Demagnetizing a Quad II output transformer - might it need doing ?

[GrimJosef]

I picked up a pair of Quad II amplifiers at RetroTechUK a week ago. One of them looks as though it may have suffered thermal runaway in the output stage. What can happen is that one KT66 switches hard on (perhaps because of a leaky interstage coupling capacitor, or simply because the valve has gone 'gassy') and the action of the common cathode resistor then results in the other KT66 switching equally hard off. The valve that's on can draw a lot more DC current than either the power supply or the output transformer primary is rated for and both the mains and output transformers can overheat as a consequence. In the suspect amp substantial amounts of bitumen (several cc in each case) have oozed out of the transformers' casings, indicating that the overcurrent has persisted for quite some time.

I wanted to check the condition of the output transformers in both the amps I bought and a relatively quick way of doing this is to put suitable load resistors across the primary and the cathode windings (I chose 2k5 and 1k5 wirewounds respectively) and then to drive significant audio power from another amp 'backwards' through the transformer under test. The above load resistor values reflect back as about 8R5 measured at the amp's speaker sockets. I used another (known good) Quad II to supply 12W at 700Hz to the transformers under test.

My first test was of the amp which hadn't overheated (or, at least, showed no signs of having done so). I measured the audio signal on its output transformer primary using a 100R tap which was part of the 2k5 resistor chain. Sure enough the performance was fine. The signal from my drive amp, fed into the transformer under test's speaker secondary, had about 0.10% total harmonic distortion as measured using my HP8903A audio analyzer. The signal on the primary side of the transformer under test had about 0.11% distortion, so the transformer appeared to be increasing the distortion by just a small amount.

When I repeated the test on the output transformer in the other amp - the one with the oozed bitumen - the input signal distortion was again low (about 0.11%) but now the distortion on the primary side was 0.3%. This still isn't a very large number. The Quad II uses significant negative feedback and it may well be that that would be capable of correcting the 0.3% distortion down below the Quad II's 0.18% THD specification (0.1% if the second harmonic is disregarded). However there seemed to be a clear difference between the two transformers and I wondered if this indicated a problem.

In 1987 Dr G.A.V.Sowter of Sowter Transformers, who know a thing or two about audio transformers, wrote in this paper https://www.sowter.co.uk/pdf/GAVS.pdf:

When a transformer has been subjected to large values of direct current or has been in the vicinity of a strong permanent magnet, such as too near a loudspeaker, it will assume a polarized state or become magnetized. Fortunately this seldom has a permanently harmful effect on the core material, although in its magnetized state it will have higher distortion and reduced audio output. The process of demagnetizing is quite simple and consists of applying to one winding an alternating current (for example, at 50 Hz) of value appreciably exceeding that required for saturation. This current should then be reduced smoothly and gradually over a period of time, such as one-half to one minute ... It is important to spend the bulk of the time on the low values of demagnetizing force, which should be reduced to absolute zero.

(My bold.) It strikes me that a large DC overcurrent through just one half of a push-pull transformer primary, as happens when one valve is switched hard on and the other is turned practically off, might lead to core magnetization. Can anyone advise whether this really is an issue and, if it is, how I might work out how much 50Hz AC current would be needed to wipe this magnetization away ? I do have some fairly high-current low-voltage mains transformers and a large mains Variac.

Cheers,

GJ

[Radio Wrangler]

It sounds feasible.

In considering the distortion figures, remember that the transformer is within the overall feedback loop and its distortion will be improved.

With one Quad II driving another Quad II transformer backwards, probing the connection between them may show distortion, but from both transformers at once as they're in parallel.

One Quad II won't be able to demagnetise the other transformer without hitting saturation itself. If you turn off abruptly you can leave both magnetised.

So mains is the right way to go. Feed via a resistor and monitor another winding with a scope. Take to a high enough level to see the onset of core saturation. The crucial part is to bring the level down smoothly without jumps or interruptions and bring it all the way down to trivial current.

It'll be interesting to see if there is any difference afterwards. You may just be seeing core to core variations not related to that one's unpleasant experience.

Normal music isn't exactly symmetrical, and there will be differences in KT66 quiescent currents so the risk is in saturation level currents during that run-away.

