Correct method for frequency measurement on Output Transformers ?

[yestertech]

This seemed to me to be a straightforward task. How to measure the audio response of any output transformer one may come across in one's 'component stores'

My initial thoughts were that if the correct primary and secondary loading was use, then a generator capable of , say 10Hz to 50Khz ( or more ) could be swept into the primary and the secondary voltages measured with respect to 1KHz. Sounds simple ?
But, my trusty Philips audio generator did not seem to be able to cope with this - the voltages measured across the primary varied substantially at certain points within the normal audio passband. This I imagined to be something to do with it's drive capability into an inductive load at certain frequencies. Luckily, since purchasing a Radford LDO with a slightly 'beefier' output, I have had better results, but only if the secondary is correctly loaded with, say 4 ohms on the 4 ohm tap and the primary is series fed via a several Kohm resistor (in this case 5k ) to simulate Ra.

Two transformers I have tested in this way give markedly different results.

1. a push pull TXF by Partridge: within 0.5 dB from 10Hz to 40KHz impressive !

2. A Grampian PA amp push pull TXF designed in all probability for KT66s:

Disappointing even for a PA amp, being -4.5dB @ 10KHz, -10db @ 20KHz and less than impressive at LF ( -2db @ 30Hz )

On this last unit, I calculated from the turns ratio ( 51 : 1 approx ) that the primary might be of the order of 10k ohms. Using a slightly higher series resistor of 10k ohms did in fact improve the HF response slightly.

The question in a nutshell is whether this 'static testing' method is in fact valid. Judging by the Partridge transformer result I would tend to think so, but the Grampian makes me wonder.
Knocking up 'real' output stages to test transformers is way too much faff !

Interested to know if anyone has successfully done this or can offer any comments ?

Andy

[joebog1]

I have this document scanned from www.AmericanRadioHistory.com.

It explains all the theory, provides what appears to be a quite comprehensive piece of test equipment and is a complete construction article.

It also breaks down each test that might be required with individual circuits for each function.

Hope it helps.

Joe

[Herald1360]

That PA amp TX doesn't sound too bad for what it is. A reasonably flat response from 100Hz to 5kHz is as good as you get from an AM radio source and plenty for basic speech and music at high level. Speech alone is perfectly intelligible with a 300Hz to 3kHz bandwidth.

[Bazz4CQJ]

Originally Posted by Herald1360

That PA amp TX doesn't sound too bad for what it is. A reasonably flat response from 100Hz to 5kHz is as good as you get from an AM radio source and plenty for basic speech and music at high level. Speech alone is perfectly intelligible with a 300Hz to 3kHz bandwidth.

Or a modulation transformer?

B

[Radio Wrangler]

The frequency response of a transformer only makes sense if the source and load impedances are taken into account. These factors interact with the magnetising inductance to set the low frequency roll-off at the LF end, and interact with leakage inductance and various strays at the HF end.

David

[Diabolical Artificer]

For a quick F response I just connect my Advance H1 to any unknown LUT and click between the 3 frequency ranges, say 100hz to 10khz, whilst scoping the secondary, if it stays flat it's worth further investigation, for that I use another Advance sig gen with balanced OP connected to 2 - 6 807's variable fixed bias/variable HT and my dummy load, takes about 20 mins.

But the first test gives a good indication from small audio IP tfmr's to big jobbies.

Andy.

[kalee20]

Originally Posted by yestertech

My initial thoughts were that if the correct primary and secondary loading was use, then a generator capable of , say 10Hz to 50Khz ( or more ) could be swept into the primary and the secondary voltages measured with respect to 1KHz. Sounds simple ?


But, my trusty Philips audio generator did not seem to be able to cope with this - the voltages measured across the primary varied substantially at certain points within the normal audio passband.

First paragraph as quoted above, yes you are right.

The second paragraph is a consequence of this. Consider measuring an output transformer for matching, say, a 6V6 to a 3Ω loudspeaker... if you have a signal generator with a nice, constant, 'stiff' output, you'd connect a resistor of about 50kΩ in series (to replicate the 6V6 anode resistance), connect to the transformer, and then connect a 3Ω resistor to the transformer secondary. Then, do your sweep.

Over the range you suggest, 10Hz to 50kHz, although the sig gen might be giving a constant output, the voltage across the primary would vary. A really good transformer would see a low primary (and secondary) voltage at low frequencies, rising and levelling off from maybe 50Hz to 10kHz. Then as leakage inductance kicks in, the secondary voltage might start to fall a bit, the primary voltage to rise... until primary capacitance makes its presence known and primary voltage falls.

Hitting the primary with a constant voltage is an unrealistic condition which is unlikely to be replicated in practice. A triode output valve might come close to this.

[rontech]

I have mains and output transformers from a Grampian PA amp which used a pair of KT66 output valves.

The transformers are "battleship" jobs; the output transformer weighs 2.8 kg and has whole range of output impedances.

Purely my own opinion but years ago a Ferrograph tape recorder only claimed something like + or - 3dB over 40Hz - 10kHz ( 7.5 ips ) and my recollection was it sounded pretty good.

[barrymagrec]

Partridge transformers were expensive and very good, obviously a PA amp would not need such a good performance but in practice the amplifier feedback would mask the deficiencies you are seeing.

[G8HQP Dave]

LF results can depend on DC current. Both ends can depend on loading. Be aware that the primary loading should be the likely valve anode impedance, not the impedance presented to the valve by the OPT+secondary load. So an OP meant to 'match' 5k to 3R should not be measured with 5k in the primary circuit. If it was intended for a pentode output then it would see something like 50-100k; if a triode then maybe 1-3k.

[yestertech]

Thanks to everyone for some useful feedback . Looks like using a real output stage may be the only way to get a definitive answer...

A.

[Oldcodger]

Originally Posted by G8HQP Dave

LF results can depend on DC current. Both ends can depend on loading. Be aware that the primary loading should be the likely valve anode impedance, not the impedance presented to the valve by the OPT+secondary load. So an OP meant to 'match' 5k to 3R should not be measured with 5k in the primary circuit. If it was intended for a pentode output then it would see something like 50-100k; if a triode then maybe 1-3k.

DC current can effect the output .A common way of providing for constant voltage on HV circuits, using a secondary winding to feed back a control voltage to keep the output constant.
Then there's the reactive effect of the transformer -possibly best fed from a sig gen where the out impedance of the genny is matched to the nominal impedance of the transformer through a matching attenuator. Just a thought.