LTP bias Q

[g1]

That's a nice new avatar LT, is there triode-states and solid-states now?

Could someone please address the question in post #40 ?

[Dave H]

Could someone please address the question in post #40 ?

The schematic is correct. The diodes prevent grid current and therefore screen current surge.

[g1]

Doh! I was thinking "how can the signal swing more than .7V then?" but I was thinking in respect to ground when I should be thinking with respect to the -DC bias voltage.

[Chuck H]

Doh! I was thinking "how can the signal swing more than .7V then?" but I was thinking in respect to ground when I should be thinking with respect to the -DC bias voltage.

Not sure why it wasn't explained as simply as that already. I missed that too. Probably because most of this thread has been about amps with a 0V grid and there's no voltages on that schematic. You need to follow the grid loads back to the bias supply to see that it's a bias supply. The grounded power tube cathodes might have been a tip off. Doh!2

[teemuk]

Oh, I don't know if it was properly addressed in this thread but generally "overdriving the grids" has also two other distinct effects in addition to bias voltage shift / clipping.

1. When we're pushing one device "fully on" the other one will be pulled "fully off". In reactive circuits (e.g. transformer coupling to reactive load) such sudden seize in load current flow will trigger resonance in the system, which manifests itself as a giant voltage surge. Current halts = a surge voltage transient is developed in the resonant circuit.

This is the one that arcs through tube sockets, PC board tracks and et cetera.

It gets worse when the operation drifts to an area where both output tubes are off simultenuously.

And this is why such surge voltages are not only interlinked to power tube clipping in general, but also to behaviour at crossovering regions. This is why everything that improves overdrive behaviour on this regard, will also improve behaviour in the crossovering regions.

2. Grid conduction will ramp up screen current draw. Consider that most guitar amps are designed in fashion that screen current draw can and WILL modulate also screen voltage. Unless we have a fairly low voltage (in guitar amp standards) and regulated screen supply (hard luck to find one from a guitar amp) the screen voltage modulation will gain compress the signal leading to further increased anounts of clipping distortion. In general, we can expect both screen and plate voltages to decrease, which would then indicate the tube needs less negative bias. The DC shift in bias, on the other hand, provides more negative DC bias.

A sustaining overload condition will therefore be pulling the power tubes towards worse and worse operating conditions.

So we have a tube that's simultenously pulling itself towards an approaching failure via excessive screen dissipation and another tube that disrupts current flow and causes high flyback voltage surges. We can see that it's no longer just about bias circuit at the grid, it's about interactive operation of the power tubes, transformer coupling, and load in various different mechanisms.

So it's at least

- DC bias shift at grids
- Kickback voltages
- Screen currents

we need to care about if we want to fully examine this issue. And then it's not just the bias circuit any more. This same reason explains why different output transformers, different speakers, zobel networks/ conjunctive filters, catch diodes, screen circuits, etc. are all as relevant to this discussion as grid circuit's properties in general. Lot of different phenomenons interact in this one and it's not just a single parameter that effects the performance of the entire system.

[R.G.]

1. When we're pushing one device "fully on" the other one will be pulled "fully off". In reactive circuits (e.g. transformer coupling to reactive load) such sudden seize in load current flow will trigger resonance in the system, which manifests itself as a giant voltage surge. Current halts = a surge voltage transient is developed in the resonant circuit.

Well, actually, one-side-off happens well before the start of grid conduction in a Class AB amplifier. That is what AB is all about - one tube turns off after the opposite side turns on, and there is overlap. The degree to which these do not sum to the exact signal is what crossover distortion is.

So the one-tube-fully-off condition happens all the time. It's no worse when the on tube is driven further on with grid conduction. The off-side tube was off a long time before in the half cycle.

This is the one that arcs through tube sockets, PC board tracks and et cetera.

I'm not sure that's the case. What arcs tube sockets and PCB tracks is the current in the leakage inductance. The fully-coupled inductance of the primaries cannot show inductor flyback, because all the turns are coupled through the M-field in the core, and are forced to a constant volts-per-turn. It's the uncoupled-by-definition leakage inductance that causes arcing.

