• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Questions About This Local NFB Scheme

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'd like to hear some comments on the local NFB scheme in this schematic I found.

I'm also wondering about its use of a 1N4007 in both the B+ and G2 supply. What is their purpose? Would there be a problem if they are omitted?

I have a SE VM that originally had the preamp section connected with an umbilical. The amp section only has a 5Y3 and a pair of 6BQ5s. I want to drive it with either a stand-alone tube pre I built or a SS pre. So I won't be using global NFB like the one that went to the stock preamp through the umbilical.

My amp does NOT have an UL tap like the one in the schematic. I want to get as much power out of it as I can so I want to stick with pentode instead of converting it to triode.

So it's either pentode with no FB or some type of local FB. Are there other local FB methods that I should consider?
 

Attachments

  • Local NFB Scheme.png
    Local NFB Scheme.png
    20.9 KB · Views: 452
Thoughts…
№ 1 — the SPDT switch could be a cetner-OFF 3 position type, too, to good effect. Would produce 'degenerate cathode' negative feedback, which I personally like.

№ 2 — if № 1 done, then also add a 100 kΩ resistor from the switch's pivot to ground. This allows the capacitor to pre-charge, which in service would allow it to be switched in and out of circuit minimizing a pop.

№ 3 — The UL/PEN switch is fine. The 1N4007 is there to prevent inflow of electrons to the screen grid. (as a reminder, the diode symbol 'points' to the blocked-flow direction).

№ 4 — for resistors P = E² / R … = 8² V ÷ 150 = 64 ÷ 150 → 0.43 W. Why the 5 W model? I'm not against it generally, but still … a solid 2 W would be smaller.​
Nothing really matters as long as the P is well below PMAX

⋅-=≡ GoatGuy ✓ ≡=-⋅
 
Thoughts…
№ 1 — the SPDT switch could be a cetner-OFF 3 position type, too, to good effect. Would produce 'degenerate cathode' negative feedback, which I personally like.

№ 2 — if № 1 done, then also add a 100 kΩ resistor from the switch's pivot to ground. This allows the capacitor to pre-charge, which in service would allow it to be switched in and out of circuit minimizing a pop.

№ 3 — The UL/PEN switch is fine. The 1N4007 is there to prevent inflow of electrons to the screen grid. (as a reminder, the diode symbol 'points' to the blocked-flow direction).

№ 4 — for resistors P = E² / R … = 8² V ÷ 150 = 64 ÷ 150 → 0.43 W. Why the 5 W model? I'm not against it generally, but still … a solid 2 W would be smaller.​
Nothing really matters as long as the P is well below PMAX

⋅-=≡ GoatGuy ✓ ≡=-⋅
Thanks for your thoughts on this . . .

#1 & #2 - I was actually planning on building this with local NFB only, so the switch would be eliminated. But I do like your idea. As I understand it, "degenerative cathode" negative feedback is just using an unbypassed cathode resistor. Correct?

Would it provide more NFB than the cap (when in local NFB mode) or less?

#3 - There will be no switch here because my OTs don't have an UL winding.

I've only used diodes in power supplies so I'm puzzled by their use here. So the end of the diode with the stripe would be connected to the screen grid. Correct?

But why would you want to block the flow of electrons to the screen? Wouldn't that result in no voltage on the screen?

And what about the diode in the B+ line going to the primary of the OT?
I'm obviously not understanding this.

I was considering the addition of a 100 ohm resistor as a screen stopper. Would that be better than using the diode or could both be used? (I've never used screen stoppers before either, only grid stoppers)

#4 - Yeah, the 5w is overkill. I'll probably use some 2 or 3w wirewounds.
 
That's more like it.And without those useless diodes.
Mona
Looks like you posted while I was typing . . .

So you're suggesting a screen stopper rather than the diode to the screen and eliminating the diode in the B+ supply.

As I said, I've only used diodes in power supplies.

What is the original schematic I posted trying to accomplish by using them? And why do you say they're "useless"?

I'm trying to understand the point of using them. :confused:
 
I was looking at the schematic. In post 1 and thinking, "there should be a resistor in there" and then seeing Ketje's schematic in post 3 reassured me that I'm not crazy after all (much).

Though, I have often just split the Rk in halves (not equal per se), bypassed the upper half of the divider only, and used that node to inject FB and wiggle the cathode ground reference around
 
Last edited:
I have tried multiple times to get someone on these threads and this forum
to give a real, proper, and formal proof, as to why those diodes are needed.

Nobody has responded with a formal proof (at least I have never seen one).
Therefore, I declare that the reason for the diodes is "An Old Wives Tale".
You may, or may not, hear an improvement in the sound of your amplifier (perhaps the use of it will convince your mind that it sounds better).
Most discussions of circuit topologies, designs, parts, etc. have a valid reason why one sounds better.
But those diodes are at the very bottom of that list of "things that improve the sound of your amp".

