Discrete opamps and gain...

[Moses]

Ok, so as I've been ferreting about i've been noticing that discrete opamps are a sort of holy grail around here... Who has the best/fewest components, which is fastest/quietest/best sounding etc...

What I'm wondering is, what is the advantage of a DOA over a non-opamp gain stage tailored to the task at hand? For instance, if I built a preamp I could build it around a monolithic and be done with it. Easy, and my local Maplin probably has all the parts. The next stage therein is to swap out the monolithic for a DOA (as per Neve, API, GML...) and get a subjective improvement.... Then I might design my own DOA for whatever reason (real back to basics DIY!!)

But why not start at the top, and design each gain stage for the job at hand? Is this just lazyness on our behalf (an opamp will do many jobs!), or is it that when we start building gain stages we naturally more or less end up with an opamp anyway, so why not just use someone elses?

Sorry for my lack of knowledge/gross over simplification!!

Mo

 

[abbey road d enfer]

What you call lazyness is called cost optimization by others. If you want to build a mixer and sell it at a price that makes it accessible to most of your expected customers, you just can't go the discrete route. Now, if you want to make a mic pre for your own usage or sell it to a clientèle that's ready to shed at least a grand, why not?
DOA's lend themselves to experimentation with VLN input devices, super high output capability, or high voltage rails, but designing a DOA as a straight replacement for a 5534 or a TL0 is not really justifiable.

 

[ppa]

Ok, so as I've been ferreting about i've been noticing that discrete opamps are a sort of holy grail around here... Who has the best/fewest components, which is fastest/quietest/best sounding etc...

What I'm wondering is, what is the advantage of a DOA over a non-opamp gain stage tailored to the task at hand? For instance, if I built a preamp I could build it around a monolithic and be done with it. Easy, and my local Maplin probably has all the parts. The next stage therein is to swap out the monolithic for a DOA (as per Neve, API, GML...) and get a subjective improvement.... Then I might design my own DOA for whatever reason (real back to basics DIY!!)

But why not start at the top, and design each gain stage for the job at hand? Is this just lazyness on our behalf (an opamp will do many jobs!), or is it that when we start building gain stages we naturally more or less end up with an opamp anyway, so why not just use someone elses?

Sorry for my lack of knowledge/gross over simplification!!

Mo

Op amps are circuit that can be used for doing a lot of circuitries, as for example eqs, dac units, RIAA preamps, mixers,.......etc.... if a company wants to reduce prices and times to market of their gears (like did gml, API, Millennia Media,......) can use op amps to reduce its gears' design and costs because parts of them (the op amps inside) are just designed and avaiables at low costs.

DOA and IC op amp are not the same things, because in DOA's design can be used the force points of the discrete amp circuits Vs. the IC amp circuits. The DOA design is different respect the IC op amp design in several aspects.

 

[Moses]

Hi guys,

Thanks for the answers! So essentially, we use opamps because they are extremely flexible basic building blocks, and on a cost level it is far cheaper to use a not-quite-optimised-opamp in a number of applications than a designed-from-scratch-gainstage for each application?

I see where the cost issue comes in, and I am asking purely out of interest, but is there any advantage to be had in designing gainstages on an application by application basis, or is the opamp so close to ideal that it negates the need to do so as far as signal integrity is concerned? I realise cost margins are a concern, but lets imagine we're in Lala land for a moment, and dream that money is no object!!

Thanks for the replies!!

Mo

 

[jdbakker]

I see where the cost issue comes in, and I am asking purely out of interest, but is there any advantage to be had in designing gainstages on an application by application basis, or is the opamp so close to ideal that it negates the need to do so as far as signal integrity is concerned?

Forget about money for a moment, this is DIY. Think about time. While it is true that on a technical level a no-holds-barred custom solution is always at least as good as an off-the-shelf part (with the potential of being better, of course), you'd have to wonder how much time you spend tweaking your gain stage and its layout before it's enough of an improvement to be worth considering. If you're not going for full-custom discrete designs for its own sake (and boy, there are enough folks out there with that mindset), you'll soon find that modern op-amps are almost always 'good enough'.

There are some corner cases where DOAs or full-custom gain stages have their place. Sometimes you really need the best possible noise performance (although IC op-amps are rapidly closing that gap). Sometimes you need more PSRR than you can get out of a device without a ground pin. Sometimes you really need crazy-high rails. And there are times that you want to drive a transformer single-ended with more power than any DIP-8 (or most 2520-DOAs) can squeeze out.

