would help. This would be a simple first order filter with a cap to ground. For this, I would need to know the input impedance of the Wavefront chip. I may also be able to add these caps in the feedback loop of the OPA's. Do you think this could help with the aliasing problem?
I added 10uF caps to the supply pins of the input OPA, but didn't have time to add them to the output. Are you suggesting adding a 'lectro to the V+ of the converter chip as well? As for grounds, the ADA8000 appears to have a big ass ground plane in stead of separate grounds for digital and analog. The caps I added to the supply pins of the OPA are connected to a nearby analog ground.
Samuel -
I am limited to the 48kHz sampling frequency of the ADA8000. I wonder if adding a pre-filter at 48kHz before the converter would help. This would be a simple first order filter with a cap to ground. For this, I would need to know the input impedance of the Wavefront chip. I may also be able to add these caps in the feedback loop of the OPA's. Do you think this could help with the aliasing problem?
Samuel Groner
Well-known member
usekgb said:
As far as I understand, you simply see the frequency response of the anti-aliasing filter of the converter (there's no actual significant aliasing happening). Adding another filter won't help unless it reverses the ripple (which is near-impossible to do in the analog domain).
Samuel
usekgb
Well-known member
So is the ripple that I'm seeing caused by the ringing at the corner frequency of the anti-aliasing filter? That would make sense. The idea of the pre-filter was to help out the anti-aliasing filter. But, if the anti-aliasing filter has too sharp of a cut off, it would cause a ring at the corner frequency. The curious thing with the ripples is that, if it is a ring, the fundamental frequency would be at 0Hz. Also, the overtones are not decreasing in amplitude. This leads me to believe that it is not caused by ringing. Very interesting.
I understand the foundations of digital conversion, and I understand the role of anti-aliasing filters. I just wish I had a more in-depth understanding of modern AD/DA converters, and the tricks used to smooth them out. My audio classes from 20 years ago didn't get very deep in to this subject. Hopefully more of this will be covered in future EE classes as I go forward.
I understand the foundations of digital conversion, and I understand the role of anti-aliasing filters. I just wish I had a more in-depth understanding of modern AD/DA converters, and the tricks used to smooth them out. My audio classes from 20 years ago didn't get very deep in to this subject. Hopefully more of this will be covered in future EE classes as I go forward.
usekgb
Well-known member
Anyone have any thoughts on just ripping out the TL074's, and replacing them with differential OPA's like the OPA1632? After reading the app notes on the PCM4222, it seems like it might be a good solution to clean up and simplify the analog front end. In the end, I was planning on bypassing the mic pre's anyway. Opinions?
Yes.usekgb said:
I've never had a ground plane, big ass or not, improve carefully routed earths in a low noise design.
Ott points out (unlike the pseudo EMI gurus that you STILL have to consider where your earth currents are flowing so if you couldn't route single track return earth currents where you want them on your ground plane layer due to breaks, adding the ground plane won't help.
But as you've already got a ground plane, you'll just have to use it
With hybrid analogue/digital stuff, its even more important to separate the analogue & digital earths .. including their ground planes.
_____________
You've done about as much as you can by swapping to OPA4134. Among the uber OPAs, its the one which is least fussy. That's why it often gives better results in 'real life' than more highly spec'd stuff.
To do more, you need to seriously revise the earthing system. This is NOT easy.
Kingston had an excellent thread on what is REALLY important for noise & THD; opamps and local decoupling of rails, some questions
http://www.groupdiy.com/index.php?topic=37307.80
Many true gurus chime in. It proves how OPA rolling takes a VERY poor second place to correct earthing, layout & decoupling. It’s a long thread but read the whole thing from #41 to find pearls of wisdom.
_____________
The frequency response ripples you see are due to crude Linear Phase FIR digital anti-aliasing filtering. Today, most CODECs are oversampling so these artifacts are much smaller in the audio range. You can't do anything about them except by binning the ADA8000.
usekgb
Well-known member
That's the impression I am getting about the ripples. They aren't huge, so I'm just not going to worry about it. ;D for now, I'm just tinkering because I can, and it's kind of fun. I am not expecting to get a high end converter out of this thing, but I am having fun squeezing a bit more performance out of it. I really don't have the budget to upgrade my converters/control surface/computer, so a little bit of extra performance from my existing equipment can help.
