Looking for a neutral DOA for summing

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

living sounds

Well-known member
Joined
Jul 26, 2006
Messages
3,930
Location
Cologne, Germany
Got a console that sums 40 channels into a CAPI 2-ACA. I've tried lot's of DOAs already (there's even audio for comparison I uploaded here somewhere) and after a lot of improvements in other parts of the console I don't like the sound of any of them in the summing position. The one that colours the sound the least is a hybrid op amp from CAPI that utilizes a NOS Phillips NE5534. Now I've read here and elsewhere that a 5534 is not really up to the task of summing so many channels.

I haven't tried a JH990 yet (smoked the last working one in an unfortunate accident in a mic pre some time ago). Will this be a better choice? Or ist there something else? I've tried several 2520, 1731, 918, even FET and germanium. Nice in a preamp or an EQ with some colour, but for some reason or other not good on the summing bus, where I prefer an accurate representation.

Thanks!
 
I think your problem is with using a DOA more than anything. They are all class B devices so you will get that harshness if you have sensitive ears. Only one way to get rid of that - go class A.

Cheers

Ian
 
Well, an SSL 4000 is not class A, but its summing bus still sounds very nice to me. The summing itself is handled by discrete components though.

I've tried higher quality (on paper) monolytic op amps as well as the input stage of the hybrid op amps in the summing position, but didn't much like the sound of that either...
 
living sounds said:
Got a console that sums 40 channels into a CAPI 2-ACA.
too much of a task for the summing amp expecting "high-fidelity" results.
suggest you operate the summing amplifier at less than "unity" , make up the needed gain in the bus amplifier.

living sounds said:
Now I've read here and elsewhere that a 5534 is not really up to the task of summing so many channels.
it may not be the ultimate solution but the 5534 based summer in SSL consoles has been responsible for many good sounding recordings.

living sounds said:
I haven't tried a JH990 yet (smoked the last working one in an unfortunate accident in a mic pre some time ago). Will this be a better choice?
no fan of the 990, but summing is arguably the only place it seems to function decently.

regarding the SSL 4k summing amp (open to correction if it is not the 4000 model),
no discrete active components.
the LM394 is truly an integrated circuit, working in conjunction with a 5534.

 

Attachments

  • summing.jpg
    summing.jpg
    59.3 KB · Views: 83
Thanks!

I naively inferred the 5534s were just buffer stages in the SSL.

The console sums through 22k resistors, so the feedback resistor and cap of the summing amp have been modified to 13k and 100p per Jeff's suggestions. Less than unity gain it is.

The hybrid op amps have the 22 pf caps between pin 4 and 8 to allow for unity gain stability. Would there be any benefit to leaving them out here?

Why do you not like the 990? It sounded pretty good in a preamp, better than the other DOAs to my ears...
 
The SSL is basically a 5532 with an enhanced front end. The 5532 is not like other op amps. Even Doug Self admits to being puzzled by its exact topology but it somehow works especially well for audio. It not in the same class  as regular claas B op amps IMHO.

Cheers

Ian
 
The Weiss DOA that I believe was designed my forum member Samuel Groner is a spectacular performer.
http://www.weiss.ch/products/op1-bp
 
ruffrecords said:
The SSL is basically a 5532 with an enhanced front end.

Yes, I had thought the summing bus was an exception, too.


ruffrecords said:
The 5532 is not like other op amps. Even Doug Self admits to being puzzled by its exact topology but it somehow works especially well for audio. It not in the same class  as regular claas B op amps IMHO.

Cheers

Ian

Which 5532? :) Even with a high quality PSU, optimized decoupling and grounding etc. they all sound different. I only use the NJM5532D and NOS Signetics/Phillipd 5534 now, they are the most transparent.


Have you tried other ICs in the hybrid amp? I would think a modern high performance opamp might be what you want.  Maybe lme49710?

Yes, I even have the coveted metal can version of the LME49710. Phillips 5534 sounds much better to my ears...
 
living sounds said:
Which 5532? :) Even with a high quality PSU, optimized decoupling and grounding etc. they all sound different. I only use the NJM5532D and NOS Signetics/Phillipd 5534 now, they are the most transparent.


Yes, I even have the coveted metal can version of the LME49710. Phillips 5534 sounds much better to my ears...

As far as I am aware, all 5532 parts are made via a mask exchange agreement with Phillips so they should be electrically identical. There will be process variations between manufacturers but they should not be gross enough to be audible.