The nice thing about the various DOAs around here is that you can treat them as semi-custom gain blocks. You know the basic design works, and now you can tweak a few parts values to make it work better in your application (with compensation and input stage current being the most common targets). By the way, ppa, that's why I feel it's a pity that your otherwise very nice DOAs are potted and proprietary: I would have loved to get under the hood. Oh well.

Moses: funny you should mention this now, BTW. I've been working on a line receiver design, and just yesterday I was wondering if it was worth the hassle to go for a full custom discrete design or if I should just put in sockets for a couple of HV DOAs. Will let you all know how that turns out.

JDB.

 

[JohnRoberts]

The last time I designed a DOA was in the '70s because I needed an opamp quieter than I could buy off the shelf, and it wasn't heavy lifting using a LM394 super pair (integrated circuit BTW). 

Today I could buy an opamp off the shelf that would kick my old DOA's ass...

JR

PS: I've done plenty of mix and match.. opamps with discrete devices in front for lower noise, or behind for higher drive current.

 

[Moses]

JDBakker and John, that was more the sort of explained and reasoned thinking I was looking for!!

So essentially, JDB you're saying there is still a time and a place for complete discrete stages, and the optimisation can be beneficial, and John you're saying that opamps are just so good that transistors are only really useful to you now as conditioning the inputs and or outputs?

Im asking because Im designing a trafo based preamp, and was looking at DOAs but then wondered if it wouldn't be better to go completely discrete and design the gain stages specifically. The reason for this is that I'd like to have a bucketload of gain and a nice, short audio signal path. Im not convinced that I can coax enough gain from a single DOA stage cleanly and without too much noise...

Mo

 

[sahib]

JDBakker and John, that was more the sort of explained and reasoned thinking I was looking for!!

The reason for this is that I'd like to have a bucketload of gain and a nice, short audio signal path. Im not convinced that I can coax enough gain from a single DOA stage cleanly and without too much noise...

Mo

How short? With current DOAs kicking around you can hit 60dB without too much noise and they do not particularly have long and winding roads.

When you are designing for your own use, or for a very small group of customer base, such as this forum, the cost is not particularly of concern. But when you are designing a commercial product for world market you'll save even a single resistor if you can get away with. There are so many designs that are full of 5534s and still kick ass with 70dB gain.

In my view the need for such "new" design is a result of getting bored with what's available off-the-shelf. I dont believe what you'll do is going to be "particularly" better than most. It will only be different. But that is still a valid reason.

 

[JohnRoberts]

Sam posted a series of preamp designs in a thread with a couple transformer based ones... perhaps check those as a starting point.

JR

 

[PRR]

> I'd like to have a bucketload of gain

WHY??

Excess gain is the bane of good sound.

Gain can always be cascaded-up, noise is permanent.

I've spent much of my life -reducing- gain.

Define what you really need. And why your needs are different from everybody else. They may BE "different"; I have obscure needs no commercial product addresses. But be sure before you chase the wild goose.

This will give a bucketload of gain. Obviously the input overloads way too easy. It has other practical problems. But you can knock it out in minutes and get a thrill.

 

[tv]

Imo you want a DOA if:
- you seek a particular signature sound/flavor
- you need a low noise opamp that can drive a specific load at higher-than-usual rails
- you want to show off your masculinity

That said, only you know exactly what you want/need.

Personally, I'm not exactly in love with opamps, my preference is with SE (class-a) "gaincards," built around a CFP... these can sound so pleasantly "slanted" (for the lack of better words) when you find the bias "sweet spot"..

 

[PRR]

This too will give a bucketload of gain, despite output loading. The input overloads still way too easy, and has other practical problems, etc. 

 

[abbey road d enfer]

Wouldn'it distort a little?

 

[clintrubber]

Personally, I'm not exactly in love with opamps, my preference is with SE (class-a) "gaincards," built around a CFP... these can sound so pleasantly "slanted" (for the lack of better words) when you find the bias "sweet spot"..

Hi,

Let's define opamps...

FWIW, as you know the difference in something CFPish & something opampish can be just one transistor, see attached picture.

Which doesn't say that there's no difference in sound of course, please elaborate more on "slanted" & "sweet spot" if you like.

Cheers,

  Peter

 

[tv]

Its a combination of bloom/timing, its a vibe thing. Has to be mis-biased _just-right_ for that. The resistor in first trans. collector is the "bloom master" - of some sort.

