PT and PS design question

[Chuck H]

Ok, I'm the least tech of the posters so far, but... I looked at the data sheet and it seems to me that at your plate and screen voltages you might be fine UNLESS this amp is class A OR your running it balls to the wall whenever you turn it on. I'm not sure how you calculated for current above, but the data sheet indicates a quad of KT88's in fxed AB1 at your voltages will draw about 250mA idle and about 600mA balls out from that secondary. Yes 600mA is more than 460mA but unless your amplifying a continuous sine wave at full volume I think you'd be aright. Maybe the transformer you have now is getting too hot because it's broken and not because the HV secondary is too small.

[krair]

R.G. - I am finding from more and more reading that the majority of transformer design and theory is indeed very loosely held, and primarily done through previous experience. Thanks for the insight on the differences between the FWCT and FWB, or similarities I should say. I agree that I should not try to push this guy into designing something he is not familiar with, at least not yet for the first few batches.

Diablo - AFAIK right now it has an iron core, and talking to the builder, he suggested I move up to a steel core (likely M3) to help with heat. Cost wise wasn't terrible, adding around $15 per transformer.

J M - That is precisely what I plan on doing, my main question pertained to the way current ratings and flow work in a transformer. Through a little more reading here: http://valvewizard2.webs.com/Transformers.pdf I realize that it is not as simple a question as that. One thing noted in the above article, is that a transformer's rating is a sum of all the windings, and if one winding is being over-worked while another is hardly utilized, there can be some "sharing" as long as the combined draw doesn't exceed the maximum that the whole thing can handle. I will have to read a bit more to clarify and confirm that, but that is my guess for why the transformer has been working and not failing, even though I am pushing a single winding beyond its limits. However I am still a little unsure about how to have him build the transformer for current handling. It seems as thought the winding used for the screens and preamp would need to be the beefiest since it will also have the majority of the load from the plates. The second winding will only be adding ~180VDC to the ~380VDC, meaning the draw is proportional to the voltage each winding supplies?

For example if I expect a draw of ~600mA, the smaller winding only needs to handle around 200mA since it is only supplying 1/3 of the voltage, while the main winding will need to handle the other 400mA, plus that expected by the screen/pre-amp. Is this math somewhat in line with what I read? Is it even possible? Enlighten me

I will mention the less turns with thicker wire and see what he says.

One thing i should note: The other day I threw in a 1k resistor between the PS and the OT, like 25W. I had the amp on and played lightly for about 20 min or so, and while the resistor got very hot (obviously way more power than 25W), the transformer itself was just barely warm to the touch, rather than too hot to touch at all - which usually happens in less than 10 min. This is kind of what got me thinking in the first place. So while I haven't yet measured the draw etc, I am pretty certain that is where the problem lies.

[R.G.]

I haven't yet measured the draw etc

I say this not out of malice, but for emphasis. This statement is the most fundamental way to say what your problem is. You do not know what you need. If you don't know, the transformer designer can't do a good job of guessing. I predict that until you measure these quantities, you will only get a transformer you like by accident. It will certainly not be by design because you don't know what you need to design.

However I am still a little unsure about how to have him build the transformer for current handling.

You cannot tell him that, even if you knew how, because you don't know what currents you need.

It seems as thought the winding used for the screens and preamp would need to be the beefiest since it will also have the majority of the load from the plates.

Unfortunately, no. This is incorrect. All plates are not equal. The plate load on the output tubes is probably 85% or more of all the total power to the amp. The plates of the preamp are almost negligible compared to that. The screens are often around 10% of the plate current for pentodes and power grid tubes.

Once again, no, the screens and preamp plates are not the biggest load. You could find out what load they represent by **measuring the current that goes out to them**.

The second winding will only be adding ~180VDC to the ~380VDC, meaning the draw is proportional to the voltage each winding supplies?

For active devices, like tubes, the current draw is NOT proportional to the voltage. You'd have to measure the load currents to know what the currents are.

For example if I expect a draw of ~600mA, the smaller winding only needs to handle around 200mA since it is only supplying 1/3 of the voltage, while the main winding will need to handle the other 400mA, plus that expected by the screen/pre-amp. Is this math somewhat in line with what I read? Is it even possible? Enlighten me

Screens on a pentode or power beam tube are often about 10% of the plate load. This skyrockets when the tube is running at high currents, but remains in a similar proportion. So to know what's happening, you'd have to actually measure the currents that the amp draws under full power operation. Then you would know.

