opa 2134 vs 2 opa627

[David Nessim]

746

Hi Motherone,

I am not too surprised about the 746. It is the dual version on the 744 which was a very fine sounding better alternative to the 711/712 about 10-15 years ago. At the time it was hard to come by, the bees-knees. Just by looking at the specs, you can already see that 627 is only better on noise, which may not be a factor. I entirely forgot these devices..

About bypassing, BB give PLENTY advice in their application notes. One time they wanted 33uF, 3.3uF, and a .33uF ceramic closest to the chip. They design the chip, and they KNOW what they are talking about. I could not stress this matter more, but have also found that basic stiff PSU design, then "fat" (47-100uF)decoupling with low ESR types somewhere near the chip, and say a 1uF tantalum stuck to the chip works very well. There is no formula for this, just that chips MUST be bypassed as well as practical close up.

David

 

[Htguy]

Decouping caps???

Hi,
I am pretty new at the audio mod and electronics in general, so I was wondering if someone could explain decoupling caps to me.

1) What do they actually do for the output?
2) What is the concept behind using them?
3) do you use polarized caps for the job?
4) Using the pins outs for the opa627 what pins would I connect the decouplers to?

5) WHy did one person suggest connecting a 100-220nf cap across the + and - on the opamp? What would this do?

Thanks for your help.

God bless...

Mark

 

[jacco vermeulen]

Exactly to the point, iamwhoiam !
That test showed no difference between AP and BP versions, difference only being the DC values.
I got the BP versions at the same cost as the AP's, a lot of them.
I have been forcing opamps in class A for ages, because of the distinct difference in sound when fets are driven in class A.
Maybe a reason why, imho, the difference between B's and A's exists.
20years ago i started using the Burr Brown OP27, in those days people who favored OP27 over 5534's already were exotic.
Telling others that BBOP27's sounded much different than PMI OP27's gave you the look, BB OP27 were very hard to come by in those days.
Now people say OP27's are no good, 5534's are trash.
It has not been that long ago when i still saw $3000 preamps running on 5534/5532's.
imo, the proper way to drive fets is in class A.
Still, the majority seems to keep running their opamps in normal mode.
Carlos is very keen on the OPA627, so am i, i am very fond of the BurrBrown sound.
I havent tried by-passing on the pins yet, but if i do i dont think i need a scope to tell the difference.
Still, it would no hurt if an existing difference could be quantified.

 

[peranders]

Re: Decouping caps???

Hi,
I am pretty new at the audio mod and electronics in general, so I was wondering if someone could explain decoupling caps to me.

2) What is the concept behind using them?

This is very basic. It's also easy to understand if you play a little with the simulation software LTSpice (freeware, just download it!) and make you ideal voltage source less ideal. When it becomes too bad you will get eventually an unstable performance. This is very real for fast circuits. The theory behind operation of an opamp is that it requires low impedance (at higher frequencys', in the MHz region) of the supply voltage.

 

[motherone]

Re: 746

You must be very hard at work !

I have OPA627's on a Stokes SDS class A headphone amplifier, before those i tried lower priced opamps.
I do hear a difference between two different Burr Brown numbers.
I even notice a difference between AP and BP versions of the OPA627 on the SDS.
Sounded like you had a cheap hobby, Motherone !
But i did read you first posting earlier, thank goodness.
Those little critters are so sensitive, an inch can be a long way.

I think you misunderstood my post (which is why I replied to the subsequent ones when Jean-Paul questioned the same thing). The comparison was *NOT* between the opamp, but rather between the *BYPASSING* methods for the opamp.

My statement was that I could hear no difference on my board that has 4x470uF Elna Cerafines + 2x 3.3uF Metallized Polypropylene versus a second board that has 4x470uF Elna Cerafines + 2x 3.3uF Metallized Polypropylene + *2x .1uF ceramics* on the PSU pins of the opamp to ground.

