[aurelius]

Mr Occam... I'm sorry... I know these topics come up a billion times and are a bit taudry.  Feel free to move it elsewhere if it does not belong.

I am interested if I can squeeze any more sonic goodness out of my Orion ASPs.  They currently have 11 x OPA2134s.  Based on others' experieinces I thought I might try AD826s.  There are, however a couple of issues with the 826s.  With bi-polar inputs, they are less ideal for filter-use due to lower input impedence (marginally) affecting the transfer function, they consume a bit more current and have a higher DC offset error.

I have read of people who prefer AD823.  With JFET inputs, they are probably more ideal for the application as well.  I am aware of the lower, 15mA, output current drive (which may be an issue in the final [line driver] stage), but is unlikely to be an issue for interstage opamps.

Does anyone here have experience with both (or with reference to OPA2134)?  Also, any comments about improvements to be gained from:
- biasing into class-A using cascaded JFETs
- PS bi-passing tricks
would be welcome.

BTW Not interested in $30-a-pop extavagances.

Thanks, Mark

[Occam]

Mark,

Let me start out with a caveat - When I refer to the subjective characteristics of a monolithic op-amp, I am well aware that it will, no doubt, measure absolutely flat from 5-30,000Hz in circuit. We shouldn't be able to discern tonal characteristics, etc.... so when I refer to the frequency characteristics, it is purely in the subjective realm. Nor should slew rate, bandwidth, etc... have an impact or be significant (with reasonable care wrt impedances, levels and offsets) That being said, I also consider the purely technical characteristics and metrics. What is an appropriate chip for a Dac I/V (LT1468) is not the same as a chip I'd choose for a 'straight wire with gain' line driver/pre (OPA671 unobtainium).

Personally, I consider the OPAx604 and OPAx132/4  mediocre, at best. Not that they do anything badly, its just that they (IMO) don't do anything particularly well. Here are 2 threads from sites that evaluate various op-amps. Please be aware that the subjective comments (by myself or others) are simply that, subjective. And these comments are generally directed towards the use of these chips as headphone amps -
http://www.tangentsoft.net/audio/op-amps.html
http://www.sgheadphones.net/index.php?act=ST&f=12&t=2628&st=0#entry43392

Addressing your query about biasing, technical considerations, etc... Tangent's audiopage has links to his own very fine articles -
http://www.tangentsoft.net/audio/
this page should be a permanent 'sticky' on the Lab Circle.

I've recently done op-amp upgrades on 2 Innersound line level crossover/amps and Zhaolu DACs. This has involved swapping appropriate chips and evaluations by myself and at least one other. This ain't rocket science, topologies were not changed, no animals were harmed. But I've mucked about with at least 10 op-amp 'families' in the last 6 months. I'll restrict my comments to duals. To a greater or lesser extent many divide op-amps' sound into the BurrBrown and AnalogDevices sound. Its not absolute and there is much 'miscegenation'. I prefer to think of it as the Great Dean Divide, Hugh vs. James. Not that Hugh doesn't gadabout on his tricked out motorcycle, pushing limits, both legal and his own.....

IMO, the safest choice is the OPA2107. Its my fave dual, a reasonably priced dual that's as close to a dual OPA627 without potential oscillation problems, as you can get. I prefer it for those DAC and crossover filter duties. I use a OPA2111 for output buffering on the DAC. The 2107 is excellent for that role, but subjectively I prefer the 2111. Then again, if your sonic Nirvana would be a Spectral/Avalon/Mit 'statement' system, these chips are probably not your cuppa.

The chips you mention -
AD826 - a classic AD sound, but not at any extremes. Subjectively, a bit 'flat' IMO. The AD8066 is my favorite AD chip, but its rarely a drop in, being smt, requiring adapters and having very high speed, and prone to oscillation if not properly routed/decoupled. I has great bass, extended treble and a vibrant mid. Excellent for new designs, not recommended as a retrofit. The AD823 is similar to the 8066 in the mids and treble (not as refined), but its bass is constrained. A safe bet in bipolar duals is the OPA2227. Linear makes an excellent series of bipolar chips in singles, duals, and quads, LT1356-1365. They span a range of bandwiths and slew rates which will accommodate retrofits and new developments where you can control decoupling and routing. You can get the sound and subjective characteristics of a high speed AD811 or a LM6171 without the proclivity to melt down.

