Compound OP-AMP

[myhrrhleine]

Hi all,

I am interested in building a series compound OP-AMP.
My first question is about compensation.
How closely do the two OP-AMPs need to be matched for bandwidth?

 

[jcx]

different reasons for different composites - lead to different gain block choices, compensation schemes

often use "mis-matching" - letting one op amp be the dominant pole source, the other so much faster that its contribution to loop phase shift is small - as in fast unity gain output buffer inside the feedback loop of a slower op amp

 

[myhrrhleine]

different reasons for different composites - lead to different gain block choices, compensation schemes

often use "mis-matching" - letting one op amp be the dominant pole source, the other so much faster that its contribution to loop phase shift is small - as in fast unity gain output buffer inside the feedback loop of a slower op amp

Thank you for the reply.
In my case, both OP-AMPs will have significant gain and similar bandwidth.
One possibility I am looking at is an OPA827 followed by a LME49710.
I don't have much control of bandwidth or compensation.
So I wonder if a dual such as the OPA2827 might be better?
TIA

 

[jcx]

I wonder what the application, system level and particular technical requirements are before stating what parts, topology I "have" to use

 

[myhrrhleine]

the application is instrumentation and the goal is THD below .0000001 at 1 volt.

 

[jcx]

1 ppm or 10ppb?

still lacking gain, bandwidth, source, load impedance, input noise, filtering requirements...

Av > 100x can use Graeme's 1/Beta technique: http://www.ti.com/lit/an/sboa015/sboa015.pdf

 

[Samuel Groner]

You'll need to show a detailed schematic of the specific opamp configuration (inverting, noninverting, source/feedback/load impedances etc.). And what's the frequency we're talking about?

Samuel

 

[myhrrhleine]

You'll need to show a detailed schematic of the specific opamp configuration (inverting, noninverting, source/feedback/load impedances etc.). And what's the frequency we're talking about?

Samuel

Intended circuit is the SVF used as an oscillator, frequency range is 10hz to 30khz, impedance is typically 5-10kohm, noise as low as possible (the OPA 827 has a noise of 4nv per rt hz), bandwidth of 0.3hz or less.

 

[AndrewT]

Have you read W.Jung?

 

[Samuel Groner]

Intended circuit is the SVF used as an oscillator, frequency range is 10 Hz to 30 kHz, impedance is typically 5-10 kOhm, noise as low as possible (the OPA827 has a noise of 4 nV per rtHz), bandwidth of 0.3 Hz or less.

This requires the opamps to be unity gain stable, but at least it's inverting (no common-mode effects). If the operating level of your oscillator is indeed just 1 Vrms, you might be able to get away with careful class A bias of a decent single IC opamp. As a starter for a possible composite opamp you might want to look at this post: http://www.diyaudio.com/forums/equi...n-audio-range-oscillator-225.html#post3454158

Of course there's much more to low distortion oscillator design than just throwing in a composite opamp. The thread linked above is good reading (if you can filter from the usual noise).

Samuel

 

[myhrrhleine]

Have you read W.Jung?

Not lately,
Does he say something about compounding OPAMPs ?

 

[myhrrhleine]

This requires the opamps to be unity gain stable, but at least it's inverting (no common-mode effects). If the operating level of your oscillator is indeed just 1 Vrms, you might be able to get away with careful class A bias of a decent single IC opamp. As a starter for a possible composite opamp you might want to look at this post: http://www.diyaudio.com/forums/equi...n-audio-range-oscillator-225.html#post3454158

Of course there's much more to low distortion oscillator design than just throwing in a composite opamp. The thread linked above is good reading (if you can filter from the usual noise).

Samuel

I have been carefully following the thread mentioned.
I have some other ideas I want to try.
I was just looking for advice on OPAMP compounding.
I thought there might be some experience in it I could learn from.

 

[Ludus Tonalis]

Have a look at these:

http://www.ti.com/lit/an/sboa015/sboa015.pdf (especially page 7 onwards)
http://www.ti.com/lit/an/sboa002/sboa002.pdf
http://www.ti.com/lit/an/slyt103/slyt103.pdf

 

[jcx]

the discussion from here suggests ultralinear amplifiers aren't the best idea depending on how thoughly you understand how the oscillation has to be stabilized
http://www.diyaudio.com/forums/equi...on-audio-range-oscillator-72.html#post3201775

also seach the thread for the AN67 circuit, comments

 

[Samuel Groner]

I thought there might be some experience in it I could learn from.

As jcx notes, the second amplifier should be as fast as possible; its noise and linearity is secondary (there are a few choices which go near 100 MHz with +-15 V rails). The first opamp can be about any good audio part with suitable noise characteristics at the given source Z.

The discussion from here suggests ultralinear amplifiers aren't the best idea depending on how thoughly you understand how the oscillation has to be stabilized.

Oliver's text does not apply directly to oscillators with linear leveling loop. You can get extremely short settling time even with ideal amplifiers.

Samuel

 

[myhrrhleine]

As jcx notes, the second amplifier should be as fast as possible; its noise and linearity is secondary (there are a few choices which go near 100 MHz with +-15 V rails). The first opamp can be about any good audio part with suitable noise characteristics at the given source Z.



Oliver's text does not apply directly to oscillators with linear leveling loop. You can get extremely short settling time even with ideal amplifiers.

Samuel

My concern is both amplifiers have significant gain.
I seems to me there would be over 180 degrees of phase while there is still significant gain.

 

[jcx]

my, ska-a Graeme article link shows the modified integrator compenstion for lower closed loop gains, can be used with both op amps in the loop ~same speed

 

[myhrrhleine]

my, ska-a Graeme article link shows the modified integrator compenstion for lower closed loop gains, can be used with both op amps in the loop ~same speed

Thank you, I actually read that article before my first post. My question is :
How closely do the two OP-AMPs need to be matched for bandwidth?
Should I use a dual such as the OPA 2827?