"Boutique" opamps from chinese vendors. Anyone tested?

[Newmarket]

Come on Dual. He has made the obvious case that ICs "sound bad" because they incorporate a Silicon Layer. Whereas proper discreet transistors function by means of ....ooh Hang On...I think I see the flaw in the argument

 

[JohnRoberts]

Be nice everybody... It is difficult to prove what people say they hear in the internets. Let them enjoy...

JR

 

[pucho812]

. The only two significant names are, IIRC, API and Quad-Eight. Most others used IC's. Actually API and QE used DOA's sparsely, only in stages where they had a proper justification. IIRC API had only two DOA's per channel, QE I don't remember but their parametric EQ couldn't have been DOA-based.
.

of the QE eq schematics I have seen they have a single DOA. For example the 444 quad eight eq is 5532’s everywhere and a single AM10(QE DOA) on the output.

 

[pucho812]

As for chip vs discrete off hand I can think of a couple of reasons to do discrete.
1. Voltage
2.current
In a discrete design you can make it so that it powers and runs off higher than the typical opamp voltages of +/-18vdc. This gives for more headroom. Ye Avalon designs runs off +/-32vdc on the audio path circuits. That stuff has headroom for days.

2. More current can be had vs a typical ic chip. This makes it nice for driving transformers. It’s often why you see a hybrid of a ic chip opamp with discrete follower giving you more current to drive the iron.

 

[user 37518]

Come on Dual. He has made the obvious case that ICs "sound bad" because they incorporate a Silicon Layer. Whereas proper discreet transistors function by means of ....ooh Hang On...I think I see the flaw in the argument

I can't remember who said it, but it goes something like this: "The highest form of ignorance is when you reject something of which you know nothing about."

 

[user 37518]

In a discrete you can make it so that it powers and runs off higher than the typical opamp of +/-18vdc. This gives for more headroom. Ye Avalon designs runs off +/-32vdc on the audio path circuits. That stuff has headroom for days.

Agreed, but I can also point out to amps such as those found in the MCI JH-500 which did use two transistors coupled with an NE5534 to run on +/- 32 V (originally it was +/- 36V but they eventually lowered it a bit), which, in terms of headroom, makes a DOA like an API to go hide in shame. You can have the best of both worlds without bashing on ICs without any evidence.

 

[pucho812]

You can have the best of both worlds without bashing on ICs without any evidence.

I concur

 

[swpaskett]

Let’s make it real. WhenI was 18 I laughed at my mom whe she said the clock radio I gave her sounded pretty good. Fifty years later my hearing is 40dB down at 4kHz. Everything sounds the same now, even that old clock radio sounds pretty good. Some people just like old tech. I wish I could find one of them to take four Altec boat anchors off my hands for $50k or so. Any takers?

 

[cbaldoni]

I’d like to share my experience with my own DOA’s…. I’ve been in audio engineering over more than 4 decades, and once I decided to build my own DOA to prove myself where is the true (technically and subjectively speaking). I just didn’t invent the wheel, I took the known 990c design, simulated it in spice with components of my choice and played a lot before building a physical device. I did a careful selection of parts, specially at the differential input (like resistors match, thermally stabilized, within less than 0.01%), kept values low to keep noise low as well, large array of matched transistors and all sort of crazyness to get the “perfect” parts. Probably I bought 20 times the amount of components needed just to make the selection. Important to say: I decided to have headroom with +/- 36v voltage applied.

What about the measured performance - yes, in a bench with equipments : 1) My DOA was able to provide far wider frequency response (GBW) and less phase shift than the market standard 5532 (just for comparison) allowing it to be used in many other higher frequency needs other than just pure analog audio; 2) I’ve got a lower noise floor than the also known THAT15xx; 3) The DOA was able to provide lower THD measurements, particularly driving transformers as a load; 4) The huge headroom swinging +/-36v simplifies the amplification stages after it, thus also contributing with lower parts count and lower THD of the amplifier as a whole; 5) Better performance measures has its price though, power consumption is higher and package is larger than a DIP8…..

