Forced Class A opamp

[bcarso]

I think maybe we are talking about two diferent animals---I was initially responding to the question of the drawbacks of loading as applied to an existing op amp. It sounds like you are talking about the situation adding an external EF.

 

[Svart]

is it just me or is the person in the article linked above making a very broad and almost too general statement insuing that opamps(all?) have BOTH NPN and PNP BJTs on the outputs, not only that but opamps(all?) have NPN highside and PNP lowside? I'm not much for one for BJT design but wouldn't it be more benficial, efficiency wise, to reverse that and have the NPN sink and the PNP source and just use an inverting stage driving them?

just wondering.. :guinness:

 

[Svart]

that's what i mean Butta, the article suggested that opamps(in general? do they mean all of them?) have BJT NPN and PNP output structures. of course we know that not all opamps are the same and a general statement such as i read could confuse and ultimately cause the subject to be dropped instead of investigated more.

the other problem is that a lot of opamps have designs that are not published and making a general statement like that simply boils down to guessing and we need to keep that from happening.

 

[bcarso]

Traditionally, lateral PNPs were all that you could make alongside NPNs on a substrate, and they tended to be very slow and deficient in other ways as well.

But there are lots of new processes that give good PNPs now, so the blanket rule is less valid.

And there are all kinds of output stages, including the quasi-complementary where a little PNP is paired with a common-emitter NPN for the lower half. In this sense the op amps have retraced the development of sand state power amps, where the first decent silicon power transistors were NPN.

BTW, I sneaked a little work in on the question of active drive of the current source to achieve constant dissipation in the on-chip output device. It's hairy.... one approach requires both a squaring operation and a division operation, and the parameters must be set up for the given load impedance. A pretty good approximation for relatively small signal swings and a given output Z can be done just by tweaking the (fixed) current sink.

 

[bushwick]

We are rechipping my board and are forcing the chips into class A. We have opted to go for tying power sup to grn via resistors. At this moment I am eating lunch and thought it a good time to drop a line. For a OPA2604 all parties here like the listening through the NPN more. We are now experimenting with the value of the resistor. 6.91? sounds better than 15? we are now about to check out 3?. Much lower than that, there are thermal issues and perhaps some headroom issues. I'll let you know what we find.

We also are using OPA627 and OPA604 together to replace some 5532's in certain spots of the console.

Tying one leg to ground on the IC made a huge difference. Huge. Bye bye nasty edge. Also, the depth of the mix was vastly improved.

Best,
joshua[/list]




Edit: I forgot to add k to the ? values.... very important. Anyway we have concluded the testing and far and away, 3k? is our favorite. Like I mentioned earlier, my tech (the brains behind all of this) thinks that going much below that is going to bring "issues" about.

 

[chrissugar]

Thank you Joshua.
That is very good real world info.

chrissugar

 

[bushwick]

I think I have heard him mention that the circuit surrounding it plays a important part in whether the NPN or PNP sounds better, and the resistor value the the IC's power supply rail has much to do with what load the IC normally sees and what the thermal issues are going to be...

I'll update as we dig further.

 

[bushwick]

My console has gone through three labor intensive modifications this being tied for the most man hours (approx). Each time I have hear the board sound better and better. This time all the chips that came in the board were removed, sockets and high end IC's were put that we had demod out. The IC chips are all running class A now and I'll I can say is

HOLY FUCKING SHIT. Why people would EVER design an IC circuit not running in class A blows my mind. And why anyone would ever use a TLO72 ever again is astounding.

I am very baffled at how great my control room sounds....fucking unbelievable. Worth ever hour we spent soldering...

Best,
j

 

[chrissugar]

This is good news.
What was the final value for the resistor? Any variation depending on the type of opamp?

chrissugar

 

[gyraf]

And don't underestimate the placebo effect of hard work / money spent..!

 

[Flatpicker]

[quote author="bushwick"]...Tying one leg to ground on the IC made a huge difference...[/quote] :? Which leg?

 

[TomWaterman]

Cool thread.

