Op-amp recommendation

[naxos]

Hi,

I've decided to build myself a custom monitor mixer. 4-8 stereo line ins to 1 stereo out.

I've used 5534's in the past but was hoping to get the consensus on what is the nicest sounding IC op-amp to use for summing. So what is your favorite IC op-amp?

(I posted this to "low-end -theory" then realized I had put it in the wrong place, sorry for the crosspost.)

Thanks in advance.

[tINY]

The Ne5534 is pretty good. You might try

LM833
LM837 (careful of the compensation as they tend to oscillate a bit at 1-3 MHz)

Analog devices and Burr-Brown catalogs have some good choices too.

Make sure that the amp can source and sink enough current for the summer! If you have a lot of channels on a bus, there can be quite a bit of current involved (especially if you are using low impedances to keep noise in check.



-tINY

[Dan Kennedy]

A summer is a special place. It requires very good performance from an opamp. Especially if it's from a large number of inputs. Remember that the amp is basically running at mic pre type gains for every signal that goes thru it, so it has to be wide bandwidth, low noise, and needs to be able to drive a fairly low impedance feedback network, plus its load.

I like the 990, or similar types, or the AD-797. LT-1468's aren't bad.

In a 4 to 8 sort of box, I'd look seriously at the 1468.

[Tim Farrant]

OP 275 (dual) - can run on +/-20V rails, will drive 600 ohm, stable in most applications, quiet enough for an 8 input type mix buss, really good DC offset performance which means less coupling caps required if you design it right.

Tim

[PRR]

You got good advice.

The Jensen-design 990 is still hard to beat in many critical audio uses. There are several pin-alike "990"s; some don't reach the 990's noise performance or VAST output current (while perhaps being better in other ways). And it isn't a chip but a chunk.

The 5534/5532 is often 98% as good for 2% of the price. But there seem to be several "vintages", some considered better than others. The balance of process specs needed for best audio are different from what most non-audio chips need: most chips want minimum currents, in audio we often want as much current as we can stand, at least in input and class-AB output stages. A foundry that does not dig audio may be "optimizing the process the wrong way" for out needs. I have not done critical comparison, but NJM makes a 5534/5532 copycat and audio is NJM's main business.

LM833 and LM837 -seem- to be reinterpretations of the 5534, generally similar though not the same (and not rated for the high power voltage). Note that 5534 is not unity gain stable (but in a summer, you would only run unity gain with all channels muted; even then you could add a dummy resistor with minimal loss of performance). I had not seen tINY's LM837 MHz oscillation but such problems are common in hot amplifiers.

I have not seen an AD797. The specs look promising. Astonishingly low noise voltage for a chip. If you do DC-couple, it has much better DC specs than most of the audio standbys: 25uV instead of 1mV. THD is low and can be gimmicked lower.

I'm also too old to know the OP275, but I know people who love it.

For eight 10K mix resistors, a fat vacuum tube can be pretty low noise, and has "glow". It is harder to get effective feedback to kill mix interaction, but you may not need theoretical zero-Z mixing in a monitor mixer. A 6DJ8 gain stage with 6DJ8 or 12AY7 (or HV NPN) cathode follower fed-back to unity gain (from one 10K input) would give a summing impedance around 300 ohms, mix interaction of a few tenths of a dB, and drive a reasonably sensitive power amp easily. If the gain stage has Gm over about 5,000 microMhos and you weed-out noisy tubes, overall noise is similar to the summing resistors alone.

Of course as a monitor mixer, mundane details like stability, grounding, and connections are more important than how cool your chip is.

[Jim Williams]

Keep in mind that the opamp's open loop gain is directly related to the amout of THD you will have and the ability to sum a large number of inputs without running out of loop gain.

Although the sum amp's signal is run at unity gain, the amplifier is running at a much higher loop gain. 8 channels wouldn't be a problem, just try it in a trident with 40 inputs to hear what I'm discussing here.

My choices for sum amps:
LT1357
THS4011
OPA134
OPA627
LT1128
Avoid the AD797 as it has a complex phase curve and tends to be unstable, besides, it sounds like crap.
The op176/275 sounds boring, try the OPA134 instead.

If you need mondo output current, a pair of complimentry transistors in the loop will allow tons of current in case you need to drive the signal into a 50 ohm load.

