Presence / NFB matched to output jack

[jojokeo]

I'm renovating a poorly made Trainwreck and the guy that built this used the incorrect transformers among many other things (300-0-300 PT, 6AR5 recto, into a steel chassis & used the entire thing as a ground plane, trannies mounted with laminations in same directions, etc.). It has two EL34s and the transformer used is 8K:4, 8, 16 taps. I've hooked it up to an 8ohm load using the 16ohm tap for a primary Z of 4K. Using the standard Presence pot (B5K) it appears that the standard 100K feedback resistor (the usual 20:1 ratio) isn't very effective. I believe I removed a 27K which was used prior?
 
Is this because it would still be considered the 16ohm tap even though I'm using it to be the new 8ohm tap therefore I should use the value normally associated for the 16? I've never paid attention as to what values to use accordingly if the NFB resistor is applied to the 4 or 16ohm taps.

[terminalgs]

If you use a 8ohm load on the 16ohm secondary,  the RMS voltage driving the speaker will be lower than if you use a 16ohm load, therefore the input voltage to the NFB circuit will be lower than expected (assuming the engineer sized the NFB resistor and presence pot for use with a 16ohm load)


The NFB resistor is the top half of a voltage divider, the bottom half is the presence pot..,,  So if you lower the value of the NFB resistor from 100K to 47K, you should get more voltage at the "top" of the presence control, or the grid of that PI triode, thus more negative feedback... (minus frequencies the presence cap "makes disappear").

... I think!

[jojokeo]

If you use a 8ohm load on the 16ohm secondary,  the RMS voltage driving the speaker will be lower than if you use a 16ohm load, therefore the input voltage to the NFB circuit will be lower than expected (assuming the engineer sized the NFB resistor and presence pot for use with a 16ohm load)

??? Thanks Term-brah! I'm using the OT according to how I want it to reflect the proper primary impedance to the EL34s' plate to plate load requirement for as close of a match as possible. Never mind what the label designations are since they are only ratios anyway. And I would be that "engineer"    I generally like the 20:1 ratio using a 100k w/ a 5K pot on the 8ohm tap.
 
I will find the proper value to use but was hoping to hear:
1) has anyone came across this situation before as well?
2) how to work the math when using either the 4ohm or 16ohm taps for the NFB resistor on a Presence control rather than "part substitution" which I'm doing now?

[Paul1453]

This isn't the Trainwreck, I didn't have access to that schemo.

This is the Roadhouse EL34 PP 50W schemo.

It has the NFB coming off the 4 ohm jack into a 220K resistor with no pot.

I don't know if this is of any help to you.   

It sometime helps me to look at similar circuits to get different ideas.   

[jojokeo]

Yes anything helps. I like to look at similar circuits too as it's all relative and goes into the "vault of knowledge" database (my head w/ tongue in cheek). I just may use something similar as my customer wants this to be a cross between a 'wreck and sort of dumble-ish. So naturally it will be called the DumbleWreck. It will have plenty of OD but rather it's the clean part that seems to be the most challenging working with the EL34s to this point and what I have to work with.

[clyde]

Nice explanation here:


http://www.aikenamps.com/index.php/designing-for-global-negative-feedback

[Paul1453]

Nice explanation here:


http://www.aikenamps.com/index.php/designing-for-global-negative-feedback

A very nice detailed explanation there!

Thank you Clyde!   

I will need to re-read this many times to let it fully take root.   

That should help answer all your questions Jojokeo,
as long as you have a better understanding of this NFB thing than I do.   

[HotBluePlates]

Which tap do you have the negative feedback connected to? Original 8Ω tap (as in the schematic), or the "16Ω" tap (where you have your 8Ω load connected)?

First try connecting the feedback loop with your normal stock values to the tap where you presently connect an 8Ω load to reflect the correct primary Z.