[note - one of the typical errors we make in using chips is often picking the correct one of 2 chips depending on the gains involved. Often there are 2 versions of a chip, with different compensation schemes roughly divided between chips that are unity gain stable and those which are only stable at gains greater that 3-10+. Examples would be the OPA627/637, OPAx227/x228]

Subjectively yours,
Paul



PS - I can source 60-80 OPA2107 and OPA2111 at very good prices. I asked the vendor to put up an ad in the Marketplace. He isn't interested. I asked him to generate an EBay add specific to the needs of AudioCirclers that would allow variable #s and/or a mix of the chips. No go.
So it appears that a bulk purchase is the only reasonable approach. But I'm not willing to run the distribution for a group buy. Heck, you can send 50 chips to Australia for $5 in Priority Mail Envelope. I'll source them and pass them on at cost to anyone willing to do the distribution. PM me if interested.

[EDIT- As you can see below, I spoke too quickly, Cordless has come through!!! Cordless - you made things easier around here, especially for me!!! Even you furr'ners can get your chips ]

[Occam]

For anyone looking for upgrade opamps for the Zhaolu (or other components), I cadged a vendor into selling 3sies of OPA2107 and OPA2111 Burr Brown Di-Fet dual opamps in 8pin dip configuration on Ebay.
Larry's current preference for the Zhaolu D 2.0 analog outputs is for 2 OPA2107 for the flanking positions and a single OPA2111. So if that were your choice note (in the ppal instructions) 2-2107 & 1-2111. Or whatever combo =3 that you want.
The price is far better than anything I was ever able to scrounge up.

Goto ebay.com
seach on 'Zhaolu'

I've no financial interest in this.

[nature boy]

Paul,

Thanks for lining up a seller for the op-amps combination for the Zhaolu on e-Bay.  I got a set in anticipation of purchasing a Zhaolu D 2.0 in the very near future, when their back in stock.

Regards,

NB

[cordless89]

Hello gentlemen,

I'm the guy with the Ebay listing for the OPA2107 and OPA2111s.  I apologize in the delay in getting them posted.  Should you have any questions, just ask.

Thanks,
cordless89

BTW - thanks for your interest.

[aurelius]

Thank you Paul,

I prefer to think of it as the Great Dean Divide, Hugh vs. James. Not that Hugh doesn't gadabout on his tricked out motorcycle, pushing limits, both legal and his own.....

  Trust me, there's more than a little "rebel without a cause" in our Hugh. 

OPA2107? AD8066? So many options, so little concensus!  There does seem a modest amount of agreement that OPA2134 is not the cat's meow.  If you were designing from the ground up, would you consider the 8066 over the 2107? 

[Occam]

Mark,

If I were designing from scratch, I'd shoot for a design that could accommodate the 'friskiest' of opamps. By that I mean the circuity and layout would be able to run an 8066 without going into oscillation. This is standard practice and one would endeavor to minimized signal line inductance, optimize ps decoupling, etc.... But this is not to say I wouldn't (subjectively) find the 2107 pref ferable. While both chips are fet input, the AD is very high bandwidth and slew, whereas the TI chip is limited in both bandwidth and slew. Inevitably, I hear the comment that the slew and bandwith of the OPA2107 or OPA2111 is too limited for audio excellence. I simply do not find that to be true. How much bandwidth and speed do we really need for line level audio? The design market for either chip, AD or TI (neither of which were intended for audio) was very different. Actually, if I wanted to try a retrofit of circuitry with something along the lines of an AD8066, I'd experiment with (which I've not done) the AD8032 (still smt only), which appears to be a similar, but more tractable implementation.

On a more practical note, Cordless blew through his original offer of 10 3sies in less than 12 hours, and has re-listed another 10 sets, selling for 1/3rd the price from traditional channels. As I composed this post, 2 of those vanished.  I don't think he has an endless supply. And someone, who shall remain nameless, has posted a link to his adverts on another board....

[Double Ugly]

As of 5 minutes ago, there were 4 sets left.  Now there are only three. 

