Uncompensated op-amps

[David]

No, I'm not talking about chips that work on a volunteer basis...   :mrgreen:


I've got the lack of compressor blues again.  I can't get a Roland to "Sustain", and each time I try to breadboard a Flatline, it does!  This is really getting frustrating.  Seems like I have a "vertical" learning curve with each new pedal type I try.  Happened with distortion, happened with the headphone amp.  Anyway, not to whine, but...

I'm kind of looking at the redraw Joel Purkiss posted of Anderton's compressor.  It looks like a good candidate, especially since I have an idle CLM-6000 that been waiting for something to do for 21 years!  Here's my question:  the circuit calls for an uncompensated op-amp, like a 201, 301, etc.  Now, I can get some 301's, but I'm wondering what's so special about them.  Couldn't I sub a 741 or a 5534 for a 301?

[brett]

I'm not familiar with the circuit, but I suspect it contains a capacitor (and maybe other components) for setting the frequency response of the op-amp.  More modern op-amps seem to simply have a flat frequency response and only require external components if you want non-flat.

But maybe I don't understand the full story, because some people substitute a NE5534, which is supposed to be uncompensated, for compensated op-amps like the TL071.  Strange....

[niftydog]

compensation is employed  to adjust the frequency response of an op amp to prevent unwanted oscillations.  It also affects the "speed" of the op amp (transient response or slew rate)

The unwanted effects of compensation include gain loss @ high frequency.  "Un" compensated op amps are more correctly called "externaly compensated" meaning that compensation can be added using external components.  This means you can tailor the compensation to a particular need.

compensated op amps include;  NE5534, 741, TL071, TL081, 1458, 4558 etc.

NB; the NE5534 IS compensated!

"un" compensated op amps include; LM201, LM301

External compensation is sometime considered a relic of times gone by, long since forgotten since IC manufacturers can now do compensation properly on the die.  You may consider a "less" compensated op amp than a 741, but not TOTALY uncompensated like the 201.  Some component adjustment may be needed.

see here for details

[WGTP]

The TL070 is externally compensated and should work.  The Rat is the most famous user of that type of op amp.

[Mark Hammer]

The "Whisper" compressor that Jack Orman and Thomas Henry marketed through Midwest Analog Products is essentially an adaptation of the Anderton EPFM project with a few extra bells and whistles.  It uses a CLM6000 the exact same way that Craig did.  The Whisper employs a TL061 in the same slot that Craig used an LM748 (uncompensated).   My gut sense is that a normal compensated op-amp like a 741 will do just fine, although other op-amps might draw less current.  Obviously you dump the compensating cap, but everything remains the same.

Were it the case that the control element in the compressor had a much faster response time (e.g., OTA, FET), I would expect the compensation to be important in shaping the reactivity of that control element.  In this case, though, the response time of the CLM6000 itself probably takes care of a lot of what the compensation was intended to do.

[David]

Mark:

Thanks for the reply!  I'm supposed to have some sample 301's on the way.  I just hate to wait.

Let me confirm what you said.  If I swap a 741 for the uncompensated 301, I also dump the compensating capacitor.  OK, makes mucho sense.  Anything else I would have to change, or should the rest of the circuit work as specified?

[Mark Hammer]

In principle it ought to work just fine.  You will note that the LED in the optoisolator turns that op-amp/CLM6000 combo into a half-wave rectifier.  Look at the envelope follower in the Dr Q/Quack/Nurse-Quacky and you'll see that all it is is an op-amp with a lot of gain and a diode after it to restrict what comes out of it to only one half cycle (i.e., nothing "below the mid-point of the waveform").  

In the case of the DQ/NQ what comes out of the other side of the diode is further conditioned so that it can drive a transistor and provide a little lag to smooth out the ripple in the half-wave rectified signal.  In the case of the circuit Craig uses, the optoisolator arrangement means the diode (LED) doesn't have to "connect" to anything else electronically, and the use of a photocell means that a lot of the ripple-smoothing is taken care of.  As a consequence, the envelope follower in the compressor looks uncomfortably simple to the naked eye - something gnaws away at you, saying "There HAS to be more to it than this".  In this application, though, that's pretty much all you need for acceptable performance.

As for *which* op-amp works out best, I couldn't say.  I don't know enough about the specific properties of the LM748 that led Craig to choose it over others.  It may be that a garden variety 741 (or half a 1458) is fine, or it may be that something like a 5534 (more current delivery capability) is better, or some other op-amp (available now but not in 1980) represents the preferred arrangement.

[David]

Thanks, Mark!

Look out, protoboard, here I come!!   :twisted:  :twisted:  :twisted:  :twisted:

[lalunatech]

compensated op amps include;  NE5534, 741, TL071, TL081, 1458, 4558 etc.

NB; the NE5534 IS compensated!

see here for details

I believe the NE5534 is internally compensated for closed loop gains of 3 or above.  They are not suitable for use as a unity gain follower without an external compensation capacitor.  Compare the NE5532 dual amp that is compensated for unity gain use.

