ampage
Tube Amps / Music Electronics
For current discussions, please visit Music Electronics Forum. New: view Recent Searches.
New: visit Schematic Hell!
The sunn still shines online!

 
Listen to great tunes streaming live right now!

ampage archive

Vintage threads from the first ten years

Search for:  Mode:  

Negative Feedback/Presence Control


 
7/12/2000 1:56 PM
Ray Ivers
email
Negative Feedback/Presence Control
Steve,  
 
The NFB-type presence control works in the negative-feedback loop of the power amp. This loop is basically a resistive voltage divider that feeds a portion of the power amp's output signal back to the power amp input (usually at the phase inverter, though not always), out of phase. Where these two nearly-identical-but-mirror-image signals meet, the parts of them that are identical partially cancel, but the signal contents that are NON-identical (i.e., power-amp distortion, etc.) do NOT cancel; instead, the signal now acquires any distortion created by the power amp circuitry, only out-of-phase to the actual distortion produced. This out-of-phase-distorted signal then has the in-phase power-amp distortion added to it on the way thru the power amp, resulting in cancellation of a certain amount of that distortion.  
 
The presence control is a frequency-dependant feedback shunt. It grounds out a certain amount of the higher-frequency feedback cancellation voltage before it reaches the feedback input to the power amp, so while cancellation of lower frequencies continues to take place, the higher frequencies are no longer cancelled out as much (depending on the setting of the presence control), and the result is more high-frequency response from the power amp (and also more high-frequency distortion).  
 
The size of the presence cap affects the turnover frequncy of this high-frequency 'boost' - the smaller the cap, the higher the turnover frequency for a given size of presence pot, and the more 'chimey' the treble boost gets. The value of the presence pot, in conjunction with the value of series feedback resistor chosen, determines the ultimate amount of treble boost available. The parallel resistor is a DC path to ground for the phase inverter; its job is mainly to prevent scratchy pot operation, but I feel that (unfortunately) it also 'dilutes' the action of the presence pot, and also can make it more 'on-off' in its operation.  
 
Randall Aiken's site has a great paper on 'Designing for Global Negative Feedback' that goes into this complex subject in a lot more detail than I've done, Steve - check it out if you haven't already.  
 
Ray
 
7/12/2000 5:08 PM
Gil Ayan
email
Trip + Presence + Why Dumble's Way?
quote:
"Gil:  
 
    Glad to hear that you're back! So how was the trip?"
 
 
Thanks! The trip was fabulous, and the highlight was definitely Spain (I hadn't been there before). It has great beaches, it is fun, cheap, and the food is great. If Tom Jobim were alive and saw me, he would re write his famous song... "Tall and fat and tan.." Yes, I made a living pig out of myself thanks to the Spanish cuisine! Saw some real guitarists out there too, seems like everywhere you go there is a great flamenco player... Maybe the Spaniards think the same about jazz guitarists in the USA, there's one in every crowd.  
 
So, highly recommended -- provided you speak Spanish and don't mind putting on a few pounds. Then, London (it was hard to be reminded how unfortunate some people are from a culinary standpoint :D ) and Paris were great too, but nothing new to either my wife and I... and much more expensive. As for Morocco, it was very different... We went to Tangier only, and I think it'd be like passing judgement on Mexico if you've only been to Tijuana, you know what I mean? Ok, so much for this.  
 
* * *  
Back to the presence control. The cap dumps highs to ground before they are subtarcted from the signal at the NFBL. So, the bigger the cap, the more midrange you "save," and so the thicker the sound of the presence control.  
 
Now, what is different about Dumble's design is that his presence control is roughly scaled by a factor of 10 with respect to Marshall. Thus, the cap is 1uF instead of .1uF, the pot is 2K instead of 22K, etc. While the ratios are pretty much the same, and the NFBL is primarily a voltage arrangement, one could expect similar responses. However, what are, in reality, the differences between the "common practice" and the "scaled down" version of a feedback loop? I have never heard an answer I thought made much sense to me. So, why does he do it that way? I have no idea.  
 
Gil
 
7/12/2000 6:10 PM
Ray Ivers
email

Gil,  
 
Sounds like you had a blast!  
 
About the 'scaling ratio' of the Dumble circuit; the actual effective Marshall pot value is roughly 4K to 5K, Gil, not 22K. The older Marshalls used the 'scratchy' circuit with the 5K pot passing the PI DC to ground, while the newer ones use a 22K variable resistance/cap series combination in parallel with a 4.7K resistor, to eliminate the DC thru the pot. So the Dumble circuit is using a lot more feedback, and the frequency breakpoint is much lower, than the Marshall 100K/5K/.1 uF circuit.  
 
