E/PCF802 2W amplifier

[PRR]

> Regarding power rating limit, does it count maximum at peaks Ia*Ua or mean value of dissipation?

What is the tube's (any SE audio device's) dissipation on peaks?

How long does it take to heat-up a tube? (Yes, transistors are different.)

[rankot]

Those are anode dissipation graphs for both amplifiers:



Mean value for PCF802 dual output is approx 0.7W; for PL84 it is approx 4W.

[PRR]

> anode dissipation graphs

Hah! Keeping your computer busy! The curves look right but I can't read the numbers.

I could throw you a fish. But let's think instead.

At idle the dissipation is V*I, no power to load.

With signal, we see three distinct extremes. Plate pulls down, low V high I. Plate kicks up, high V low I. If V and I can be made near-zero, dissipation at these extremes would be *zero*. We can't build a tube big enough to pull to zero, but dissipation will be low at these extremes. In between it must pass through the idle point, a higher dissipation.

Here's just a power amp, twice. 250V 75mA(avg). For reference the input signal is shown below, 2 cycles of Sine.


TU2 has no input signal so feels 250V*75mA= 18.7W dissipation. Blue down-triangle trace. Dead-constant because no signal.

TU1 has a big input signal and is driven near max. Red diamond line. When input goes through zero the dissipation is (near) idle dissipation. When input is *either* plus or minus the dissipation falls to near 6W. The time-scale is milliSeconds and some tube-abuse will tell you that it takes much longer than that for a tube to heat/cool. It is reasonable to find the Average over the wave, which is about 12 Watts.

The output power is 370Vpp in 3K or about 5.7 Watts. Using round numbers:
Idle --- 18W diss, 0W out === 18W total
Roar -- 12W diss, 6W out === 18W total

The amplifier runs *cooler* when pushed LOUD.

Now you have to ask yourself.... do you play loud ALL the time? Never stopping for more than a few seconds?

Radio transmitters and motor-drivers might only work maxed-out. (And they need protective relays to cut-out if signal fails.)

Speech/Music audio "never" runs flat-out ALL the time. It gets boring. And you might drop your pick. Also beer-breaks and drum solos.

So if you design an audio amplifier for the lower dissipation at FULL output, it will melt if you ever stop.

You design SE power amps for IDLE.

[rankot]

Great explanation, PRR!

OK, I am really safe with my PL84 design, it will dissipate 9.68W on anode, which is below it's maximum of 12W. And I"m safe with my 5W 5k:8 transformer here, right? 

Dual PCF802 will dissipate 2.38W, which is too close to 2x1.2W, so I will have to alter this design. I recalculated everything and found that it is OK to have 150V for B+, 10k primary OTF, 2k4 for Rg2 and 100 ohm for Rk. It makes approx. 1.0-1.1W per tube, which is fine. All I have to do is to find 10k:8 SE transformer for cheap

[StarGeezers]

  Ran , have you run this up the flagpole at Wattkins.com yet ... ( toob amps )  This is a Great Pedal sight  !!!

[rankot]

No, I didn't. It's not easy to maintain topics on many sites, so only DIYSB is fine for now

[rankot]

I've found a very nice article about OTF calculation here: http://www.turneraudio.com.au/se-output-trans-calc-1.html

[Marcos - Munky]

It isn't too hard to calculate your own transformer. In fact, here's a nice calculator that does it pretty well. It's in brazilian portuguese, google translator do a decent translation to english.

[rankot]

Indeed, translation is really good.

This is what I get when I enter my params:


If using core with 1cm middle leg width, it requires only one sheet of metal? Is it normal, as I've never seen such transformer before. Does it mind if I use bigger core - I suppose no?

Also, it proposes 0.005mm wire for primary, isn't it a little bit too thin?

[Marcos - Munky]

That's strange, the translated page shows more parameters to insert than the original page.

Anyway, you should use something like 60Hz as the minimum frequency, 18000Hz it's the max frequency. Some transformer makers use 40Hz as the minimum frequency to get more bass signal, and some use 120Hz to offer a cheap transformer option. With a 60Hz transformer, you'll need a center leg of about 1.3 cm.

And the 40AWG wire seems fine for the primary, since the low power. I got 40AWG using the original page too. I suggest you to oversize the power a little, as a safe measure. This will also make the wires thicker.

I'll do the maths by hand later.

[rankot]

It seems it works better in Portuguese

[Marcos - Munky]

Here's the math:


Notice the results are somewhat different from the ones from the calculator. This is because different books use different parameters for the math. For example, the book which the calculator is based uses 7.5 for the core calculation, while the book I used for the math use 10. That results in a core a little bit bigger.