Carlos has been spouting that without bypass caps on the pins, all the burrbrown opamps (2132/627 included) sound awful and midfi. I'm just tying to state that I think that's BS Listeing on a headphone amplifier, there was absolutely no difference in sound between the two boards that were bypassed differently.

Hi Motherone,

I am not too surprised about the 746. It is the dual version on the 744 which was a very fine sounding better alternative to the 711/712 about 10-15 years ago. At the time it was hard to come by, the bees-knees. Just by looking at the specs, you can already see that 627 is only better on noise, which may not be a factor. I entirely forgot these devices..

You're probably right. The thing that impressed me about the 746 is that I found the opamp in a 15 year old piece of gear, and was really impressed by it. After doing a little digging, you're correct that it's comparable to the BB in spec, just giving up a little on noise. I would happily recommend it to anyone who wants a "less expensive" 627/637.

About bypassing, BB give PLENTY advice in their application notes. One time they wanted 33uF, 3.3uF, and a .33uF ceramic closest to the chip. They design the chip, and they KNOW what they are talking about. I could not stress this matter more, but have also found that basic stiff PSU design, then "fat" (47-100uF)decoupling with low ESR types somewhere near the chip, and say a 1uF tantalum stuck to the chip works very well. There is no formula for this, just that chips MUST be bypassed as well as practical close up.

David

I agree that bypassing correctly is essential to getting the circuit to function correctly. That doesn't mean bypassing directly on the pin is neccessary. Someone pointed out Tangent's site above. I'm actually working on a PPA amp, which is a pretty impressive piece of kit. Up to 9 PSU rail caps (all of mine are 220uF Elna Cerafines), 2x 100uF + 1uf Metallized Poly's at EACH opamp for bypasses, plus another set of polys around each set of buffers.

Furthermore, Tangent, PPL and Morsel and the other guys at head-fi actually try to make an effort to make sure their boards are stable with all sorts of opamps. I believe that some of them also make a point to try and *measure* the differences between them. They do state caveats on that site to get some picky opamps working correctly.

My whole statement on this thing (and not trying to derail this thread) is trying to show that yes, proper bypassing is essential for opamps, *BUT* proper bypassing DOES NOT mean that you have to solder caps between the pins of the opamps and ground to get them to "sound" right. While this *MIGHT* help on very high speed opamps operating in high-gain scenarios, I highly doubt it will make a difference on a properly designed boards or standard (non-phono stage) gain device in the audio chain. It is certainly not a requisite!

I apologize (again) if my original post was cryptic and made it sound as if I was comparing the sounds of those opamps, rather than the bypassing methods. I'll reiterate that I *tried* all those opamps with the bypassing method carlos mentions and could not hear a difference between the carlos bypass method (.1uF on the pin) and the standard bypass method (3.3uF Metallized Polyp ~ 1/2", which I just measured, from the opamp). It's not a comparison between the opamps.

Unfortunately, I can't go back and edit my post to clarify what I was saying due to the "1 hour" limit on this forum.

Mike

 

[jean-paul]

Mike, before you get mad of the remarks that are to come I can edit your original post if you wish.

 

[jacco vermeulen]

i heard you, Mike.
Carlos must have stated that everyone so far has been listening to cra... sound.
Which i have not.

 

[motherone]

Mike, before you get mad of the remarks that are to come I can edit your original post if you wish.

Jean-Paul,

I wasn't mad at anyone.. I was just trying to emphasize my points so there would be no further confusion. I apologize if the tone came off badly. It certainly wasn't my intention. Just tryint to alleviate the confusion on the original post.

If you could just change the line:
"I heard absolutely NO difference between any of the chips that I tried in the socket (OPA2132/4, 627/637, AD8620, etc.). "

To:
"I heard absolutely NO difference between the socket with 2x .1uF bypass caps going from the V+/V- pins to ground versus the socket without. None of the chips I tried (OPA2132/4, 627/637, AD8620, etc.) changed sound with the bypass directly on the pin versus 1/2" away from the socket"

I would most appreciate it! Then you can probably delete my confusing follow up posts.