Formal Proof, anybody?

Credit to Mona:
A 100 Ohm screen resistor is a good idea.

All of the above is just my opinion.
 
Last edited:
That's more like it.And without those useless diodes.
Mona
When I looked at the alternate schematic you posted I was focussed on the lack of diodes and the addition of the screen stopper.

Just now I noticed that the bypass cap in yours is connected to the negative speaker lead and the 8 ohm lead is grounded.

Why is that?

Is this accomplishing the same thing as having the secondary connected in the normal way but with the cap to the 8 ohm lead and the diode in the B+ supply?

Or is the original schematic I posted just wrong?
 
Last edited:
The feedback has to have the right phase, if wrong you get positive feedback and oscillations.
One could as well invert the primary but if was easyer to do it like this :sorry:
Mona
So is the phase reversal only necessary with this type of local NFB? It's obviously not needed when using "normal" NFB.

It that because the "normal" variety goes back to a previous stage while the "local" variety affects the stage where it originates?

Earlier, GoatGuy suggested using a switch that would allow a choice between the local NFB, no NFB (standard cathode bypass cap scheme) and removing the cap from the circuit altogether (unbypassed cathode reisistor) - which would produce degenerative NFB.

Questions:

1) which NFB option produces more NFB? The cap tied to the OT output or the unbypassed resistor?

2) if the amp is switched to degenerative NFB mode, will the amp oscillate since the phase of the OT has been reversed?

3) If the switch is thrown to put the cap to ground, thereby eliminating any NFB, does the reversal of phase cause any problems?
 
Disabled Account
Joined 2013
To answer all your questions it will depend on the a lot of components with optimised value for each. If only compared using the values you listed 1) cap tied to OT gives more NFB or less distortion 2) No 3) you make sure both channels have the same phase otherwise it can cause phase cancellation at low frequency.
I put together a circuit for you review, it has best of all options, if you have a spare secondary winding for nfb.
 

Attachments

  • Local NFB Scheme-2.png
    Local NFB Scheme-2.png
    66.6 KB · Views: 221
I've been reading this thread with interest and the OP experiments and ideas are similar to mine only more developed!

I have recently used an inverted secondary winding to supply global feedback to the input stage, in a 3 stage amp, and was impressed by the result.

All my other attempts at global feedback have been very flawed, and unstable.

However, I have not yet used an isolated secondary winding, which I would expect would perform better than the loudspeaker secondary.

But I would also perhaps think that an additional primary (accepted this is not normally available) would be a better feedback signal source, at least because the amplitude of available signal is larger.

Given the little I have learned using an inverted secondary, I now wonder if an additional isolated primary could be a better way to derive local plate to grid feedback in the output stage, if the driven primary winding is DC free, then no capacitor needed in the feedback loop. Then global feedback could be added from a separate inverted secondary.

Really interesting developments
 
To answer all your questions it will depend on the a lot of components with optimised value for each. If only compared using the values you listed 1) cap tied to OT gives more NFB or less distortion 2) No 3) you make sure both channels have the same phase otherwise it can cause phase cancellation at low frequency.
I put together a circuit for you review, it has best of all options, if you have a spare secondary winding for nfb.
Unfortunately, I'm somewhat limited because my OTs do not have an UL tap and they only have a single secondary. The schematic I posted in my initial post was just something I found online that used local NFB.

Re: your comment on #3. I understand that if one channel is out of phase with the other it can cause phase cancellation of low frequencies.

The earlier advice by @Ketje was that it would be necessary to reverse the phase of both channels when the local NFB to the OTs is used. Otherwise there would be positive feedback and oscillation.

To rephrase my question #3: If the amp was wired to allow a switch from the local NFB (with the cap to the OTs with their phase reversed) to a standard cathode bypass setup (cap to ground), would it also be necessary to have the switch reverse the phase of the OTs at the same time?

In other words, in a normal cathode bias setup with bypass cap to ground, if you reverse the phase of the OTs, wouldn't this result in positive feedback and oscillation?
 
Disabled Account
Joined 2013
Code:
To rephrase my question #3:  If the amp was wired to allow a switch from  the local NFB (with the cap to the OTs with their phase reversed) to a  standard cathode bypass setup (cap to ground), would it also be  necessary to have the switch reverse the phase of the OTs at the same  time?  

