Voltage drop/sag under load

[pdf64]

'It makes sense that the fixed biased amp moves towards class-A when overdriven b/c the sag in the power tranny makes the bias voltage shift to a less negative value, which is like biasing the amp hotter.'
I don't think that it works like that; heavy current demand on B+ supply causes that to sag but the bias supply current demand stays constant. The bias winding voltage will droop a little due to the overall transformer loading, but from the perspective of the tubes, the change to the bias supply voltage will be far less significant than the change to the plate voltage; the overall effect being that the effective bias changes so as to move closer to class B (this is assuming a regular type amp, not regulated supplies etc). Add this to the grid conduction / blocking distortion / bias excersion thing, and it's almost universal to see crossover effects on an overdriven amp's waveform, even though the bias is set to cover it on a clean wave. I guess the the crossover effect that's visiable could be due to a OT back emf inductive spike though. Pete.

[guitarmike2107]

but from the perspective of the tubes, the change to the bias supply voltage will be far less significant than the change to the plate voltage; the overall effect being that the effective bias changes so as to move closer to class B.

I think that’s what I was getting at in my usual non clear way, I think Kuehnells approach ignored power supply sag in this chapter, so the bias supply circuit does charge and cause it to move towards Class A, but like you say and as is evident in real life, when the power supply sags its effect appears to be far greater than the bias supply excursion alone. It has to be on those Marshall amps otherwise they would red plate when driven hard. The ones that do red plate when driven hard are probably not sagging enough.

The book may cover this later on , I haven’t read it all, but it would be good to get my head wrapped around it all.

Steve.. I may just pop on your doorstep one day!

[redelephant]

Well that's kind of odd since those caps charging will decrease bias. I think I'll stick with my claim.
I am not quite convinced the bias shifts so much either since the reference voltage often has plenty capacitance, but I did mean any eventual shift in bias is from tranny droop. If the secondary that gives b+ sags 15%, the tap that supplies the bias will too.

[EFK]

You all kind of lost me, and I'm not certain whether this is relevant to the discussion, but while it sits at idle at @ 24W dissipation, under full load the bias must shift somewhat because calculating the dissipation using the "full bore" voltages (plate, cathode etc.) then yields 35W dissipation, which is max for the 7581 tubes I currently have in there.

If I just made myself sound stupid trying to play with the big boys, never fear, I shall go find myself a beer.

[Steve Conner]

Well, if you measure the B+ and multiply it by the current draw, you get the total DC power delivered to the output stage. But not all of that gets dissipated in the tubes, like it does at idle. Because the amp is making noise, some of the power goes to the speaker.

The amp I was talking about earlier has about 400v B+ and draws 300mA cranked. 400 x 0.3 is 120w. 50w goes to the speaker, leaving 70w, so the tubes must dissipate 35 each. Before I fitted the regulator, the B+ was over 500 at idle.

[redelephant]

You all kind of lost me...

If I just made myself sound stupid trying to play with the big boys, never fear, I shall go find myself a beer.

...and I just poured me a fat glass of Makers Mark so cheers. I hope I'm not the one appearing to put u down or me above. I find this an interesting discussion thats all, and I dont have the answers carved in stone, just trying to put my way of seeing it on here. I havent noticed any 'stupid' postings, but I tend to be rather bland and not surprised if I missed it.

Hey Steve, when you regulate the screens, do u have some screen resistors so they can drop some voltage to protect the screens? How do u like the amp with such stiff supply?

[EFK]

Oh I don;t think anyone's trying to put me down! I just like beer. I'm putting myself down for not reading more on the important aspects of amp design, which I should certainly do! I appreciate all of the help and discussion very much, even if I don;t understand it all. I can't tell you how much I've learned just reading the posts here on a daily basis.

[guitarmike2107]

I can't tell you how much I've learned just reading the posts here on a daily basis.

Me too on a daily basis, even those who appear to know everythign still learn things on the forums, just our rate of learnign is much higher

I mostly actaully find my foot in my mouth .. like now.

to be continued...

[guitarmike2107]

Well that's kind of odd since those caps charging will decrease bias. I think I'll stick with my claim.
I am not quite convinced the bias shifts so much either since the reference voltage often has plenty capacitance, but I did mean any eventual shift in bias is from tranny droop. If the secondary that gives b+ sags 15%, the tap that supplies the bias will too.

Ok just had a quick look at the book and that was a little porky,(thanks for calling me out) to start with the whole section is discussing grid bias supply excursion. From reading a couple paragraphs quickly he says the grid bias voltage rises, my understanding of what he says is thats its due to the grid current through the grid load resistor effectively placing Rg in parallel with the bottom resistor from the bias circuit. Thereby reducing the voltage divider and causing a bias voltage to rise. His example grid voltage raises from –34v to –28v

[pdf64]

'those caps charging will decrease bias'

From Blocking Distortion

'the problem with your standard resistance-capacitance coupled amplifier stage is that the onset of grid current makes this forward biased diode out of your previously high impedance grid circuit. This results not in diode clipping, as mentioned above, but rather diode _clamping_, which is the cause of both the transient distortion, and an increase in another type of distortion known as crossover distortion.
What happens is, the forward biased diode clamps the tops of the grid waveform to a relatively fixed point. Since the previous stage is AC coupled to the grid, the tops are fixed at the clamp point and the entire waveform then shifts downward as gain is increased, pushing the amp more into class B operation, with a resultant increase in crossover distortion.'

