Output tube impedance

[DaBreeze]

Is there a chart somewhere showing the output impedance for various power tubes in push-pull. I've seen various differing opinions in other forums, just would like to have a chart for reference on future projects

Thanks

D

[MWJB]

Common values are:

6L6/KT66/6550/KT88/KT90 - 4K/tube nominal for Fender (though the 3x10 bandmaster ran 2.67K), 3.6K to 4.2K are typical, sometimes as low as 2.5K, maybe up to 5K+ for an amp that has a compromise primary Z & can run 6V6 as well. Some old PA amps ran 7.5K. I'd normally go with the Fender value as this gives some leeway with ext cab jacks. Halve the speaker load with a 2.5K primary and you can wave goodbye to your power tubes!

EL34 - typically a shade lower than 6L6 (3.5K to 3.6K) but to all intents and purposes interchangeable with 6L6.

6V6/EL84 - Centre value might be considered around 8K/tube, some old PA amps ran 4.7K, early low voltage SE amps might be as low as 5K-5.5K, BF Deluxe is around 6.6K, up to 8.5K at the higher end for Princeton/Tweed Deluxe.

In short, stick to 3.6-4.2K for the bigger tubes & 6.6-8.5K for the smaller tubes if you intend to wire the amp Fender style, with parallel ext. jack, fed by the same OT secondary - an amp set up like this will normaly tolerate halving or doubling of the matched speaker load. If you swing to more extreme hi/lo primary Z, make sure you use a multitap OT & always match speaker loads.

[Old Tele man]

...footnote at the bottom of page 188 in F. E. Terman's ELECTRONIC & RADIO ENGINEERING, 4th ed., 1955, McGraw-Hill:

"To the extent mu (µ) is constant and that Eq. (6-10) is true, the plate resistance depends only on the plate current Ib, and is inversely proportional to the 'cube-root of Ib'."

Eq. (6-15) rp = (2/3)*mu / (Ib*K^2)^(1/3)

where:
rp = tube dynamic plate resistance, ohms
mu = tube triode amplification factor, dimensionless
Ib = plate current, amps
K = tube constant Perveance, amps-per-volt^(3/2)

[R.G.]

The short answer is that yes, there is a chart somewhere that does this. I don't have it, and it may well not be accurate, or not all accurate.

First of all, if I don't misunderstand you, you're interested in what plate load to apply to the plates of a pair (or two, or three pairs) of output tubes. There was a large amount of work done on what was the "best" load for a pair of output tubes back in the Golden Age of Tubes. They came up with answers, but the answers weren't simple.

If you had a magic output transformer which could dial in a load on a pair of output tubes while you drove it, and you looked at things like maximum output power before distortion and the curve of distortion versus output power level, you could make charts of output versus loading and distortion versus loading, so you could pick off the "best" loading points. That was done back in the Golden Age, and I've seen a few of these charts.

For power pentodes (like the EL34) and power beam tubes (like the 6L6) there is a broad maximum of power output at a specific loading level. For 6L6s this is about 4K to 4.4K plate to plate. For EL34s this is at a higher loading. However, the "peak" in power output is not really a peak - it's a broad hump. Missing the magic maximum only loses you a little power output, and your ears do not hear all that much change in output level. So it's entirely possible to use the same loading for 6L6 and EL34 and not miss much in loudness.

But then there's that other curve of distortion versus loading. It turns out that the distortion before clipping starts is minimized at some point of loading. This loading is not the same point as maximum power. For the 6L6, it's up at about 6.6K plate to plate.

The result is that hifi tube amps tend to (or tended to, back when people knew how to design amps with tubes, not easter-egg in parts) use the lowest-distortion loading, and guitar amps tend to use the highest-power loading.

[tubeswell]

...footnote at the bottom of page 188 in F. E. Terman's ELECTRONIC & RADIO ENGINEERING, 4th ed., 1955, McGraw-Hill:

"To the extent mu (µ) is constant and that Eq. (6-10) is true, the plate resistance depends only on the plate current Ib, and is inversely proportional to the 'cube-root of Ib'."

Eq. (6-15) rp = (2/3)*mu / (Ib*K^2)^(1/3)

where:
rp = tube dynamic plate resistance, ohms
mu = tube triode amplification factor, dimensionless
Ib = plate current, amps
K = tube constant Perveance, amps-per-volt^(3/2)

Yeah but what is "Eq. (6-10)"?

And how does that differ from "Eq. (6-15)"? - (Apart from "5"??)

[Old Tele man]

Yeah but what is "Eq. (6-10)"?

And how does that differ from "Eq. (6-15)"? - (Apart from "5"??)

Eq. (6-10) is the "Child-Langmuir 3/2's Law" for triodes:

Ip = K*(Vg + Vp/mu)^(3/2)

...which, with proper messaging also works for tetrodes and pentodes:

Ik = (Ip+Is) = K*(Vg + Vs/mu.triode + Vp/mu.tetrode)^(3/2)

where:
Ik = cathode current
Ip = plate current
Is = screen current
K = Perveance
Vg = control grid voltage
Vs = screen grid voltage
Vp = plate voltage
mu.triode = tube triode amplification factor (mu1)
mu.tetrode = tube tetrode or pentode amplification factor (mu2)

[DaBreeze]

A couple more points/questions

1) Contrary to what's been posted, the Weber site apparantly figures the impedance of KT66's at about 8k, as the shift the OT taps in their kits when a KT66 is used. Are they wrong?

