SE KT66 amp - which OT?

[Pete1950]

Which SE OT best matches a KT66 with regards to primary impedance?  2.5k, 5k, anything in between?

[kagliostro]

B+ 250v 2.2k load

http://tdsl.duncanamps.com/show.php?des=KT66


Franco

[PRR]

Pentode SE optimum load depends entirely on Voltage and Current.

You have a w-i-d-e range, limited mostly by tube ratings.

Abusing a 6550, I could work from 316V @ 125mA and 2.5K load, to 447V @ 88mA and 5K load, to 632V @ 63mA and 10K load. All give 40W Pdiss. Power output was the same within 1dB.

KT66 at 25 Watts Pdiss would run maybe 80% of these voltages and currents for the same loads. .

So: what voltage do you like? Then, can you find a suitable OT?

Over 440V makes e-Caps double-up. Even 100mA is hard to find in a lo-cost OT. Unless you found some odd parts, this narrows you to say 400V raw (360V across the tube after smoothing and bias) at 70mA, 5K load. Expect ~~10W near-clean audio output.

[Willabe]

Abusing a 6550, I could work from 316V @ 125mA and 2.5K load, to 447V @ 88mA and 5K load, to 632V @ 63mA and 10K load. All give 40W Pdiss.

Do all 3 set ups sound the same? Or should I ask; do they all shift the same along the load line? 

[shooter]

Do all 3 set ups sound the same

I'll leave sound to you guys , however, your plate volts sorta determines how BIG a signal you can make before it distorts, so a plate of 250 will give you clean *sound* of maybe 200Vp-p?  where 400 plate will give you clean *sound* of maybe 340Vp-p

[silverfox]

Do all 3 set ups sound the same

I'm going to guess, and that is the word I intend, guess- They could all sound the same depending on the bias and grid voltages involved as long as the output impedance remained the same for each comparative case. It should be proportional.

Wondering,

silverfox.
 

[HotBluePlates]

Do all 3 set ups sound the same

... your plate volts sorta determines how BIG a signal you can make before it distorts, so a plate of 250 will give you clean *sound* of maybe 200Vp-p?  where 400 plate will give you clean *sound* of maybe 340Vp-p

Yes, but this ain't a preamp tube (voltage amplifier). Where changed supply voltage and OT primary impedance cause the voltage swings get smaller, the current swings get bigger. So you get the same output power.

Which is why PRR asked what the supply voltage will be (tube, OT and PT are not selected in a vacuum, but are balanced among each other).

[shooter]

the voltage swings get smaller, the current swings get bigger

I get how power would stay the same, but wouldn't the *swing* dictate sound?, if I'm making a square wave at 200 V-plate at 30w, wouldn't sound be different than a clean sine wave at 300V-plate at 30watts

[HotBluePlates]

Explain to me how the size of the voltage swings changes the waveshape, given that in either case the voltage swing is within the capabilities of the power supply voltage. If you do that mental exercise, you'll have your answer.

[shooter]

you'll have your answer

I understand a clean sin wave at 200 will sound the same as a clean sin wave at 300, yet louder.
what I was doing, probably wrong  , was holding the PA drive constant - call it 50.
50 into the PA, with plate at 300 = 30w clean to the spkr, now drop the plate to 200, still 30W but it's a square wave because it swings to the PS max.
I'm Still on the learning curve

[HotBluePlates]

So there's 2 assumptions there you might not realize: drive into the output tubes stays the same, and power supply voltage stays the same.

What PRR was getting at (and what I'm emphasizing) is that the ideal OT impedance (and therefore voltage swing) changes if you change the supply voltage but wish to keep the same power output.

Got 300v? Load impedance goes down and current goes up to get the power output. Got 600v? Load impedance goes up to keep current swings small while staying class A. The unstated assumption (almost a rule) I'm making is that single-ended (class A) idles at a given current and with input signal swings up to double-idle current and down to just-touch no-current (o at least come close).

You already know intuitively that at higher supply voltages you need a bigger bias voltage to keep the current from overheating the tube. You might not realize that a basic assumption is that the incoming drive voltage doesn't cause the grid to go beyond 0v at any instant, so no positive grid drive and no heavy grid current. So if our 2 output stages above have bias voltages of -20v (300v B+) and -45v (600v B+), the 300v amp doesn't get a drive signal bigger than 19-20v peak while the 600v amp can accept a 44-45v peak drive signal (and will likely need all that). Sure the 600v amp's plate voltage swing will be larger, but its grid drive is also larger meaning the output tube's voltage gain stayed essentially the same.

So we wouldn't just change the OT impedance arbitrarily while keeping the supply voltage the same, nor would grid drive be the same in each case. The lower B+, lower impedance power section would also have a smaller bias voltage and easier grid drive for full output power. That might matter for your preamp design...

Again, this is why I said:

... tube, OT and PT are not selected in a vacuum, but are balanced among each other ...

[PRR]

Power is Voltage times Current.

My 12V tractor starter motor delivers much more power than my 230V A/C motor, because it sucks boat-loads more current. (12V@200A vs 240V@5A)

My 60V welding transformer can do more damage than my 30,000V furnace spark transformer (100A vs 0.030A).