One thing i should note: The other day I threw in a 1k resistor between the PS and the OT, like 25W. I had the amp on and played lightly for about 20 min or so, and while the resistor got very hot (obviously way more power than 25W), the transformer itself was just barely warm to the touch, rather than too hot to touch at all - which usually happens in less than 10 min. This is kind of what got me thinking in the first place. So while I haven't yet measured the draw etc, I am pretty certain that is where the problem lies.

Did you happen to think of measuring the voltage across that 1K resistor and thereby producing a number which would tell you how much current was flowing?

In case I forgot to mention it above, YOU NEED TO MEASURE THE CURRENTS THAT THE AMP DRAWS WHEN OPERATING AT IDLE AND FULL POWER SO YOU CAN FIGURE OUT WHAT VOLTAGE AND CURRENTS THE POWER SUPPLY HAS TO PROVIDE.

Any further noodling around without getting some real information is putting off a successful transformer design, not helping it. Discussing the philosophy of copper loss versus core loss is wasting time when you have no real idea what loads have to be supplied.

Did I mention that you need to measure the actual loading?

[J M Fahey]

Sorry but you seem to be confusing *power* with *current* in the following paragraphs:

current handling. It seems as thought the winding used for the screens and preamp would need to be the beefiest since it will also have the majority of the load from the plates. The second winding will only be adding ~180VDC to the ~380VDC, meaning the draw is proportional to the voltage each winding supplies?

No, it's not, if you connect them in series as suggested the current througnh them will be essentialy the same, with the lower one carrying more because of screens and preamp.
YOU'LL HAVE TO MEASURE ACTUAL LOAD CURRENTS.
To start with something, even if improperly rated you'll have to test with the transformer you actually have now.
Not a good choice, but it's the only poker game in town
Measure those actual currents, and consider the definitive transformer will need to supply, say, 20% more, since the one you have today is crippled by definition.
Even then, order *just one*upgraded transformer and run the tests again, don't commit yourself to any quantity yet.
Consider it still an experimental one.
After, say, a couple iterations, you'll have the right middle point between adequate and over-rated.

Going the datasheet route:
I checked only the Svetlana one, which I hope is representative.
For Vg1=0V and VScreen=300V the plate current, when saturated (Vp=100V) reads around 370 mA, meaning both tubes will draw around 700 mA.
If the amp is overdriven (which is the normal state in a guitar amp, he he), one half (2xKT88) will pass 700mA during its half cycle, the other half will do the same in their time slot.
The Power supply couldn't care less who's loading it, will see a continuous load equivalent to around 700mA DC ... only for the plates, plus whatever ths screens need.
Now I'm finding the Svetlana datasheets *very* poor, will check others when I have some free time. Sorry.
In a nutshell, I'm finding you'll need a monster power transformer, plus, yes, you saw it coming: measure those currents !!!

[R.G.]

You know, J M, I think you may be right. He may actually need to measure the actual operating currents.

[Chuck H]

Ehem... Hmmm.

[R.G.]

Ehem... Hmmm.

Yeah, I coughed a few times too.

Another aphorism comes to mind. "When you don't know where you're going, any direction will do."

I personally sometimes get amazed at the amount of time people will spend sniffing out even very dubious information on the internet to avoid getting out their meter and making a measurement.

[Chuck H]

1) I agree. measure the current wherever you want to know how much current is being used.

2) as an alternative to an entirely new transformer (I think I read that this is not going to be an amp for sale or manufacture) why not buy a second, single secondary (400-0-400) transformer good for 300mA and hook it up parallel with the existing HV plate winding. This should be much more economical than having a 'super' transformer custom made.

[J M Fahey]

The second transformer should have the exact same AC voltage as the first one, within, say, 200mV, if it's going to be used in parallel, or it will be the same as having a couple shorted turns.
Of course, using 2 400-400V transformers of the same brand and model, bought at the same time, can be reasonably expected to match safely.
Now, *different* transformers can be used safely in series.

Never asked before: what is the intended use of this super-amplifier?

[krair]

Sooooooo as I expected:

The KT88 plate winding draws ~190mA at idle. 520mA at full without massive distortion. With pre-amp and master volumes at max it drew 630mA.
The screens drew 20mA at idle, 90mA at "full", 116mA at max.
Pre-amp drew 10mA at idle and 30mA at max.

So now can I finally get help with the design?

[R.G.]

So now can I finally get help with the design?

Sure. Whether you realize it or not, that was a huge help in your design. Now you know what you're aiming for.

You didn't mention the voltages on those outputs, though. Need those.