At this point, I'm really just trying to improve the analog sections of this box. The Wavefront stuff was neat in it's day, but it's performance is seriously lacking compared to modern converters. I understand this, and I'm fine with it. I've mixed plenty of good albums through this thing, and I've always made it work.
In the end, I've just opened up a can of worms. My original intention was to just get data on an OPA swap, as I've never seen real measurements of this procedure. Now that the box is open, I'm just trying to fix the sections where Behringer took shortcuts. I'm still interested in the OPA1632 idea though. Maybe I'll build up a small input PCB and give it a try.
At this point, I'm really just trying to improve the analog sections of this box. The Wavefront stuff was neat in it's day, but it's performance is seriously lacking compared to modern converters. I understand this, and I'm fine with it. I've mixed plenty of good albums through this thing, and I've always made it work.
In the end, I've just opened up a can of worms. My original intention was to just get data on an OPA swap, as I've never seen real measurements of this procedure. Now that the box is open, I'm just trying to fix the sections where Behringer took shortcuts. I'm still interested in the OPA1632 idea though. Maybe I'll build up a small input PCB and give it a try.
living sounds
Well-known member
usekgb said:
My 2 cents: I wouldn't bother, having myself wasted far too much time with op amp swaps. Ricardo's advice is spot on: Clearing up the existing circuit by seperating analog/digital/dirt grounds and adding proper decoupling caps and damping resitors will get a good op amp like the OPA4134 closer to its limits.
These uber-op-amps have theoretical performance far exceeding the capabilities of the rest of the unit, so even if they were to work as advertised (which usually requires quite a bit of tweaking and good measuring gear) they would be total overkill, or in other words not the limiting factor here.
usekgb
Well-known member
Fair enough. What I have already done is quite an improvement. The only reason I wanted to try the opa1632 is because it avoids the whole invert and invert again that is happening now. And, I've never used one before. 8) I use the opax134 all the time, and it always does a great job.
living sounds
Well-known member
usekgb said:
That's something else, I somehow confused it with the OPA1612. :
I did a comperison with my RME hdsp the other day
http://geir-music.blogspot.no/2014/04/converter-test-behringer-ada8000-and.html
My practical conclusion is: I can't really hear the difference between the two,
so I'll leave mine unmodded and focus on making music
http://geir-music.blogspot.no/2014/04/converter-test-behringer-ada8000-and.html
My practical conclusion is: I can't really hear the difference between the two,
so I'll leave mine unmodded and focus on making music
usekgb
Well-known member
G-Sun said:I did a comperison with my RME hdsp the other day
http://geir-music.blogspot.no/2014/04/converter-test-behringer-ada8000-and.html
My practical conclusion is: I can't really hear the difference between the two,
so I'll leave mine unmodded and focus on making music
LOL..... I should take your advise and just make music!
Too much of a tinkerer I guess.
usekgb
Well-known member
I finally got back to this thing after putting it aside to work on things that actually bring in money! I tried out the differential op-amp thing on the input, bypassing the mic pre all together. I used the TI OPA1632 chip used in the same configuration as used on the OPA4222 Evaluation Board. It worked very well, but I really didn't see enough improvement to justify completely reworking the entire input section of this thing.
Next, I tried the OPA1644 op-amp. This chip measured nearly identically to the OPA4134. I believe that, at this point, I am at the limits of what can be done to this circuit though improving the analog section. The only reason I would want to go with the OPA1644 instead of the OPA4134 is the lower power consumption of the newer chip. They do sound slightly different, but they are both very good, and I am happy with both of them sonically.
Thus ends my journey of tinkering with this cheap, but surprisingly useful, converter.Thanks for all the input everyone!
Cheers,
Zach
Next, I tried the OPA1644 op-amp. This chip measured nearly identically to the OPA4134. I believe that, at this point, I am at the limits of what can be done to this circuit though improving the analog section. The only reason I would want to go with the OPA1644 instead of the OPA4134 is the lower power consumption of the newer chip. They do sound slightly different, but they are both very good, and I am happy with both of them sonically.
Thus ends my journey of tinkering with this cheap, but surprisingly useful, converter.Thanks for all the input everyone!
Cheers,
Zach
