How to Bias SF Twin Reverb (135W)?

[matt h]

Everything? Hardly. But seeing multiple attestations from people who *do* know their shit... yes, I'm inclined to believe those until I hear a plausible explanation otherwise. I don't really have command of the terminology enough to explain it, but from what I do understand, it's not hogwash. For lack of a better descriptor, it's a net vs momentary-directionality issue. Where the net change, both plate and screen increase under load-- at a given moment, the current change directionality opposes. This is what I took it to mean.

[pdf64]

Bear in mind that the way a tube responds to test conditions on a curve tracer rig may not align with how it gets used in an amp, eg we wouldn't hold Vg1 and Vg2 constant whilst reducing Vp.
As used in an amp, Ig2 is related / proportional to Ia, so both are affected by Vg1, and as Vg1 tends to increase towards 0, the proportion of Ig2 to Ia tends to increase.

[gingertube]

There are 2 well recognized tapping percentages that are in common use for Ultralinear connection.
43% is what most HiFI guys use and is known as the "minimum distortion" tapping percentage.
20% (or 15%) is less common and it is referred to as the "maximum power" tapping point, it gives most of the benefits of the 43% tapping but at more power and a bit more distortion. Many smaller HiFI Amps using 6BM8 or similar tubes in push pull used the 20% tapping for best compromise between distortion and power out.

There are less common HiFi output tannies which use 50% tapping points.

There are also a variant where the Ultralinear connections are done by separate push pull windings - these are handy when you want to run screen voltages at lower voltage than the anode connections, these trannies tend to be expensive.

Note these are percentage of turns or percentage of voltage tapping points (same thing).

Some of the older papers (in particular the Fritz Langford-Smith papers) on Ultralinear use impedance tapping percentage. For those you need to take the square root of the impedance tapping percentage to get the voltage (turns) tapping percentage. This has lead to confusion in the past, particularly because a 20% impedance tap is pretty much the same as 43% turns (voltage) tap - make sure you are talking the same terms in any discussion.

Ultralinear is a particualr version of Distributed Load. There are others, most notable, those that distribute the load (primary windings) between anode and cathode circuits, The Quad HiFI Amps were of this type and amoungst those types there is also another specific version called Unity Coupling (50% of winding in the anode + 50% in the cathode) which was used in McIntosh HiFi Amps.

I use 43% turns taps in guitar amps with zero global feedback - this works well. I would not hesitate to use 20% but these output trannies tend to be not as common.
NOTE That as the anode voltage swings downward so does the screen voltage (and vv) which helps reduce the stress on the tubes and that is why the maximum voltage ratings for the tubes tends to be higher for Ultralinear connection than for triode or pentode mode. Ultralinear mode tends to be easier on the screens of the output tubes.

I use 10% cathode feedback + 33% screen feedback in HiFi amps, note that the screen voltage is with respect to the cathode so 10% cathode feedback effectively make s the 33% screen feedback into 43% Ultralinear.
These trannies:
http://shop.plitron.com/specs/2100CFBH.pdf
driven by a qaud of KT88.


Cheers,
Ian

[two tone]

schematic of the famous Quad 2 ,I have two of these,great tone and looks

[tubeswell]

For a pentode or tetrode in non-UL mode, under most signal conditions the grid curves are nearly horizontal so the sum of Ia and Ig2 is constant. For the most part, the proportion of Ia:Ig2 is constant at any given plate voltage (except when Va drops right down and the Ia:Ig2 ratio increases). So for calculating gain and gm, you can pretty much rely on tube current in a pentode/tetrode being constant. What's more, any series Rg2 modulates Vg2 causing a screen voltage swing in opposite phase to the plate voltage swing, and increasing Rg2 reduces gm (but increases linearity) under dynamic conditions (and that's a side debate, although it serves to illustrate what's happening with Ig2 in relation to Ia).

This changes in UL mode where hooking g2 up to a tap on the OT primary puts an in-phase signal on g2, which is proportional to the point at which g2 is tapped into the primary. Because Vg2 and Va are now in phase, tube current is no-longer constant. But then again UL mode is less like 'pentode/tetrode mode', and more like 'triode mode'. I could have missed some vital turning point in the thread, but then that would serve me right I suppose.

[matt h]

Tubeswell, thanks for chiming in! (and thanks for not being mad that I quoted you here hahah)

[martin manning]

For a pentode or tetrode in non-UL mode, under most signal conditions the grid curves are nearly horizontal so the sum of Ia and Ig2 is constant. For the most part, the proportion of Ia:Ig2 is constant at any given plate voltage (except when Va drops right down and the Ia:Ig2 ratio increases). So for calculating gain and gm, you can pretty much rely on tube current in a pentode/tetrode being constant. What's more, any series Rg2 modulates Vg2 causing a screen voltage swing in opposite phase to the plate voltage swing, and increasing Rg2 reduces gm (but increases linearity) under dynamic conditions (and that's a side debate, although it serves to illustrate what's happening with Ig2 in relation to Ia).

This changes in UL mode where hooking g2 up to a tap on the OT primary puts an in-phase signal on g2, which is proportional to the point at which g2 is tapped into the primary. Because Vg2 and Va are now in phase, tube current is no-longer constant. But then again UL mode is less like 'pentode/tetrode mode', and more like 'triode mode'. I could have missed some vital turning point in the thread, but then that would serve me right I suppose.

Since you say "the grid curves are nearly horizontal," you are looking at Ia and Ig2 vs. Va at constant Vg1 and Vg2. In fact the slope of the Ia curve is always positive and the slope of the Ig2 curve is always negative under those conditions (ignoring the kinky part), so the ratio of Ig2:Ia increases as Va decreases, and it increases dramatically to the left of the knee. Running with a screen resistor or distributed load compresses the curves at low Va by reducing Vg2, but the relationship still holds (you can see this in the traces I posted above). What you are missing is that Vg1 has to increase to drive Va down, so the operating point moves across the Vg1 curves. The screen current then goes from perhaps 5% of anode current at zero signal to more than 2x that figure when the signal peak reaches Vg1=0. Adding a screen resistor, or operating with a DL will reduce that effect but not eliminate it.

[pdf64]

What's more, any series Rg2 modulates Vg2 causing a screen voltage swing in opposite phase to the plate voltage swing

When you next get chance, please check this with an amp and twin channel scope.
It will only take 5 mins and you will find it very beneficial.

EDIT. I checked the OT of a F75.
With 1Vac sine into the 8 ohm secondary, there's 23.45V across the primary, 2.94V across the taps.
So ~4k4 primary with ~12.5% screen grid taps.

The amp is arranged for pentode operation.
In use, the signal at the screen grid and plate is in the same polarity, in opposition to the control grid signal.

[tubeswell]

EDIT. I checked the OT of a F75.
With 1Vac sine into the 8 ohm secondary, there's 23.45V across the primary, 2.94V across the taps.
So ~4k4 primary with ~12.5% screen grid taps.

The amp is arranged for pentode operation.
In use, the signal at the screen grid and plate is in the same polarity, in opposition to the control grid signal.

Yes my earlier statement was a bad choice of words. The Rg2 resistance modulates the voltage in g2, but the overall swing will still be in phase with the plate in UL.