David

[kalee20]

I can't see there being a problem really - the iron in decent transformers will be 'soft' so although it could retain magetisation (high remenent magnetism Br), it's easily demagnetised again (low coercive force Hc).

In use, particularly at high-amplitude low frequencies, the flux in the transformer will swing around significantly. If it starts off one-sided, the first half-cycle swing may well be curtailed if it tries to drive the core further towards saturation, but the subsequent half-cycle will take the magnetisation considerably away from saturation - the core will 'walk' towards zero, very much what can happen with (particularly) toroidal mains transformers on switch-on.

Radio Wrangler's post - feed the Quad transformer from a Variac via a resistor and monitor with a 'scope - is bang on. You might need a lot of volts, the Quad amplifier is designed (may be wrong!) to deliver full output at 20Hz, so at 50Hz you'll need 2.5 times the full-output voltage to give the same flux swing. Adding to that is that (hopefully) a decent margin away from saturation will have been designed-in - you could be looking at 4x normal output voltage, so you might have to choose carefully the winding you inject the demagnetisation voltage into.

How much current? It'll be a few milliamps - you know from the KT66 data what the anode current swing is, and on the basis that almost all of this will go to the load and only a small fraction (<10%) will be 'wasted' in core magnetisation current, you can see this to be true.

[GrimJosef]

Thanks for your thoughts gentlemen.

I re-ran the experiment a couple more times using two other Quad IIs I have here and the distortion on the primary side of their output transformers was in each case 0.11%, suggesting that the earlier transformer i.e. the one with signs of an overcurrent incident, is an outlier at >0.3% (although I accept that a sample of four still isn't very large).

Originally Posted by Radio Wrangler

... One Quad II won't be able to demagnetise the other transformer without hitting saturation itself ...

Yes. However I do somewhere have one with a 100V-line output transformer and if the current requirement really isn't going to be very large then that might serve to deliver enough volts. It should also allow me to do the demagnetizing at 20Hz. I confess I'm a bit nervous about a huge voltage breaking down the transformer's insulation and if I can achieve the required H at lower applied or induced volts then that would be altogether more comfortable.

I'm lucky that Keith Snook has done such a detailed job measuring the physical and electrical parameters of the Quad II transformer (all the info's at the bottom of this page https://keith-snook.info/quad-ii-val...amplifier.html). I'll see if I can work out (any help much appreciated !) a connection to the windings which should do the job.

If the 100V-line signal doesn't seem appropriate then I also have a large jack-of-all-turns-ratios Gardners output transformer to hand (pictured, with 20p in the distance for some sort of scale). It contains a great deal more iron than the Quad II transformer so I might hope that it could deliver a clean 20Hz signal without serious saturation.

Cheers,

GJ

[retailer]

Interesting thread - my thinking may be all wrong here but if the residual magnetism in the Tx core is say 0.25 Tesla why do you need to take it saturation, which may well be greater than 1.6 Tesla ? surely all you need to do is take it to say 0.5 Tesla and gradually reduce it from there, if the Tx is not already installed in an amp then a variable 50Hz mains supply is the obvious choice but if the Tx is in a working amp would not a sine wave into a dummy load work just the same - gradually reducing the amplitude.

[Radio Wrangler]

You need to take it to saturation with AC so that the saturation characteristic itself provides centring of the DC component of the residual flux. Another view is that the AC field takes the core around the periphery of its B/H curve right into the saturation level... The smooth and progressive reduction of the AC field causes the core to be exercised over a shrinking B/H loop until the loop closes with minimal hysteresis and then it finally shrinks to a point centred on 0,0 on the B/H plot.

Standard demagnetisation technique. Driving the core into non-linearity and hysteresis, then bring ing it out. If you don't take the core outside its comfort zone you can leave it with any residual magnetisation intact.

In an amplifier, it is unlikely that notrmal operation will take a core as far as needed unless something big goes wrong, as in the case of the amp G-J is looking at, and this is where the suspected residual magnetisation is suspected to have come from, one output valve running away to an effective short and creating enough ohmic heating to melt the tar out, so the current had been at damage level for that amplifier. The demagnetising process needs to be comparable, but thankfully of limited duration.

David

[trobbins]

It may also be worthwhile testing the insulation resistance of the primary to core, and to the secondary, in case the fault related high temp has caused some internal dc leakage.