Also, you'll need to redo the math for the number of turns based on the core you'll use. Bigger cores means less turns, but also means a bigger transformer with more weight. You don't need to redo the wire gauge, since it depends on the transformer power and not on the transformer size.

For the value of B, grain oriented iron can stand bigger values before the core saturates. To get the maximum value you iron can handle, you'll need the datasheet of the iron. The guy who taught me how to do the math told me to use 10000 G as a safe value if you don't have more info on the iron.

Question: is the 2.24W power the audio power or the total dissipation (audio + heat)?

[Marcos - Munky]

It seems it works better in Portuguese

Yeah, I found that out too. There were extra parameters when using the calculator along with google translator. It probably messed up with the code.

[rankot]

I used 2.24W which is total anode dissipation (for two paralleled tubes in the circuit), and I know that real audio power is approx. 40% of that, but I wanted a little bit bigger core.

[PRR]

> it requires only one sheet of metal?

Parameter H, Height of stacked sheet: 1.35cm (in Portuguese).

Measure the thickness of your lams and figure how many to make 1.35cm (finger-width).

Half-inch (my measure) stack seems cheap but not unreasonable.

But foo on math. The SE audio transformer needs about 4X the iron of a straight AC transformer for the same frequency. A little bigger for low distortion. However guitar bass is higher than power frequency. So for 2.4W 82Hz audio, find a 10VA 50Hz power transformer and gut it.

The primary wire probably can be super-thin. This will not normally be stocked at a transformer rewinder (we rarely need small power at high voltage), but is common for guitar pickups.

[rankot]

PRR, since it is actually 0,9W of audio power (2,25W total plate dissipation), may I use 220/6V 3VA transformer in this case? It will loose some of the bottom end, but could be good enough for testing?

[PRR]

> it is actually 0,9W of audio power

Ah, you are correct. 6VA core is ample.

> It will loose some of the bottom end,

In principle, the bass response may be "the same" on a smaller core by winding more turns.

In practice, you are already near the smallest (and weakest!) wire you can easily buy. If the core is too small you won't be able to fit more turns.

Audio transformers usually get better the bigger they are. Not if you try to make a million for profit! But in DIY it is often affordable to be generous, and that may lead to a better, and easier-to-wind, design.

For example: most home heat/cool systems use a 24V 40VA transformer for thermostat and relays. To a heating contractor, these are like 100K pots to a pedal-builder: they accumulate under the bench. You may be able to get one just by asking. Or I can buy them at the home improvement store for $10.

Cowles says experimental SE outputs can be extra-large power transformers of appropriate ratio. An optimum design would have an air-gap, but if the iron is over-size it is not critical. Your 220V:6V 0.5A transformer looks to me like 16K:12 impedance. It may work as-is.

[rankot]

One question about winding interleaving - probably not important for such a small OTFs, but in case I decide to make something bigger and need to order custom made OTF - I saw few articles about that, and what I don't get is if it should be done always, or only if we need more than one secondary (4 and 8 ohm for example)?

And what happens with DC core saturation if I use simple power transformer for output?

[Marcos - Munky]

I saw some discussion on interleaving. Some people says you should always do it, some people says it doesn't make a difference in small transformers and some people says it doesn't make a difference at all for guitar amp transformers because of the limited frequency range guitar amps have.

For DC core saturation, if you're winding an push pull transformer you should arrange the lams alternately (one E, one I, one E, one I...). The DC signal of one tube will cancel with the other, and you'll have no DC saturation. But in a single ended transformer, you'll have only one DC signal with no "other tube" signal to cancel, so the core will be under DC saturation. To avoid this, you need to put all the E lams at one side and all the I lams at the other side (check the last image of the calculator screenshot you posted). Between the E pack and the I pack, you need to put an "air gap", which is basically a thin piece of paper.

Also, when you're winding the output transformer, you should do one layer of wire, then one layer of insulation, then one layer of wire... Basically, a layer of insulation after each layer of wire.

[PRR]

Secondary over primary (no interleave) works fine up to 1K impedance, and higher for guitar bandwidth.

At 13K I might try half-primary, secondary, half primary, the first mode of interleaving. However it greatly increases winding troubles. And depending on gauge and turns-count, you may not have two layers of secondary.

Insulation between turns was THE standard for 400V windings. Modern enamel is tougher, and you are only 200V. And may be cramped. I wouldn't insulate between primary layers. (Do wind as neat as you can.) DO insulate between primary and secondary!!