'If the secondary that gives b+ sags 15%, the tap that supplies the bias will too'
I take your point but it doesn't seem to work like that in practice, can't think why though. Pete.

[redelephant]

This is excactly what I am saying. Someone stated the amp would go toward class-A when overdriven, which makes sense if bias droops toward gnd, which in theory it does. Then someone mentioned the cap charging effect also exaggerating the effect. I claim if cap charging (blocking dist) is ocuring the opposite will occur, since the charged caps will present a bias voltage that goes away from gnd, thus biasing the amp colder, into class-B or even C.

I see the bias staying rather steady even tho the tranny droops b/c of the filter capacitance being large enuff to keep it steady. The cap in the bias circuit is usually large compared to the current drawn from it so it holds the voltage pretty constant even if we get some droop. If the droop is over a long enuff period of time however, the bias should eventually shift towards gnd.

I am about to finish modifying my little 6V6 PP job and have included a SS rectifier option. Will do some testing this weekend. (Tho I use transformer coupling to the output tubes, so can't test the cap charging effect)

[guitarmike2107]

I see the bias staying rather steady even tho the tranny droops b/c of the filter capacitance being large enuff to keep it steady. The cap in the bias circuit is usually large compared to the current drawn from it so it holds the voltage pretty constant even if we get some droop. If the droop is over a long enuff period of time however, the bias should eventually shift towards gnd.

What I was trying to discuss has nothing to do with the bias circuit excursion, I got that wrong, I wrote this above to correct myself.

From reading a couple paragraphs quickly he (Kuenhel) says the grid bias voltage rises, my understanding of what he says is thats its due to the grid current through the grid load resistor effectively placing Rg in parallel with the bottom resistor from the bias circuit. Thereby reducing the voltage divider and causing a bias voltage to rise. In his example grid voltage raises from –34v to –28v

This is still not taking any power amp sag into account

I am about to finish modifying my little 6V6 PP job and have included a SS rectifier option. Will do some testing this weekend. (Tho I use transformer coupling to the output tubes, so can't test the cap charging effect)

Actually I was just thinking that measuring it on real amps would be the most straightforward way to prove anything.

[pdf64]

Ahh, I think there may be some confusion (by me at least) with how 'bias' is used, in that it seems to be used to cover both bias voltage and idle dissipation, also adding to the confusion is the negative polarity (right word?) of bias voltage.
So, if bias is said to increase, does that mean that the bias voltage increases or the idle dissipation increases? If bias voltage increases, does that mean that the magnitude of it's negative voltage increases, or that it becomes less negative, heads in a positive direction towards 0V?
To my way of thinking, 'bias' has to mean bias voltage, the C voltage supply. It would be illogical to think of it as meaning idle dissipation when the amp isn't idling, eg at full power output, the idle dissipation decreases!
And to increase the bias means that the magnitude of the bias voltage increases, becomes more negative, eg increases from -27V to -30V. Pete.

[guitarmike2107]

ha ha good point.

In this case (I think) we are talking bias voltage and variation of this voltage.

Apparently the bias voltage doesn’t change much until overdriving of the power amp / grid conduction etc start

For me increase means addition, so -27 + 1=-26 , like Redelephant has been saying "moves towards ground"

If you think about this in terms of the load lines it’s effectively pushing the load line up the graph.

The dynamics of what I am trying to wrap my head around is how bias excursion and power supply sag work together. i.e. if the bias voltage increases from -28v to - 26v when overdriving the power stage, which effectively moves the load line up the graph and the operation closer to class A, now if there is power supply sag and at the same the same time you have power supply sag which pushes the grid curves down and effective lowers the load line moving towards to class B.

It becomes evident as to how important the design of the power supply is in terms of getting the dynamics of the overdriving power section the way you want it.

[Steve Conner]

You have it the wrong way round. The bias excursion from grid current pushes the bias colder. To avoid the above confusion with "increasing" a negative number, I always use the terms hotter and colder instead, referring to more or less idle current. Colder means an increase in bias voltage magnitude, ie from -27 to -28.

Power supply sag should also push it colder. The screen voltage will sag more than the bias, because the screen voltage comes from the main HT winding, but the bias voltage comes from a separate winding. And, screen voltage has the same effect on bias current as grid voltage, in a proportion equal to the tube's "inner mu".

So to me, both effects point in the direction of "colder", towards Class-B. That is why a crossover notch appears under heavy overdrive.