Simplest discrete op-amps

[tk@halmi]

[quote author="RogerFoote"]I don't get it sometimes... This forum sucks up to people who can't even solder and then chase away veteran audio guys. Does that make sense?[/quote]

My apologies. I did not mean to hurt John's feelings, but his post was rather ignorant in my opinion. I don't think we are chasing anybody, but his post basically stated that one should not bother and use off the shelf components. This entire forum must be pointless using the same argument.

I agree that simple LTPs make for poor inputs unless you make them complementary reducing distortion and helping slewrate symmetry.

 

[NewYorkDave]

Cost/benefit is at the heart of all engineering. Of course if the benefit desired is expanded knowledge or pride of design I can understand the mental math.

Agreed. I'm a broadcast engineer by trade, and oftentimes it's more expedient or cost-effective for me to replace a piece of equipment rather than repair it. But I was also a bench tech for years, so I could repair it if I had to... Being familiar with the inner workings helps one to make more effective "worth doing or not worth doing" decisions.

Good post, by the way.

Loft...

At one of my old bench tech jobs, I used a neat little signal generator made by Loft. Same company?

FWIW Heathkit went from $100M to zero in the same time frame.

I used to get Heathkit's catalog in the mail; my best buddy and I would thumb through it and drool at the offerings. I regret that I never got my act together to actually buy one of their kits before it was too late. I own a lot of Heathkit gear but I never had the pleasure of putting one together myself--same with Eico, et. al.

I remember the '70s as being a really hot time in the electronic kit market. It's a shame that kids today just don't seem to be interested in building things.

 

[JohnRoberts]

[quote author="tk@halmi"][quote author="RogerFoote"]I don't get it sometimes... This forum sucks up to people who can't even solder and then chase away veteran audio guys. Does that make sense?[/quote]

My apologies. I did not mean to hurt John's feelings, but his post was rather ignorant in my opinion. I don't think we are chasing anybody, but his post basically stated that one should not bother and use off the shelf components. This entire forum must be pointless using the same argument.

I agree that simple LTPs make for poor inputs unless you make them complementary reducing distortion and helping slewrate symmetry.[/quote]

My feelings are not hurt, but I fear you don't get my point. If you want to roll your own so you can say you rolled your own go for it. If you assume you're going to come up with a 10 or 20 transistor opamp that compares to the current technology in integrated circuits you are to put it kindly “optimistic”.

I love discrete circuit design and it has been good to me but I can't ignore practical realities. When I sat down to design my current product I could have done it with discrete analog components but it would be huge, clunky, more expensive, and far less powerful than what I was able to implement using a modest 8 bit micro...

I'm an old analog dog but it was worth teaching myself how to code when I can get a single micro with a half dozen 10b a/d inputs, 2 PWM outputs, flash memory, etc for like $5 in small quantity... and being a micro, I can dramatically change how it acts later with software updates.

Opamp design is of academic interest but for any who truly want to learn I suggest studying the massive body of work that has gone before. I have run into the mindset in consumer hifi where they tend to ignore the common knowledge of decades ago expecting to stumble upon some near magical breakthrough. If you want to advance the art, stand on the shoulders of those who have gone before. If you just want to craft something yourself intentionally keeping yourself ignorant of prior art may lead to reinventing the wheel, or worse yet making square wheels that don’t roll very well.

I’m sorry if my practical observations are dissonant with a DIY fior the sake of DIY vibe. I don’t intend to subvert other peoples fun.

JR

 

[JohnRoberts]

[quote author="NewYorkDave"]

At one of my old bench tech jobs, I used a neat little signal generator made by Loft. Same company?

[/quote]

Yes, I designed the Loftech TS-1.. back in the early '80s.

Yet another mostly analog design that if I were to revisit today I'd do it with a micro.

JR

 

[tk@halmi]

I do not ignore prior work. I am very thankful for Self, Slone, Pass and many other publications on my bookshelf for making good information available. To me, the forum is about DIY and fun though.
It is most deliberating to shed most concerns of commercial design and manufacturing requirements. I mean we have people winding their own transformers, etching PCBs and engraving fromt panels. Why not make your own opamp? Dave should design his own opamp and have more fun than using little chips.