 In other words, in a normal cathode bias setup with bypass cap to  ground, if you reverse the phase of the OTs, wouldn't this result in  positive feedback and oscillation?
There is no "dot" marking the phase of OT wire so it's difficult to tell if it's reversed or not. In my attached sch, the phase is marked with dots. So if you wire up accordingly it shouldn't oscillate. But if you reversed the OT secondary wire, it might or might not oscillated but definitely will if output level is increased (so more than 360 degs out of phase). So first to make sure the wire phase is correct, usually the 0V is ground.
So the ans to your question is yes, it will oscillate and increase distortion.
This you need to remember: The input and cathode is same phase, adding same phase (say from OT) to the cathode will cause the cathode signal to become higher and simulated a high cathode resistor apparent value thereby reduced the stage gain and distortion.
 
Last edited:
Aren't you just describing a secondary with lots of turns?

Hehe it could be taken that way yes.

In my experiments, I have a dual primary dual secondary toroid - since I am misusing a PT as an OPT, best results are using one primary and secondary.

This is parallel feed with no DC in the primary.

This would leave me an isolated primary and secondary to use for FB

Or Tertiary and Quaternary winding if one wishes to pick :D
 
The issue of multiple windings to use them for negative feedback is the Leakage Reactance and the Distributed Capacitance of the windings.
The plate winding and negative feedback winding may not be accurately in phase (or may not accurately be out of phase).

It can even be an issue for Ultra Linear primaries. If the leakage reactance is high, from the UL portion of the primary to the rest of the primary winding (to the plate), then there can be a delay from the plate action to the UL screen. That means the phase of the UL negative feedback is not accurate.

How about that?
 
Last edited:
I have been selling SSE amplifier boards for nearly 15 years. They use a common audio pentode like the 6L6GC, EL34 or KT88 and can be wired for triode, UL or pentode mode with or without Cathode Feedback. Switches can be used to change modes if desired.

The most popular configuration is pure triode without CFB, or UL with CFB. The CFB can do wonders for some OPT's, but have little effect on others. Of course the speaker, and its damping factor requirements make a big difference too.

As stated, not all OPT's have the same phasing, so you must try CFB one way, then reverse the secondary and try again. Choose the connection that gives the lowest gain (volume level for a given volume pot setting) to ensure negative (not positive) feedback.


I have tried multiple times to get someone on these threads and this forum to give a real, proper, and formal proof, as to why those diodes are needed. Nobody has responded with a formal proof (at least I have never seen one). Therefore, I declare that the reason for the diodes is "An Old Wives Tale".

The actual origin of the screen grid diode was an old Eimac application note for a RF transmitting TETRODE where under certain conditions the the screen grid could experience a negative resistance region and emit electrons, causing spurious oscillation (illegal in a radio transmitter).

The diode prevents this. This is rarely seen in any tetrode, and usually only seen in small geometry RF tetrodes where the screen is at a far lower voltage than the plate. I have seen this happen in a screen driven pentode amp where overdrive can make the grid red hot, then it will emit electrons that travel to the plate causing a tube arc. I put a diode in series with the screen grid to prevent this. This should NEVER happen in a normal G1 driven pentode, therefore the diode is unnecessary, and can cause distortion under some conditions.

Then an "old wife" created a tale that garnered a following about 15 years ago. Once a "guru" gets a "truth" it can propagate forever. Anyone remember the "Optimized Electron Stream Technology?" I tried it, got a blown EL34 for my effort. If one wants to read a long winded BS story, one can dig as deep into the substance as one can stand here:

Tubes

The magic diodes begin to appear about a third of the way down the page, magic chokes come later. Both get more mention elsewhere on the web site if I remember right. I haven't looked in at least 10 years.
 
Tubelab_com,

Thanks for the explanation of the screen diode.
RF is often different than audio.
EIMAC, a Giant in RF.

But I will try to remember the issue for screen drive and low screen voltage versus plate voltage; and regular g1 drive where the screen voltage is much lower than the plate voltage.
I hope I got it right.

And thanks for the info on triode wired with no cathode feedback, and ultra linear with cathode feedback.
Interesting. Using cathode feedback on ultra linear output gives a plate resistance, rp, that is closer to triode wired mode.
Seems to make sense.
I will try and remember that too.

Millimeter is a whole different world than RF or audio.
I once measured 500GHz using a Tek 2782 and a waveguide mixer.
Getting my hands on the 2782 and waveguide mixer was easy; getting my hands on the 500GHz signal was a little harder to do.
What made that experiment work, was the small waveguide dimensions that cut-off below 220GHz.
The infinite ultimate rejection to the 100GHz source, meant that only the 300, 400, and 500GHz harmonics could get through (and a little of the 200GHz too). The "Identify" function on the analyzer that moves both the 1st and 2nd local oscillators verified which harmonic was on screen, and that it was not a false mixing product (like from another harmonic LO number of the waveguide mixer).
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.