putting feedback around coupling caps

[mikep]

Reading the servo thread reminded me of this little idea I had:

the offset will not be perfectly zero because of the DC offset coming from the node that includes the opamp's minus input pin. however, the source-Z, if you will, of that offset is quite high and with a low-z load it is pretty much swamped. it is also not a problem (stability wise) to short the output to ground or a DC voltage. there are some interesting effects at LF into low-z loads. for one, the AC feedback around the caps tends to lineraize them, but if you slightly undersize C2, it can be made to overcompensate, you actually get a bit more LF energy into stiffer loads.

comments? Id like to see other examples of output caps included in a feedback loop if anyone has one.

mike p

 

[BradAvenson]

http://www.maxim-ic.com/appnotes.cfm/an_pk/3171

Here is an example of feedback around the input cap from maxim.

I think I have seen a circuit similar to yours in one of the EE mags, but I don't remember which one.

 

[JohnRoberts]

[quote author="mikep"]Reading the servo thread reminded me of this little idea I had:

the offset will not be perfectly zero because of the DC offset coming from the node that includes the opamp's minus input pin. however, the source-Z, if you will, of that offset is quite high and with a low-z load it is pretty much swamped. it is also not a problem (stability wise) to short the output to ground or a DC voltage. there are some interesting effects at LF into low-z loads. for one, the AC feedback around the caps tends to lineraize them, but if you slightly undersize C2, it can be made to overcompensate, you actually get a bit more LF energy into stiffer loads.

comments? Id like to see other examples of output caps included in a feedback loop if anyone has one.

mike p[/quote]

I’m too lazy to check, did the old 60 W Dynaco power amp wrap HF feedback around their output blocking cap? If they didn't they should have.

I don't recall seeing a dedicated circuit to do exactly what you propose, I have seen series electrolytics with DC bias applied. While C3 and C4 will be well controlled inside the feedback loop, C2 is still very much in the audio path so you may want to look at the impedances and dielectric used there.

One obvious observation is that your circuit does not block DC at it's input. IMO a primary function of using a DC blocking capacitor.

Sorry, I don't mean to sound dismissive, I am very much in favor of using clever feedback configurations to make electrolytic caps inconsequential or shift impedance to nicer dielectrics. Your circuit will indeed protect the output from say a single PS rail failure assuming the preceding stage holds zero.

IMO there is a "there" there, but tricks like these are best rolled into other functional circuit blocks. For example in some of the hybrid mic preamps, electrolytic caps are effectively inside the feedback loop and their non-ideal characteristics minimized.

JR

 

[Gus]

IIRC in the mid to late 70's there was a solid state hifi preamp made using feedback "around" the output cap. It also claimed a very high slew rate using a feedforward circuit in the phono section. I seem to remember a name like ap something. Anyone remember this?

 

[mikep]

[quote author="JohnRoberts"]
IMO there is a "there" there, but tricks like these are best rolled into other functional circuit blocks.[/quote]

exactly. I'm already using this concept in a bigger, more useful circuit. I just sketched it with the opamp to make it as simple as possible for show and tell purposes. that and I dont want to give away ALL my tricks just yet. point taken about C2.

 

[burdij]

Kensonic Accuphase, perhaps?

 

[EZ81]

(part of a voltage controlled fader, that's why the LDR is in there...)

If i understood this right, the influence of the 1000uF output cap is nulled by the feedback loop :?: :?: :?: . The output transformer is stepup, resulting in minimum 100 Ohm load for the output amp.

 

[Guest]

[quote author="mediatechnology"]mikep; Let me think about another example of where I've seen this...

I do remember feedback being taken from the output of transformers
[/quote]

I come too late...

Duh - Didn't see Brad's post.

I decided to read answers before answering, and found your answer already answered...

I recently took a positive feedback by current from bridged transistor amp using a line input transformer connected across 0.2 Ohm resistor in series with speakers. It works! :thumb:

 

[JohnRoberts]

[quote author="EZ81"]

(part of a voltage controlled fader, that's why the LDR is in there...)

If i understood this right, the influence of the 1000uF output cap is nulled by the feedback loop :?: :?: :?: . The output transformer is stepup, resulting in minimum 100 Ohm load for the output amp.[/quote]

IMO no. Circuit is anchored to 12V nominal output by second opamp, 1000uF just provides AC ground while keeping DC out of transformer. If anything grabbing a early 12V reference from there for earlier inverter may possibly increase apparent cap LF rolloff. The feedback is wrong polarity to help and varies with LDR..

JR

 

[Guest]

[quote author="JohnRoberts"] The feedback is wrong polarity to help and varies with LDR..

JR[/quote]

It is negative by current (increases output impedance on a very low end)

 

[Guest]

[quote author="mediatechnology"]Here's the cite I was thinking about:

http://www.maxim-ic.com/appnotes.cfm/appnote_number/1036/

[/quote]

I like this clone with extra cap more than mikep's original idea with resistor after the coupling cap.

 

[mikep]

Here's the cite I was thinking about:

cool, I also came up with this arrangement on my own. but there are 2 problems I see w.r.t. audio quality. the feedback is being "equalized" to compensate the LF response. the error signal has been thru an extra cap, therefore feedback cannot reduce capacitor nonlinearity at the output jack, rather the cleanest signal will be at the node where Rfa and Rfb join!

it does a good job of reducing cap size, but at the expense of headroom in the low end, where it is quite important, IMO.

 

[dale116dot7]

The Soundcraft 2400 uses feedback around its LF EQ coupling cap - which becomes an insert send when the LF EQ is switched in. It causes problems when driving some transformers - it'll become an oscillator. It happens with my Gates Solid Stateman comps. You have to be careful sometimes.

 

[Guest]

[quote author="mikep"]
it does a good job of reducing cap size, but at the expense of headroom in the low end, where it is quite important, IMO.[/quote]

In any case you loose this headroom, that depends on the capacitance value. The trick is, the feedback cap is loaded by higher resistance than output one, also if the output current through the load load is asymmetrically nonlinear in your case the opamp will be biased by rectified output current, with the additional feedback capacitor this rectification will be filtered out.