[Valvepower]

Hi,

Just picked up on this thread as I’ve been putting some leg work into the next BVWS article…

I’m not sure if there is any ‘sensible’ relevance here, but back in the guitar amp days in the 1970’s I remember a guitar amplifier coming in for repair where it just sounded ‘bad’ with a ‘discordant’ musical sound.

Replacing valves didn’t make a vast difference.

We had limited test equipment, so it was case of just measuring the output power and frequency response which measured as one would expect, however the amp just sounded ‘bad’ and ‘discordant’.

We did suspect this amplifier had been run into an incorrect load impedance for some considerable time.

One of the chaps said…. Well lads, it sounds if the output transformer is cooked. We replaced the output transformer, and it sounded fine.

This has stuck in my mind, and I think I’ve experienced varying degrees of this with valve amplifiers over the years, when you get to work on amplifier that ‘seems OK’ but is just sounds under par, where one suspects it’s had a hard time possibly driving incorrect loads.

Looking back at this amplifier, I feel there was a level of asymmetric and/or odd order harmonic distortion or such like of the notes of the guitar… I sometimes hear this when a [preamplifier] stage is incorrectly biased thus affecting the operating point or an ‘imbalance’ fault in a LTP phase splitter.

I’ll admit it’s a tad out there and wacky, but it’s my two penn'orth.

...Ooh, Radio Caroline is playing the Motors, Dance the Night Away... best go and turn up the volume! Pub Rock as it best!

Terry

[Gabe001]

I've come across this thread too, and I'm following it with interest. I like GJs threads, they're very instructive.

I'm keen to see how the de-magnetization goes and what effect this will have on THD and the overall sound

Good luck.
Gabriel

[kalee20]

Originally Posted by retailer

Interesting thread - my thinking may be all wrong here but if the residual magnetism in the Tx core is say 0.25 Tesla why do you need to take it saturation, which may well be greater than 1.6 Tesla ? surely all you need to do is take it to say 0.5 Tesla and gradually reduce it from there

Good question, but... how would you know there's a residual internal field of 0.25T? Unless you saw a slot in the core, and slip in a Hall-effect sensor / snatch out a search coil and measure the integrated induced voltage / (insert your favourite field-measuring technique).

No, the way to demagnetise is to supply enough volt-seconds to take it to a defined hard limit like saturation, cycle it symmetrically, and gradually reduce the swing to zero.

(What happens when 'done' and you then switch on your amplifier, with valves mismatched to 2mA, probably negates the exercise. But here, I'm going to be interested in the outcome!)

[retailer]

A figure of 0.25T is just hypothetical, taking a magnetised object to saturation during demagnetisation doesn't tie in with my experiences of magnetising and demagnetising, but then I'm not talking about transformer cores but ordinary objects found in a workshop the type that are made of hard steel and pickup nuisance metal filings, objects like lathe tools and dial/digital calipers.
The worst is dial/digital calipers - any foreign objects may affect the measurements taken, I had been successfully demagnetising them with an old Akai tape head demagnetiser, this works even though the magnetic field from one of these is miniscule, it does though take a long time going back and forward over the length of the calipers, if I concentrate only on the caliper jaws time taken is reduced but the magnetism soon returns. To overcome this I cut the centre section from an old bathroom exhaust fan motor frame - plug this into a variac and go over the whole calipers, it's been almost 12 months and the calipers are still not picking up metal filings. In using this as a demagnetiser I've found that the variac voltage required to demagnetise seems to depend on how hard an object is magnetised, so I assumed one only has to swamp the residual magnetisation and then gradually reduce from there, however this is all just conjecture as even though I have a gauss meter and a magneto magnetiser I haven't tried to confirm this with experiments.
Over the last few years I've noticed an odd thing with my AC stick welder, the breaker (not the RCD) in the workshop will trip when I turn on the welder, thereafter no matter how many times I switch it on and off the breaker won't trip, this only happens if I haven't used the welder for some months, at first I thought it may just be coincidence that I switch on the welder when the mains voltage is hitting a positive or negative peak, but since I became aware of it, I now take notice and can say with a good amount of confidence if I haven't used the welder for 4 or 5 months the breaker will trip when the welder is switched on.