Assessing the 'Quality' of AF output transformers

[Bazz4CQJ]

In recent days, two threads on the forum have touched on the question of the quality, or lack of it, that vintage (valve) AF output transformers might possess.

Can anyone suggest what sort of tests would needed on such a transformer to assess it in terms of where it sits on the spectrum between "kitchen radio and quality Hi-Fi"? I'm guessing that you'd ideally measure its efficiency both as a function of frequency and function of power level. However, I wonder what the simplest check that could be done that would give a useful indication?

I'm not sure if this observation carries much weight, but my impression is that really good transformers were often fully enclosed in a 'can' (thus preventing visual assessment of their construction).

B

[Diabolical Artificer]

The weight of an OPT is the simplest indication of whether it will be likely to perform well - more iron more flux density. Induction is a reasonable indicator too, the higher it is, the better the bass response.

Andy.

[Hartley118]

Victim of autocorrect methinks? It's primary inductance that's important for LF response. Too little and the primary winding becomes a shunt impedance to the signal at low frequencies - and it's a non-linear shunt impedance too.

Look too for interleaved primary and secondary - that reduces leakage inductance. Leakage inductance behaves like a small inductor in series with the signal and reduces HF response whilst increasing phase shift - bad news for negative feedback circuits.

Martin

[G8HQP Dave]

Primary inductance sets the LF rolloff. Self-resonance sets the HF rolloff. Measure these two and you have a rough idea how good it is. As others say, weight is a rough guide to quality too - although it can't distinguish between a good low power type and a poor high power type.

[Wendymott]

If anyone remembers the "Williamson" amplifiers, using either PX4's or the Later PX25's, these were considered the "bees knees" at the time. The Output transformer was quite large, thus lots of iron. I remember we had to replace one and obviously in the mid 60's a genuine replacement was not possible, thus a Gardners transformer was retro fitted. Now the Gardners was reputed to be a good grade transformer, but not to the "ears" of the customer, who quickly rejected the repair. I was the recipient of the now transformerless amp. Now heres the perverse bit... the customer then went and bought a Quad system. It may have been that the "repair" engineer, was just that, and not an enthusiast who should have tried tweeking the feedback etc to make the Gardners "fit", but in those days, thats how it was.

[trobbins]

5VAC across primary P-P and measure current (in uA) to measure inductance. Above approx 50H for push pull is above average.

Sig gen sine wave through say 1k resistor - self resonance when voltage across resistor is min. Above 30-40khz is good.

Equal resistance of primary half windings in push pull is sign of excellent interleaving.

[kalee20]

Actually, self-resonance isn't a good guide to HF roll-off. Suppose you have a transformer with 100pF effective primary self-capacitance (a good figure), and it also has super-exotic-unobtanium core material giving a primary inductance of 1,000 henries (thus the transformer will also have a really good LF response), the self-resonant frequency is only going to be 500Hz... But it's a good transformer!

High primary inductance is desirable, plus the ability to support full voltage at lowest frequencies without core saturating (the two are distinct, though related), for lid LF performance and low distortion.

Low leakage inductance and low self-capacitance is desirable for good HF performance.

In push-pull, good coupling between primary sections is desirable in class AB and class B, slightly less so for class A.

Optimising any one of these tends to hurt the others - transformer design is a compromise!

[Radio Wrangler]

For a serious hifi output transformer you need to check that it operates to quite a bit higher than the top of audible frequencies. The amplifier loop gain has to roll off down to below unity before too much phase shift builds up in the feedback loop. You can have a perfect transformer across the audible range, but unless it has the bandwidth to also handle the roll-off region at higher frequencies, the amplifier will be unstable.

One method is to load the secondary, drive the primary and check gain and phase shift right up to 100kHz or more.

Another method is to measure the primary inductance, and then do it again with the secondary shorted. This will give you an idea of the amount of leakage flux/ leakage inductance.

David

[Edward Huggins]

Originally Posted by Bazz4CQJ

I'm not sure if this observation carries much weight, but my impression is that really good transformers were often fully enclosed in a 'can' (thus preventing visual assessment of their construction).

The point of the "can" was to fully shroud/screen the output transformer particularly with regard to its proximity to other nearby transformers e.g. mains tranny or a large choke. Also read up on "grain orientated laminations" used in high quality transformers in the late 1950s. Also a Push-Pull transformer with ultra-linear primary taps is another sign of more careful design - aka "ultra linear". Edward

[G8HQP Dave]

Originally Posted by kalee20

Actually, self-resonance isn't a good guide to HF roll-off.

I meant the HF resonance beyond which the output drops off rather sharply and the phase shifts. Probably related to leakage inductance more than primary inductance, but my understanding is that as the inductance and capacitance are both distributed through the windings you can't simply measure an inductance and a capacitance and then calculate a frequency. Instead you have to measure the frequency.

If no feedback is used then the OPT can be used up to near its HF resonance. With feedback you need more room for extra phase shift, which is why good hi-fi transformers are expensive.

[Radio Wrangler]

Grain-oriented laminations are just that non-isotropic. They have different properties in different directions.

With E and I or U and T laminations, the flux has to flow in different directions in order to circulate. Grain oriented materials really cme into their own with C-core transformers and later toroids and 'R' cores.

David

[kalee20]

Originally Posted by G8HQP Dave

I meant the HF resonance beyond which the output drops off rather sharply and the phase shifts.

Ah yes... Agree. trobbins's post above would actually measure the self-resonant frequency that I allude to, rather than what you describe.

Leakage inductance is more troublesome with triodes than pentodes or beam tetrodes - triodes tend to hit the transformer with a lowish source impedance, so any leakage inductance leads to a reduction in secondary output (and the associated dreaded phase shift). Whereas tetrodes/pentodes, the leakage inductance just appears in series with the near-constant-current drive, so has little effect apart from limiting power output at higher frequencies as the valve 'bottoms'.

[tri-comp]

About Audio Transformer Theory, Crowhurst put it all nicely together in this 1952 article.

[tri-comp]

I knew I had his article on acquiring param's for audio transformers as well.
Just couldn't find it yesterday.
Crowhurst is back 'splaining.

[Bazz4CQJ]

Many thanks to everyone for their very useful responses. Sadly, I now think that the few OP transformers that I have in my "junk" box are indeed, just junk .

B

[Hartley118]

Before condemning your transformers, remember that the proof of the pudding is in the eating. Just connect each in turn into an actual valve output stage and run a frequency sweep. You may be pleasantly surprised.

Martin

[Bazz4CQJ]

Maybe... guitar amps?

B

[turretslug]

Ooooh..... I'd hesitate before slinging them out! Provided they're not occupying too much space and/or provoking domestic disharmony that is. Even if they're not Ongaku-grade, hand-made from isotope-certified silver wire by blind monks on the slopes of Mount Fuji, they'll have a use somewhere by someone. Quite often, there'll be a help, my (modest domestic set) has a duff OPT here- and it might otherwise get written off. Also, quite a few of us run military/professional sets with "odd" output stages to line or hi-Z, or just headphone-compatible, and giving these even modest loudspeaker capability is a boon for real-world use. There seem to be fewer and fewer people making them and they ain't getting cheaper, either....

[trobbins]

Ooh yes - guitar amps love lack of bass response (lowest note is way above 20Hz) and lots of distortion component complexity (saves having to design it in via circuity). No finesse needed - some enjoy re-utilising 12V power transformers with 0-120-240V primary.

[G8HQP Dave]

Yes, don't destroy old OPTs. They can be used in radios and guitar amps, where the requirements for sound quality are much lower than hi-fi.