2) When doing calcs, is it better to use 3.2 for 4 ohm speakers and 6.8 for 8 ohm speakers which seems normal when reading the speakers?

thanks

[Bruce / Mission Amps]

Common values are:...

6L6/KT66/6550/KT88/KT90 - 4K/tube nominal for Fender (though the 3x10 bandmaster ran 2.67K), ...

With respect to the 3x10 Bandmaster, your # of 2.67 is right but that would be secondary load in ohms not primary zed in Kilo-ohms.
Actually the tweed Bandmaster and tweed Super ran the same impedance OT... which was not 4K but between 6200 and 6500 ohms (depending on the year and all that rot) so the Bandmaster, at 2.67 ohms was around <4K35, very close to the 4K you mentioned above.
The tweed Pro also used a 6200/6500 Ohm OT but with an 8 ohm secondary.
To be honest, all those amps sound really good even with the wrong 4K output in them but they do sound a bit different and more classic with the right impedance.
What's really cool is you can use a softer rectifier tube and sub 6V6s right in those old tweed amps with no change to the OT or speaker loads.

[Old Tele man]

...Bruce, you're referring to the original Triad/Stancor transformers and not the later Woodward-Schumacher transformers, correct?

[Wakculloch]

If you check out the KT66 data sheet, it calls for a 8K load, but this is for minimum distortion, so no it's not incorrect but max power calls for a load between 6 and 8K.
See here; http://www.mif.pg.gda.pl/homepages/f...086/k/KT66.pdf
Speaker impedance includes inductance, so it is higher than the measured DC resistance.

[soundmasterg]

It would be nice if there was a sort of tutorial out there in basic terms so everyone could get a handle on it that said something to the effect of

"If you use an OT with a primary impedance of 4k with a 6L6GC pair, and you change to a 6k6 load, THIS is what will happen to the sound. Or if you use an EL34 on those same transformers, THIS is what happens to the sound."
But of course everything changes with applied voltage and current and what not, so I know its not that simple......

Greg

[MWJB]

"1) Contrary to what's been posted, the Weber site apparantly figures the impedance of KT66's at about 8k, as the shift the OT taps in their kits when a KT66 is used. Are they wrong?" Marshall used 4K, to all intents & purposes KT66 & 6L6 are interchangeable in most aspects, except heater current draw (some current production KT66 even match that - check with mfrs notes).

2) Transformers work on AC impedance not dc resistance, so go with the nominal impedance rating. However ratings are just that "nominal" & float with freq response, also the discrepancy between DCR and AC-Z (20%?) won't be enough to cause damage in a middle of the road design.

[Old Tele man]

...the published speaker impedance values are "measured" at 1,000 Hz, which is just one point on their spectrum, so obviously actual values will be different, but "...at..." 1KHz they will roughly (within tolerances) produce their rated impedance. As the frequency varies, so does the XL reactance and thus the speakers impedance. 1KHz is just the standardized point of measurement.

...just remember, the DC-resistance (R) measurement of a speaker will typically be about 80% of it's rated AC-impedance (Z) value...give or take a little.

...and +100% on tube rp variance vs. (a) minimum distortion and (b) maximum power...commonly 10% or more apart in value, with a lower value yielding lower distortion and a higher value yielding higher power.

[MWJB]

"...the published speaker impedance values are "measured" at 1,000 Hz, which is just one point on their spectrum," - Not so. Just around the speaker's resonant frequency there is a spike in impedance, nominal impedance is usually taken at the foot of the trough following this spike, for many guitar speakers this will be somewhere around the 100Hz mark. Often by 1KHz an "8ohm speaker" is nearer 10ohms.

Not that I have idled away far too many hours gawping at speaker charts, or anything like that...:-). (Still no substitute for hearing them though).

There was a thread recently where the OP was querying AC voltages & primary Z on current Fender schems, it transpires that when you work out turns ratios that the 8ohm amps seem to be operating at 10ohms...suspiciously like running the test into a speaker rather than a 8ohm dummy load?

[hasserl]

On the old forum Speedracer (Joe from Obsolete Electronics) posted this as "Common Output Tranny Primary Z ratings"

Amplifier Primary Impedance

Marshall, 50W 2xEL34 3,400 Ohms

Marshall 100W 4xEL34 1,750 Ohms

Marshall JTM45 2xKT66 8,000 Ohms

Vox 30W 4xEL84 4,000 Ohms

Vox 15W 2xEL84 8,000 Ohms

Fender 50W 2x6L6 4,100 Ohms

Fender 100W 4x6L6 2,000 Ohms

Tweed-Spec Cathode Bias 6L6 p-p 6,600 Ohms

Matchless 15W* 2xEL84 4,000 Ohm

Matchless 30W 4xEL84 4,000 Ohms

Park 50W 2xEL34 5,000 Ohms