Bias vary trem & Fixed bias - what optimum combination?

[tubeswell]

My latest build is a 5F6A with a 6G16 trem circuit (and a reverb) - schematic attached.

I've got most of the 'trouble' out of the build, now I just want to 'shoot' it .

Anyway I decided to start a new thread cos the one on eth debugging BB was getting too long to follow, and this focusses on a new aspect so...

(Reverb aside for the moment) the trem is not very intense, and I am wondering what to do about getting a more hypnotic intensity out of it. (I love the way my 5G9 sounds - granted its a different amp with different tubes and different trem circuit, but I figured that the 6G16 trem circuit would've been about right for a 5F6A running 6L6GCs to get a real shimmery intense trem - correct me if I'm wrong, something is obviously not as I expected)

The output tubes are JJ 6L6GCs with plate voltage of about 444, screen voltage of 441, and tube current of 34mA. (470R screen grid resistors)

The trem circuit LFO plate voltage is 210 (cathode voltage 2.28) and the CF stage has a plate of 440, grid of 210, and cathode of 217.

1) So can I bias the 6L6s in such a way as to increase the intensity of the trem?

2) Or do I have to start pulling bits out of the trem circuit? The 6G16 circuit has a 22nF cap from the plate to the cathode of the LFO stage, and I built this in to my amp. (I guess this is some kind of intensity bleed cap?) Would this cap be better out? or reduced in capacity?

[txstrat]

The Deluxe Reverb II circuit has a 500pF cap between plate and cathode to prevent the circuit from oscillating. Don't know if the .02 cap in the 6G16 circuit is there for the same reason. The Vibrolux 6G11 has no cap at all at that place, a different value resistor before the intensity pot and no cap (.1) to ground. You might have a look at the schematic. All other values seem to be equal at a quick glance. I'd try to change the specs for 6G11 values and see what happens.

[tubeswell]

Okay I'll try a range of cap values in there. Thanks txstrat

What bout the bias? 50% too cool? (and if they were hotter, would that in theory make the trem wiggle more effective? - due to the grid voltage swing being closer to zero, in order to drive the tubes into cut-off?)

[txstrat]

You remember my 6g3 clone?
I biased the tubes in that amp at 70%. I think with the bias vary trem ON, the tubes get into hotter areas but I wouldn't bother, since the tubes can stand 80% as permanent setting as well (OK at least the JJ 6V6s I chose). And 6L6 are sturdier than 6V6 from their specs alone.

[tubeswell]

Okay I experimented progressively going from the 22nF to a .01uF, then a .0047uF, then finally a 500pF cap between the plate and the cathode of the LFO, but I couldn't tell much difference. Then I rebiased the output tubes to 17W and I think that has increased the intensity a wee bit. Still not the same as my 5G9 tho'. So what else can I try?

[txstrat]

Following the signal path from the cathode of the 12AX7 to the depth pot. Right after the 470k resistor is a .1 cap to ground (another .1 leading to the pot). I would lift it on one side to see what happens, since it might shunts some of the signal to ground.

[Tubenit]

You might find some useful information here:

Hoffman AB763 Tremolo 6L6 mods?

I would experiment with lowering the 470k resistor to 220k. I'm refering to the 470k between the 470k cathode resistor & the .1uf/.1uf juncture.

With respect, Tubenit

[MWJB]

Regarding bias, often running the bias a shade cooler will affect the intensity of the vibrato, see what happens around 28mA per tube.

[tubeswell]

Originally Posted by Tubenit

You might find some useful information here:

Hoffman AB763 Tremolo 6L6 mods?

I would experiment with lowering the 470k resistor to 220k. I'm refering to the 470k between the 470k cathode resistor & the .1uf/.1uf juncture.

With respect, Tubenit

I lowered the 470k in question, and it is more intense, however when the speed pot is on any setting under about "4" a clicking sound (in time with the Tremolo) starts up (and goes away when I speed it up again). is that something to do with the LFO waveform and the output tube grid bias?

