Adjusting phase inverter coupling caps

[hesamadman]

Does adjusting the CC on the output of the PI affect the frequency response like as in the preamp or should these be left alone? I heard some talk on this before but couldnt find it.

[Willabe]

Does adjusting the CC on the output of the PI affect the frequency response like as in the preamp

Yes, same thing.

The smaller the coupling cap the less mids/bass can pass through to the the next amplifying stage. 

[VMS]

It seems to also have an effect on something called bias excursion when overdriving the power tubes:

http://ampbooks.com/mobile/amplifier-calculators/bias-excursion/

[jjasilli]

Good stuff!  But note: caps pass highs and cut lows.  Lows can be cut only if they are they are still present in the signal.  I.e., the overall tone of the amp is principally set early in the preamp, especially the first stage.  Smaller cap values downstream will have less and less effect, because there are less lows left to attenuate.  And larger cap values downstream cannot add lows that are already gone.

[HotBluePlates]

... the overall tone of the amp is principally set early in the preamp ... cap values downstream will have less and less effect...  And larger cap values downstream cannot add lows that are already gone.

This is all true.


If the amp has negative feedback from the speaker, it often returns to the phase inverter. Too many reactive components, and their resulting phase shifts, inside the feedback loop could turn negative feedback into positive feedback. Phase shift due to that R-C is zero where attenuation is zero, 45-degrees where there is 3dB of attenuation, and approaches 90-degrees at some low frequency. If there are 2 roll-offs (and therefore phase shifts), the circuit could ring; with 3 phase shifts inside the loop (and parasitic characteristics of the OT will contribute shifts), the amp will oscillate if care isn't taken to select the right frequency range and amount of roll-off.


With a feedback loop, you really have 2 choices:

• Make the coupling cap to the output tubes very, very large to place the roll-off and the resulting phase shift below the intended audio range. You already cut out this frequency range earlier in the tone shaping of the preamp. But this also makes bias-shifting due to output tube overdriving, and the resulting grid blocking distortion, more likely to sound objectionable.
• You take the opposite approach and keep relatively large coupling caps earlier in the circuit, and save your smallest coupling cap and highest bass roll-off for the final point inside the feedback loop. The goal here is really to keep the roll-offs at very-different frequencies, which reduces the rate of attenuation and hopefully keeps your audio from having enough phase shift to cause the feedback to turn positive.

Of course, you can also reduce the amount of negative feedback to reduce chances of oscillation due to phase shift, or eliminate output stage negative feedback altogether.

Still, I agree with JJasilli that your best option is to do most of your tone shaping in the preamp. You will note looking at old Fender amps that the biggest coupling caps are typically in the preamp, while the smallest is usually the caps to the output tube grids.    {The coupling cap going into the long-tail inverter doesn't count, as the effective resistance to ground is on the order of ~5MΩ or more, due to bootstrapping effect; often a 500pF cap will be found to pass most/all of the guitar range.}

[Fresh_Start]

When you talk about multiple phase shifts inside the feedback loop, would a post phase inverter master volume with two pairs of coupling caps and a dual-ganged MV be an example?

Thanks,

Chip

[hesamadman]

This is all very informative. The reason for the post originally was not to adjust them to attenuate frequency. I had done a CC swap on an entire build recently. Not all at once. Swapped a CC and tested amp one at a time. I was talking with someone who said that in their AC30 style builds they used .001 caps a lot as the CC. So I tried it everywhere. Even in the PI. But all was well until I got to the PI. When those were changed...it was very very shrill and almost static sounding. The info received in this post almost explained that all to me. Im assuming, the whole preamp had killed all the lows given their low values. By the time the PI had saw any of the signal, it must not have had much frequency left to pass through the 500pf or .001 CC that were installed.  It was on of those two values I dont remember.


So anyway. To hijack my own post a little I had a similar thing to discuss. The bright cap commonly seen on a preamp volume. My current build does not have this. And as the preamp volume is cranked it seems to add (allow) more bass. Based off of the information above, if I were to add a bypass cap across my pot, it should keep the cranking of the pre amp volume less bass heavy when maxed. Is this correct?

[sluckey]

if I were to add a bypass cap across my pot, it should keep the cranking of the pre amp volume less bass heavy when maxed. Is this correct?

No. The bright cap on a volume control will allow the amp to sound brighter at low volumes. When you turn the volume to max the bright cap has no effect.

[hesamadman]

if I were to add a bypass cap across my pot, it should keep the cranking of the pre amp volume less bass heavy when maxed. Is this correct?

No. The bright cap on a volume control will allow the amp to sound brighter at low volumes. When you turn the volume to max the bright cap has no effect.

So is more low common as a pre gain is turned up?

[hesamadman]

The bright cap on a volume control will allow the amp to sound brighter at low volumes. When you turn the volume to max the bright cap has no effect.