Jim Williams
Audio Upgrades

Remember! They don't work if you let the smoke out.

Jim Williams
Audio Upgrades

[Kev]

welcome to PRR

[Dan Kennedy]

I dunno, the 1357 is edgy and brittle.

The 797 doesn't sound like crap in a properly implemented circuit.

Maybe just modifiying doesn't give enough insight into proper application?

[Tim Farrant]

Originally posted by Jim Williams
The op176/275 sounds boring, try the OPA134 instead.

Boring? please elaborate....

Could it just be that it does not impart too much of it's own sound onto the signal?

Cheers
Tim

[naxos]

Thanks so much for the advice. I'm going to go off and digest it all now. I'll post again and let you all know how it went... probably in a few weeks (job pressures, ya know), and maybe ask a few more questions in the meantime.

[jdunn]

Maybe just modifiying doesn't give enough insight into proper application?

Jim has been designing mic pres and other gear for years, so he does a lot more than just "mod" gear. Maybe you didn't know that he sells his own mic preamp, the Audio Upgrades High Speed.

[dasbin]

Originally posted by jdunn
Jim has been designing mic pres and other gear for years, so he does a lot more than just "mod" gear. Maybe you didn't know that he sells his own mic preamp

Well, so does Dan.
No need to argue credentials in here, I think. Theory and application!

I'm a fan of OP627's, 2134's, AD825's, and I have to admit that the lowly old TL072 ain't that bad in a well-designed circuit, too! Don't know about in a summer, though. It probably wouldn't behave with the impedance levels.

[dcollins]

Originally posted by Jim Williams

Avoid the AD797 as it has a complex phase curve and tends to be unstable, besides, it sounds like crap.

I think you must have a problem with your implementation of the 797!

DC

[tINY]

The "sound" of an Op-amp usually has a lot more to do with circuit implementation than with anything inherent to the device.

If you are getting any charateristic "sound" from a summing amp, you are doing something wrong....



-tINY

[Jim Williams]

Originally posted by dcollins
I think you must have a problem with your implementation of the 797!

DC

Not really. I discussed the implementation of the part with Walt Jung and other AD engineers when it came out. Since there is a "knee" in the Bodie plot of open loop gain it has a couple of querks. AD recommends using a 100 ohm resistor to isolate the feedback loop from a pure integrator function to enhance stability.

I just found the part lacking in any depth or low level detail, to my ears it's a dry filter of the dynamics. I just don't care for most of AD's opamps. Like anything else, it's great until you find something better. I have.

Jim Williams
Audio Upgrades

BTW, I would never be so bold to question the credentials of any member of this forum. I've learned that if I don't know the whole story, I better find it out before I put my foot in my mouth.

[Tim Farrant]

Originally posted by Jim Williams
BTW, I would never be so bold to question the credentials of any member of this forum. I've learned that if I don't know the whole story, I better find it out before I put my foot in my mouth.

That's fine Jim, but you do not quantify your submissions. You say the OP275 sounds boring, I find that our users like the sound of it - otherwise they would not be buying our products some of which use this device. So, I am looking for a definition of "boring".

Cheers
Tim

[dcollins]

Originally posted by Jim Williams
Since there is a "knee" in the Bodie plot of open loop gain it has a couple of querks. AD recommends using a 100 ohm resistor to isolate the feedback loop from a pure integrator function to enhance stability.

Yes they are quirky, and more likely to break-into-song than other parts, but they can have fantastic performance sonically, imo. A "kink" in the Bode plot -- well it wouldn't be the first time...

I just found the part lacking in any depth or low level detail, to my ears it's a dry filter of the dynamics. I just don't care for most of AD's opamps.

We'll have to agree to disagree here.

I'm thinking of a well regarded digital audio product that has nothing but AD parts. Renowned, as it turns out, for its superior low-level performance.

Even with the dreaded 797! How can it be?

In your experience, is there any parameter we can look to for better "low level detail" or "dynamic filtering?"

I know what pleases the ear is all that _really_ matters, but sometimes measurements can also be useful, imo.


DC

[Jim Williams]

Originally posted by dcollins
Yes they are quirky, and more likely to break-into-song than other parts, but they can have fantastic performance sonically, imo. A "kink" in the Bode plot -- well it wouldn't be the first time...

Yep, the 5534 comes to mind...