[antonis]

After 20 years (from initial thread), they might made NE5534 unity gain compensated , also..
(or relied on about 22pF stray capacitance needed..)

[duck_arse]

and a welcome, Antonis?

[antonis]

and a welcome, Antonis?

Definitely, NO...!!

[Rob Strand]

I'm kind of looking at the redraw Joel Purkiss posted of Anderton's compressor.  It looks like a good candidate, especially since I have an idle CLM-6000 that been waiting for something to do for 21 years!  Here's my question:  the circuit calls for an uncompensated op-amp, like a 201, 301, etc.  Now, I can get some 301's, but I'm wondering what's so special about them.  Couldn't I sub a 741 or a 5534 for a 301?

It's an uncompensated opamp but it has a compensation cap connected which then turns it into a compensated amplifier!  You can use any opamp, it will work.  Just leave off the compensation cap (pins 1 and 8 ).

[BJF]

Hi there,

Some thoughts

Compensated OP amps  are stable at amplification 1(unity gain stability).This means that the OP amp can be used as buffer without any further compensation. Because an OP amp has very high Open Loop Gain ( amplification without feedback loop) and internally several stages that while those basically are a differential input stage a Voltage Amplifier stage and an Output Current Amplifier stage, the phase shift through circuit will bend and when it reaches 180 degrees oscillation is a factor. This means negative feedback becomes positive feedback.
There are several ways to compensate ( really making amplification diminish to 1 before phase becomes critical and with a marginal
This goes for any circuit that has negative feedback: the easiest way is to use a single slope lowpass filter but then open loop gain at high frequencies may suffer and so will slew rate of output and so there are other ways such as feed forward compensation that is often used with LM318.
Actually the circuit idea for Honey Bee I based on a schematic that Bjorn Isheden showed me using an LM318 in feed forward mode followed by a variable single pole lowpass filter and a buffer made with an LM310 ( dedicated buffer OP amp) and then negative feedback taken from the output of LM310. This way slew rate could be adjusted and of course open loop gain- and years later it proved to be a nifty way to remove certain overtones that my pedal hating friend Gronis Suprodude von Danikken didn't want to hear .
Slew rate is a measure of how swift the output can swing without deforming square wave ; or at how high frequency can a square wave be reproduced accurately.
In feedforward mode LM 318 gets double  the bandwidth and an increase in slew rate far beyond almost any other OP amp

As mentioned using one dominant roll off results in the least complicated phase response. In an OP amp the VAS (Voltage Amplifier Stage)
Is often a transistor with a rather high collector load producing the bulk of voltage gain; in high power tube push pull power amps with negative feedback a dominant roll off is often made across the anodes of the phase splitter, while other methods like placing a shunt capacitor to ground at each control grid of each power tube will also work but is more a brut force.
The reason for mentioning tube power amps is that they are somewhat similar to OP amps as are most high power transistor power amps the difference being that transistor solutions typically have much larger Open Loop Gain (amplification without feedback).

So frequency compensation is used to ensure stability by making sure open loop bandwidth falls with a safe margin to phase shift  and especially at 100% feedback which is worts case. For this reason the typical OP amp only has extreme open loop gain at very low frequencies
but negative feedback trades raw gain for bandwidth.

With tubes open loop bandwidth is low and so negative feedback can only be applied cautiously with enough margin to oscillation.
If margin is not enough oscillation can be triggered by an incoming signal. Compensation can be used as mentioned above to increase safety margin.

Sometimes over compensation can be desirable and when using CA3130 I overcompensate that to reduce some unwanted  distortion and some noise. Likewise when I use NE5534 I may over compensate to gain slightly lower noise

Decompensated OP amps are available such as the LF357 and these are not stable at 100% feedback but loop gain must not be lower than 5 with a margin. It is a very fine OP amp if some gain can be allowed.

Typically in data sheets on OP amps it would be stated that the OP is stable at unity gain or that a capacitor is needed to connected between two terminals and this is known as a compensation capacitor.
Sometimes manufacturers offer unity gain stable version and decompensated version I e TL071 and TL070.
The TL070 needs an external capacitor to be stable at 100% feedback ( when used as buffer) but at higher gain and less feedback performance can be increased such as higher slew rate and less distortion at high frequencies.

OP amps are often used with 100% feedback even if not buffers such as when feedback elements allow 100% feedback at some frequencies and therefore industry has called for unity gain stable OP amps.

Have fun
Bjorn Juhl

[Rob Strand]

But maybe I don't understand the full story, because some people substitute a NE5534, which is supposed to be uncompensated, for compensated op-amps like the TL071.  Strange....

There is more to the story. 

The suggested compensation caps are usually for unity gain compensation.  If you operate at higher gains you can use a smaller compensation cap.  That applies to most uncompensated opamps.  The NE5534 is a little unique as it is actually internally compensated for gain of three (and above) instead of unity,
https://www.ti.com/lit/ds/symlink/ne5534.pdf

That's why some circuits can sub an NE5534 without adding the compensation cap.