It gets worse when the operation drifts to an area where both output tubes are off simultenuously.

Well, parasitic ringing certainly gets worse when that happens. The question there is how much energy is left in the M-field to be coupled out to damage things.

All this is very familiar to me from switching power supply design, where the core is left un-driven for significant times during every cycle. The approach there is to snub it. And both-sides-off is what AB biasing is supposed to prevent. If it works.

And this is why such surge voltages are not only interlinked to power tube clipping in general, but also to behaviour at crossovering regions. This is why everything that improves overdrive behaviour on this regard, will also improve behaviour in the crossovering regions.

I think it's more complicated than that.

we need to care about if we want to fully examine this issue. And then it's not just the bias circuit any more. This same reason explains why different output transformers, different speakers, zobel networks/ conjunctive filters, catch diodes, screen circuits, etc. are all as relevant to this discussion as grid circuit's properties in general. Lot of different phenomenons interact in this one and it's not just a single parameter that effects the performance of the entire system.

I fully agree with that. Mother Nature is **complicated**.

[loudthud]

If there was an inductive spike from the output transformer, you would see it with a dummy load although it wouldn't be very big. The inductive spikes that arc tube sockets and kill OTs come from speakers. Look at the waveforms in the zip file in this thread: http://music-electronics-forum.com/t35493/ Hint: The amp doesn't have an output transformer because it's solid state.

The scope photo below shows the plate Voltage of a 6L6 swinging 600V below ground (1000X probe).

[nickb]

If there was an inductive spike from the output transformer, you would see it with a dummy load although it wouldn't be very big. The inductive spikes that arc tube sockets and kill OTs come from speakers. Look at the waveforms in the zip file in this thread: http://music-electronics-forum.com/t35493/ Hint: The amp doesn't have an output transformer because it's solid state.

The scope photo below shows the plate Voltage of a 6L6 swinging 600V below ground (1000X probe).

Interesting about the speaker inductance. I had never considered that before. Did you actually do the same thing using a dummy load for comparison? The leakage inductance of the transformer will have an effect too and it would be interesting to see the true relative magnitudes.

[nickb]

I hate to go tete a tete, especially since my neurons are still entangled by all the quantum physics malarkey the other day. But..here's the rub, I did simulate it before I spoke and found that it does seem to work:

[ATTACH=CONFIG]34644[/ATTACH]

The average DC level on both sides of he cap is the same aside from a constant offset so there is no net charging of the capacitor.


I think...

Well, for the record, it turns out I thought wrong. The sim was perfectly correct but I had mis-interpreted the results...it happens, I guess.

[loudthud]

Interesting about the speaker inductance. I had never considered that before. Did you actually do the same thing using a dummy load for comparison? The leakage inductance of the transformer will have an effect too and it would be interesting to see the true relative magnitudes.

When you compare a dummy load in the setup used to create the zip files, you just see a straight line because you are only looking at the load. To see the effects of an output transformer's inductance, you need to look at a tube with a resistive load connected. In the attached pic the left shot just shows how things are setup relative to the plate curves you see on a data sheet. Plate Voltage and Cathode current are being monitored. The next pic shows the bias point. The last two show a resistive load as seen on the primary of the transformer. The signal source is a guitar so inductive effects don't show up at all frequencies or drive levels. The amount of looping (effect of inductance) in the last pic is about the most I saw in that setup. I'll have to look for the source file.

I might have overstated things in my previous post. You can see the effect of transformer inductance, but it is quite a bit less than that generated by a speaker. That might be because you don't want to look at a transformer without some kind of load and a resistive load dominates (dampens) the inductive spikes that the OT generates.

[loudthud]

This zip file may not be the exact source of the above resistive load snaps, but it was done at the same time with the same setup. You can hear me banging on the strings of the SG used as a signal source.

[nickb]

Nice data LT. Thanks. Good to know the theory is right after all!