(Also, in deference to MWJB - I don't see why increasing the negative bias voltage would make it more intense. Can you explain that please?)

Despite this, it is still not as intense as the 5G9 trem, and it must be able to get more intense

[MWJB]

"(Also, in deference to MWJB - I don't see why increasing the negative bias voltage would make it more intense. Can you explain that please?)" - a more noticable swing from low to high current (being a bias trem, this is how it works, if it made no difference, you wouldn't get any trem).

[pdf64]

The 470k plate resistor on the oscillator section might be limiting the voltage swing, ie amplitude.
Most of the old designs show it as a 100k or 220k.
And the cathode resistor of that stage usually smaller than 4k7, try 3k3 - 1k5.
And the cathode resistor of the following buffer as less than your 470k, try 220k.
The aim is to get the biggest signal out of that oscillator / buffer to the intensity pot, so check out on the scope, or with meter, the effect of each change and combination of changes.
So as above, reducing that series resistor to 220k will help, but also the DC blocking cap could be increased in value to let more oscillator signal pass - a 1uF film cap would help, or just parallel up some 0.1uFs.
A low pass filter somewhere in the oscillator signal path would help get rid of the click - maybe a cap between the plate and ground of the oscillator, and / or re-instate a cap to ground on the hot end of the intensity pot. The exact value wil be a compromise between the amplitude of the oscillator signal and the click.
The click might be an artifact of the oscillator or buffer going into cut off, so biasing them a bit hotter, as suggested above, might help anyway.
Peter.

[tubeswell]

Okay I took out the SS rectifier and put the GZ34 back in. This brought the B+ down and the 6L6 plate voltages are now about 420 with the tubes biased to 35mA. This has opened up the LFO a bit more. And the amp has a nicer tone. (Only thing is now I'm getting some unpleasant kind of intermittent static buzzing on the bass notes and as I turn the vol up - is that crossover distortion?)

I changed the 470k series resistor to 220k and that opened up the LFO a bit more too.

I'm just looking at the Tremolux 6G9B circuit and I see it doesn't have a 0.1uF to ground before the blocking cap to the depth pot, but has a 0.05uF cap from the depth pot wiper to ground. Would that also open it up more? It also doesn't have a cap between the plate and the cathode of the LFO. Why is that 0.02uF cap there in the Vibroverb 6G16 circuit?

[MWJB]

"(Only thing is now I'm getting some unpleasant kind of intermittent static buzzing on the bass notes and as I turn the vol up - is that crossover distortion?)" Unlikely at 35mA, you could even go lower (28-29mA) without crossover disortion becoming a problem at moderate volume settings.

[Merlinb]

Originally Posted by tubeswell

It also doesn't have a cap between the plate and the cathode of the LFO. Why is that 0.02uF cap there in the Vibroverb 6G16 circuit?

The various caps to ground (before or after the depth pot, or between anode and cathode) are all there to make the trem sound 'smoother' by filtering out higher harmonics from the trem wave, making it more like a proper sine wave. They're not essential, and as you have found out, they usually don't make a lot of difference to the trem sound. You can build the circuit without them all, and then try adding one before or after the depth pot to see if it makes any useful difference.

[tubeswell]

Originally Posted by MWJB

"(Only thing is now I'm getting some unpleasant kind of intermittent static buzzing on the bass notes and as I turn the vol up - is that crossover distortion?)" Unlikely at 35mA, you could even go lower (28-29mA) without crossover disortion becoming a problem at moderate volume settings.

Thanks MWJB

Actually after having read Chapter 2 in Merlin's book, I wonder whether maybe its blocking distortion. I don't have a scope to see for sure, but now I am armed with equations , so I'll take new voltages again and try and figure out mathematically where its coming from. Come to think of it, I have a 220uF bypass cap on the reverb recovery stage - maybe putting a 22uF one there will help it (but that stage is somewhat loaded down with the reverb tone control, and I even get the distortion with the reverb recovery entirely mixed out). Maybe 1k5 grid stoppers on the output tubes would help too.

