[Staxnuts]

Yesterday i tought i knew how transformers worked.
I wired an hammond Push-Pull tranny backwards to build a stax adapter. The  impedance ratio was 1:1900 when using the 4ohms tap.
I figured that at 50K the stax headphones would then be reflected as about 26 ohms. 
I feel so dumb now, sitting next to my dead sugden amp breathing burned electronics fumes.
 
My understanding now is that the amp saw only the low-DCR of the tranny 4-ohm winding, which was extremely low. Should i have put a resistor in series with the winding, like stax boxes have (or is it a thermistor?) ? But then again, i see somme designs without it, for example Andrea Ciuffoli's: 
http://www.audiodesignguide.com/my/stax_transf.html
 
Can someone explain me what went wrong?
Thanks in advance.
"Depressed".

 

[wakibaki]

So sorry to hear about your experiment gone wrong. Don't be depressed, it's a temporary setback.

It is not uncommon to see resistors used in the amplifier circuit of Stax transformer drivers. This is intended to 'isolate' the amplifier from the transformer, these are frequently in the range of 2 ohms.

The Ciuffoli design uses transformers with a quite substantial DCR. An 8 ohm speaker will typically exhibit a DCR < 6R5.

We'd like to mimic the speaker as a load, I think, but I don't think we can do that with any kind of efficiency with a passive shunt or series network modifying a non-ideal transformer characteristic, although I stand to be corrected here.

An amplifier should withstand a dead short if it has no DC output offset and there is no input signal. This, however, is a very strict set of conditions, and may not be met at all times with a DC coupled amplifier. Many amplifiers have S/C protection, but not all. Transient conditions at switch-on may be responsible for your problem. Having a couple of ohms resistance in the circuit may have prevented a short-term overheating, even though it costs some efficiency.

Exactly what went wrong is not clear, however. When I'm building something like this I use an eBay amp for experimenting that is cheap enough to be considered disposable.

w

 

[nikongod]

You may have better results with a lower stepdown ratio. If the transformer has a 16 or even 32ohm tap, for example, try that. You wont get as much voltage at the output, but you dont really need that. As much as the STAX mafia likes to talk about freakishly high voltage swing, nobody uses much more than 100vrms in real life. It is not uncommon to see electrostatic headphones singing proudly with only 5-10v. 
 
Transformers are not perfect devices. They do not (generally) work as well as one would hope backwards. Yea, a theoretically perfect device would work exactly the same forwards or backwards, but you know. 
 
The lower stepdown ratio would appear as a higher impedance to the amp, which is nice. Lower step down ratio transformers are said to work better (I dont know why) than their higher ratio brothers. The higher impedance primary (well, really the secondary wired backwards...) would show a higher DCR to the amp AND show a higher inductance - both of these things are niceeee here. 
 
You mentioned the Ciuffoli article - the transformers he used *only* had a voltage ratio of 1:14.4 which is an impedance ratio of ~1:207 - a FAR cry from 1:1900(impedance). Again, you don't really need as much voltage as many people like to say and 1:14.4 (voltage) gets us ample output voltage with even a relatively low voltage input. 2.6v from the amp (which is basically nothing for any reasonable SS speaker amp) becomes ~37.5v at the output - which is a bit more thana more than we really need - with WAY more voltage available from the amp and also through the transformer.
 
Regarding stability -  The vast majority of solid-state speaker amps will be stable driving 32ohms. They will be stable driving 300ohms. They will be stable just swinging voltage with no load at all! Some amps do have output protection, but I would never count on it. Shorting an amp (or even showing it a load impedance BELOW some acceptable limit) is a surefire way to kill it. Don't worry that you are not "loading the amp down" - that is leftover BS from certain tube designs. Put another way I would worry FAR more about shorting the amp out than I would about "underloading" it. There are no doubt exceptions, but they are insanely rare. 

 

[jcx]

xfmr also don't like V offset which causes DC current - completely safe is to include a big DC blocking cap if your amp is DC coupled output - the blocker often has to be electrolytic to get the high C value
 
 
you should really look at the (loaded with ES equiv C) high frequency response too - high bandwidth xfmr at high Z often need the right drive/load R to damp their parasitic winding C + load C and the leakage L resonant peak

 

[FrankCooter]

I've played around with transformer-based electrostatic  amps off and on for years, but I really can't add anything to the 3 excellent posts above. By far the most likely cause of your problem was the low DCR of your Hammond.  In actual construction, an output transformer is a very complex device that while theoretically can be run backward, was never designed for this purpose and will almost certainly have degraded performance if asked to do so.
 
A  cheap but better solution would be to  tap off the secondaries of a push-pull tube amp using appropriate  blocking capacitors and a dummy load on the transformer secondary. Better yet, just buy yourself a cheap "normal bias" Stax transformer box off Ebay. You'll need to build yourself an appropriate bias supply in either case.
 
If you are really interested in building an electrostatic amp with output transformers, you will need purpose-designed output transformers. Mine come from Electra-Print.

 

[Staxnuts]

Thank you for the great replies.
 
Now that i have swallowed the pill, i realized that using the 4 ohms tap was not the best bet. I wanted the highest gain, but as pointed out by nikongod, that may not have been necessary. I will try the 16 ohm tap, probably with a small series R.
 
My error was to take for granted that the reflected impedance would be high-ish, even at low frequencies, and not taking the time to make the actual calculations. That's what happens when i venture in a "quick project" without much thinking Hell, i didn't even measure the static resistance of the windings on amp side... And since the old Sugdens are known for their "limit" design, it could not cope with such a low DCR.
 
Now, off to the thrift store to find an old amp for testings, as i should have done first.
 
Thanks again!