Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[maxdimario]

This is a split from the EQ thread, continue op amp based discussion here.
B.


Yikes!

what are all those little triangles with 5532 written beside them doing in a piece of professional audio gear!

I wonder what massenburg uses for his parametric.

[bblackwood]

maxdimario wrote on Fri, 12 January 2007 08:32

what are all those little triangles with 5532 written beside them doing in a piece of professional audio gear!

Yah, there's no amazing sounding gear with 5532's in them.

Yeesh.

Quote:

I wonder what massenburg uses for his parametric.

I dunno, what do you put in a piece of gear to give it that harsh upper midrange?

[bruno putzeys]

maxdimario wrote on Fri, 12 January 2007 15:32

what are all those little triangles with 5532 written beside them doing in a piece of professional audio gear!

EQ'ing, mixing, compressing, you name it.

[maxdimario]

yes! I am all too aware of the lovely audio gear full of them!

and to think how many people actually bought it!

there was plenty of money to be made in those days huh?

[maxdimario]

Quote:

I dunno, what do you put in a piece of gear to give it that harsh upper midrange?

harsh upper midrange?

FETS, fet-input opamps?

that is one thing 5532's will not do.. make things harsh..

they go one step beyond and make things wimpy and overly smooth on the top..

hmm..

[bruno putzeys]

http://en.wikipedia.org/wiki/Redundancy_%28information_theor y%29

[zmix]

Dave....

Well, you tried at least.

CZ

[dcollins]

zmix wrote on Fri, 12 January 2007 14:28

Dave.... Well, you tried at least. CZ

Max can run a variety of opamps here, they don't have to be some kind of evil IC!

Probably no simpler than this, even:

http://www.collinsaudio.com/selfdsctamp2.gif

This starts to get a little sticky:

http://www.collinsaudio.com/quad8am10op.jpeg

And if you like Analog Devices obviously this:

http://www.burwenaudio.com/Sound_System.html

DC

[maxdimario]

Not all OPAMPS are little 8-pin chips.

discrete opamps understandably will sound better, and can be made to work...

they are not the BEST but they are easy to implement..

OPAMPS are not the guilty culprit... engineers and manufacturers are.

virtually every 80's piece of audio gear made use of the tl072 or 5532 chips and their close relatives..

manufacturers finally had ONE or TWO components which could replace the MAJORITY of components in a piece of audio gear..

every mixer from soundcraft, allen&heath, auditronics broadcast, eq's, compressors.. you name it.. they all implemented said opamps NOTICEABLY reducing production costs..

buying the same component in very large numbers, and not having to select (opamps do not need selection like discrete transistors) made things CHEAPER..

using opamps also meant using younger cheaper engineers..

it also meant NOT being forced to use TRANSFORMERS, CLEAN POWER SUPPLIES, INDUCTORS (gyrator circuits or parametrics) etc..


every audio function as well as anything else was done with the wonderful, cheap, do-it-all OPAMPS...

So I don't think all opamps are the same... but THOSE opamps are no good..

as far as audio, asymmetric class A discrete still beats most everything else.

[bruno putzeys]

Although EQ's are not my gig I can see this is an elegant and efficiently executed circuit. I've got something of a weak spot for state variable implementations, precisely because of this. It's one of those designs that are meant as a tool, not as an effects device. Those looking for EQ's that warm up or euphonically colour the sound even when flat should look elsewhere.

[dcollins]

Bruno Putzeys wrote on Tue, 16 January 2007 23:36

"Massive distortion"? Wow. I sometimes get the impression the entire audio world is in a pissing contest about who can say the most negative thing about the 5532/4. I'm not too much of a fan of the grainy (dusty) mid and top end that these chips tend to produce under circumstances of less than ideal decoupling, but to speak of massive distortion is a bit uncharitable imho.

From time to time, I may exaggerate a tiny bit just to make a point......

Those that think analog IC's are the worst idea ever, will love this:

http://www.collinsaudio.com/opamp.pdf

I wonder who wrote that paper?