What about the listening experience with DOA in different places ? The really noticeable improvement (to my 62 years old ears) was in the front end pre-amplifier of my turntable, and in digitally recorded guitar play using it as a microphone preamplifier. Places like line level input preamp or EQ circuit just didn’t make any difference to me. My son helped me to run it playing guitar and not telling me what was installed in the amps tested during the blind test.

I hope everyone can keep enjoying this hobby with your own truth….

 

[Ricardus]

What about the measured performance - yes, in a bench with equipments : 1) My DOA was able to provide far wider frequency response (GBW) and less phase shift than the market standard 5532 (just for comparison) allowing it to be used in many other higher frequency needs other than just pure analog audio; 2) I’ve got a lower noise floor than the also known THAT15xx; 3) The DOA was able to provide lower THD measurements, particularly driving transformers as a load; 4) The huge headroom swinging +/-36v simplifies the amplification stages after it, thus also contributing with lower parts count and lower THD of the amplifier as a whole; 5) Better performance measures has its price though, power consumption is higher and package is larger than a DIP8…..

We'd all love to see that schematic and that test data.

 

[JohnRoberts]

I designed exactly one DOA for use as a summing amp inside a small console back in the early 80s. It was based on the LM394 (not exactly a discrete device) for the input LTP. Since the op amp was designed to only be used inverting and at modest high noise gain, the design was simpler.

Since then I have been able to buy off the shelf ICs to cover most needs, with superior performance. The 5532 is not exactly the industry standard for other than "good for the money" op amps, while it has been widely used by many (including me) and when applied properly it doesn't suck.

JR

 

[Gold]

That may be your perception, in regard of a business that has been under many changes in the last 20 years (DAW, plug-ins,, almost infinite track count...) but when big analog behemoths mixers were ruling the studio world, not many were equipped with DOA's. The only two significant names are, IIRC, API and Quad-Eight. Most others used IC's. Actually API and QE used DOA's sparsely, only in stages where they had a proper justification. IIRC API had only two DOA's per channel, QE I don't remember but their parametric EQ couldn't have been DOA-based.

Neumann used the OA10 in a lot of products including mixers. Personally I’m not a fan. I’d take a 5534 over it every time. Same with a Melcor 1731. Not a fan.

 

[inkster]

I have tried some and they were not anything special and one of them died after a couple of years...the sparkos lab stuff is nice but I would only use those in extremely high quality circuits with a minimum of components in the signal path.
In recent years I have become a fan of well designed discrete smd circuits, there's a clarity I can only attribute to reduced stray inductance and capacitance as well as tight tolerances associated with smd components.
Emphasis on "well designed"!

 

[Newmarket]

Wrt reduced stray reactive impedances - do you mean in comparison to thru hole discrete or to ICs ?

 

[user 37518]

I designed exactly one DOA for use as a summing amp inside a small console back in the early 80s. It was based on the LM394 (not exactly a discrete device) for the input LTP. Since the op amp was designed to only be used inverting and at modest high noise gain, the design was simpler.

Since then I have been able to buy off the shelf ICs to cover most needs, with superior performance. The 5532 is not exactly the industry standard for other than "good for the money" op amps, while it has been widely used by many (including me) and when applied properly it doesn't suck.

JR

The 5532 was, for several decades, the unquestioned audio champion when it comes to an opamp having all good attributes in terms of low cost, low noise, large SR and GBP, driving capabilities, low distortion, and not being a nightmare to stabilize. In my opinion, it wasn't till the LME49720 or the LM4562 (which seem to be the same part under a different part number, the former now defunct) that the 5532 got some competition (and lost), the price tag being the truly defining factor to choose a 5532 over a 4562. However, the price of the LM4562 has reduced dramatically over the years, by now, the LM4562 is around 3x to 4x more expensive than a 5532, somewhere in between $1.5-$2 USD.

IIRC the LME49720 had some issues, some users reported and measured some weird artifacts, but I believe they were a manufacturing process dependent kind of thing that was corrected ever since, and the LME49720 is now retired, but the LM4562 remains and it seems to be the same thing as the 49720, although the 49720 was more expensive than the 4562. AFAIK the LM4562 never reported such issues.

There are now some opamps that have even better specs, but are usually considerably more expensive than the LM4562. In any case, practically none of them are as expensive as a 990 or a 2520.