I'm still yet to try this out after Brad's advice in the other thread.
For those who want to use a constant current sink or source check out horowitz and hill. There's a couple in there that may be helpful. p72 and p125 onwards.

Hey Joshua, what is the consle you are upgrading there - sounds like your really digging the transformation?

-Tom

 

[Svart]

I'm also interested in how Bushwick tied "the leg" of the IC to ground.. which leg through 3k?

 

[Samuel Groner]

And don't underestimate the placebo effect of hard work / money spent..!

I wonder why I never heard a listening comparison of console mods. It would be very easy: you record a signal trough a moded and an unmoded channel, make a mp3 and upload it somewhere.

Samuel

 

[Svart]

i would prefer a shorter clip in WAV format myself. MP3 just kills too much detail in these critical listening tests.

For the forced class A, would it be more beneficial in a summing area, or is it more important in a preamp area? I am sure most would say both, but with the boards that *need* this type of mod, there is usually a fixed amount of current available and concessions must be made..

 

[bushwick]

Hi all-

First of all, I pecked that last emil out to quickly and didn't realize my gaff. You tie the audio to the power suppl not the gaff as some one noted. The the resistor value was about 3k?. Doing this largely depends on whta load the IC sees after it but yes, tie the audio output to one of the power supply rails depending on which transistor(s) you want to hear, NPN or PNP. We like the sound of NPN better for OPA 2604, 604, AD797, OPA 627. The 627 was a toss up really, we ab'd everything we did and with the NPN/PNP issue for the 627 it was tough to call.

Hi Gyraf.... yeah for all of these test that he have done, we have ab'd stereo pairs of modules in the board against the modded version so I have been fortunate enough to hear the change in real time. All of these tests were done with an ABX box in case some of the changes were subtle enough that we wanted to pick blindly. Almost none of them were that subtle.... like being able to pick after one flip from unmodded to modded or vice versa.

Testing was made much easier with John Chester's Audio Precision device. John is the formidable brains behind all of the mods we have done. He is very, very sharp, an intellect you would all be glad to know.... cool dude too.

So no Gyraf, there is absolutely no placebo effect going on here. I have heard the changes, some pretty profound all along the way and while it often was obvious, we went about choosing things blindly. This has been an incredibly invaluable experience, a very rare opportunity to hear first hand many of the things people talk about practiced in front of me to experience first hand.

The console started out as an Allen & Heath Saber. Soul II Sould mixed a muli-platinum album on one, but I can tell you, this thing is an order of magnitude better than when I got it. We have performed three modification sets to it and each time is was, "Wow!".

This might have been the most profound of them all though.

Oliver Archut is making some transformers for the inputs of the console right now that we liked and those should go in in August. We are also going to make some various discrete amplifiers to A/B to see what we like the most and possibly replace the input amplifiers with one of Johns designs.

Gotta run....
j

 

[Svart]

so you pretty much just tied the outputs to - via 3k?

 

[gyraf]

Cheers, J, :thumb:

That seems like the right way to evaluate things!

If only more people made this effort of actually listening to "upgrades" - we'd have much less bs to fight regarding what opamp is the best drop-in replacement for what and so on..

Good work! And thanks for sharing..!

Jakob E.

 

[Svart]

I have primarily opa2604 opa404 and OPA627s in my console now so I am very interested in trying this! I plan on doing this to a stereo summing channel and master channels.. however I forsee using some kind of heatsinks on them as some of the OPA627s in the high gain areas get a little warm to the touch..

what was your estimated current draw increased by and how many opamps did you try this to?

 

[jdiamantis]

Back in the early '90s (geez alert!), I did the "class A mod" to the audio consoles in the radio station I was engineering. (Pacific Recorders BMX II's, and III's)

We used Analog Systems chips (no longer in business, but incredible sounding devices) for the audio path, and increased the bias on the output driver transistors as well (had to add heatsinks there!)

There was a marked difference in in clarity between modded and un-modded modules. The output driver, which was Jensen Transformer coupled to the outisde world, really made a big difference.

While it may be difficult to hear a single opamp changed to class A output, when there are several in line, it becomes easily discernable to even the untrained ear.