[jojokeo]

Thanks Clyde I haven't read that Aiken chapter in a while and his site is different than before too. The 100k/5k combo is working fine after testing at a good volume level tonight but there may be a parasitic to deal with which could be playing tricks interacting with the Presence control or vice versa causing me to think it was ratio/value related? I'm only using two stages at this point where I've added a couple small value caps across the load resistors and nothing changed so next step is to disable the NFB entirely but can't test until tomorrow...

Which tap do you have the negative feedback connected to? Original 8Ω tap (as in the schematic), or the "16Ω" tap (where you have your 8Ω load connected)?

First try connecting the feedback loop with your normal stock values to the tap where you presently connect an 8Ω load to reflect the correct primary Z.

I haven't posted a schematic HBP but it's nothing out of the ordinary. Think of a 'wreck/Fender basic preamp with the tone stack between only two stages. I disconnected & bypassed the other tube and have V1B wired directly going into the pi through a .02uF coupling cap. The pi has basic values 470r/10K with a ppimv that was already installed. It's a A250k double pot using 2M2's as it should. Since there was no shielding anywhere I've also shielded the inputs, return from tone stack and the ppimv wires which helped a great deal.

To answer your question I'm only using the 16ohm tap from the 8K OT - which becomes the new 8ohm output jack to provide the EL34s a 4K primary impedance. Yes, this is where the 100k resistor comes wired from.

Another piece of info of note: The whole amp has cap cans installed. I did everything I could to re-wire the entire grounding system and there is no hum or noise with all controls full up where before it couldn't be powered up halfway without it oscillation, hum, and much noise. However - is there any problem with the screens and pi sharing the same cap can ground??? (since that has to do with the feedback loop & divider)

[jjasilli]

However - is there any problem with the screens and pi sharing the same cap can ground??? (since that has to do with the feedback loop & divider)

If it doesn't hum, then there's no problem. 

For DC operation the ground can be anywhere and the amp will function.  For AC (signal) operation the location of grounds is important: to avoid ground loops, and/or to avoid AC power hum getting induced into AC signal.  The point of grounding schemes is to use a recognized plan that is likely to avoid hum in the finished product.  But whatever ad hoc things you've done in a given amp, if it doesn't hum, you win.

[jojokeo]

Thanks JJ, no hum or noise except what would be considered normal very low floor noise speaker hiss only when turned full up. However the only ground not going to a cap neg point is the speaker jack's. I know when using NFB it's best to go back to the pi's cap ground but didn't do this one since I was making all the other corrections to the point to where I am currently. It appears now would be a good time to make that connection either way?

Describing what I think is a parasitic: no matter the MV & other vol levels (amp mainly set clean) if I barely pluck the low E & A strings I can hear a very small fuzziness aside from the actual note. It's easiest to hear & most noticeable on the lower strings. It's very slight but think it may be affecting the sustain quality a bit and if the amp had a recording mike in ftont of the speaker it likely could be audible. This amp will be transported and reside in a studio when finished and why this is important (aside from my anal perfection nature).

The amp is in such a basic form at the moment that I should look to troubleshoot this now before I add the additional stages & switching. I have separated & isolated all signal grounds back to their respective filter cap ground to this point except the output jack's and the only circuit ground location is at the resevoir/B+ CT which is made at the PT bolt.

[Ed_Chambley]

Joe, I emailed you some info, but attached is a proven Wreck Schematic.  Most make the NFB switchable.  Here is photo.

[jjasilli]

no hum or noise except what would be considered normal very low floor noise speaker hiss only when turned full up


That sounds normal.  Also, hiss is not a grounding issue.  Hiss might be caused by a tube, and not the circuitry.  Though it's nice to have no hiss at all, some hiss is normal in tube amps -- especially with vol up and no signal. 