[Merlinb]

Originally Posted by tubeswell

Maybe 1k5 grid stoppers on the output tubes would help too.

Woah woah woah! I didn't realise you'd copied the 5F6 like-for-like! That's one of Fender's design blunders- you absolutely need grid stoppers there, or you WILL get blocking! Personally I would up them to 4.7k minimum. I would be very surprised if that isn't where your problem is coming from, those 100nF coupling caps are HUGE!

[MWJB]

+1 on the grid stoppers...but 4.7K might be overkill (depends on how good your layout is, it is possible to "get away" with none, Fender didn't install them as a matter of course on fixed bias, LTPI, 2x6L6 Fenders until the BF era), use the smallest value that gives you proper performance. I'd start at 1.5-2.2K.

[Merlinb]

Originally Posted by MWJB

use the smallest value that gives you proper performance. I'd start at 1.5-2.2K.

I don't wish to start a flame war, but that's a bit like saying "use the smallest brake discs that will stop the car".
You shouldn't be using the minimum value (there no logic in it), you should be using a "healthy value". And before you worry, Tubeswell, they have to be pretty enormous before they will have any audible effect (except the reduction of blocking of course! ).

[MWJB]

"I don't wish to start a flame war", & I don't wish to propagate one, but...

A car that doesn't stop is a car that you cannot drive. It is dangerous. An amp that sounds fine & doesn't suffer blocking with a smaller grid stopper is an amp that works & is probaby laid out well...in short, what the amp designer should be trying to do in the first place. So it's NOT the same, it's more like using lighter engine materials/reducing unsprung weight.

What are the size of the grid stopper on a 59 RI Bassman? (47R just to save you time looking it up).

Do I fit them? Sure I do, usually 1.5K, if I run out I use 2.2K...if you typically need bigger than that in a 6L6 Fender style circuit it's because you have done something wrong/new/different somewhere else.

[Merlinb]

Originally Posted by MWJB

An amp that sounds fine & doesn't suffer blocking with a smaller grid stopper is an amp that works & is probaby laid out well...

But the problem is that layout has very little to do with grid stoppers on the power valves. Whether the amp suffers problems there is mainly dependant on the gm and interelectrode capactiance of the power valves, and on the primary and leakage inductance and capacitance of the OT, over which the deigner has no control. (And we probably aren't using the same OT Fender was using back then...). Just because the "seems" to work doesn't mean it isn't oscillating at HF (remember, HF oscillation can be inaudible, but will still burn out voice coils). And it doesn't mean that it will stay stable as the valves age or when you plug in a different set of valves with slightly different characteristics.
Using the smallest grid stopper you can 'get away with' is like using the smallest brakes on the car that will "just do the job". It may be fine now, but long term reliability is questionable. You should always use values which are large enough to protect against natural component (mostly valve) tollerances.

if you typically need bigger than that in a 6L6 Fender style circuit it's because you have done something wrong/new/different somewhere else.

That simply isn't true. Many power valves need values as high as 10k to make them stable with a particular output transformer (particularly when new = high gm) and no two output transformers are alike either. If you find you need bigger than Fender used it's not because you've done something wrong, its because you're not using the exact same valves or OT that Fender used. (Marshall discovered this pretty quickly!)

I also don't see any logical reason for using the smallest possible value. If zero is not enough, 1k does the job, and 2k also does the job and doesn't sound any different, then why would you want to always use the 1k "borderline" option?

[MWJB]

I think that there are enough, well used, 40-50yr old amps, with the original OTs & no grid stoppers, to suggest that there's more going on than simply perameters in the power tubes, over which the designers "have no control".

We're not talking about "many power valves", we're talking about 6L6 style tubes in a largely proven topology...one in which I have never seen grid stoppers over 1.5K. Indeed, I own a EL84 amp with a 27K grid stopper & I see no reason to change that.

