Behringer ADA8000 Op Amp Replacement - RMAA Measurements

[usekgb]

Well, I was bored and had a few OPA4134 op amps kicking around and I decided to see what they would do in a Behringer ADA8000.  I only had enough to do the input/output of channels 1 and 2, and the input of channels 3 and 4.  I thought I would post my results here in case anyone is curious.  The left channel has both the input and output op amps replaced.  The right channel is unmodified.  I did not run tests for the input-only change.  The numbers don't lie.  The op amp replacement does actually show some improvement over stock.

 

[usekgb]

Next...

 

[usekgb]

IMD + Noise.....

 

[usekgb]

Noise.....

 

[usekgb]

THD.....

 

[usekgb]

I am curious about the uneven frequency response, and if something can be done about it.  My first impression is that it looks like ringing, but I can't be sure without a scope.  It is present on all channels, modified or not.  Anyone have any thoughts on this?

 

[shabtek]

flip the leads/switch I-O connectors

see if change follows

describe your testing set-up

where do you think there is ringing?

 

[usekgb]

This was a simple loop through test from out to in using ~1m Mogami 2534 w/ Neutrik connectors.  Response is the same on all channels, even using different cables.  Not sure what is causing ringing.  I'm planning on bypassing power supply pins to ground with .1uF caps, but ran out of time today.

 

[usekgb]

Since the odd frequency response is uniform across all channels, this leads me to believe that the issue is either:
1) Power supply issues
2) Converter implementation
3) Clocking

I can only really look at 1 or 2 with the equipment that I have.  This curve could also be due to the way RMAA does it's measurements, but these oddities don't appear on my other sound card, leading me to believe it is an issue with the ADA8000.

 

[Walrus]

Did you run the same tests before you replaced the opamps so you could definitely say your change is the new opamps....?

 

[usekgb]

Yes.  It is also on the unmodded channels.  I will have to run it again being fed from a different source, and see if it is on the AD side, the DA side, or both.  The variation is only a few tenths of a dB, so it isn't audible.  I would just like to smooth it out.

 

[ricardo]

Is the first measurement (showing difference in gain) of the EXACT same channel before & after?

What OPAs did you replace with the OPA4134s?
______________________

The ripple is the anti-aliasing filters on the ADA8000.  All DACs have this 'feature' .. but some are better than others.

 

[usekgb]

No.  That was as close to matching as I could get two different channels with the gain knob on the ADA.  The left trace is the modded channel, and the right trace is an unmodded channel.  I replaced the stock TL074's with the opa4134's.

Anti-aliasing filter.  Got it.    No easy fix for that one, as it would be in the AD converter.  I guess that answers my question.  I would still like to mod the other channels, as I got good measurable improvements with the new op amps.  While I'm at it, I'll probably do the fix to the power supply as well.

See what happens when you have a shop attached to the studio?  When you get bored, you try to fix things that aren't broken! ;D ;D

 

[mitsos]

I was going to ask if you added extra PSU decoupling caps but you said no.  Try going with larger than 0.1uF low ESR electrolytic caps, like Oscons (I think that's what BLA use for decoupling). 

Also, you may have already seen these:
http://www.wavefrontsemi.com/DataSheetsFolder/WavefrontAL1101.pdf
http://www.wavefrontsemi.com/DataSheetsFolder/WavefrontAL1201.pdf

The 1201 datasheet says you can add further filtering "if it's critical". 

While you're in there, maybe check decoupling caps at all chips, including the AD/DA? It'd be nice to know.  Not sure what BLA does but they make mention of their special decoupling techniques for digital parts of equipment they modify.

 

[usekgb]

I noticed something about that on the Black Lion site as well.  The converters do have decoupling caps on the PCB, as stated in the Wavefront app note.  They are small and parts, and I have no idea what their value is.  Probably 0.1uF ceramics.  I was planning on using 0.1uF poly's for decoupling.  I've never used 'lectros here before.  I ran out of time to look for decoupling caps near the op amps, but the opa4134 usually doesn't need them.  It's a pretty stable amp.  It will be a couple days before I can get back to this, but I will report more as I go.

 

[Samuel Groner]

The odd frequency response looks more like a measurement artefact than actual circuit behaviour. First verify the validity of your setup e.g. by using other AD and DA converters.

Samuel

 

[usekgb]

My other converters did not display the wavy frequency response at all.  That is why I was so curious about this.  I'll keep investigating.

 

[mitsos]

but the opa4134 usually doesn't need them.  It's a pretty stable amp.

I don't think it's about stability per se, it's about fixing the environment the opamp is working in, to allow it to work better. There was a thread started by Kingston where he did some detailed testing of opamp swaps, and another more recent one about "opamp rolling" where Ricardo mentions the benefit of using 10uF caps at the opamp pins.  I don't think BLA goes into detail on their site but if you take a look at the youtube video explaining their digi 002 mod, you'll see they use Os-con electro's all over for decoupling.  In fact, every other word out of the guys mouth seems to be "decoupling caps". BTW, those Oscons are very expensive caps (for electros). Something like $1 a piece, not much for a one off DIY mod, but if you're doing tons of mods like they are, they severely cut into the bottom line as compared to 0.1uF ceramics.  Leads one to believe there is a difference.

Of course, I also think that swapping opamps makes a difference.

 

[usekgb]

I see what you're saying.  It's not about removing oscillation, but more about providing a close low Z power reservoir for the amps.  Makes sense.  I've got a boatload of 10uF Panasonic FC's, but no Oscons.  Maybe I'll try those on one amp and see what happens.

 

[usekgb]

I read a comparison review from Sound on Sound today of the ADA8000 vs ADA8200, and the rippling frequency response I observed showed up in their test results as well.  So far, it looks to be a part of the Wavefront AL1101 AD converter.

I had a little time today to replace the power supply caps with Panasonic 105deg caps.  I felt that this was important because they are sitting right next to the heat sinks on the voltage regulators.  I replaced the coupling caps on the first two channels with Panasonic FC's.  I added 0.1uF poly decoupling caps to the AD converter according to the Wavefront app notes, and I removed the two ceramic decoupling caps that were on the PCB.  The new poly caps are soldered directly to the converter chips legs.  Doing this to a surface mount chip was a royal PITA, but I wanted to see if this would improve performance at all.  I am planning on adding 10uF electrolytics to the OPA4134's power pins as well as 0.1uF poly's this evening when I get back to the shop.  If I still have time, I'll take another measurement and report my findings.

Thanks for the tips everyone!