Let me know if you require anything else of me, cordless.

[Occam]

Ther are 2 opamp sets left listed on Ebay
http://cgi.ebay.com/Opamp-Set-for-Zhaolu-DAC-upgrade-OPA2107-OPA2111_W0QQitemZ250014488249QQihZ015QQcategoryZ36332QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
plus I believe Cordless has 2 additional that you could purchase if you contact him.
A set of 5 opa2107 just went for $33 on an ebay auction, so it makes cordless's 3/$9.50 quite a baragain (you can make the mixture of 2107/2111 any breakdown you want)
Just buy the sukkas, if you intend to upgrade your Zhaolu or need some excellent upgrade opamps. Cordless ships internationally, for $2 more than US postage.

[mgalusha]

I bought a set of them as well. Not that I have a Zhaolu but who knows, I might one of these days. Gotta sell my house first and free up some cash.

[NagysAudio]

I have experimented building solid state amplifiers capable of up to 3MGz frequency bandwidth. The very best results sonically and measurement wise are achieved by using jfets in the input stage. More importantly millitary metal can style dual jfets. They have a very high input impedence which is desirable and can lead to very high slew rate circuits. The following stages would comprise of high frequency bipolar video transistors.

The higher the slew rate, the wider the frequancy band of the amplifier. The wider the frequancy band, the lower will be phase and time errors.

What differentiates live music from recorded is the transient response, the closer phase error and time error is to zero, the closer the sound will be to a live performance.

[Occam]

Schematics (or the specific part # if monolythic ic) to aid the discussion, please.

Thanks in advance,
Paul

[aurelius]

Hi Paul,

Just thought I'd say thanks and give you an update.  Finally gotten round to dropping in 11 OPA2107s per side into my Orion ASPs (thanks Tommie!).

You're dead right... with the specs that these things have across the audio band, we shouldn't be able to hear any difference, but the 2107 is subjectively quite a different beast to the 2134.

First off, the percieved background noise level fell away to almost nothing after dropping in the 2107s.  The noise specs are almost identical for the two devices, so there shouldn't be this level (or any level) of difference, but to my ears, it is quite marked.  The lower noise level gives an impression of improved dynamics.  Especially percussive sounds seem to sound closer to the real instrument (not just drums, but the percussive edge of guitar notes, etc).

Having said that, I also a percieve a slight brightning of the upper mid-range and a flattening of the mid-bass.  Mr Linkwitz has said, sometimes upgrades don't sound better, they just sound different.  I am still reserving judjement on this aspect... 

Across the board, I am enjoying the difference especially the dynamics, the enhanced sense of spaceousness and position... might need to look as PS bypassing next (at the moment each chip is bypassed with 100nF greencap only).

Thanks for the tip.

Mark

[Occam]

Mark(us),

Thanks for the update. You're right, the specs between the 2134 and 2107 shouldn't account for the perceived subjective differences. Even voltage noise and looking at the PSRR graphs shouldn't produce that oft commented on lower noise. Nor do I think the consistently mentioned spectrum differences would track the ruler flat audio band production. Sadly, the terms we use to describe subjective performance are also the same terms we can objectively measure, and they don't jibe.

If you're decoupling the PSs at the chip's ps pins, you might look at the ubiquitous EPCOS/Siemens stacked Mylar film caps, not only to ground by across + and -. These B325xxx caps have very low inductance for their size. RS, Farnell, Maplin Aus carries them in the nekk'd unencapsulated versions and the Stateside suspects all carry the encapsulated versions, all reasonably inexpensive. If you've got the coordination for smt and the wallet, there are stacked film PPS smt caps available.

Dan Banquer is of the opinion that once you've gotten the PS right, the differences between opamps diminishes. Given that I've heard the dreaded JRC5534/2 s, with Herculean PSs, sounding quite exceptional, he might be right......