I just want to apologize again -- I'm not trying to come off sounding gruff, but unfortunately text isn't the best way to try to communicate the tone in which you're trying to explain something

 

[carlosfm]

First, trying op-amps on circuits like those very bad "Cmoy" style headamps that populate the net can be very misleading.
That tangentsoft.net link is a good example of how to muck things up.

Second, from the OPA627 datasheet, page 9:
"Power supply connections should be bypassed with good high frequency capacitors positioned close to the op amp pins. In most cases 0.1mF ceramic capacitors are adequate."

Notice: "In MOST cases". Remember that most op-amps are not specifically made for audio use, and the datasheets just have generic information.

And it continues:

"Applications with low impedance loads or capacitive loads with fast transient signals demand large currents from the power supplies. Larger bypass capacitors such as 1mF solid tantalum capacitors may improve dynamic performance in these applications."

One should learn to read between the lines and think.
Is this a perfect description of audio use or what?
Isn't music a case of "fast transient signals" instead of steady signals?
The OPA627 NEEDS capacitance on their pins, not just small bypass caps.
The OPA2132/4 needs them too, as much as the OPA627.
Using them for music, with just small bypass caps the bass is untight and mucks up all the rest of the spectum.
It really sounds horrible.
Why?
Because they are starving for PSU current and stable voltage.

If you guys just look at an op-amp specs (faster, slower, etc.) on the datasheet and decide what kind of PSU arrangement to use just by doing that you are very wrong.

 

[David Nessim]

Bypass..

Mark,
The purpose of decoupling is several. When we think of conductors, we imagine PCB track to be of zero impedance, because our Fluke-11 measures them as perfect. But dependent on frequency, they exibit an impedance, like having an inductor and resistor in series. This gets worse as frequency goes higher.
We also imagine power supplies to have zero output impedance. But in the real world they become worse as frequency rises. What is true for a few hundred herz is not so for higher frequencies.
To make things worse, we now draw currents into the opamp in a pulsive manner, with peaks and spikes in our demand. These traks could be feeding yet other opamps, and the opamps cross modulate each others power line.
PSU's output noise, from the rectifiers, from the regulating circuits inside. The big fat smoothing caps work on the 50/60 herz sine, but do not react that well to higher frequency energy, to a diode close to cutoff and brieffly flickering on and off, with all the micro noise associated.
So in a way, your circuit can be thought of as your "ideal" opamp, with a noise source on the PSU, and inductors with resistors in series feeding all this to the supply rails.
What the bypassing does is "short" much of this to ground, between supply rails and earth, or even between themselves (someone suggested a cap across the rails). This make the PSU look like a low impedance source at higher frequencies, and capacitor types and sizes tackle this at higher frequencies. In a sense, like have the same "smoothing" fat caps duplicated near you IC to act the same for the higher frequencies, and quickly discharge dump current when the opamp wants it and is trying to pull it through the PCB conductors.
Hope this is all legible..
David.

 

[iamwhoiam]

Re: Re: 746

I agree that bypassing correctly is essential to getting the circuit to function correctly. That doesn't mean bypassing directly on the pin is neccessary. Someone pointed out Tangent's site above. I'm actually working on a PPA amp, which is a pretty impressive piece of kit. Up to 9 PSU rail caps (all of mine are 220uF Elna Cerafines), 2x 100uF + 1uf Metallized Poly's at EACH opamp for bypasses, plus another set of polys around each set of buffers.

Furthermore, Tangent, PPL and Morsel and the other guys at head-fi actually try to make an effort to make sure their boards are stable with all sorts of opamps. I believe that some of them also make a point to try and *measure* the differences between them. They do state caveats on that site to get some picky opamps working correctly.

Mike

You should see the capacitor arrangement they have in PPAv2 (especially C2) with onboard discrete diamond buffers, tis mad..