Transformer "impedance" selection interchanges voltage and current.

Getting a lesser plate swing at more current, then using a lesser step-down ratio, tends to give about the same Power. You don't hear plate swing, only speaker swing.

In audio SE design, the *real* limit is always Pdiss, plate dissipation. And over a wide range of V and I giving the same Pdiss, we can get about the same speaker Power (around 40% of Pdiss).

I may be cross-threading here. A Power Pentode just sitting there does NOT have an optimum load. And its internal resistances (both of them) are very-far from any high-power point. Indeed that is why a pentode makes more power than a power triode. We can select a load which is very far away from either internal impedance. In most power pentodes, the spread of the two tube impedances is so wide that we may select many different loads with hardly any difference.

TWO internal impedances??

Yes. KT66 data is partial; take 6V6 on GE 1957, page 4, bottom.

(Ignore grid-lines above Ec1=0V; we don't go there in audio.)

The conventional "plate resistance" is the slope of the near-horizontal lines. Taking the Ec1=0V line as typical, from Vp= 75V to 375V the plate current rises from 100mA to 115mA. 300V change makes 15mA change, so Rp= 20K ohms. If I take a narrower swing at a lower current I see slope nearer the 50K the sheet says.

But the "other impedance" happens below the knee. Plot from 0V 0mA up the slope of the below-knee lines. 20V and 80mA suggests 250 Ohms. If you plot a lot this runs 300r to 200r for 6V6.

For best power transfer we want a load much higher than 250r, and much lower than 50K or so.

If "much" is "3", then we want over 750r and under 16K. That's a 21:1 zone!

What we really do with "good" (minor parasitics) devices is match the power stage V and I ratio to the load impedance. The numbers I gave above do that. "316V @ 125mA and 2.5K load"; 316V/125mA is 2,528 Ohms. Franco cites 250V 85mA, 2200 load. This actually computes to 2.9K. However once you get close, you may find that a slightly different (often lower) load gives "better numbers" (higher Watts and lower THD). But the "lower THD" is often higher 3rd harmonic (and low 2nd harmonic), which is lower on the THD meter but higher (harsher) on the ear.

Look at HK257, page 3, Class A for Audio, 1 tube operation. You have your choice of 1000V 75mA at 12K load or 500V 150mA at 2.6K load.... 75 Watts either way. Double the voltage, half the current, 4X change of load, no real difference.

Well, differences. A good 12K transformer is a lot of fine wire, 2.6K winding is cheaper. 1000V supply needs more heroic insulation (but less-fat conductors). However in H-K's market (big radio!), a designer may find 1,000V laying around left over from a bigger stage; or may find the lower grid drive voltage more convenient. As a "video" (extra fast audio) amplifier, where transformers are problematic and direct-drive is common, he may need the large swing of a 1,000V stage.

[HotBluePlates]

... In audio SE design, the *real* limit is always Pdiss ...

I'm not sure, but I think we just got P-diss'ed... 

There ya have it folks, one to save for the files. Thanks PRR!

[shooter]

but its grid drive is also larger

That's where I was coming from, *will it sound the same*, if you redo, impedance, B+ AND drive.
Thanks all for your help.  Achieved and re-reading, twice

[HotBluePlates]

but its grid drive is also larger

That's where I was coming from, *will it sound the same*, if you redo, impedance, B+ AND drive. ...

They should sound the same, as long as the test isn't rigged to cause one version to distort more than the other.

Abusing a 6550, I could work from 316V @ 125mA and 2.5K load, to 447V @ 88mA and 5K load, to 632V @ 63mA and 10K load. All give 40W Pdiss. Power output was the same within 1dB.

I think PRR may have built a 6550 Champ a long while back. Maybe he can advise.

There's a problem though: We've got another assumption to make the design process workable, but it makes it so the amp doesn't sound the same all the time even at a single chosen B+, load & drive. That is, we assume the speaker acts like a resistor, but really the impedance above & below ~400Hz, where the impedance is the nominal value, is actually much higher. So you always have a varying load & varying reflected impedance in any guitar amp (though negative feedback can help squash that some).

[shooter]

built a 6550 Champ

That's where my *knowledge* came from, I built a 4 gain stage 6SN7 into a single 6550.
dialed the PA to 85%Pdiss, was quiet, so I plugged in and got the *sound* I wanted, think smooth jazz.
BUT, when I dimed the PRE vol, with the PA drive on 7ish, it sounded like a bad marshal wearing a mask, carrying a chainsaw! Swapped to KT88, about the same, put a scope on it and found my PA drive was clean but 96Vp-p!(self biased around 23), the signal at the 6550 plate had a triangle for the top lobe and a severely compressed bottom lobe. Re-did the pre drive (took out last gain stage and made it a CF), and all was good, nice smooth jazz and dimed, good breakup, 'old skool rock sound.  Re-dialed the PA, with less B+, more current, sound stayed about the same.  Re-dialed the PA with more B+ and current for the KT88, Pdiss at 38W and sent it off gigging!