We'll have to agree to disagree here.

I'm thinking of a well regarded digital audio product that has nothing but AD parts. Renowned, as it turns out, for its superior low-level performance.

Even with the dreaded 797! How can it be?

If their using an AD dac chip, like a 1853/5, they're missing a lot of low level details that can be revealed with the BurrBrown 1704 or 1792 dacs. The AD1955 doesn't get there either.

In your experience, is there any parameter we can look to for better "low level detail" or "dynamic filtering?"

I know what pleases the ear is all that _really_ matters, but sometimes measurements can also be useful, imo.

I don't know of any test gear that can qualify fidelity of sound.
I use good stuff like Audio Precision, but it only tells you if something is wrong, your ears will tell you if something is right.

Jim Wiliams
Audio Upgrades

P.S. The boring quality I called the 275 is due to the masking of some low level details, particularly fast and small transients like bell trees buried in a mix. I don't like dual opamps in general compared to single one's due to the increased crosstalk but I do use a lot of them in other people's stuff. Just personal taste, like music.

DC

[manning1]

jim. i think i have tested every op amp in the market in building my own diy mic pre's. i would appreciate your comments on one op amp type. to my ears the best op amps i found for my little mic pre were op37 and varaietie like max 437.
do you have any comments why - from a technical perspective ? thanks.

[Jim Williams]

Originally posted by manning1
jim. i think i have tested every op amp in the market in building my own diy mic pre's. i would appreciate your comments on one op amp type. to my ears the best op amps i found for my little mic pre were op37 and varaietie like max 437.
do you have any comments why - from a technical perspective ? thanks.

The OP-37 is just a decompensated version of an OP-27 precision opamp. BurrBrown probably makes the best one, and they make a dual version that's used on the lynx cards. Low dc offset and bias current but pretty slow. The OP-37 needs about a gain of 5 to be stable, and any capacitor across the feedback loop will create an integrator and will make the opamp oscillate. Most mic pres use some form of feedback cap, the Jensen transformers require one as it's part of the "tuning" circuit. The OP-37 would cause problems there. In your case it may be the low level oscillation is adding a subtle "aural excitement" to the preamp. Have you seen the output on an 'scope?

Jim Williams
Audio Upgrades

[tINY]

Actually, putting a small capacitor across the feedback resistor reduces the closed loop gain at higher frequencies and can make the loop stable if chosen reasonably.

The difference you hear (perhaps) has to do with the gain-bandwidth of the op-amps. Since the typical GBW of the OP27 is 8MHz, the most gain you can have at 16kHz is about 54dB. If you run the gain higher than that, you will actually attenuate the high-frequencies relative to the rest of the audio spectrum. The OP37 is "decompensated" and has a GBW of almost 64MHz. That will give you about another 18dB before this starts happening.


-tINY

[manning1]

thanks for your comments jim and tiny. much appreciated.
to be frank i got away from using scopes many moons ago.
because i found they dont reflect the "sound" of the device.
so i record a track and listen to playback.
maybe its how the 37 works with a follow on ashley compressor that i use. purely for level control and NOT over compressing.
i run the "37" at not high gains....then into the ashley.
particularly i find one part - the max 437 is practically indestructible. and it has a sound i like. frankly i just did not like 5534's. or op 275 in my playback tests. nothing i can quantify technically. just the playback sound quality to my ears.
heres another "weird" thing. i once rigged up a crazy mic pre circuit using a max 1001 op amp i think it was (not known to be a low noise op amp) and got a great vocal track from it.
it sounded quite warm and round. in fact this brings me to another oddity. it seems when i try ultra low noise op amps
with one to five nanovolt root hertz noise "something" gets lost in the "sound picture" i can put my finger on with the exception of the "37" types i mentioned.

[zmix]

Originally posted by The Buzz
OP 275 (dual) - can run on +/-20V rails, will drive 600 ohm, stable in most applications, quiet enough for an 8 input type mix buss, really good DC offset performance which means less coupling caps required if you design it right.

Tim

I want to make a few notes here:

I have listened carefully to nearly every serious DIP packaged opamp available, and at great length, too.


I really do like the sound of the OP-275, and I can pick out it's sonic signature a mile away.
But it is vital to examine the circuit to determine what conditions the device may encounter.


