SOT: Do all transformers work step-down and step-up?

[SaBer]

I'm building a tube preamp and using two 2*12V transformers.
One's 20VA and has the secondaries in series (center-tap-style) and is regulated to +/-12,6V for filaments and op-amps. The center-tap is grounded.
The other one is 6VA and is used as a step up transformer. Its hooked up the same way as the other transformer: secondaries (I'll refer to the windings as they were meant to be...) in series, "center-tap" grounded and it's connected straight to the other transformer.
I can measure 12V AC on both secondaries, but I get only 180V AC on the primary winding and around 250V DC after filtering.

Can the transformer have such a big voltage loss when used as a step-up transformer, is it broken or could it be something else?

The next thing I'll do, is hook it up normally and see if i get 12V out. It's just a shame taking it out when it's wired so neatly 

[R.G.]

Mother Nature is teaching you about transformers and resistance. She taught me this same lessons some years ago, so maybe I can point you in the right direction.

Transfomers (a) do not have voltages; they have MAXIMUM voltages at any specific frequency above which they'll saturate (b) do not care which winding is driven (c) have resistance in their windings

The voltage rating on a transformer is the manufacturer promising that they have tested their device to do AT LEAST thus-and-so. It may do more, and for small transformers, will.

Small transformers have many more turns of wire per volt of exciting voltage than big transfomers, because they have so little iron and they must have a big enough inductance to limit no-load current to a safe value and keep from saturating. Lots of turns of wire, especially small wire, means high resistance. So a small transformer (and 20VA is small) has big resistances.

To see what's happening, you can model the transformer as a perfect transformer with some imperfections outside of it. Your transformers are 120Vac to 12Vct, yes? And one is at 20Va, presumably the other is smaller. 12Vct at 20VA is 1.67A. You'll find that the transformer is only 12Vct at exactly 120Vac input and exactly 1.67A output. If it's unloaded, it is likely to be something like 13.6Vac (guessing at 15% losses in the wire resistances ), the difference being eaten up in the primary and secondary wire resistances at full load.

So you are using a 120 to 13.6V(probably) transformer loaded down to about 12V to drive another 13.6(probably) to 120Vac transformer. The second transformer is not getting 13.6V. It's getting 12vac, so its output at most is 12*120/13.6 = 105.8.9V. Clearly, my guesses about exactly how much excess voltage the secondaries were wound for was a bit pessimistic, but not much.

So yes, what you're getting makes perfect sense. Mother Nature is telling you that transfomer makers have pre-compensated their transformers to account for the voltage losses inside the windings.

The 108Vac will give you 152Vdc after rectification. If you're only driving a few 12AX7s at about 2ma each, you could probably use a voltage doubler/rectifier and get something like 300V out of that.

[SaBer]

I guess I forgot to mention that I'm in Europe, so I should be getting 230 volts.

I tried the smaller transformer by itself and I got 12,9 volts. That means I should get 12x230/12,9=213 volts. I still only get 180 
Doubling that would be too much... 213 would be enough though...

Thanks, R.G., for the information. That's the way I thought things were, but after things not working the way they should, I lost my self-confidence.

I wish I had an oscilloscope...

[SaBer]

OK, I have an idea!

Since I'm running only three 12AX7:s, I don't need terribly much high-voltage power.
I already have my current transformer mounted in the case, so I thought I'll make something up, that would enable me to use it.
Since it's hard to increase the voltage going into the tranformer (and it would be too much for the transformer), I figured I'll have to lower it, and then have a doubler at the output.
So I thought I could use a ~5W amplifier-chip to drive the transformer with a voltage of about 7V or so, and then double the output to get a suitable voltage.

The only problem I can think of is the tranformer being a hard load for the amp.
Could this possibly work? Am I nuts?

Now for some math:
I'll want about 115V at the output of the transformer. A 12AX7 will take 2mA of current at the maximum (probably less). As I'm using a voltage doubler, I need double the current, right?
So:
3*2mA*2=12mA
0,012A*115V=1,38W --->  5W is plenty of power for the amp.

My transformer is 6VA, so the maximum current through the primaries is:
6VA/12V=0,5A

and while I have 12mA @ 115V, the current in the primaries is:
(230/12)*0,012A=0,23A ---> I'm fine!