From head-fi.org regarding why read (massive decoupling discussion with ppl, tangent and morsel):
http://www6.head-fi.org/forums/showthread.php?t=110747&page=4

 

[motherone]

One should learn to read between the lines and think.
Is this a perfect description of audio use or what?
Isn't music a case of "fast transient signals" instead of steady signals?
The OPA627 NEEDS capacitance on their pins, not just small bypass caps.
The OPA2132/4 needs them too, as much as the OPA627.
Using them for music, with just small bypass caps the bass is untight and mucks up all the rest of the spectum.
It really sounds horrible.
Why?
Because they are starving for PSU current and stable voltage.

If you guys just look at an op-amp specs (faster, slower, etc.) on the datasheet and decide what kind of PSU arrangement to use just by doing that you are very wrong. [/B]

First off, the amp I'm referring to isn't a "CMOY" style amp -- it's actually identical to the OPA627/BUF634 preamp with the addition of a buffered ground channel.

What I'm trying to state here, Carlos, is what you do -- "empirical study". I have kept an open mind and tried your decoupling method.

I wired up two buffered amps, BUF634's on the Left, Right and Ground. I used an OPA627 on the ground channel (they "buffer" the ground return currents on the Pimeta that I tried this with). I then tried both the OPA627 and 637, using a dual SOIC to single-dip converter for the left/right channel. Finally, I wired one board with 2x .1uF ceramics directly to the pins of the L/R opamp socket, and left the other one as the standard way.

The result? With my AKG K 401, there was absolutely no audible difference between the two setups. Everything else was identical (I have 5 of these pimeta amps sitting around, some for preamps and the others as gifts for friends). I tried many different opamps, including the AD8610 in the ground channel and the AD8620 in the L/R socket. Still, there was no difference.

In my opinion, the AKG's will be much more revealing of any audible changes than a standard speaker setup, since they are basically bolted to your head. Others may disagree with me, and they're certainly entitled to their opinons

Like I said, I think that the .1uF on the pins MAY make a difference on very high speed opamps (like the AD8065/66), but I certainly don't think that the OPA627 qualifies in that arena.

Please note, Carlos, I am not attempting a personal attack here. I agree with you that the sound of the 627/637 is fantastic, and I also agree that proper decoupling on ANY power supply is neccessary. But, I disagree with your assertion that the only proper way to decouple these opamps is by putting the caps directly on the pins. While it doesn't hurt anything, I'm simply trying to show that decoupling ON the pins isn't mandatory.

If you wish to contend that they require a larger decoupling capacitence AND a bypass directly on the pins, I will be happy to hack one together and test it, but I don't think that it will make a difference. The resevoir capacitors and traces on this board are fairly tight given the parts.

Mike

 

[iamwhoiam]

Then again, the PIMETA pcb is a very tight design..

If track lengths were longer, it might introduce some distortion, but even with many mm's between cap and chip pin, it doesn't affect it audibly..

 

[iamwhoiam]

First, trying op-amps on circuits like those very bad "Cmoy" style headamps that populate the net can be very misleading.
That tangentsoft.net link is a good example of how to muck things up.

Firstly testing the different characteristics between chips won't require such a high end device.. he is simply comparing between different chips. It isn't hard determining differences between OPA2132/4 and OPA627s/637s on that amp at all.

Secondly, have you ever listened to a cmoy amp?

Thirdly, look around the rest of the site. Tangent knows what he's talking about, you should see the rest of the amps and his pcb layouts before criticising his methodolgy of "mucking things up". I'd like to see you head over to head-fi.org and telling people how well he screws up his amps..

 

[motherone]

Then again, the PIMETA pcb is a very tight design..

If track lengths were longer, it might introduce some distortion, but even with many mm's between cap and chip pin, it doesn't affect it audibly..

I agree, it's a very well done layout. Carlos is of the opinion that they bypassing must be done on the pins of the opamp. I'm trying to show that's not the case, at least with most of the opamps being used for audio.

Firstly testing the different characteristics between chips won't require such a high end device.. he is simply comparing between different chips. It isn't hard determining differences between OPA2132/4 and OPA627s/637s on that amp at all.