Agreed difference between 1K & 2K may be moot, but if you have to use nearer 10K you are likely just masking symptoms, not dealing with the cause. Likewise if a different brand of 6L6/OT (assuming that it's of a suitable rating & close enough primary-Z) causes you trouble.

[Merlinb]

True, beam tetrodes have lower gm than similar pentodes and therefore can use smaller values. But I would argue that plenty of those 50-60 year old amps are not stable, never were, and shouldn't be copied because of it. They are not all 'proven' designs, they are just bad amps that seem to work, either by dumb luck, or because their symptoms were masked by something else (such as less sensitive speakers than we have today, and by all that background hum and hiss).
There are innumerate vintage amps which just sound terrible today (because they weren't designed to be overdriven), and desighers then were just as liable to overlook something (or cut corners) as they are today. JUst because an amp is old, or came from the sacred house of Fender, does not mean it is well designed or beyond improvement.

Grid stoppers are not a band-aid; they do not just cover symptoms, they are the solution. Unfortunately many enthusiasts don't recognise them for what they are and what/how they do it. (See neutralisation in any Terman textbook). When you copy an old amp using a new OT and new components and trouble arises, it's not because there's something wrong with those components, it because the old design wasn't designed to cope with a sufficiently wide spread of component tollerances. It was designed to use that OT with those speakers and those high ESR PSU caps, etc. Modern electronics has changed all that. Today we (should) design things that are univerally reliable.

[MWJB]

We're getting of topic here...& I should know better but...

The tweed bassman IS a proven design, yes, Fenders have their shortcomings & yes there is sometimes room for improvement. You might also want to consider skill & experience as well as "dumb luck". What background hum, & hiss? Many vintage Fenders hum & hiss less than most modern amps.

The modern, hi quality, repros don't seem to need excessively large grid stoppers either.

This is becoming a discussion about a discussion, thus this will be my last post on this matter.

[tubeswell]

Thank you MWJB and Merlin for this information. 1k5 (right on the grid pins) seems to have removed the audible 'blocking' distortion (which was only very slight on some of the bass notes anyway) although I have not 'dimed' the amp yet. I will experiment with higher values to see if I can detect changes in the sound.

FWIW it is quite a 'bright' sounding amp (at the moment it has a Telefunken ECC83 in V1) altho' it seems to have plenty of bass response as well. FWIW the OT (custom made) is fairly robust and was ordered as 50W, but in reality can probably handle quite a bit more than that.

[TheTinMan]

I'm curious having just finished an AB763 Super Reverb with 6G16 bias vary tremolo using Hoffman's approach.

There are two "extra" components in your tremolo circuit. There's a .02uf cap connecting the plate & the cathode on the first triode. Someone just recommended that change to me but couldn't explain what it does other than "improve the sound". The second change is the 470k resistor connecting the grid and the plate of the buffer triode. What is its purpose?

There's nothing new in the tremolo circuit I used. It's straight off Hoffman's diagrams with the tweaks Tubenit linked to (i.e. reduce 470K off buffer cathode to 220K and change .1uf to .047).

Don't know if this helps,

Chip

[Merlinb]

Originally Posted by TheTinMan

There's a .02uf cap connecting the plate & the cathode on the first triode. Someone just recommended that change to me but couldn't explain what it does other than "improve the sound".

It filters out harmonics to make a 'purer' sine wave. You can read about it at Aiken: Designing Phase Shift Oscillators
I don't think it makes any useful difference, personally.

The second change is the 470k resistor connecting the grid and the plate of the buffer triode. What is its purpose?

That's just the load resistor for the oscillator! It's just drawn in a funny way.

[tubeswell]

I ended up taking that substituting that .02uF cap for a .01uF, then a 500pF, and then I took it out completely. I also reduced the 470k series resistor (from the CF cathode coupling cap to the depth pot) to 220k. (So the LFO circuit is now closer to a 6G9-B than a 6G16). Man what a difference! Now the trem has a great range of depth, even when the output tubes are biased hotter.