 

[Guest]

[quote author="JohnRoberts"]I have run into the mindset in consumer hifi where they tend to ignore the common knowledge of decades ago expecting to stumble upon some near magical breakthrough. If you want to advance the art, stand on the shoulders of those who have gone before. If you just want to craft something yourself intentionally keeping yourself ignorant of prior art may lead to reinventing the wheel, or worse yet making square wheels that don’t roll very well.
[/quote]

Your words sound very pessimistic, John!
To stay on shoulders of others not necessary means to repeat their mistakes, right? And one of such mistakes is to repeat the same topology called "Operational Amplifier" that usually means a differential pair on input and complementary emitter follower on output...

 

[JohnRoberts]

[quote author="JohnRoberts"]

Your words sound very pessimistic, John!
To stay on shoulders of others not necessary means to repeat their mistakes, right? And one of such mistakes is to repeat the same topology called "Operational Amplifier" that usually means a differential pair on input and complementary emitter follower on output...

Pessimism or reality may be a matter of POV.

I don't recall signing on as apologist for conventional design topology but for the record I don't consider operational amplifiers or existing prior art as flawed. That said there are numerous examples of poor execution. I have routinely used opamps as useful building blocks in combination with discrete parts either in front of to optimize some input or behind for some desirable output characteristic.

The result is what matters and if you are pleased and get good results with unconventional topologies enjoy.

JR

 

[NewYorkDave]

So, Wavebourn, do you have anything to contribute to the discussion beside your hatred of op-amps?

Anyway, back on topic... I've been a satisfied user of IC op-amps for a fairly long time, and I've done loads of stuff with vacuum tubes over the years. But I never got very deep into design with discrete semiconductors. Lately, I've been developing an interest in back-filling that gap in my experience.

Excellent op-amp ICs for audio applications do indeed exist--and new ones still come out every now and then; National released one recently that looks promising. But if you want high supply voltage capability or the ability to deliver significant power into low-impedance loads, discrete still seems the way to go in many instances. (And the chip makers are going after the mobile devices market, which means operating voltages will continue to drop and through-hole parts will gradually be phased out...).

 

[Martin B. Kantola]

Here's a couple of questions I've been wondering about:

- whenever an OP-amp is used as a buffer, wouldn't it be better to use a device that's specifically designed for a gain=1 and even lacks an inverting input? Using massive amounts of feedback (>100dB...) doesn't make any sense to me. OP-amp type I.

- in another typical application we might need some gain, but not necessarily more than say 20dB, and in this case a higher output impedance might be perfectly acceptable. OP-amp type II.

- in a third application we might instead need a generous amount of gain, 60 dB or so. Low noise is important, since compared to the two examples above we're probably dealing with low level signals. Performance could be optimized for these conditions. OP-amp type III.

These three types could be designed in very different ways, each one optimized for its task. They might even run on very different supply voltages.

Martin

 

[Guest]

[quote author="NewYorkDave"]So, Wavebourn, do you have anything to contribute to the discussion beside your hatred of op-amps?
[/quote]

Hatred? :shock:

I love them! But when properly used in right places.

Can you specify your requirements to the opamp that differ from what you can buy in cheap chip, except "discrete", so we may contribute something real according to your requirements?

[quote author="Martin B. Kantola"]

These three types could be designed in very different ways, each one optimized for its task. [/quote]


Exactly.

 

[burdij]

I think we need to divide up the waterfront here and I agree with the categories presented. There are reasons why different designs for op amps are used and some sub-categories of those uses. To summarize:

I think the biggest reason why there was a divergence between the use of integrated opamps and discrete opamps in audio back in the 70's was the problem of driving a line at +0db. The then available integrated amps (lm741, 4558, 5534, etc.) just couldn't do it so the big console guys had to build with discretes. They all have the basic characteristics of the negative feedback amplifier, an input section with at least a negative input, a gain section, and a driver section. Some are simple, like Neve, some are fully balanced differential amps like the API with a 2520 (API sold the opamps competing with Philbrick-Nexus, Burr-Brown and others for non-audio use, too).

The ability to drive a line also contributes to the selection of power supply voltages. For various atomic-semiconductor reasons its difficult to make an integrated amp that can run on voltages higher than +- 15V which is not quite good enough to provide the +20db headroom that is expected in pro-audio. You can even see the discrete opamp manufactures grappling with that problem. The API preamp (and line driver) both get free voltage gain out of the output transformer.

What is a little hard to explain is why the integrated opamps don't seem to perform "sonically" as well as the discrete opamps in certain applications. I don't think that anyone here will argue with this difference. And that is, in part, why we go back and examine (and recreate) those discrete devices to try and understand the reasons they sound the way they do.

