Measuring (power) output transformer

[Disco Volante]

Pardon me, this is probably old hat to some of you magnetoheads:

Having wound my first OPT for a small PP tube power amp (8-10W) (see this thread Who in (Germany?) made these? Valve rack units, small power amps and a mystery box, I would like to measure its frequency response.

Can someone sketch up a basic rig that ensures it sees reasonable impedances in and out?
RDH5 talks about a zero-impedance voltage source and a pure resistance 0.4xthe standard distribution voltage. How would this translate into a real-world diy setup?
I am a little unsure about how to handle the BH-curve of it, that is, how do I ensure that I am loading the transformer properly, so that the measurements make sense?
I have a basic setup with a function-generator, scope, DMM. Also have REW and a measuring microphone.
I don't mind "reading the manual" and I'm slowly making my way through the RDH chapter on transformers. Also reading old posts here on transformers, but none so far that details how to measure FR in practice..

Any advice appreciated;-)

 

[CJ]

try this if possible run two tests, loaded and unloaded,

 

[abbey road d enfer]

RDH5 talks about a zero-impedance voltage source and a pure resistance 0.4xthe standard distribution voltage.

The set-up would be different according to the type of transformer and circuit it is connected to.
Valve output xfmrs are typically driven at relatively high impedance and at a much higher level than what a signal gen delivers.
Short of driving it with proper level and impedance, you can drive it with the signal gen via a 10k resistor; that would give you a good appreciation of the HF response. Since the output level would be quite small, you must ensure your measurement set-up is noise-free. The LF response would be actually somewhat better with higher signal level.
Of course you should load the secondary with the nominal impedance.

 

[CJ]

Not to many people have a 500 volt signal generator.

 

[Disco Volante]

Thanks, that pretty much pinpoints my thoughts about impedance and voltage.
I will have to experiment a bit, make a few measurements. Maybe I could even find out what output impedance the function-gen has.

In a PP OPT the 320VDC is supposed to be cancelling out, right? does that leave only the voltage swing of the output tubes on the transformer? Or does the HVDC still affect the core?

@CJ: is your skteched setup (comparing P and S with a dual-ray osc) the way you measure your OPT's? Or do the true gurus use something more sophisticated?

Thanks guys ;-) merry christmas

 

[abbey road d enfer]

Thanks, that pretty much pinpoints my thoughts about impedance and voltage.
I will have to experiment a bit, make a few measurements. Maybe I could even find out what output impedance the function-gen has.

Most sig gens have an impedance of 50r, which is negligible vs. primary impedance (usually 3-10 kohm)

In a PP OPT the 320VDC is supposed to be cancelling out, right? does that leave only the voltage swing of the output tubes on the transformer?

More than the voltage, its the current that counts, because it's what magnetizes the core.
You are right, in a PP xfmr, if the output tubes are properly balanced/biased, the two currents cancel out.

 

[Disco Volante]

Yes, current of course. Reading Merlin Blencowes excellent book on power supplies, a number of things about transformers are clearer to me now..

 

[emrr]

You want to look up the PP plate resistance of the output tubes and look at that input Z in your measurements.

 

[MisterCMRR]

When making impedance measurements on transformers, the excitation level (generator output voltage) must be considered. Both inductance and impedance are indirect measurements of the slope of the core material's B-H curve. M6 steel (6% silicon steel) for example, has considerable magnetic hysteresis at low signal levels and its inductance will measure much less than expected at low signal levels. At higher levels, inductance will measure higher until magnetic saturation is approached, when inductance will drop again. With most core materials, inductance will vary considerably with frequency - generally dropping as frequency increases. Unfortunately, most manufacturers do a very poor job of stating the operating conditions under which any of their measurements are taken (and performance claims made). And, of course, since transformers only reflect impedances among the windings, the input impedance of a transformer's primary applies only when a stated load is connected to its secondary.

 

[Disco Volante]

Thank you MisterCMRR for this clarifying comment.
I suspected that would be the case.