[George Giblet]

> I tried the smaller transformer by itself and I got 12,9 volts. That means I should get 12x230/12,9=213 volts. I still only get 180

You have missed something here.

Suppose you want a transformer with 12VAC out with 230VAC in.  In an ideal world you would have a turns ratio of 230/12.  As RG mentioned in the real work the transformer windings have resistance which cause a voltage drop.  For a small transformer the voltage drop is usually in the order of 20 to 30% .  So in order to get your 12V under load you have to design the transformer with a higher output voltage, to compensate for the drop, which means using a lower turns ratio.  Sticking with RG's example of 13%, 13% on 12VAC is 13.6VAC, so the turns ratio for the transformer will be 230/13.6.  With no load you should get 13.6VAC and under load you should get 12VAC.  The reason transformers get hot is because power is being lost.  (The real story is there's not only winding resistance losses but also losses in the iron core you have to adjust the output to compensate for both.)

Now suppose you wire two of these transformers back to back, also suppose the input transformer is loaded so it produces it's nominal output of 12VAC. The 12VAC is applied to the second transformer the output for the second transformer with no load is  12 * (230/13.6) = 202V.  Now if you put a load on the second transformer the 13% loss is present - you lose it twice, the loss is not gained back when you have the transformer in reverse.    That will give you 202AC * (100-13)/100 = 175VAC.  Which is about what you are getting.     What's  going on in your situation is more complicated in terms of how much drop is being caused by each transformer, it will also depend on how much load is on the 6VA transformer,  but the basic idea still applies.

If you use a doubler with you 180VAC that will put you up around the 500VDC mark.  Presumably you wanted around 350VDC so you will have to lose 150V.  If you follow the old tube designs this is done with a resistor on the DC rail.   12mA * 150V = 1.8W, you could do that with a 5W resistor.  It's throwing away power but I guess that's the penalty of not haveing the luxury of changing the transformers!

[SaBer]

If you use a doubler with you 180VAC that will put you up around the 500VDC mark.  Presumably you wanted around 350VDC so you will have to lose 150V.  If you follow the old tube designs this is done with a resistor on the DC rail.   12mA * 150V = 1.8W, you could do that with a 5W resistor.  It's throwing away power but I guess that's the penalty of not haveing the luxury of changing the transformers!

Would this also work, with a big enough heatsink on the mosfet?
http://headwize.com/images2/opamp19.gif

[R.G.]

Yes. You can still use a resistor to eat up most of the voltage and current, still leaving plenty for the MOSFET to regulate. Be certain to use a high enough voltage MOSFET and to use a gate protection zener.

[puretube]

watch out:
high-potential heatsinks are dangerous for your health

[SaBer]

watch out:
high-potential heatsinks are dangerous for your health

Thanks for the heads-up! On the other hand, it's easier to discharge that section, and certainly easy to put an alligator clip on it to keep it discharged 

I'll have to do a bit of R&D before I get back to the workbench...

[R.G.]

... so you always insulate the high voltage devices from the heatsink and ground the heatsink for safety reasons.

[George Giblet]

If you are only driving a tube preamp I'm not sure there's that much to be gained by using a regulator (ripple can be lower).  At the end of the day to get from 500V to 350V using linear means something has to get hot, either the MOSFET or a resistor - the resistor is much simpler!

Here's an example of circuits which use the gate protection RG mentioned:

http://www.normankoren.com/Audio/TENA.html

[SaBer]

If you are only driving a tube preamp I'm not sure there's that much to be gained by using a regulator (ripple can be lower).  At the end of the day to get from 500V to 350V using linear means something has to get hot, either the MOSFET or a resistor - the resistor is much simpler!

Here's an example of circuits which use the gate protection RG mentioned:

http://www.normankoren.com/Audio/TENA.html

The thing I never got is how you calculate the average current flowing through a tube. Since the resistance of a tube is dependent on the plate voltage, and voltage drop across resistors is dependent on current, how do I figure it out? I've tried googling around alot, but never found anything. If it's some sort of complex math, go ahead and spew it out. I can handle quite alot (derivation, integtration,  differential equations etc...).

If I dont't know the current, I can´t calculate a suitable resistance....

I kind of figured out the protection zener idea after it got pointed out. So if my max Vgs is +/-30V I could use a 27V zener, right?