Some comments/questions:

The KT88 plate winding draws ~190mA at idle. 520mA at full without massive distortion. With pre-amp and master volumes at max it drew 630mA.The screens drew 20mA at idle, 90mA at "full", 116mA at max.

Was this taken with guitar signal into it, or sine wave driving the output to a heavily clipped semi-square wave?

Pre-amp drew 10mA at idle and 30mA at max.

It's odd that the preamp varied at max. Does this include the phase inverter?

The schemo you reference looks reasonable. If the full/massive situation currents you mention hold up, it'll be three windings, one for bias and trivial. The other two will be whatever voltage, but rated for 650ma for the upper/boost high voltage and 780ma for the lower winding.

And, of course, heater windings rated for the current and voltage for the number and kind of tubes you intend to use. I recommend you use at least two and perhaps three heater windings; one for power tubes, one for large signal preamp tubes, and one for input tubes. Each will have a different current rating. You could spec the input tubes winding for a bigger voltage and go to DC on your input tubes if you want. If not, having a separate winding for them lets you pull just those tube heaters positive for lower hum.

Some questions:
- how concerned are you about fluctuation of the AC power line from -10% to +10%? This is really a question about whether a +10% high line pushes the output stage into sudden death by overheating. If you're worried about that, design for a bit lower nominal case. If not, design all the voltages for nominal line and cross your fingers.
- Do you want to worry about any reserve capacity for modification and/or extra reliability?

As for total power from the transformer, once you have the DC voltages, the AC winding voltages and current ratings fall out, as you'll be getting RMS currents in the windings of about 1.8 times the DC average currents. Designing for that as a target and class A/105C insulation will get you a cool transformer. At least not overly hot.

[J M Fahey]

Fully agree with the analysis and add: being a transformer winder and thinking as such, I would specify *both* HV windings for 800mA (rounding to next thicker gauge if between two normalized ones) simply to avoid resetting the machine wire advancing screw (and simplifying stocking).
Both will be quite close anyway, remember current carrying capacity is proportional to the square of the diameter, or to see it from the other side, diameters will be proportional to the square root of current capacity.
I would rather wind two, say, 1 mm diameter secondaries, one after the other, than winding, say, one in 1.05mm, readjusting, and winding other in .95 mm.
Difference in parts cost is nil; difference in time/handwork is not.
Your final design will have one secondary which when bridge rectified, will provide around 480VDC or whateve you need for screens and preamp, plus another which when in series with it will provide the remaining plate voltage (extra 180VDC?)
Both rated for 800 mA.
A secondary benefit will be that it will run slightly cooler.
And voltage will be stiffer.

[krair]

Voltages were lightly mentioned. 560V on the plates, ~370V supplying the screens (before the 1k screen resistors), 360V on the preamp, which later dropped slightly. yes in "pre-amp" I am including the PI. I realize that is not politically correct, but from a layout and voltage supply standpoint it makes more sense. Also, MAX was literally massive amounts of distortion (like beyond playable).

I don't have a signal generator or o-scope yet (looking for a decent but affordable USB one if you have any suggestions). So yes, it was with a guitar with high output humbuckers.

Not concerned about fluctuating line voltage, there is a lot of room for fluctuation. I didn't design anything at its max. For this design, there will not be any additional power.

So from what you guys have told me: 3 windings, 1st is 8-10A heaters, 2nd at 275-60-0-275 (800mA), 3rd at 125-0-125 (800mA). Connect the center tap of the 125-0-125 to the rectified side of the 275-0-275. This would float the ground and hopefully give me a 560V HT for the plates and ~380V for the screens and pre-amp plates. Or a FWB on the 125 (upper) winding... Probably use high grade steel (M3) to keep size roughly equal, should only increase costs by ~$20 according to the supplier.

[krair]

Yes, I could use a separate winding for the bias supply. I will ask about cost on that one... see if it costs more to center tap the 275-60-0-275 winding or to add an additional winding for the bias.

[J M Fahey]

Almost there.
I would specify a 275V and a 125V main windings (as calculated by you, didn't re-check) , and diode bridge rectify each of them.
No sensible reason to "save 4 diodes" and spend 100 times that in copper and double winding time.
2 diode full wave rectifiers are a DNA link to our reptile past, not needed today.
And they don't sound better/vintagey/whatever either.
You can always have a dedicated bias tap. In fact, you don't even *need* it, but I don't want to complicate you by now.
But you are almost at the end of the race.
Now go talk to your winder.