PP Output Transformer Testing Puzzle

[LinuksGuru]

Hi, folks,

Recently I've got 70W PP output transformer, designed by myself, and decided to put it in bench test. I have not finished prototype amp yet, so its impossible to test trafo in real circuit.

Factory I ordered this device from did not have core of the size I wanted, instead, one was too small, and another was too big (probably enough for 150W unit). So 2nd.option was the obvious choice. The resulting product is a 8 kg monster.

I used a generator 10Hz - 100 KHz with 10V 5 Ohm output (yes, 5 Ohm not 50). I connected it to primary (shunted with 5K resistor), and loaded secondary with 10 Ohm resistor (not wire-wound).

Transformer has quite unusual -4dB drop at 40 KHz and 1.5 dB peak at 50 KHz, the rest is fine up to 100 KHz. I do not have equipment to test above 100 KHz, or below 10 Hz, and do not I think this is necessary.

Voltage on primary was always constant, so this is not issue of generator output impedance. Frequency response from 10Hz to 20 KHz is linear +/-1 dB.

Do you have any idea what is the reason for so unusual behavior (or may be there is nothing at all to bother ?)?

If anyone have any other ideas of bench-testing (e.g. with square wave), I will be very welcome

[schiller]

First of all your testing method is not the right one for a PP trans. You need a differential source for the primary. Then resistors IN SERIES at each side with the same value as the Rp of the tubes you want to use.

Secondary loaded with the nominal speaker value, grounded (ground the lead which gives the better FR.

Right now you are testing a SE transformer with Rsource of 5 Ohm !! (no tube with such low Rp as far as i know), and loaded with 5 Kohm // with the reflected load from the secondary.

[LinuksGuru]

Originally Posted by schiller

First of all your testing method is not the right one for a PP trans. You need a differential source for the primary. Then resistors IN SERIES at each side with the same value as the Rp of the tubes you want to use.

Secondary loaded with the nominal speaker value, grounded (ground the lead which gives the better FR.

Right now you are testing a SE transformer with Rsource of 5 Ohm !! (no tube with such low Rp as far as i know), and loaded with 5 Kohm // with the reflected load from the secondary.

Hi, schiller ! Thanks for suggestions! I have not connected tested transformer backward (secondary to the generator) as it have been done by many, I've connected primary (shunted with 5K resistor) straight to generator with 5 Ohm output impedance. So generator should "see" 5k shunt in parallel with 5k transformer input impedance. Please correct me if I'm wrong.

PS. One fellow noted a roll-off from 12 to 20 KHz. However, the whole chart is ONLY 6dB range, and this roll-off is just -1dB, its within norm.

[DF96]

A differential test would be good, but you should be able to get a reasonable idea just by using the right resistive loading. If testing single-ended this is 2xRp in series with the primary. Why are you worried about what impedance your sig gen sees? As long as this is high enough not to damage it the real issue is what impedances the transformer sees. The frequency response is secondary voltage/ sig gen output (not primary voltage).

What you are seeing is the HF resonance of the transformer.

[tubelab.com]

I've connected primary (shunted with 5K resistor) straight to generator with 5 Ohm output impedance.

As mentioned the 5K resistor should be connected in series with the transformer, not in parallel to it. The generator should behave with any load even an open. You want the transformer to see an impedance representative of what the output tubes look like. It is rather hard to determine what the tubes look like since that depends on the class of operation, how they are connected, and how the feedback is implemented.

It is rather safe to say that 5 ohms does not represent any common tube. If you get the same peaking only more pronounced when the resistor is connected in series, you are looking at the usual resonance between the leakage inductance and the distributed winding capacitance. Since this is out of the audio band, build your amp and tweak the feedback as needed for proper square wave response.

[SY]

As long as you don't have DC running through it, you absolutely can test it single-ended. Use a resistor of 2x the plate resistance of the tubes you're going to be using in series with the generator (no need for a primary shunt) and a load resistor on the secondary equal to the nominal load resistance (8R, if that's what the transformer is spec'ed for).

You will almost surely see an HF resonance- going to a larger core just about insures that your leakage inductance will be higher, so you're likely to have some limits on the high frequency response.

[20to20]

As mentioned the 5K resistor should be connected in series with the transformer,

I believe the first step in this excersize is to determine the true input Z of the transformer at some recognized test freq. like 1kz. Then that resistance is put in series to the primary for the following bandwidth test.

[LinuksGuru]

Originally Posted by tubelab.com

As mentioned the 5K resistor should be connected in series with the transformer, not in parallel to it.

Originally Posted by SY

As long as you don't have DC running through it, you absolutely can test it single-ended. Use a resistor of 2x the plate resistance of the tubes you're going to be using in series with the generator (no need for a primary shunt) and a load resistor on the secondary equal to the nominal load resistance (8R, if that's what the transformer is spec'ed for).

Many thanks, tubelab and SY! What is the purpose of resistor (equal to Raa or 2xRaa) connected in series with primary? In my setup generator will "see" 5K of primary anyway, and output impedance of this generator is only 5 Ohm. And what this resistor should be - equal to Raa or 2xRaa? Different people suggest different things, so I have to ask again.

PS. Yes, transformer is indeed too large, and those peaks and dips at 40-50 Khz may be result of this. On another side, even at max power flux is below 8000G, so distortions caused by the core are minimal. Additionally, this over-sized monster can handle 10mA DC imbalance of output tubes idle current without degradation of LF response. At least, it looks so on my calc sheet.

[hemgjord]

You are not very interested in what impedance the generator will see are you? To get a somewhat representative frequecy plot you will need to assure that your transformer "sees" a resonably representative driving impedance i.e. the output impedance of your PP stage.
In Class A the output impdance of a PP stage is approx 2xRp. To simulate this you need to put a resistor of 2xRp in SERIES with your generator.

/Olof

[DF96]

Don't think about what impedance your generator will see. This is irrelevant. The issue is what impedance the transformer sees. At present it sees the 5 ohm output impedance of the generator, which is far too low. The 5k you added in parallel has no effect whatsoever - 5 in parallel with 5k is 4.995 !

You can add either two separate Rp resistors for balance, or one 2xRp resistor for simplicity. We are all telling you the same thing! The purpose of this resistor is to model the source impedance of the valve anode. Note that this is the anode impedance from the data sheet, not the matching impedance. I suspect that 5k may be the matching impedance, as it sounds too high for a triode and too low for a pentode?