Cheers

Ian
 
living sounds said:
Got a console that sums 40 channels into a CAPI 2-ACA. I've tried lot's of DOAs already (there's even audio for comparison I uploaded here somewhere) and after a lot of improvements in other parts of the console I don't like the sound of any of them in the summing position. The one that colours the sound the least is a hybrid op amp from CAPI that utilizes a NOS Phillips NE5534. Now I've read here and elsewhere that a 5534 is not really up to the task of summing so many channels.

I haven't tried a JH990 yet (smoked the last working one in an unfortunate accident in a mic pre some time ago). Will this be a better choice? Or ist there something else? I've tried several 2520, 1731, 918, even FET and germanium. Nice in a preamp or an EQ with some colour, but for some reason or other not good on the summing bus, where I prefer an accurate representation.

Thanks!
A 40 input sum amp is running at a noise gain of 41x (32dB+)... A decompensated 5534 will work adequately but won't have excessive loop gain margin at 20kHz. 

Depending on the topology (is bus always loaded), a decompensated DOA might squeak out some extra open loop gain if not required to be unity gain stable.  This may be as simple as removing (reducing?) the compensation cap inside a competent DOA.

I wouldn't rule out some of the newer uber IC opamps. There are some good off the shelf ICs that weren't available last century.  No I don't have suggestions.

JR
 
ruffrecords said:
As far as I am aware, all 5532 parts are made via a mask exchange agreement with Phillips so they should be electrically identical. There will be process variations between manufacturers but they should not be gross enough to be audible.

Cheers

Ian

Hey Ian, there are threads here and elsewhere with people stating that everything is cooking their own thing with 5534/5532's. They even measure differently, Signetics and NJM have lower hf distortion than the TI variant.
 
JohnRoberts said:
A 40 input sum amp is running at a noise gain of 41x (32dB+)... A decompensated 5534 will work adequately but won't have excessive loop gain margin at 20kHz. 

Depending on the topology (is bus always loaded), a decompensated DOA might squeak out some extra open loop gain if not required to be unity gain stable.  This may be as simple as removing (reducing?) the compensation cap inside a competent DOA.

I wouldn't rule out some of the newer uber IC opamps. There are some good off the shelf ICs that weren't available last century.  No I don't have suggestions.

JR

Thanks John. I tried some of the uper IC opamps and didn't like what they did. I could remove or reduce the compensation cap from the 5534 though, does that make sense here?
 
living sounds said:
Thanks John. I tried some of the uper IC opamps and didn't like what they did. I could remove or reduce the compensation cap from the 5534 though, does that make sense here?
The 5534 is stable for closed loop gains greater than 10dB (roughly 3 inputs) without any compensation cap.

There are also ways to trick decompensated op amps into thinking they are running at > 10dB by using a RC to ground from - input. The pole frequency for this RC can be >>20kHz so will not affect audio passband, just needs to look like 10dB closed loop gain by several hundreds of kHz to prevent oscillation.

JR

PS: I am a fan of empirical measurements, if you bench tested the uber op amps that you didn't like the sound of, the deviation from linear might reveal a way to minimize that deviation.  I have seen some data sheets claiming linearity down below gnat fart levels (-140dB). If you hear something (anything at all) it might not be operating linearly. Certainly not meeting spec.
 
 
living sounds said:
Well, an SSL 4000 is not class A, but its summing bus still sounds very nice to me. The summing itself is handled by discrete components though.
Why not build it then? The schematic posted before is the right one afaik.

Michael
 
I couldn't stress enough the importance of proper setup of power supply rails and their filtering. Sounds to me like you're just listening to bunch of opamps performing poorly. This goes down all the way to basics of proper hierarchical ground and well isolated and localized filtering for all stages. Then you can start attacking the system with modern regulators (not LM317/337).

I know you have bad power setup for the fact you are hearing the opamps at all.

Also without measurements all threads like this - and all that came before - is just shooting in the dark.
 
Kingston said:
I couldn't stress enough the importance of proper setup of power supply rails and their filtering. Sounds to me like you're just listening to bunch of opamps performing poorly. This goes down all the way to basics of proper hierarchical ground and well isolated and localized filtering for all stages. Then you can start attacking the system with modern regulators (not LM317/337).

I know you have bad power setup for the fact you are hearing the opamps at all.

Also without measurements all threads like this - and all that came before - is just shooting in the dark.

The PSU is a really high quality discrete one with ultra low noise and very low impedance. Regardless of the op amps used, the console sounds way better using it than the upgraded one made to be used with this console. Everything else has been taken care of, local decoupling, grounding, filtering, dampening etc.

Regardless, I hear these differences between op amps almost regardless of piece of gear, position, specific task, PSU, whatnot. And people use these DOAs for their specific sound, after all. I just don't like it on the 2bus.
 
Back
Top