If you want to operate at unity gain then you need to add the compensation cap.   The compensation cap is 22pF, not the common 33pF for unity gain you see on other uncompensated opamps.  If the look at the internal schematic you can see there is 12pF cap inside the chip.  If you add 12pF to 22pF you get 34pF, not unlike other opamps.

The more common dual NE5532 is unity gain compensated.  Pretty much a dual NE5534 with compensation.

The reason the NE5534 is offered with less than the usual unity gain compensation is so it can gain speed performance in circuits not requiring unity gain compensation.  Unity gain compensated opamps are convenient but it comes at a cost of reducing the speed.  Uncompensated opamps come at the cost of wasting pins for the compensation cap and the need to add the cap.  A lot of old analog circuits which required fast response would use uncompensated opamps with an optimally selected compensation cap value.

[antonis]

FWIW, when someone troubleshoots oscillating output transistors in their power amp, it might worth to take a look at the unity gain op-amp driver..

[Rob Strand]

FWIW, when someone troubleshoots oscillating output transistors in their power amp, it might worth to take a look at the unity gain op-amp driver.

Output transistors can have *parasitic* oscillations.  That's where the transistor itself forms an oscillator due to the *local* circuit conditions.  That's very different to the stability of the overall feedback loop and the use of compensation networks and caps.

You can actually get parasitic oscillations anywhere.   The oscillating buffer is a common one.

[antonis]

Output transistors can have *parasitic* oscillations.  That's where the transistor itself forms an oscillator due to the *local* circuit conditions.

Yeaapp..

I was talking about unity gain uncompensated op-amp acting as voltage follower driver..

[BJF]

Hi there,

Dedicated IC OP amp buffers with no other function and that I have used through the years

https://www.farnell.com/datasheets/105178.pdf

http://www.ic-ts-histo.de/sehenswertes_histo-datenblatt/TL068.pdf


NE5534 is stable down to gains of 3 and the relevant gain is noise gain
For instance an inverting OP amp with two alike resistors as feedback resistor and input resistor has a voltage gain of 1 but a noise gain of 2.
When using NE5534 as buffer it needs a compensation capacitor of 47pF but if it is an inverting buffer compensation needed is 22pF.
These are minimum values for safety distance. Typically I would use 100pF with NE5534 because I would rather have less noise than the extra bandwidth and some distortion can be accepted.
LM 741 makes a good buffer but it is not recommended at higher than 10. Because of the negative feedback just about any OP makes a good buffer but while most OP amps can drive 2K Ohms NE 5534 can drive 600 Ohms.
Current output can be increases by paralleling OPamp  buffers inside feedback loop and by placing a resistor of 100 Ohms in series with each output to force equal current distribution
If extremely high input impedance is neededa JFet or MOS fet type can be used and reference resistor from noninverting input to ground/ virtual ground/ voltage reference can be divided in two series resistors and a capacitor of say 470nF can then AC couple the junction of the resistor with inverting input and effectively bootstrap reference resistor so that the resulting input impance is
that of the OP amp- aside from making IZ up in the Terra's it is an effective method of increasing IZ of an LM741 and other bipolar OP amps. The total value of the reference resistors should then be held around 200K to avoid output offset. Resulting IZ can then for the LM 741 reach a couple of Mega Ohms.at 10KHz.
For dimensioning input capacitor it would then only need to be in the tens of nanoFarads for a low enough time constant: it takes only 100nF to allow destructive discharge to an unprotected non inverting input- very common fault on Valvestates

Murphy's Law
Amplifiers always oscillate, Oscillators more seldomly

Have fun
Bjorn Juhl

[BJF]

No, I'm not talking about chips that work on a volunteer basis...   :mrgreen:


I've got the lack of compressor blues again.  I can't get a Roland to "Sustain", and each time I try to breadboard a Flatline, it does!  This is really getting frustrating.  Seems like I have a "vertical" learning curve with each new pedal type I try.  Happened with distortion, happened with the headphone amp.  Anyway, not to whine, but...

I'm kind of looking at the redraw Joel Purkiss posted of Anderton's compressor.  It looks like a good candidate, especially since I have an idle CLM-6000 that been waiting for something to do for 21 years!  Here's my question:  the circuit calls for an uncompensated op-amp, like a 201, 301, etc.  Now, I can get some 301's, but I'm wondering what's so special about them.  Couldn't I sub a 741 or a 5534 for a 301?

Hi There,


So very specifically for the Craig Anderton compressor I would personally choose a higher performance dual OP amp for the compressor amp and recovery amp such as LF353 or OPA 2134 because handle the audio signal and any distance to noise that can be gained is good but for the LED driver an LM 741 would do just fine it is pretty modest application and though gain is higher than 20dB there is no audio quality needed.
LM301 can be used without compensation in this application because noise gain is high enough and it saves a capacitor

Have fun
Bjorn Juhl