Regards,
Paul

[Occam]

Well, after Legarem posted on the Paradisea thread about the LM4562 (thanks Lagarem!) -
http://www.audiocircle.com/index.php?topic=33534.msg299780#msg299780
and finished my power amp rebuild, I actually decided to try these National duals.
So I replaced the OPA2111 dual opamp that was in the center, output position for a LM4562NA (8 pin dip) in my Zhaolu 2.0 dac. Shazzam! Not as warm sounding as the OPA2107, but very resolving and excellent extention in the bass as treble. Its working marvelously with the remaining 2 OPA2107s feeding it from the dac.
What I really like about this chip (like the OPA2107) is its sensible bandwidth and slew rate, 55mHz and 20v/uS, which means its unlikely to misbehave when swapped for another dual opamp. (not that most circuits don't benefit from improved ps decoupling, but its nice when you can just swap without jumping through hoops)

So now I've got 2 'standardized' dual general purpose opamps in my arsenal. For a slightly warm (tubey) sound I've the BB OPA2107 and for absolute neutrality (whatever that means) with resolution and extension (the 2107 isn't far behind) I've the National LM4562. This isn't to say that they're aren't other chips better for specific applications, but I will say that I can't see a reason for using/leaving any OPA2604/2132/2134, other than a small cost saving.

[JoshK]

I have experimented building solid state amplifiers capable of up to 3MGz frequency bandwidth. The very best results sonically and measurement wise are achieved by using jfets in the input stage. More importantly millitary metal can style dual jfets. They have a very high input impedence which is desirable and can lead to very high slew rate circuits. The following stages would comprise of high frequency bipolar video transistors.

The higher the slew rate, the wider the frequancy band of the amplifier. The wider the frequancy band, the lower will be phase and time errors.

I am inclined to believe that is probably true.

What differentiates live music from recorded is the transient response, the closer phase error and time error is to zero, the closer the sound will be to a live performance.

This I am not.  There is a hell of a lot more than that to differentiate live from recorded. 

[JoshK]

Well, after Legarem posted on the Paradisea thread about the LM4562 (thanks Lagarem!) -
http://www.audiocircle.com/index.php?topic=33534.msg299780#msg299780
and finished my power amp rebuild, I actually decided to try these National duals.
So I replaced the OPA2111 dual opamp that was in the center, output position for a LM4562NA (8 pin dip) in my Zhaolu 2.0 dac. Shazzam! Not as warm sounding as the OPA2107, but very resolving and excellent extention in the bass as treble. Its working marvelously with the remaining 2 OPA2107s feeding it from the dac.
What I really like about this chip (like the OPA2107) is its sensible bandwidth and slew rate, 55mHz and 20v/uS, which means its unlikely to misbehave when swapped for another dual opamp. (not that most circuits don't benefit from improved ps decoupling, but its nice when you can just swap without jumping through hoops)

So now I've got 2 'standardized' dual general purpose opamps in my arsenal. For a slightly warm (tubey) sound I've the BB OPA2107 and for absolute neutrality (whatever that means) with resolution and extension (the 2107 isn't far behind) I've the National LM4562. This isn't to say that they're aren't other chips better for specific applications, but I will say that I can't see a reason for using/leaving any OPA2604/2132/2134, other than a small cost saving.

I've got some LM4562's that I was planning to roll into my UcD's.  Gordy has some too and he is more likely to complete it first.     But these are SMD, SOIC iirc.

[Occam]

Well iff'n you want soic you need the MA suffix chip rather than the dip versions which have the NA suffix.
Now this comment is purely second hand.... if you're using the CRD (current regulating diode to bias the opamp into class A) on the Hypex board, you want to tie it to the + rail, as opposed the the way it comes configured, tied to the - rail.

[JoshK]

Well iff'n you want soic you need the MA suffix chip rather than the dip versions which have the NA suffix.
Now this comment is purely second hand.... if you're using the CRD (current regulating diode to bias the opamp into class A) on the Hypex board, you want to tie it to the + rail, as opposed the the way it comes configured, tied to the - rail.

I think I read that too.  I ordered a couple from digikey, they were the soic ones.  I also have another pair acquired by other means, I beleive they are SOIC as well.  I don't recall the suffix, but I looked at the type and there were dip ones and soic ones and I went for the soic. 

[Gordy]

Josh, the 4562 samples I have are dip, so you may just beat me to the UcD swap.  I'll be using mine in an EZdac build I a'rekkin.... a most kind L.A. AC'er sent me a .999 version board    I also have the 2107/2111 opamps from the offer that Paul set up so, much to play with!