However, you might want to checkout http://geofex.com/ > tube amp degub page > faint hiss or hum


****
Fuzzy trailing on certain notes is a PIA.  I think it can relate to a number of causes to trace down & debug:


* parasitic isolation -- lead dress, or snubber cap after the PI
* bad El Caps(s) or bad cap grounding (or maybe your ground solution is ah issue in this context)
* amp / speaker relationship -- try different speakers
* Not enough NFB (again this may vary with different speakers)
* bad tube(s)


If you want to experiment with different amounts of NFB, I suggest Doug's trick:  use a pot in place of a fixed series resistor in the NFB loop, at least temporarily.

[HotBluePlates]

... attached is a proven Wreck Schematic.  ...

I was looking at something similar when I posted my comments. Specifically, the feedback is 100kΩ to  5KΩ Presence pot, fed from the 8Ω tap of a 50w amp.

Why does that matter? Your current subject amp is a 50w amp, and you're using 100kΩ/5kΩ as your feedback loop, running from the tap to which you've connected an 8Ω load to reflect the correct primary impedance. All that being true, your feedback loop needs no change to operate in the stock manner.

What's important in that determination is this kind of feedback  is voltage feedback from the secondary; the full output power of the amp is assumed to appear across a load impedance tap, which implies a voltage input to the loop. The feedback loop resistors then determine how much of that starting voltage is actually applied to the earlier stage.

In both the schematic case and in your "jiggered" case, you have 50w across 8Ω for √(50w * 8Ω) = 20v RMS as input to your feedback loop. It doesn't matter that the tap you connect to is not "the real 8Ω tap" but that with 8Ω connected to that tap, and the reflected primary impedance which results, and given your supply voltage and tubes, you have 50w of output on that tap. If you used a different tap (resulting in a different voltage-input to the loop) or had a different power throughput (resulting in different voltage-input to the loop), then you'd change the ratio of the feedback loop resistances to compensate.

... Describing what I think is a parasitic: no matter the MV & other vol levels (amp mainly set clean) if I barely pluck the low E & A strings I can hear a very small fuzziness aside from the actual note. It's easiest to hear & most noticeable on the lower strings. ...

Good chance it's oscillation. Also good chance of just a bum solder joint or a component with a fractured lead (even not visible because it's just inside the case). A scope or a listening amp would be nice to have in this situation.

[jojokeo]

HBP thanks for confirming as you have it understood clearly and what I'm doing exactly.
 
This amp's sensitivity has been a little like riding a bull at a rodeo but nothing I can't handle. Once I've repositioned the OT, fixed up the grounding, and shielded those wires it's mostly behaving well now. After re-reading the intro to what Ed sent me it specifically states not to use a steel chassis. It's what I have been given so can't change now. Reason is that it has been ordered and made to fit an old Blues Breaker cab with original black black Celestions (so jj I can't change these either).
 
Plus again, this amp is very quiet for a 50 watter. The tiny bit of speaker hiss is completely normal when 50watts is ready for that "Back To The Future" power chord from Michael J Fox. :) All things considered, it's almost perfect at this point aside from that very small fuzz. The big question now = oscillation or parasitic? I will put the speaker jack's ground where it should be and then short out the Presence control and see what it does?
 
JJ - are you saying to remove the pi 47pF cap? Change it? I've already done that and raised it to 470p and then put down to 100p as it sits.
 
In the end the desired result is not to have a Trainwreck amp. Rather my objective is to simply create as clean of a channel as possible using the first two gain stages similar to the way shown in the TW/Fender BF schems., (likely use Plexi biasing values 2.7k/.68uF)  then use the 2nd tube to cascade the stages to create an OD/Lead channel. Of course the two channels are not separate so the OD will be dependent on the clean side working well. My client has a wall of Marshall stacks in his studio so another reason why a need for something different.
 
As I type this it just occurred to me that another one of the dimwit previous builder's methods was to run all of the B+ wiring close under the length of the board and if I'm remembering correctly Fender wrapped those wires for a shielding affect? It's been a little while since I worked on a Black Face but I think if those wires are not wrapped/shielded this caused the amp to oscillate? Can anyone confirm this please?