 

[JohnRoberts]

The 5532 was, for several decades, the unquestioned audio champion when it comes to an opamp having all good attributes in terms of low cost, low noise, large SR and GBP, driving capabilities, low distortion, and not being a nightmare to stabilize. In my opinion, it wasn't till the LME49720 or the LM4562 (which seem to be the same part under a different part number, the former now defunct) that the 5532 got some competition (and lost), the price tag being the truly defining factor to choose a 5532 over a 4562. However, the price of the LM4562 has reduced dramatically over the years, by now, the LM4562 is around 3x to 4x more expensive than a 5532, somewhere in between $1.5-$2 USD.

IIRC the LME49720 had some issues, some users reported and measured some weird artifacts, but I believe they were a manufacturing process dependent kind of thing that was corrected ever since, and the LME49720 is now retired, but the LM4562 remains and it seems to be the same thing as the 49720, although the 49720 was more expensive than the 4562. AFAIK the LM4562 never reported such issues.

There are now some opamps that have even better specs, but are usually considerably more expensive than the LM4562. In any case, practically none of them are as expensive as a 990 or a 2520.

I used truckloads of 5532 and TL07x often inside the same SKUs. The 5532s were better when heavy lifting was involved because of their stronger drive capability. The TL07x were fine for line level stages. Inside Peavey we had a specified 5532 with its own house number (3678?) that was graded for modest 1/f*** noise. When used as a summing amp with high noise gain, the graded parts had a better sounding noise floor when the customers did their WFO listening tests. Yes, customers do that, and infer sound quality from that noise floor. High 1/f noise just sounds nasty.

JR

*** The noise grading was not for an absolute low level of LF noise, but IIRC it specified a noise level ratio between 10Hz and 1 kHz. This pretty effectively culled out parts with process issues. Today chip processes are much cleaner than they were several decades ago so I don't know if these part are still in use.

 

[user 37518]

I used truckloads of 5532 and TL07x often inside the same SKUs. The 5532s were better when heavy lifting was involved because of their stronger drive capability. The TL07x were fine for line level stages. Inside Peavey we had a specified 5532 with its own house number (3678?) that was graded for modest 1/f*** noise. When used as a summing amp with high noise gain, the graded parts had a better sounding noise floor when the customers did their WFO listening tests. Yes, customers do that, and infer sound quality from that noise floor. High 1/f noise just sounds nasty.

JR

*** The noise grading was not for an absolute low level of LF noise, but IIRC it specified a noise level ratio between 10Hz and 1 kHz. This pretty effectively culled out parts with process issues. Today chip processes are much cleaner than they were several decades ago so I don't know if these part are still in use.

The TL07x is pretty much gone for pro audio, his cousin the TL06x is an exception since it consumes very little current, so it is still one of the favorites amongst stomp box pedal builders. The NE5532 is still very popular, in some applications it is one of the best options. I remember reading the books by Douglas Self on phono amplifiers, and apparently the specific values of the voltage and current noise specs of the NE5532 make it an almost unmatched part in terms of noise when it comes to IC op-amps for that specific application (phono amps). For other applications I still think the 5532 is great, IIRC old SSLs are full of them.

I am sure I could go to any pro studio with outboard gear and with a 95% certainty find at least one NE5532 somewhere in the signal chain... The NE5532 is like the Teflon of audio, almost all music has been touched by one at some point of the process.

 

[thor.zmt]

I was looking for some smd components on Aliexpress and came across some "weird" opamps. Have you guys tried some?

What you guys think? some of them are really expensive (U$15 per unit)

As with all things Cheena - CAVEAT EMPTOR. They are often (I cannot say always because I have not tested everything under the sun, but only everything that got into my hands) bad knockoffs of often uninspiring designs to start with using incorrect/fake parts.

As for "discrete OPA vs Integrated", we get rather excellent and inexpensive +/-18V capable parts from TI these days (OPA1678/OPA1679 and others), which makes discrete OPA's a rather narrow niche.

In the days of TL072 and NE5532 one could easily argue in favour of discrete Op-Amp's or of (for example) extending a NE5534 with external input and output circuits.