I have noticed that a lot of the "sound" of these devices is greatly affected by the magnetics involved as opposed to the electronics and this is where the integrated opamp is going to have some difficulty because of their lack of ability to generate raw power. So we create the sub-categories or "hybrids" to allow the integrated amps to perform well with transformers and high capacitance loads like audio cable.

 

[Ian MacGregor]

[quote author="NewYorkDave"]
By the way, Brad, that first QE op-amp you posted is almost exactly the topology I was describing earlier (but with collector bootstrapping added). Thanks for the link.[/quote]

Sorry to be off-topic (or on topic??), but I am curious about the boostrapping mentioned above. I guess I am just curious how it works and what it accomplishes?

I sorta have a guess, but I'm note sure, so here goes: C8 is a near short at audio, so the ac signal on each side of R12 is about the same, making R12 appear much larger than it is. Does this greatly increase the voltage gain contributed by Q3 (since the gain is approx -gm*R12)??

Ian

schematic in question here: http://www.avensonaudio.com/tech/Quadeight/mm100schematicsm1.jpg

 

[NewYorkDave]

Yes, that's right. Well done.

 

[Martin B. Kantola]

It was already mentioned, but I think that designing an OP-amp with very few parts that takes care of DC-offset could be a challenge.

Looking at the big picture, there's a lot of good sounding equipment that uses capacitors so maybe it's not even necessary to stay at exactly 0 volts DC at all times.

Still, for the Type I DOA I suggested it would be an issue, because we'd be working with low impedances and many times interfacing to other equipment. In this case no DC offset can be tolerated, and we'd need a clever servo circuit.

Any suggestions?

Of course, it could also be a dedicated, fourth type of building block. Sonically transparent naturally...

Martin

 

[Martin B. Kantola]

So, here's a simple buffer idea to start with, is this the kind of stuff you had in mind with this thread Dave?

It has current generators drawn, but not necessarily needed, and I would love to do this SE only...

Martin

 

[Guest]

[quote author="Martin B. Kantola"]
It has current generators drawn, but not necessarily needed, and I would love to do this SE only...

Martin[/quote]

Hmmm... You wanted the first current generator in the source instead of in the drain, right?

 

[Martin B. Kantola]

Right, thanks, I started with an N-channel FET in that first stage and never thought about reversing it. Also, my knowledge is very limited since I come from audio engineering...

Do you think there's any benefit over a simple resistor anyway?

Martin

 

[Guest]

[quote author="Martin B. Kantola"]

Do you think there's any benefit over a simple resistor anyway?
[/quote]

Yes. More voltage swing on given load with less distortions consuming the same power.

 

[JohnRoberts]

[quote author="Martin B. Kantola"]Right, thanks, I started with an N-channel FET in that first stage and never thought about reversing it. Also, my knowledge is very limited since I come from audio engineering...

Do you think there's any benefit over a simple resistor anyway?

Martin[/quote]

In a basic fet follower using a resistor from source to ground to set the device operating current the Vg-s will change with signal swing causing distortion as this change in Vg-s subtracts from the signal at the source follower output. If you instead load the source with a current source the Vg-s is more constant so you eliminate that particular distortion source.

=====
I did a phono preamp design years ago where I used a low noise Nch J-Fet as the input gain stage. I used something like a 1k ohm resistor from the source to ground to set first stage gain, then hung an additional current source (3-5 mA) at that point to increase the current density for lower noise and improved linearity as the delta current due to signal was small relative to the constant current source.

I fed the drain output to a capacitor to ground and a resistor into the inverting input of a (heaven forbid) opamp. That RC established the 75 uSec pole for the RIAA playback eq. One nice benefit of this topology is that there was no negative feedback involved in this first gain stage so you could literally hit the input with thousands of volts/uSec and never slew rate limit. The output would just harmlessly be rolled off by the passive pole (approx 2 kHz). The rest of the RIAA eq was in the opamp gain stage and a DC servo fed back to a second current source attached at the drain of the input stage to balance out the lower current source and provide a dc coupled final output that was within mVs of 0VDC.

I couldn't measure distortion with my bench which while crude by today's standards could resolve components down 100 db or so. Despite the fact I couldn't measure any distortion I had a next generation design on paper that linearized that first Fet stage further, but around that time I lost interest in the tweaky audiophile market. It was too non-linear for me.

JR

 

[CJ]

One good thing about discrete is that you can screen parts, pick vendors, where as, with a fab chip, your limited to whats on the wafer.

Maybe price had something to do with it, opamps were a lot more expensive in the early seventies, compared to pennies on the dollar for a 2N product.
Space would be the tradeoff.