[jjasilli]

JJ - are you saying to remove the pi 47pF cap? Change it? I've already done that and raised it to 470p and then put down to 100p as it sits.

Sorry, missed your prior info there. The 47p - 100p snubber cap fully addresses the snubber cap approach.  If there's still fizz/fuzz it must be something else.  BTW I had this issue with 3 amps:   Hot Rod Deluxe (much modded) -- different speakers was the solution.  Hotrodded VibroChamp & Prniceton -- more NFB was the solution.


Other causes I forgot to mention above: bias too cold; partially bad speaker - another reason for swapping speakers.

[2deaf]

Adjusting the PPIMV changes the feedback ratio with a constant PI input.  The presence depends on NFB, so turning the PPIMV down progressively destroys the presence control.

[2deaf]

2) how to work the math when using either the 4ohm or 16ohm taps for the NFB resistor on a Presence control rather than "part substitution" which I'm doing now?

For the case where the speaker stays connected to the same tap and the voltage divider ratio is large (like your 21:1 example) you would multiply the feedback resistor by the square root of two when going from the 4 ohm tap to the 8 ohm tap.  You would do the same when going from the 8 ohm tap to the 16 ohm tap.  If you went from 4 ohms to 16 ohms, you would multiply by the square root of two and then do it again which is squaring a square root so you are multiplying by two.

If you mismatch the output and you want to have the same feedback with a 5K pot., you can use this relationship to determine the feedback resistor when the PI input is constant:

5000V_s / (R_f + 5000) = V_f    rewritten:  R_f = (5000V_s / V_f) - 5000

V_s = voltage at secondary tap
V_f = desired voltage at the center of the feedback voltage divider
R_f = Feedback resistor in series with the 5K pot.

If you check the desired voltage at the 5K pot. before the mismatch and then measure the voltage on the tap you are using when mismatched, you can plug those numbers in to get the feedback resistor value.


 

[2deaf]

I removed an earlier post where I had confused doubling the load on the same tap with putting the same load on a different tap.  In the meantime, I hauled a 100W amp in and put an 8 ohm load on the 8 ohm tap and then put the same 8 ohm load on the 16 ohm tap.  The voltage was less when connected to the 16 ohm tap, but not significantly.  It was also less by about the same amount when connected to the 4 ohm tap.  I could easily be convinced that the lower voltages were the result of reduced efficiency as I moved away from the optimum match.

[jojokeo]

Adjusting the PPIMV changes the feedback ratio with a constant PI input.  The presence depends on NFB, so turning the PPIMV down progressively destroys the presence control.

This helped me better understand what was changing as I made various changes. Thanks.

JJ - are you saying to remove the pi 47pF cap? Change it? I've already done that and raised it to 470p and then put down to 100p as it sits.

Sorry, missed your prior info there. The 47p - 100p snubber cap fully addresses the snubber cap approach.  If there's still fizz/fuzz it must be something else.  BTW I had this issue with 3 amps:   Hot Rod Deluxe (much modded) -- different speakers was the solution.  Hotrodded VibroChamp & Prniceton -- more NFB was the solution.


Other causes I forgot to mention above: bias too cold; partially bad speaker - another reason for swapping speakers.

Damn JJ if you didn't nail this! I think one of the old blackbacks is causing this and haven't taken anything apart in the cab but plugging into another cab it disappeared. I made a couple sound examples of it to document. Saved me time & added frustration, thanks!

[jojokeo]

This isn't directly related to the thread topic but a quick update on the amp as it sits.
After another dropping resistor adjustment to the filter string, removing the GZ34 to diodes, & since I had an extra cap in the last capcan I shifted everything to parallel the resevoir cap. This along with the signal grounds being isolated from chassis ground and the net result is success. Quiet, responsive, super punchy, headroom (finally)...can now move on to the remaining parts of the project.