I'm trying to wrap my head around this. Trying to figure out why it is volume dependent. Obviously the bright cap is in parallel. At lower volumes there is more resistance in the pot. Higher volumes less resistance. I don't know much about resistance in capacitors but I know they are frequency dependent. Thinking about this, I'm going to make an assumption that at higher frequency, a capacitor will have less resistance? Which is why the highs are passed at lower volumes (more resistance in the pot) and has no affect at higher volumes (pot has less resistance than cap)??

Hope this doesn't make me sound stupid. Trying to get a grasp on this.

[shooter]

I'll try and guess here, I'm trying to get this in my brain also, so  I'm using the formula(without #'s) freq = 1/2piRC.  when the vol is low, R is low, allowing the highs(determined by cap value) to pass thru the cap, when the vol goes up R goes up causing the F part of the formula to move lower.  I'm probably all wet, which is fine cuz it's hot and sticky here

[sluckey]

If you look at a schematic that has a bright cap on the volume control you'll see the cap is connected between the wiper and the top (hot) side of the pot. When you turn the pot up to max the wiper actually contacts the top side of the pot. This puts a dead short across the bright cap, therefore it has zero effect at max volume.

As you turn the pot down from max the resistance begins to increase and now you have two parallel paths for the guitar signal, highs will begin going thru the bright cap and the rest of the guitar signal goes thru the resistance between the top side and the wiper. As you turn down even further the highs still go thru the bright cap but the rest of the guitar signal goes thru an increasing amount of resistance between the top of the pot and the wiper. So the high signals are still strong but the rest of the signal is getting weaker due to the increasing resistance. This effect is most noticeable at low settings on the volume pot.

[shooter]

Thanks Sluckey, I'm adding that to my scrapbook

[hesamadman]

Excellent. This gave me a better visual of how a potentiometer functions too.

[PRR]

There's a "right" way to pick a brightness cap. And maybe another way for guitar.

The pot wiper goes to a tube (usually). The input of a tube is a capacitor, around 100pFd.

If we have a high-value pot, say 1Meg, and turn it to half-volume, the wiper comes off the pot with about 500K each way, about 250K effective source resistance.

100pFd across 250K gives 3dB down at 6.6KHz. Bad for hi-fi or general studio gear. For guitar, it makes the highs mild.

Putting a physical 100pFd from to wiper semi-balances this treble loss. Cancellation can be perfect at half-volume, and moot at full-up or full-down.

For guitar, we try a larger cap. For most useful knob-settings this gives some treble *boost*, though not at full-up. This suits post-1950s guitar styles.

[jjasilli]

Another issue is the human ear: it hears Mids as louder than bass or treble, even though a Meter measures them all at the same volume level.  This is especially true at low (and also at very high) volume levels.


(Also at low signal levels I suspect the amp may not be putting out enough power to generate bass notes well).


Anyway, this is why there is a Loudness control on home stereo's: to easily boost bass (and maybe hi's too) to compensate for their drop-off at low volume settings.

[tubeswell]

When you talk about multiple phase shifts inside the feedback loop, would a post phase inverter master volume with two pairs of coupling caps and a dual-ganged MV be an example?

Thanks,

Chip

L-C reactance occurs in the OT between primary and core, and then between core and secondary. This is accentuated if your OT has more interleaving. Since the NFB is sourced from the OT secondary, then you risk positive feedback at some frequency band if the NFB signal is too large in relation to the open loop signal.

[HotBluePlates]

When you talk about multiple phase shifts inside the feedback loop, would a post phase inverter master volume with two pairs of coupling caps and a dual-ganged MV be an example?

The more reactive components in series (meaning R-C or R-L or L-C pairs) inside the feedback loop, the more potential for trouble. The amount of feedback is critically important to whether the loop will be stable, along with the number of shifts and the amount-of-shift (or relative spacing of phase shifts) at a given frequency or range of frequencies.


Because it can be a complex problem, there are often simple rules followed to reduce the risk of negative feedback turning positive.

L-C reactance occurs in the OT between primary and core, and then between core and secondary. This is accentuated if your OT has more interleaving. Since the NFB is sourced from the OT secondary, then you risk positive feedback at some frequency band if the NFB signal is too large in relation to the open loop signal.

What he said. 


There was a period of time when designers sought to apply as much feedback as possible. The problems we're mentioning are part of the impetus for designing output-transformerless amps (to get rid of a source of frequency/bandwidth limitations as well as a complicating factor for feedback) and for all-direct-coupled amplifiers (eliminating flaws of caps as a limiting factor on amp performance as well as eliminating a source of phase shift limiting the use of feedback).


If you're not trying to increase the amount of feedback everywhere, then you generally don't have to worry about these issues.