Buzz, please look at the OP-275 data sheet very carefully, take particular notice of what happens to the distortion when the load drops below 2k5...Yikes!

The load impedances in a simple parametric EQ could be a real problem here, for example.

Tom Graphey fixed this rather unfortunate condition in the OP-176.


I don't think an FET input opamp is best for a summing application, either.

Your mileage will certainly vary....

[Jim Williams]

Originally posted by zmix


I really do like the sound of the OP-275, and I can pick out it's sonic signature a mile away.

Bingo! That's why I don't like it. I find nearly all SS opamps don't add anything musical to the playback, it's just how much you want them messin' with your "sonic signature". I found I like opamps best when I don't hear them. Rather being enamoured with a particular part's sound, I consider them to be a necessary evil. This is why I choose the most neutral of the batch. The 275 is anything but neutral.

Jim Williams
Audio Upgrades

P.S. This brings up an interesting point, one that may be touchy with designers. How much of the designer's "sound" should be put into a product and does it serve the end user? I often ask myself "what the hell would this thing be useful for". It's if the designer's have a better clue what your're recording and how it should sound, even though they don't know what you are using it on. Sort of reminds me of encounters in top LA LA studios; "I know I just met you, but I have a black box that can make you sound better than you actually are".

[naxos]

Wow, now I am well and truly confused...

I didn't realize that summers had to have low input impedances; I thought the impedance was set by the series resistor going into the minus input.

Can anyone recommend a good book on op-amp circuit design? There used to be one, something like 'complete idiots guide to op-amps' or something like that but it is out of print now.

A friend of mine who is generally clueless recommended using video op-amps because they have a higher bandwidth; is there anything to that?

[shikawkee]

This is a great thread.

[Tim Farrant]

Originally posted by zmix

Buzz, please look at the OP-275 data sheet very carefully, take particular notice of what happens to the distortion when the load drops below 2k5...Yikes!

Agreed, one has to watch out for the loading, but this applies to any device really. With a 1k5 load and +/-20V power supplies the distortion remains pretty good right up to +22dBu and typically we get distortion figures well below 0.01% from the OP275. I tend to design with higher value feedback resistors to keep the combined loading within the circuit above 5k minimum, usually higher.

Yes, this is an interesting thread - funny how everybodys opinion seems so opposed!

Debate on.....
Tim.

[Tim Farrant]

Originally posted by naxos
Wow, now I am well and truly confused...

I didn't realize that summers had to have low input impedances; I thought the impedance was set by the series resistor going into the minus input.

Can anyone recommend a good book on op-amp circuit design? There used to be one, something like 'complete idiots guide to op-amps' or something like that but it is out of print now.

A friend of mine who is generally clueless recommended using video op-amps because they have a higher bandwidth; is there anything to that?

A good book that covers many aspects of circuit design is The Art of Electronics by Horowitz and Hill.

Some Video opamps may have high distortion figures because this parameter does not matter so much with analog video. But, I know Jim makes a preamp using a video amplifier, is this right Jim?

Tim.

[zmix]

In so many ways, arguing about opamps is like arguing about the word "the"...

Someone may well construct a sentence thusly: "My studio is covered in the green fabric" and argue that the word "the" is the stupidest thing that they ever heard.

Caveat Emptor.

It is a great thread, though.

[Jim Williams]

One of the best books on the subject is "Audio Opamp Applications" by Walt Jung. Out of print but worth looking for.

There are lots of video bandwidth opamps that sound crappy. Most have poor thd specs but sometimes it's a bit hard to weed out the contenders as the data sheets usually don't show anything below 100k hz. The opamp's Bode plot can show more.

What you want to see is a large open loop gain in the audio bandwidth. Most audio opamps have very high open loop gain (+100 db) in the 100 hz area, but fall off rapidly in the high frequencies, typically only 60 db at 10k hz. Compare to some "video" opamps which have the open loop gain at around 90 db all the way up to 10k hz. This translates to lower thd in the high frequencies were it's the most nasty. Some of these devices eliminate loading problems by being able to drive 50 ohm loads. I find some of these opamps scraping the residual distortion of my Audio Precision analyzer regardless of the load.

As to dropping in these devices in audio gear: be aware that you will need a scope and some applications chops. Remember, they don't work if you let the smoke out!

Jim Williams
Audio Upgrades