G2 power testing and conduction angle

[Cliff Schecht]

What I see is a goofy looking voltage divider, but there is one nonetheless. The 2.2k resistor is the series element and the two 220k resistors to ground are the shunt elements. Yes they are connected strangely in parallel with the capacitors because they also double as bleeder resistors, which is quite clever IMO.

[Gaz]

Andy, I may know how the power measurement discrepancy happened - I was measuring clean output before clipping. Were you perhaps measuring max power until you could get no more? I would have definitely gotten a lot more power if I were doing that.

I have JJ 6L6s and E34Ls on hand, and am happy to do some more testing for 'our cause'. I currently have individual 1ks hardwired, but can easily (and somewhat safely) bypass them, or parallel them with other to obtai smaller values.

I know it's pretty elementary, but I'm not sure how to calculate power dissipation when using a voltage divider, and if it's different than calculating the power rating for dropping resistors. When I can figure that, I'll feel more comfortable doing tests galore as I'm currently unemployed! I did know someone who used to do this setup with a 2.2k/25W and a 7.5k/25W resistor to ground bypassed by a big Solen to feed the screens. I just wonder why he made the resistors so large. He's the one who told me regulation gets better as the bottom leg of the divider get smaller.

[Cliff Schecht]

With that voltage divider, it's actually really easy to calculate the power dissipation (at least if you are using DC conditions). For the series element , 2.2k in our case, you do (Vin-Vo)^2/R (Vin being voltage going into the divider and Vout being the voltage coming out) and for the shunt element you do (Vo-Gnd)^2/R (just output voltage squared divided by resistance, Gnd=0).

Andy said he was using KT88's. Those will easily do 50W clean (they are more efficient, or better stated have a higher transconductance than EL34/6L6 IIRC).

[Andy Le Blanc]

Yes ... I was looking at a peak output voltage

and that is exactly correct the regulation does get better as the better as
the bottom leg of the divider get smaller, but I think we only need a
measure of regulation, there's an economy with DIY builds and it is a guitar
amp and not HI FI, its more a matter of knowing and controlling the G2
voltage

I'm technically un-employed too, I usually end up working so hard I don't
have time to look for a job, hopefully some gainful opportunities will
surface for you, I know how frustrating it can get.

I've got 6l6, el34, 6550 , kt88... Ive a couple gigs the next couple days
but I think I have some time to un-mod and measure 6l6 per a typical
fender circuit, and then back to a voltage divider.

I've been collecting static readings, and then reading at a peak output
the static bias current and peak current, measured over a 1r at the
cathode give data to figure the conduction angle.

[Gaz]

Thanks for the clarification, Cliff - I was unsure about the shunt element in the divider. Funny, when learning about tube amps - sometimes the fundamental come later.

Okay, so that clarifies the power thing as well. I knew it was too good to be true at those voltages, despite the tube type. I thought maybe you were working some magic there

When you get around to taking some more measurements you should note power at clipping as well, which might be telling. Can't have enough data! Do you have a spreadsheet template by the way, I could put in some more detailed data. I can easily measure bias too in this amp because it has test points.

Haha, no need for a job when I can play with screen grids. I was supposed to be living in Egypt right now, but had to leave... Just fun with amps until I go back in October. Screen grids are more frustrating than finding a job, IMHO

[surfsup]

Warning: noob question coming....

Thats ok... the resistors do form a divider. You can set it any way you want. the voltage drop is whats important.

Don't you get a voltage drop between the node supplying the plates and screens from the choke already?

[Andy Le Blanc]

yes...

In a simple amp the entire power rail can be seen as one series resistor.

each stop, tap or node along the way has a filter cap to decouple each tap
from the other. Many dont have bleeders, the tubes are the bleeders.

This is a convention based on the economics of manufacture.

Going back into the development of "electronics", the earliest convention
was to use one big power resistor to ground, with taps to provide the
various voltages for the intended circuit, but the issue was that you needed
a rather large and costly part that had to handle twice the max current
of the rig, beyond a few watts it becomes impracticable, so the most economical
way to get around it was a series rail with the big current handling choke
or resistor at one end where its needed the most, and the filtering is pretty
good down the other end where the voltage has been dropped the most

Its practical, its economical, and has been a design assumption for a very very long time.

But there are places when dealing with tubes where its not so good.
The screen grids after a series screen grid resistor is just such a place.

Through a series resistor, a change in current also means a change in voltage
so as your screens are pushed the current they draw changes the voltages
across the tube elements between the plate and cathode...

the potential gradients across the device are thrown off and it puts things
into a risky spot the harder you push your tubes.

So... you see a lot of talk about screen grid resistors, in the attempt to
"fix" the issue, in the hifi world this done with separate power supplies
and stabilized and regulated power modules etc...

but in the DIY guitar world you deal with "tradition"... "cloning"...
and in general , upside down design practices and assumptions...
fender did it so must be good... same goes for any other design or famous maker you can think of.
The focus on tube bias as a design priority alone I think has mislead far to many DIY builders,
the same with clones, and there has been some rather harsh and nasty exchanges...