Secondly, have you ever listened to a cmoy amp?

Thirdly, look around the rest of the site. Tangent knows what he's talking about, you should see the rest of the amps and his pcb layouts before criticising his methodolgy of "mucking things up". I'd like to see you head over to head-fi.org and telling people how well he screws up his amps..

Yes. If you look closely, you'll see that Tangent did the opamp testing with a buf634 between the opamp and the headphones. Furthermore, it's not just Tangent working on this stuff -- I believe PPL (Philip Lorocco) did the actual design work, along with Morsel. Tangent did a lot of the testing and some of the external power supply design.

These guys are not idiots. They have been doing the headphone amp stuff for at least 5 years now, and have hundreds if not thousands of people using/tweaking their designs and trying to improve on it. Most of the headphone amplifiers you see for sale on the head-fi forum and other headphone forums are based on these designs (there are a few notable exceptions).

 

[carlosfm]

Then again, the PIMETA pcb is a very tight design..

If track lengths were longer, it might introduce some distortion, but even with many mm's between cap and chip pin, it doesn't affect it audibly..

Jesus!
I once looked at it and got horrified!
No way Jose.
The OPA627+BUF634 must be used in a much better layout to give their best.

I insist, it's not just a question of bypassing with small caps on the pins, the chips need some capacitance there too.
EACH chip, because sharing caps for several chips is no good.

Just a remark: even with the best headphones, if the source component is not on par, it will defeat any reliable evaluation.
Because the source component dominates, it is the most important element on a system.
Trash-in, trash-out, no miracles.
The source dominates, the rest of the system just has to follow.

I know some don't agree with me and give the most importance to the speakers.
For me, that's like building a house starting from the roof.
It won't work.

 

[carlosfm]

Please note, Carlos, I am not attempting a personal attack here.

I've read the whole thread before posting.
I didn't see any personal attack, that's out of the question.

But just try, instead of small 0.1uf, 22~100uf directly on each PSU pin to ground, on each chip.
And also 100~330nf directly across V+ and V- pins, as J-P already said.

 

[motherone]

Jesus!
I once looked at it and got horrified!
No way Jose.
The OPA627+BUF634 must be used in a much better layout to give their best.

There are many measurements done on this amp that would say contrary. If you think you can create a more optimal layout, you should and generate a product out of it. If there truely are the audible differences that you advocate, then the market will certainly buy these in droves. I am not saying this sarcastically -- I mean it in all seriousness. I would gladly purchase a board from you to try it out and help you beta test it.

Have you looked at the implementation on the PPA? That does include 100uF electros and 1uF polys at each opamp.

I insist, it's not just a question of bypassing with small caps on the pins, the chips need some capacitance there too.
EACH chip, because sharing caps for several chips is no good.

What is your recommendation for putting on each chip? My modified SuperDAC uses 10uF cerafines + .1uF ceramics at each opamp, on each rail.

How much is sufficient, in your opinion, to keep them from being current-starved? Also, do you have anything to show (measurement wise) that would prove that the opamp is current starved or having issues with transient response?

Just a remark: even with the best headphones, if the source component is not on par, it will defeat any reliable evaluation.
Because the source component dominates, it is the most important element on a system.
Trash-in, trash-out, no miracles.
The source dominates, the rest of the system just has to follow.

I know some don't agree with me and give the most importance to the speakers.
For me, that's like building a house starting from the roof.
It won't work.

Well, I'll give you a little background on my source. Presently, my "high-end" CD rig consists of a NEC CD-602 CD-ROM drive feeding a heavily modified M-Audio SuperDAC 2496. I'll be posting the details of my mods, but the gist is that it's gone through a total cap replacement with quality parts (Panasonic FM/Cerafine), moved the bypass caps closer to the opamps, and all the opamps have been replaced with OPA627's. I'm still working on modifying the transport, and will probably add a Monarchy DIP to provide a better clock to the DAC and attenuate jitter from the transport.