Concerning HV insulation, I usually think all or nothing, since I'll have to be careful poking around with my dmm as long as there's something to poke at. If I'd be selling the product I probably would insulate everything in the end. But I guess a heatsink is big enough to make an exception...

[George Giblet]

>If I dont't know the current, I can´t calculate a suitable resistance....

The easiest way is to check out some old amplifier schematics which have the voltages on them, like Fender schematic.  Find one that uses a supply near 350V, read off the plate voltage (it will usually be around 0.6 times the supply say 210V).  You then calculate current = (Vsupply - Vplate) / Rplate.  R plate is usually 100k so you end up with about 1.4mA per stage.  That will probably be about 10% accurate which should put you within a resistor value from the exact value.  If it's off a bit do you really care?  It's usually not that critical.    If you want to be 100% accurate put a close one in measure the voltage then find tune up or down with the next value resistor.  The complicated way is looking up tube charateristics, the values you get here will vary depending on what data sheet you use, and that may not match the actual tube you are plugging into the circuit - often it doesn't work out any more accurate.

[R.G.]

The thing I never got is how you calculate the average current flowing through a tube. Since the resistance of a tube is dependent on the plate voltage, and voltage drop across resistors is dependent on current, how do I figure it out?

Just like with any active device, you set it to what you want, or at least to what you can stand. You pick the current - that's why you can't find it by Googling.

Designing your own circuits, there is no "right" to hit. There's only you and the part and Mother Nature. How well you know your parts and Mother Nature's rules determines how well the circuit works.

Looking at a 12AX7 datasheet will quickly lead you to conclude that it can't conduct much more than 2ma, and it can't "saturate" to lower than about 50V between the plate and cathode. So you take your plate supply voltage (f'rinstance, 300V) and your desired set current (let's pick 1ma) and your desired plate voltage (let's pick 200V) and you calculate the plate resistor (Rplate = (300V-200V)/1ma = 100K. Then you make that prediction come true by adjusting the plate-cathode bias to make the current be 1ma. We know that a 12AX7 is substantially fully off if Vgk is more than -2V, so the cathode resistor at 1ma is less than 2V, or Rk< 2V/1ma < 2K.

Oddly enough, most preamp circuits for the 12AX7 have Rplate = 100K to 330K and Rk = 680R to 1.8K.

All of the old tube service literature also offers the note "readings may vary 20%" to indicate that the actual circuits vary. They had to take into account tube variance and resistor variance too.

So what I'd do is to decide that I was going to set the tube currents to something reasonable like 1ma each plate, then design the power supply to do that. Once you're there, you can easily mess with the circuits.

[SaBer]

The thing I never got is how you calculate the average current flowing through a tube. Since the resistance of a tube is dependent on the plate voltage, and voltage drop across resistors is dependent on current, how do I figure it out?

Just like with any active device, you set it to what you want, or at least to what you can stand. You pick the current - that's why you can't find it by Googling.

Designing your own circuits, there is no "right" to hit.........

The problem is that I'm not really designin my own cicuits, I'm just combining two different designs (two sepeate channels) and putting them into one box.
But now that you've pointed me to the right direction, I think I'll be able to figure things out with a easonable amount of iteration. Getting pointed to the right direction is often the best way to learn. Thanks alot R.G and George! (and puretube for possibly saving my life....) 

[SaBer]

UPDATE:
Well I went and bought myself a 2x9V toroid. So after driving the "secondaries" in series with 12V, and doubling that, I get a nice 330V without load. I'll have to see what happens when I get the tube-circuits ready...

Now I have a +/-12,6V regulated supply, a plate voltage delay-circuit, the step-up trannie and the voltage doubler+filter.

I was wondering about the plate voltage delay I have... This is what happens when I power up:
1. Both high and low-voltage rectifiers and regulators fire up.
2. The plate-voltage delay circuit gets its power. The heater voltage rises gradually.
3. The heater voltage reaches 12,6V and about the same time the plate voltage gets switched on.

So, now I have it hooked up so that the ss-relay in  plate-voltage delay-circuit is placed after the voltage doubler. i.e. the plate voltage rises very fast after it's switched on. Is there any reason to hook the ss-relay before the doubler, so that the voltage would rise gradually? I know filaments' resistance changes as they heat up, so it's good to power them up slowly, but is there any reason to do this to the plate-voltage?

BTW: now I have one extra hole in my chassis