But everyone wants to build a better amp. Putting a simple divider on the
screens in a guitar amp seems a great way to focus more on the issue
of how best to handle some of the design problems that confront DIY makers
and do it very economically, and not stray too far from "tradition"

[pdf64]

Some amps used volt dropper tubes such as OC3 between the plate and screen nodes, see Leslie 147.
http://www.captain-foldback.com/Leslie_ ... cs/147.GIF
Nowadays a power zener could be used instead.
That arrangement would avoid the high power sag on the screen node, in the case of condition 1, post #1.
Given that, what's the advantage of the potential divider method? Pete.

[Andy Le Blanc]

control... and some regulation

The high power sag is the absolute max condition for the circuit.

the control as part of static circuit set up is a good thing, tweak the bias
tweak the feed back, tweak the screens, they all interact .

and every little bit of regulation helps

[Ian444]

I have often thought of using a voltage divider from the B+ to feed a mosfet to regulate the screens, so the screen regulation would vary with the B+ sag, I have not got around to trying it yet. For example, if the B+ is 450V at idle, and you want to run the screens at say 400V at idle. Set up a voltage divider to give 400V at the mosfet gate, then as the amp ramps up to full power, the B+ will drop a bit, and the regulated screen grid voltage will also drop as a percentage of the B+, in this case it will always be 88% of B+, as determined by the voltage divider feeding the gate. Because its a series regulator and the current draw is small, the heatsink requirements would be minimal.

[Andy Le Blanc]

reestablished the original circuit... fender 50w

022798 pt...nsc018343 opt...125c1a ck

5u4g, choke, 470r screen grids, the pi filter caps are series with 220k balancing resistors

@ 100ma tot. PS draw on pair of 5881/6l6wgc

Plate voltage is around 414... with a 20w dissipation per tube.

the screen grids are riding about 1.250 v positive relative to plates

the rig is fizz city, it sounds like a fender and it doesn't even have fender pre

this the condition that got me on the the G2 issues. there is fizz at the
onset of distortion, you get to a certain point and the only resolution is
to bias the thing cold or frig with the screen grid resistors.

evidence to support the use of voltage divider appears in the 1954 STC
6l6g data sheets and in the writing of Norman H Crowhurst.

[Andy Le Blanc]

going for the quick and dirty bench fix I clipped in a couple 220k resistors
between the g2 pins on the socket and ground, this reversed the relationship
@100 ma tot. dissipation the g2 were ring about 100ma negative relative to
the plate

Ive been wanting to preserve some value of screen grid resistor.
having the voltage divider/ bleeder on the other side gives a few more
options at control and have some sort of screen grid resistor helps to
buffer the differences of the tube set, and possible oscillatory issues.

The amp plays great, no fizz, solid and a nice top, etc

It doesn't take much to address the g2 at the bias you intend for your tube type
at set up, you can put them exactly where you want. Its the screen voltage
relative to the plate that's important, not the screen grid resistor value.

it can be as simple as a couple extra resistors...

[Structo]

Can you dumb this down a bit for somebody not as experienced with this stuff?

On my last 100w build I have the screen node come off of the choke to a 500R 5W resistor on each socket ( 6L6GC x 4)

When I measure the voltages of this amp at idle, the plate and screen voltages are the same.
I thought that was weird given the choke and screen resistors that are in series with each tube.
So I measure 436v on pin 3 and pin 4.

Someone explained before that the screen voltage will sag down a bit once you start playing the guitar through it.
Is that true?

There has been some high end harshness in this amp since day one that bugs me and I have not been able to figure out where it is coming from.

[Andy Le Blanc]

measure the voltage difference between pins 3 and 4 at Idle
its more telling than referencing to ground

Put the red probe on pin 3 (plate) and the black probe on pin 4 (G2)
If the grids are positive the meter will read a negative (-) value

even if the screens are only a few hundred mv positive, when signal is applied
the grids goes negative relative to the plate, just as you've said
and that swing shows up as a slightly audible distortion.

I was surprised when I heard it out of a nearly text book circuit, fizz
I was more surprised it was the screen grids relationship to the plate
if the plate is always slightly positive there's no +/- swing between the
plate and the screen grids.... no fizz

using a voltage divider instead upping the screen grid resistor gives you
control over exactly where you "place" the screen grid voltage and
has the bonus of a measure of regulation, or more stability as you push the amp.

[Andy Le Blanc]

more tinker time on a simple bleeder and bypass cap applied directly to the
screen grids of 6l6.

I applied a 100k and 10uf/450v after 470r screen grid resistors
from the g2 pin on the socket to ground.
the effect is same, it moved the g2 voltage to about 2v negative to the plate

bleeder wattage becomes an issue, I'd start with 3-5w bleeders...

It starts to show a choice in economy, applying this at the socket pin you
be in for 2 higher wattage bleeders, before screen grid resistors the bleeder
can be a lower wattage/higher value with one higher value dropping resistor.

similar for bypass caps, after the screen grid resistor you need two
before the resistors you can get away with one .

the advantage of a dropping resistor right at the socket pin looks like
a better degree of regulation right on the screens