Transformer Phase Inverter?

[Rhodesplyr]

I've been looking at the schematic for a Gibson GA-20 RVT guitar amp, and I'm puzzled by (at least) one aspect of the design.

It uses a transformer phase inverter driven by one 12AU7 section, but it appears to me that the secondary of the PI transformer is loaded only by the stray/interelectrode capacitance of the 6BQ5 control grids. Put another way, it seems to me that, in terms of reflected impedance, it would look like almost an infinite load, raising the question of what kind of plate load the 12AU7 is seeing on the primary side.

As a Hammond tech, I work with transformer phase inverters all the time, but there are usually grid leak resistors on the driven tubes, providing a defined load resistance.

In terms of the GA-20, is it a matter of bad design, or is there something about transformer phase inverters I don't fully understand?

ga20rvt.pdf

[martin manning]

Interesting question. As I recall from other discussions about this amp, the impedance ratio of the transformer is not too far from 1:1, but I think that the losses in the transformer will reduce the effective load resistance to something less than the impedance seen at the grids of the power stage. It'd be interesting to make some measurements on one of these.

[MWJB]

The interstage transformer is centre tapped on the secondary, so each leg will see a dcr to ground from the output tube grid.

Fender Musicmaster Bass Amp, Gibson BR9, Gibson EH150, for example, are the same.

[martin manning]

The Musicmaster schematic shows a 15k resistor across the transformer's primary. On the Gibson GA-20, both sides of the secondary have the grid input impedance to ground in series with the DCR.

[Rhodesplyr]

The interstage transformer is centre tapped on the secondary, so each leg will see a dcr to ground from the output tube grid.

I had considered the DCR of the secondary windings, but my mathematical/circuit analysis skills aren't up to the task of calculating how the DCR of the windings would appear as an impedance on the primary side. I'll try direct measurement when I get my hands on the amp.

Also, I was wrong in stating that most Hammond power amplifiers had the input tube grids ground-referenced by resistors. Many of the earlier amps have no such resistors, relying on the grounded center tap of the preamplifier's output transformer, connected to the amplifier by a 30-40 foot cable. Later amps had blocking capacitors and resistors, possibly to avoid damage should someone plug a Hammond tone cabinet into an organ that had had a Leslie kit installed. A few Hammond amps (like the PR-40) used their own DC control voltage scheme to switch reverb on and off.

[R.G.]

I had considered the DCR of the secondary windings, but my mathematical/circuit analysis skills aren't up to the task of calculating how the DCR of the windings would appear as an impedance on the primary side.

The resistances can be modeled as a discrete resistor in series with the primary and secondary. These resistor transform to either primary or secondary just like a secondary load does. So if the transformer has turns ratio Np to Ns and DCRs of Rp and Rs, then the total resistance of the windings appears in the primary as a series resistor of Rp + (Np/Ns)*(Np/Ns)*Rs .This resistor looks like it's just inside the primary lead wires where you can't get at it.

You can also equivalently refer it to the secondary, where it appears as Rs+ (Ns/Np)*(Ns/Np)*Rp .

[martin manning]

RG, in this case the secondary circuits each have the power stage input impedance in series with the DCR of the winding as the secondary load. Since that is quite high, that means that the reflected impedance seen at the primary will be very high, no?

[R.G.]

RG, in this case the secondary circuits each have the power stage input impedance in series with the DCR of the winding as the secondary load. Since that is quite high, that means that the reflected impedance seen at the primary will be very high, no?

Mostly, and at middle frequencies only.

On the primary side, the signal source sees the transformer as
- a series leakage inductance composed of the primary leakage plus the reflected secondary leakage
- a series resistance as noted in my previous post
- a capacitance to ground formed from the primary shunt capacitance and the reflected secondary shunt capacitance
- a capacitance to the actual (not reflected) secondary load composed of the primary to secondary winding capacitances through the transformer
- a nonlinear resistor across the primary made up of the nonlinear exciting current for the core and the core's eddy current losses
- a big and nonlinear primary inductor
That's just the transformer - and simplified at that.
Then the secondary load, whatever is connected outside the secondary leads, is reflected to be in parallel with the primary side inductance, but this is inside of the primary side leakage inductances.

At mid frequencies, the leakage inductances and capacitances are too small to be very significant. The primary inductance and core loss resistances are too big. This lets you ignore them at mid-band and only worry about the wiring resistances and reflected secondary loading.

If you never drive the grid on the secondary positive, then yes, it's a high impedance and the wiring resistances don't have much effect.

However (you knew I was going to say that, right? ) if you put a big enough signal in to drive the grid on the secondary positive, it starts conducting and changes from a nearly open circuit to a several-K nonlinear resistance. That's reflected to the primary as it happens, too, and the transformer suddenly has to start driving current. In fact, one way driver transformers are used is to step down the drive signal voltage to be able to drive more current into a grid when it goes into conduction in Class AB2.

At low frequencies, the impedance of the primary inductance gets low enough that it starts shunting primary current away from the reflected secondary load; the main determiner of the low frequency response is the relative size of reflected secondary load and primary inductive impedance.

At high frequencies, the leakage inductance reactances get big and impede current from going into the secondary load. The shunt capacitive reactances get small, so they shunt current away from the secondary load. The leakage inductance is the biggie for falling high frequency response. The combination of the leakage inductances and shunt capacitances can cause resonant peaking at the high end as well.

In the middle, it's the resistances and any grid current.

[martin manning]

Excellent! Thanks very much for the detailed response.