NFB, presence and THD: some questions.

[freefrog]

So, NFB is supposed to reduce power amp distorsion...

... and increasing a typical presence control is meant to reduce NFB in the high frequencies by "darkening" the signal fed back thanks to a cap - in such a way that high frequencies are less altered by NFB and become louder...

Correct me if I'm wrong.

If my simplistic explanations are right, there's something that I don't understand.

I was recently toying with presence cap values in one of my amps (the HRD that I use for practice)... I've measured its frequency response and THD through a SM57. And to my surprise, my ears and my screens noticed more distorsion with the presence pot @ 0/10, where NFB is supposed to be extended to the whole freqency range and to limit the distorsion...

Is it due to the loudspeaker (Jensen C12N in this case)? To a poor OT?

Why is the sound crispier (and less distorted on my screen/to my ears) with the presence control full up while it's supposed to diminish the NFB in the high range? Is it due to a change in the overall NFB ratio (because of the variable relationship between the 25k pot and 4,7k resistor to ground)?

I'm a self taught hobbyist so feel free to school me.


Also: when my 25k pot is between the presence cap and ground, how big is the influence of the presence control on the sound? Is the 25k pot a brickwall or does the cap keep altering the high range beyond the resistive barreer?

Finally: how do I calculate the "knee frequency" due to a presence cap? I've changed the .1 capacitor for a .068 network in my HRD and I'm pleased by the sound but I'd like to know on which frequencies it operates exactly.

BTW, if there's any harm to change the value of the presence cap like that, please, tell me, because I don't want to create problems that my amp hadn't.

Thx!

[Jazz P Bass]

When the 'Presense' control is on 10 (full On) there is less feedback action.

[freefrog]

When the 'Presense' control is on 10 (full On) there is less feedback action.

Firstly, thx for your answer.

Secondly, I agree. Reason why I don't get how a presence @ 0/10 (more NFB) can distort more than the control full up, as it appears to be the case in my HRD.

[Chuck H]

NFB in guitar amps is primarily used to tighten the bottom end and low mids. Because of how guitar amps are voiced and the nature of the speakers I'm sot surprised that there could be less HD at higher frequencies with the presence in use. It's also possible that because guitar amps don't usually enjoy all the same considerations in design that hifi stuff does that there are phase errors in the NFB loop due to the OT and tone shaping circuits. Consider that the "presence" cap itself is tied to it's own ground. This should cause a phase shift that is inconsistent with what the loop needs to reduce HD. Possibly even causing more HD as you've seen. If you ground this cap so nothing phase shifted is fed back the distortion disappears with it. Violla!

There is certainly a formula for figuring the knee frequency of a cap based on it's operating impedances. I don't know it though I do well with on line calculators and other SPICE based tools though. If someone posts the formula I should write it down so I can still design amps when the zombie apocalypse kills the internet

[freefrog]

NFB in guitar amps is primarily used to tighten the bottom end and low mids. Because of how guitar amps are voiced and the nature of the speakers I'm sot surprised that there could be less HD at higher frequencies with the presence in use. It's also possible that because guitar amps don't usually enjoy all the same considerations in design that hifi stuff does that there are phase errors in the NFB loop due to the OT and tone shaping circuits. Consider that the "presence" cap itself is tied to it's own ground. This should cause a phase shift that is inconsistent with what the loop needs to reduce HD. Possibly even causing more HD as you've seen. If you ground this cap so nothing phase shifted is fed back the distortion disappears with it. Violla!

There is certainly a formula for figuring the knee frequency of a cap based on it's operating impedances. I don't know it though I do well with on line calculators and other SPICE based tools though. If someone posts the formula I should write it down so I can still design amps when the zombie apocalypse kills the internet

Thanks a lot for your answer: it contributes to my feable understanding. :-)

If you or someone else recalls how to calculate what I mentioned, I'll be grateful.

EDIT: is is risky to put an on/off switch between the presence network and the ground? I see how it would change the nfb ratio but I wonder if it would traumatize the amp...

[jazbo8]

Handy calculator(s) can be found at ampbooks.

[Dave H]

Also: when my 25k pot is between the presence cap and ground, how big is the influence of the presence control on the sound? Is the 25k pot a brickwall or does the cap keep altering the high range beyond the resistive barreer?

Finally: how do I calculate the "knee frequency" due to a presence cap? I've changed the .1 capacitor for a .068 network in my HRD and I'm pleased by the sound but I'd like to know on which frequencies it operates exactly.

The ‘ampbooks’ link doesn’t work for me. I think it goes something like this. When the 25k Presence pot is shorted. The frequency response is flat up to a frequency given by f = 1/(2*pi*RC) where R = 4k7 and C = 0.1u. It then rises until it hits the ‘knee frequency’ caused by the amp reaching open loop gain. There isn’t much feedback so the knee frequency is only a couple of octaves higher. (see graph below)

The 25k pot is almost a brick wall. The cap has 25k in series so the combination will only have a small effect in parallel with the 4k7.

[nickb]

One thing that should be mentioned is the foregoing only applies when the amp is not driven to clipping. NFB will tend to increase the distortion once clipping starts as the output stage will be driven harder.

[nashvillebill]

Try this link for ampbooks: Cathode Bypass Capacitor Calculator

[freefrog]

Thx to each of you gentlemen!

Yes, the ampbooks website is extremely useful.

Dave, special thanks for your explanation, that I find totally clear. :-)

[mhuss]

EDIT: is is risky to put an on/off switch between the presence network and the ground? I see how it would change the nfb ratio but I wonder if it would traumatize the amp...

It depends how you do it. You don't want to disconnect the PI tail. :-)

I actually built an '18watt' amp that has an electronic switch in the loop (disconnecting the resistor going to the OT.) It's a two channel amp with a 18watt style channel (which sounds better without NFB) and a wreck-ish channel (which definitely needs NFB). The channel switching logic also controls the NFB switch.

http://mhuss.com/18watt/schematics/18wTRex-S.pdf

[freefrog]

It depends how you do it. You don't want to disconnect the PI tail. :-)

I actually built an '18watt' amp that has an electronic switch in the loop (disconnecting the resistor going to the OT.) It's a two channel amp with a 18watt style channel (which sounds better without NFB) and a wreck-ish channel (which definitely needs NFB). The channel switching logic also controls the NFB switch.

http://mhuss.com/18watt/schematics/18wTRex-S.pdf

Extremely interesting. Thank you!

[Tooboob]

One thing that should be mentioned is the foregoing only applies when the amp is not driven to clipping. NFB will tend to increase the distortion once clipping starts as the output stage will be driven harder.

Folks don't say this enough.