I think the scaling used depends a lot on the choice of the resistance from the PI 'tail' to ground (whether pot, resistor, or combination of both), which affects PI bias conditions. I would think that you would set your PI static conditions and then adjust feedback loop values depending on desired presence control action/damping factor/etc. I would personally prefer to keep feedback loop resistances as low as possible, to minimize the chance of oscillation in the output stage.  
 
Ray
 
7/12/2000 8:50 PM
Gil Ayan
quote:
"Gil,  
 
Sounds like you had a blast!"
 
 
Thanks, we sure did.  
quote:
"  
 
About the 'scaling ratio' of the Dumble circuit; the actual effective Marshall pot value is roughly 4K to 5K, Gil, not 22K."
 
 
I was talking about the physical pot itself. As you will see, the whole feedback loop is scaled:  
 
1. Feedback resistor: (M) 100K, (D) 4.7K (off which tap in the Marshall? Important point, and I don't know!)  
 
2. Ground resistor on feedabck side: (M) 4.7K, (D 390  
 
3. Pot: (M) 25K, (D) 2K  
 
4. Cap: (M) .1uF, (D) 1uF  
 
Even if the Dumble has roughyl twice the negative feedback, I wonder why the choice of "scaled down" values. The PI's "tail" is (M) 10K, (D) 24K...  
 
quote:
" The older Marshalls used the 'scratchy' circuit with the 5K pot passing the PI DC to ground, while the newer ones use a 22K variable resistance/cap series combination in parallel with a 4.7K resistor, to eliminate the DC thru the pot. So the Dumble circuit is using a lot more feedback, and the frequency breakpoint is much lower, than the Marshall 100K/5K/.1 uF circuit."
 
 
Right. But the breakpoint is not that different... In the extreme case, i.e. with the presence wide open, the Marshall sees 4.7K//.1uF, and the Dumble 390//1uF, the Dumble's frequency breakpoint being HIGHER, the way I see it...  
 
quote:
"Well, agreed. But look at the overall scheme, and the Marshall has a total of 10K I would think that you would set your PI static conditions and then adjust feedback loop values depending on desired presence control action/damping factor/etc. I would personally prefer to keep feedback loop resistances as low as possible, to minimize the chance of oscillation in the output stage."
 
 
I guess I don't see that quite in the same way. The feedback loop is negative, so having a higher impedance wouldn't result in more noise in this particular case, would it? I guess if it was a regular gain stage, sure, lower impedance would definitely mean more stability.  
 
As I said, I am not sure I readily see a good "reason" for doing things that way. :) Probably nothing more than Fender Blackface legacy, where the feedback resistors were all "scaled down" with respect to the Bassman, as well.  
 
Cheers,  
 
Gil
 
7/12/2000 9:55 PM
Ray Ivers
email

Gil,  
 
Yes, you're absolutely right; this is what I get from not having the print in front of me while I type. Forgot completely about the 390 ohm resistor, which changes everything, of course.  
 
About the stability thing; I just meant that I think I would rather have an 820 ohm/100 ohm feedback node (some blackface Fenders) than a 100K/5K (Marshall) one, just in case lead dress or stray capacitance causes some earlier amp stage to pick up the signal from this node (where it may constitute positive feedback to that stage). It wouldn't actually be the output stage that was oscillating - sorry about that.  
 
Ray
 
7/13/2000 2:18 PM
Steve Ahola
email

Gil:  
 
    Thanks for the explanation; 10:1 scaling seems to nail it to a "tee"... One more question: how would the effect of a 1.0uF tantalum cap be different from a poly cap? Any idea on why it would be used? —other than the price factor, which is also scaled down 10:1... :D  
 
Steve Ahola
 
7/13/2000 6:03 PM
Gil Ayan
email

quote:
"Gil:  
 
    Thanks for the explanation; 10:1 scaling seems to nail it to a "tee"... One more question: how would the effect of a 1.0uF tantalum cap be different from a poly cap? Any idea on why it would be used? —other than the price factor, which is also scaled down 10:1... :D  
 
Steve Ahola"
 
 
I can only say: I use tantalum caps there (the small yellow bullet-shaped ones from Mouser), while reports indicate that Dumble uses an NTE electrolytic there. How would a plastic cap sound? I don't know, never tried it! :)  
 
Gil
 
<<First Page<Prev Page 2 of 3 Next> Last Page>>