Next question is: What excitation level do I measure my new transformer at? And how to go about it?
I have a reasonably linear power amp I could amp up my sig-gen with, but power amps have very low output impedance..

Perhaps simply installing it in the relevant circuit I'm building/modding, but that would of course include the entire frequency-response complexity of the amplifier:-/

Primary is 10k a-a, for use with ecl86,el84, el95 and the like. UL taps and CFB taps included just for the possibility of experimenting.

 

[abbey road d enfer]

When making impedance measurements on transformers, the excitation level (generator output voltage) must be considered. Both inductance and impedance are indirect measurements of the slope of the core material's B-H curve.

I agree with that, however, the OP's objective is not to measure the unloaded impedance.
When a load is presented at the secondary, the reflected impedance quickly dominates the measured primary impedance, so the error is no so big.

M6 steel (6% silicon steel) for example, has considerable magnetic hysteresis at low signal levels and its inductance will measure much less than expected at low signal levels.

That is true certainly for the unloaded impedance, which is supposed to be much higher than the reflected or nominal impedance.
At higher levels, inductance will measure higher until magnetic saturation is approached, when inductance will drop again. With most core materials, inductance will vary considerably with frequency - generally dropping as frequency increases.

 

[Admiral-dk]

My prefered methode (not perfect - but ballpark) :
I connect the Signal Generator to the highest Output Impedance (Secondary) - 16 Ohm (on a normal 4, 8, 16) + an Oscilloscope across the same spot.
Then try to see what the Max. Un-Distorted Voltage I can get there @ 500Hz. on the Scope - this very much depends on the quality of the Output Buffer in the Sig-Gen + Transformer Inductance. When I have established this - I measure the AC Voltage across the Secondary and the Primary (across Anode to Anode - in a Push-Pull Tube Amp).

Now it's just a simple Calculus : Output Impedance x (Primary Voltage / Secondary Voltage)^2 = Primary Impedance.

Just to confirm you got the right points - measure the Voltage on the other Impedance Outputs.
4 Ohm should be close to 0.5 times the Voltage on the 16 Ohm - 8 Ohm should be close to 0.7 times the Voltage on 16 Ohm.
The Primary should have approx. same Voltage measured between the VHT (Very High Tension) and each of the Anodes).

I hope this is helpfull.

Per

Edited a few errors ....

 

[Disco Volante]

Ah, that makes sense actually. Sig-gen has a low-imp output so better paired to the secondary. Then measure the primary with a high-impedance meter (scope). I'm going to try this.
Sweeping/varying the frequency and comparing on the scope should give me a fair idea of the transfer function of this thing.

 

[Admiral-dk]

Oh and NO - the Scope is across the Secondary to confirm that I have No Distortion on the signal from the Sig-Gen when it drives the Transformer ..!!!
If you have Distortion here => you can't trust the measurements !!!!

I measure the Voltages with a Fluke DVM - but most other ones should do the job just fine.

Per

 

[CJ]

Placing the sign gen on the sec has been bad news for me.
I think it liked driving the primary better.

Are you after the voltage ratio-impedance ratio or the pri inductance?

Using a lower level pri voltage should give you a rough idea of the freq response. Then the real test is to install the XFMR and listen to it. True, there is a flat spot at the bottom of a B-H curve but a 20 volt p to p or better signal should get you into the initial perm part of the curve.

To see how your generator does not like the sec, measure the before and after voltage of the generator.

It might work at 1 kc but it will crap out at 40 Hz because reactance goes down with freq.

Note that there will always be a DC imbalance even in a push pull because tube matching is never perfect. Some amps have a balance pot but how long will that work?

Measuring a SE transformer is tricky because you must bias the core with DC. This requires a high z source for the DC as your AC signal will disappear into a low z source. So yeah, a vacuum tube is a good hi z source for DC.

I am not even going to tell you what Jensen used for ultra low freq arc-tan testing because I was instructed not to do so.