Output transformer theory question

[gui_tarzan]

I'm trying to understand how impedance matching works with different types of tubes and OTs.

The attached chart is self-explanatory but I came across some OTs that were designed to run a pair of 6GW8s in push-pull on each side in a stereo and I would like to use them with different tubes. The plate resistance (according to tube data) is 48k ohms and the transconductance is 10k umohs. A 6V6 shows a plate resistance of 80k and TC of 3750.

The chart lists OTs of 8k ohms in a PP 6V6 configuration and 5k in a SE config. So putting the tubes in PP effectively almost doubles the primary impedance load? If that's the case are the OTs from this radio supposed to handle a 20k primary load or am I looking at this wrong?

[sluckey]

I came across some OTs that were designed to run a pair of 6GW8s in push-pull on each side in a stereo and I would like to use them with different tubes.

Here's a real world example...

The Hammond AO-44 amp uses a pp pair of 6GW8s. The Hammond AO-43 uses a pp pair of EL84s. Both amps use the exact same OT with an 8Ω speaker load.

Just based on that info I have used that OT with 6GW8s, EL84s, and 6V6s. I would also use it with 6AQ5s or any other pp pair that wants a 6K to 8K load. And I would use your OT with any of those tubes also.

[Stevem]

Well you could run the 6v6s on those, but here is a basic rule of thumb.
If the ideal spec for the 6v6 is 3750 you can go up in OT impeadance by a factor of 1.5 to 2 times and things will work, but you will drop off 60 to 70% of the output power you would have had for less distortion.

[tubeswell]

The optimal reflected load for a tube is related to power and potential. The formula for impedance for centre-biased Class A operation is

Z(out) = Va/(Pa/Va)

where:
Va = the plate voltage (at idle)
Pa = the maximum power dissipation rating of the tube

So for your typical 12W 6V6 idling at 340 Volts, this would be 9k6.

If you run the tube at a lower load everything will run hotter and this increases the chance that components will burn up more quickly. If you run the tubes at a higher load, the tubes will have to work harder to make the same output.

However ^this^ is an oversimplified response. In reality impedance reactance is different for different frequencies so what might be optimal for one band of frequencies is sub-optimal for another. Most output transformers in guitar amps reflect a lower than optimal load overall. A lower load puts the load line above the knee of the grid curves and in a tetrode or a pentode this makes the response more linear and gives better fidelity.

Also the classic fender amps have smaller than ideal output transformers which saturate bass frequencies sooner, so you can get away with a lower load without it harming your speakers when you are driving everything at full bore.

[gui_tarzan]

We all know Leo occasionally pushed tube voltages far over what the original specs called for, the GE data sheet shows a max of 315v in single tube class A, 285v per tube in a pair for class AB.

According to my calculator, 340v at 12w would be:

Z(out) = Va/(Pa/Va)

(Pa/Va) = (12/340) = .035294

1.038062 = 340/(.035294)

So if I ran a pair of 6V6 at 285v at the rated 7w (on the data sheet) I get:

7/285 = .024561 / 285v = 8k6

I did the calculation from the Marstran website on my transformer in question and it comes out to 9k5 so it would be safe to run these OTs on a pair of 6V6s, correct?

By the way, I HATE math, especially formulas.

[martin manning]

Measuring the OT was the right thing to do- you now know exactly what you're dealing with.

9k5 will be fine for a pair of 6V6 up to 400V, and beyond if you want to push them. Just bias accordingly. Screen voltage at 85% of the plate voltage would be nice for running the load line through the knee of the Vg1=0 curve.

Tubeswell's optimum impedance formula is for SE, BTW. Have a read through Valve Wizard's SE and PP articles to get a handle on this.

[pdf64]

Most output transformers in guitar amps reflect a lower than optimal load overall. A lower load puts the load line above the knee of the grid curves and in a tetrode or a pentode this makes the response more linear and gives better fidelity.

I agree that (following Leo's lead) most guitar amps use a load which seems lower than optimal (ie that which gives max power?).

I'm basing that on bench tests, incrementing the secondary load whilst monitoring the waveform and load Vrms, and also from the interactive valve datasheet tool http://bmamps.com/Tech_tds.html

However, I can't see how a lower than optimal load makes the response more linear etc; could that be clarified?

[JMFahey]

Saying such and such tube loves such and such impedance with no further clarification is a blanket statement as accurate as : "if you play Country you MUST use a Strat or Tele; if you play Jazz youMUST use a LP or 335, the opposite is IMPOSSIBLE"

Many will argue: "but the tube datasheets suggest certain impedances ... are you saying that they are wrong?"

Not at all, I'm saying read the full datasheet specs, not single ones.

They will say, for example,
6V6GT
Class A1 amplifier
Load resistance 8500 ohms
Maximum signal power output 5.5 Watts

so people read a couple dozen datasheets and compile tables such as posted above, or simply repeat "6V6 love 8k loads" .

Fact is, that 8k5 mentioned is true ... if all other parameters are respected, such as:
Plate voltage 315 Volts
Screen voltage 225 Volts <-- nobody never ever respects this
Grid number 1 voltage -13V
Zero signal plate current 34 milliamperes.

If some of these parameters are altered, optimum load impedance also changes.

The proper way to calculate it is to use the tube plate curves, also supplied in the datasheet, using the actual values used in that amp and using whatever results you find.

Any other, is just a more or less educated guess.

EDIT: somebody said that MI amps in general use load impedances lower than what's usually seen in datasheets.

True, but nobody offers an explanation ... although there must be some justification, most classic amp designers can't just have committed the same "error", simply by chance.
Plus we know that although many were not Engineers, many not even Techs, they were certainly no fools and must have tested their amps to death, so .......

Fact is that all (maybe out of convenience or simplicity or cost) run screens at almost +V plate voltage (with just a few volts loss in the choke or filter resistor) which in general is quite higher vthan suggested in the datasheet.

That alone increases available plate current which of course lowers optimum impedance.

This applies to over 90% MI amps out there.

[pdf64]

Doesn't the constraint of plate dissipation make low load values a problem?

It may be that 'lower than optimal for max power' load values help to deliver more power to real speakers?

most classic amp designers can't just have committed the same "error", simply by chance

There's the possibility that popular / successful designs were blindly copied, without much verification of the validity of the original design rationale (even if that could be ascertained).
The dreadful kludge of standby, hot switching a capacitive load on tube rectifiers (amongst its many sins) being a possible example of this.

[tubeswell]

However, I can't see how a lower than optimal load makes the response more linear etc; could that be clarified?

Merely that the grid curves are less linear below the grid curve knee, therefore a steeper load line (above the knee) will have more-even spacing between the grid curves on the load line.

http://www.ampbooks.com/mobile/classic- ... -grid3.png

[pdf64]

Thanks!