As for your assertion of "garbage in, garbage out": At one of the audio shows, Wilson Audio demo'd two "$10,000" systems. If I recall correctly, on one, they used $4,000 speakers, plus a $1,500 CD Player, a $3,500 amplifier, and $1,000 worth of interconnects. On the other, they used their $8,000, a $1,500 amp, some zip cord and asked everyone which system they thought sounded better. Everyone in the audience resoundingly said the second system (with the Wilson speakers) sounded better. They then revealed that the source was an iPod.

Their point was that the speaker affects the sound far more than anything else in the signal path, and regardless of how good the rest of your system is, if the speakers can resolve, then it doesn't matter how good your source is. I guess it's a "chicken and the egg" type problem.

Sorry, I feel bad that we've derailed this poor thread to kingdom come. I was only trying to state that the original poster (htguy) should not freak out about bypassing the opamps when he replaces them. I would say that unless he goes for highspeed opamps, the bypassing in his present unit is probably sufficient. It wouldn't hurt to double-check and see if the manufacturer did provide proper bypassing, but he most likely does not need to worry about it.

 

[carlosfm]

Everyone in the audience resoundingly said the second system (with the Wilson speakers) sounded better. They then revealed that the source was an iPod.

They sell speakers, right?
Do you really trust that test?
Also, do you blindy believe that a CDP, just by costing $1500, is a good one?
The cheapest Linn CDP (which costs around 1,600€) eats the Sony SCD-1 for lunch, anyday, anytime, anywhere.

And I don't talk because I've read reviews on audio mags, or because others say.

Their point was that the speaker affects the sound far more than anything else in the signal path, and regardless of how good the rest of your system is, if the speakers can resolve, then it doesn't matter how good your source is. I guess it's a "chicken and the egg" type problem.

One of the worse sounds I've heard from a "suposedly good" system (like the owner liked to belive) had an old, mid-range Denon CDP, two cheap Denon integrated amps (bi-amping) and big Kef Reference speakers.
The sound hurt my ears in a way I had to turn down the volume.
Digitalitis at it's best.
Show me a speaker that cures this.
One day I went there with a Marantz CDP that I extensively modded.
The guy started compulsively listening to his favourite CDs and I could only get out of his home at 3 AM.

 

[motherone]

They sell speakers, right?
Do you really trust that test?

Yup. And some of them are arguably the best in the world, according to not just the audio mags, but the forums and everything else I've read. I've never had the opportunity to actually listen to anything from Wilson, though.

Also, do you blindy believe that a CDP, just by costing $1500, is a good one?
The cheapest Linn CDP (which costs around 1,600€) eats the Sony SCD-1 for lunch, anyday, anytime, anywhere.

I don't blindly believe anything. I'm just stating what was written in the article. I believe the demo was done at CES in Vegas, with quite a few people in the room.

And I don't talk because I've read reviews on audio mags, or because others say.

Wow. Apologies if I can't offer something first hand to contradict it, since I don't have many "World Class" sound systems sitting about my apartment.

One of the worse sounds I've heard from a "suposedly good" system (like the owner liked to belive) had an old, mid-range Denon CDP, two cheap Denon integrated amps (bi-amping) and big Kef Reference speakers.

The sound hurt my ears in a way I had to turn down the volume.
Digitalitis at it's best.
Show me a speaker that cures this.

Interesting. I don't think I've ever heard "Digitalitis" that has required me to turn down anything. I've heard bad speakers that made me plug my ears, but never a source component.

One day I went there with a Marantz CDP that I extensively modded.
The guy started compulsively listening to his favourite CDs and I could only get out of his home at 3 AM.

Well, that's good for him I did the same thing once I purchased my SuperDAC, and again once I modified it.

I don't have any audio friends in the area that to swap gear with, so I'm stuck with what I have on hand to test and compare with, and what I've heard in shops in ideal and not so ideal setups. All of my friends prefer to spend their money on Computers, Cars, Golf and Women rather than audio.. Though I have managed to pull one of them into the fringe.