Breadboard one... it's tamildaa - or tasildaa in this case ("s" for "sound")

 

[PRR]

> Wouldn'it distort a little?

At 1V peak out, 0.6%THD, nearly pure Second Harmonic.

HARMONIC   FREQUENCY    FOURIER    NORMALIZED    PHASE        NORMALIZED
    NO         (HZ)     COMPONENT    COMPONENT    (DEG)       PHASE (DEG)

     1     1.000E+03    1.069E+00    1.000E+00   -1.798E+02    0.000E+00
     2     2.000E+03    6.258E-03    5.855E-03    9.054E+01    4.501E+02
     3     3.000E+03    3.626E-05    3.393E-05    1.121E+02    6.515E+02
     4     4.000E+03    2.029E-05    1.898E-05    8.013E+01    7.993E+02
     5     5.000E+03    2.584E-05    2.418E-05    1.260E+02    1.025E+03
     6     6.000E+03    1.968E-05    1.842E-05    8.939E+01    1.168E+03
     7     7.000E+03    2.598E-05    2.430E-05   -1.586E+02    1.100E+03

     TOTAL HARMONIC DISTORTION =   5.855462E-01 PERCENT

This is, of course, at 0.25 milliVolts input, 1.25mV at Base. Since naked BJT distortion gets past 25% 20mV input, working 10 times lower should give ~~~10 times less distortion, near 1%.

I can not think of the last time I had an audio path opened-up so that 1/4mV gave "zero VU" or anything near that.

 

[PRR]

> Wouldn'it distort a little?

At 14V peak out, 7%THD, and the 3rd+ are starting to come out.

This is 3.3 milliVolts input, 17mVmV at Base.

 

[Moses]

Hi guys!

Thanks for the guidance Prr...

Basically, I have an ISA428 which keeps overheating (think uncomfortable to touch front panel...) during recording sessions, which leads to the line outs splatting, popping and making generally unpleasant sounds, and it's just gotten to be too much! Im planning on selling it and putting together something homemade.. I've built a few of the JLM baby animals, and I may just go that route for 8 channels.. But before I jump on the band wagon, Im looking at doing my own thing, possibly with an assignable mixbus for mixing channels on the spot ala that API job...

So basically Im wondering if I can do something with similar gain to the 428, but less noise and greater fidelity. Not that I don't like the 428, but the JLMs with lundahl 1538, 99V, and a 1517 output stage is a different league, imho... Planning a listening test between JLM's input trafo, the lundahl 1538, and an OEP, as well as the 99V and OPA 2604 opamps, and a DRV134, the 1517 output stage (like calrec polar, as posted elsewhere by Gyraf) and the valve Lundahl 5402 stage Gyraf uses in the pultecs (with modified gain...) in the near future. Say next month or so, should anyone be interested...

So that's where my thinking's coming from!! Is it worth my while designing my own preamp, or more hastle than it's worth, especially with the JLM kits so affordable?

Thanks!

Mo

 

[abbey road d enfer]

Your question is somewhat similar to asking a singer if he wants to do covers or write his own songs. Some do covers very well, some write more or less good songs, it's a matter of personality and creativity. But all songwriters have started by singing other peoples' songs. Same for electronics construction; you need to be familiar enough with other designer's stuff to start designing your own. There are so many rookies mistakes to be made, it is nice to have this forum to submit the results of your thinking for evaluation/critics.
If you want to design your own stuff, you must know what you're aiming at. You shouldn't start by saying you want transformer A, input mic B, output stage C; you should write down what you want in terms of performance and design in accordance with it.

 

[PRR]

I don't mean to be rude, just bluntly asking things which IMHO must be asked....

> I have an ISA428 which keeps overheating (think uncomfortable to touch front panel...)

Why is a solid-state preamp "overheating" that much?

> line outs splatting, popping and making generally unpleasant sounds

How does overheat make splatt?

> Im planning on selling it 

Are you going to tell the buyer that it is incredibly defective?

Have you asked Focusrite about it? It -could- be a known problem with a known solution.

Overheat is usually a simple problem and cure. If you have not dug into it to fix it, it seems silly to instead get swamped in a major scratch-build, or even a kit-toss.

> ISA428 1538, 99V, 1517 1538, OEP 99V OPA 2604 DRV134 1517 5402

Alphabet-soup overload.

> Im wondering if I can do something...

Yes, you can do something.

Can you beat the Focusrite? If you have to ask, then it is uncertain. Just looking at the numbers, I could not pay myself to design/build a better-number box.