Using items like OPA828, OPA1611 (single), OPA2828, OPA1612, OPA1642, OPA1656, OPA1652/78 (dual) plus OPA1644, OPA1654/79 (quad) as replacements gives exceptional performance that few discretes can match and the low distortion is by design for inherent linearity, not massive amounts of NFB and only valid at 1kHz with a fast rise above, but commonly shows very little distortion rise towards 10kHz.

Discrete "Op-Amp type structures" are still valid and I often find them more interesting and less expensive than canned op-amp's while allowing superior performance in certain areas, such as extra low noise or the ability to drive very low loads or to operate on rails higher than +/-18V and the ability to be optimised for GBWP for the actual gain they are employed at (externally compensated and decompensated Op-Amp's seem to have disappeared, sadly).

But at this point we have passed beyond "universally applicable unity gain stable gain stage" and are into specialised designs, which naturally in the specific areas we aim for outperform "canned" Op-Amp's, or at least SHOULD do so. And of course at this point there is absolutely zero requirement to stick with "Operational Amplifier" derived circuits and we can get far more radical.

As "plugin replacements" discrete OP-Amp's are more about getting specific "vintage" sounds or just simply about selling overpriced, objectively underperforming stuff for obscene profits using voodoo - jedi mind tricks clouding the weak minds of those susceptible.

Thor

 

[leadbreath]

For fucks sake, we didnt just enter the ic v discrete debate on gdiy????!!!
I thought we were beyond these childish games....

 

[leadbreath]

Probably why I sadly skip coming yo tjis place no more...

 

[Newmarket]

For fucks sake, we didnt just enter the ic v discrete debate on gdiy????!!!
I thought we were beyond these childish games....

There's no obligation to join in...

 

[Newmarket]

I’d like to share my experience with my own DOA’s…. I’ve been in audio engineering over more than 4 decades, and once I decided to build my own DOA to prove myself where is the true (technically and subjectively speaking). I just didn’t invent the wheel, I took the known 990c design, simulated it in spice with components of my choice and played a lot before building a physical device. I did a careful selection of parts, specially at the differential input (like resistors match, thermally stabilized, within less than 0.01%), kept values low to keep noise low as well, large array of matched transistors and all sort of crazyness to get the “perfect” parts. Probably I bought 20 times the amount of components needed just to make the selection. Important to say: I decided to have headroom with +/- 36v voltage applied.

What about the measured performance - yes, in a bench with equipments : 1) My DOA was able to provide far wider frequency response (GBW) and less phase shift than the market standard 5532 (just for comparison) allowing it to be used in many other higher frequency needs other than just pure analog audio; 2) I’ve got a lower noise floor than the also known THAT15xx; 3) The DOA was able to provide lower THD measurements, particularly driving transformers as a load; 4) The huge headroom swinging +/-36v simplifies the amplification stages after it, thus also contributing with lower parts count and lower THD of the amplifier as a whole; 5) Better performance measures has its price though, power consumption is higher and package is larger than a DIP8…..

What about the listening experience with DOA in different places ? The really noticeable improvement (to my 62 years old ears) was in the front end pre-amplifier of my turntable, and in digitally recorded guitar play using it as a microphone preamplifier. Places like line level input preamp or EQ circuit just didn’t make any difference to me. My son helped me to run it playing guitar and not telling me what was installed in the amps tested during the blind test.

I hope everyone can keep enjoying this hobby with your own truth….

Thanks for that post. Would be great to see the schematic(s) of how you implemented the DOA in the preamp etc. and interfacing to the recorded guitar.
Observations / notes (no criticism): Understand the use of 5532 as comparison - against a 5534 and LM4562 woud be interesting. GBW - well tbh if I were looking at high frequency appplications I'd probably be looking at an opamp more suited than 5532/4.
Cheers.

 

[leadbreath]

There's no obligation to join in...

With dicks like this I rest my case...

 

[abbey road d enfer]

For fucks sake, we didnt just enter the ic v discrete debate on gdiy????!!!

With dicks like this I rest my case...

Do you think that using foul language adds to your credibility?
It makes you just looking like a f...ing d..k.