DC

[bruno putzeys]

John Roberts  {JR} wrote on Wed, 17 January 2007 17:49

It was a little weak wrt negative PSRR (...)

PSRR of any miller compensated (I mean the dominant pole is set by a cap between collector and base of a transistor) op amp is limited by available loop gain. The rail that powers the transimpedance stage is an implicit voltage reference input. Any noise there gets coupled 100% to the output through the miller cap and is attenuated only by loop gain. For a simple gain stage, PSRR on the "bad" rail (from supply to output) works out as PSRR < GBW/(ftest*Acl)
The other rail is substantially less sensitive. Anyhow, this is a good reason for using faster op amps.

John Roberts  {JR} wrote on Wed, 17 January 2007 17:49

The output stage was the only thing I didn't recall seeing elsewhere, but I don't study old schematics.

This output circuit has appeared, for example, in Cirrus logic ADC reference designs to improve signal symmetry before feeding a the converter chip.
(ADC chips do insist quite strongly on voltage symmetry - their "differential" inputs should not be used to reject common mode)

[John Roberts {JR}]

dcollins wrote on Wed, 17 January 2007 02:03

From time to time, I may exaggerate a tiny bit just to make a point...... Those that think analog IC's are the worst idea ever, will love this: http://www.collinsaudio.com/opamp.pdf I wonder who wrote that paper? DC

That was an invited paper for the IEEE written by J.E. Solomon who was a National Semi employee at the time. It is one of several classics on the subject and a good overview. It is worth note that it was written in 1974. National Semi was a big dog in things analog back then.

JR

[John Roberts {JR}]

Bruno Putzeys wrote on Wed, 17 January 2007 13:08

PSRR of any miller compensated (I mean the dominant pole is set by a cap between collector and base of a transistor) op amp is limited by available loop gain. The rail that powers the transimpedance stage is an implicit voltage reference input. Any noise there gets coupled 100% to the output through the miller cap and is attenuated only by loop gain. For a simple gain stage, PSRR on the "bad" rail (from supply to output) works out as PSRR < GBW/(ftest*Acl) The other rail is substantially less sensitive. Anyhow, this is a good reason for using faster op amps. ======= This output circuit has appeared, for example, in Cirrus logic ADC reference designs to improve signal symmetry before feeding a the converter chip. (ADC chips do insist quite strongly on voltage symmetry - their "differential" inputs should not be used to reject common mode)

IIRC the 5532/4 series used a somewhat unconventional compensation scheme. I think they called it "split-pole" or something like that. It also involved compensation Cs across a second internal long tail pair. I think the bode plot had the typical 6 dB per oct falling characteristic but there was a nonuniform region in the gain and phase plots due to this scheme. This was all from memory and I don't wish to add ammunition to it's detractors. I just did a quick search for bode plots and both I found show the gain artifact occurring above 10kHz and I couldn't find a phase plot. So IMO this is really a minor detail when applied properly. My recollection was that it occurred at a lower freq but I won't argue my memory over what I can find in data sheets.

I have used lots of them or similar parts with IMO good results, so this is just a little esoteric trivia.

JR

[bruno putzeys]

For a schematic, see:
http://www.dself.dsl.pipex.com/ampins/webbop/5532.htm

The 5532 has three explicit poles and a zero (C1, which compensates for one pole). The dominant pole (dominant as seen from the outside) is by necessity referred to one rail - all miller op amps are created equally in that respect. Here it is around C3+Q8+Q9. The integrator is a transimpedance amp i.e. current in (ICQ7+ICQ4) voltage out (VCQ9-VBQ8). The reference voltage for the output is VBQ8, which is pretty much the same as the voltage on the emitter of Q9 = the negative rail.

So far the only real difference between IC op amps and most discrete op amps is that the latter have historically been faster and simpler because the typical problems with monolytic transistors were absent (see the national paper earlier in this thread). My reason for building discrete op amps is to get round the PSRR problem. Given the THD performance of animals like the LM4562 I cannot see another valid reason (bar headroom or drive capability perhaps).