Testing Power Transformers

[catalin gramada]

Thanks for You answers gents. It is not the case to blame the product as I didn't tested yet. But simply this transformer is suspect by its size in respect with similars. I ordered it by curiosity (due to its claimed specs) to serve in a DIY project. But I want to perform a test in respect with most tech agree. I can put it with all secondaries on resistive batteries bench (have a full shelf of power resistors laying around), I can do precise voltage measuring and temperature aso. It is a regular Marshall 100 size, the specs are:
Pri: 120V, 230V, 240V @ 50Hz
Sec1: 356 VCT @ 600mA
Sec2: 6,3 VCT @ 7,5A
Sec3: 98 V @ 150mA
Sec4: 12,6 V @ 1A
Should be able almost 300va by its nominal rating...will see
Any ideas to perform a objective proper test fully appreciated. Should be very interesting if I can collect enough data to have a picture of its characteristics. It will be very instructive aso (for me at least). Thank You.


...and No, it is not a MM btw....

[R.G.]

If you're interested in pursuing this further in pursuit of the knowledge here's how I would set up the test.

Arrange a place where it is safe for the test to run to completion. This is an issue because a full test will run many hours, even a day or two. Trobbins is correct - it is amazing how long a proper test to stable temperature will take. The thermal time constant for that much mass of iron and copper is likely hours, and a simple exponential rise is assumed to be close enough at five time constants. This is not a simple exponential rise, given that the real underlying heat transfer is not a simple exponential rise, but better described by a complex differential equation involving the fluid thermal dynamics of air moving past the outside of the tranformer to carry heat away and whatever amount of radiation can happen. It's going to run for a long time. That's why I mentioned approximating the end temp by continuously mapping the temperature rise curve so you can make a decent estimate of the final temperature and call off the test when you get a "good enough" answer.

Since it will run a long time, and have exposed high voltages, you need to exclude other people and pets from the test area so they don't get electrocuted. Also, unattended high power tests always involve some danger of fire, so worry about that a little. I've worked with AC mains power long enough to be paranoid about leaving anything plugged in and running when I'm not there to deal with a smoke-and-flames situation.

Second, think about how you will measure the resistance. The typical ohmmeter isn't good enough for the low voltage windings. These things need a four-wire milli-ohmmeter to get usable accuracy. You might be OK with a primary or high voltage winding that gets up into the 100's of ohms, but for a complete test you want to measure them all. The windings will generally not rise at the same rate or have the same final ending temperature, as power transformers do not try to distribute the windings together like OTs do. They're simply layer wound in chunks, and so the have different thermal resistances to the outside. One will be hotter than the rest, and there is generally a hot spot, usually the innermost winding, that runs 10C to 15C hotter than the outer winding. So you need to measure them all. You don't need to worry about polluting the test by the time you spend un-powering the transformer, disconnecting the leads, measuring all the resistances, and reconnecting. If this takes a few minutes, so be it. The very long thermal time constant works for you because it is just as slow for the thing to cool down as it is for it to rise. The time spent measuring is negligible compared to the time of the whole test. This may not be true for small transformers, though.

Third, yes, as Trobbins mentions, you need to get the RMS current up to what it would be in an actual use. Rectification does involve RMS currents in the windings that are larger than the DC out of the rectifier/filter combination. For heaters, this can be ignored (mostly) because the heater load is usually resistive AC powered heater windings. For windings that drive rectifier/filters, you need to compensate by taking into account the DC loading AND the RMS transformer current to get that much DC.

To find out how much transformer RMS current results from the DC output current, you need to do some math. Rectifier circuits with capacitor filters make the heating (RMS) current seen by the transformer winding be different from the DC current pulled out of the filter capacitor. This is worst for the full wave bridge rectifier set-up, not too bad for the full-wave center tap setup. You can either replicate the rectifier/filter/load you would see in an amp, or compute the transformer RMS current for a given DC load after rectifier/filter and use a resistor load to give the transformer AC the same loading as it would see with the rectifier/filter load. Both work. RMS is RMS for heating.

[J M Fahey]

FWIW , now knowing your core size, 114x95 mm EI lamination, stacked 50mm, "absolute maximum 100% perfect" VA capability is 361 VA ; Real World "good quality" commercial stuff (such as Drake, Dagnall, Hammond and equivalent) about 80% of that, so some 290VA .

I wind transformers all day long using that lamination, buy it in 50kg and 100kg lots; we call it N° 155 following its US denomination (roughly 1.5" center leg width)

It is fine for a standard 4 bottle 100W Tube amplifier ... or a 300W RMS SS amplifier.

Not sure about your KT88 project.
If two of them, it should easily cope; if four .... very unlikely.

Sec 1: 356 VCT @ 600mA 213VA <-- this sounds about right for an unefficient 100W Tube amp , no way you can pull 150 or 200W RMS out of it; I wouldnīt expect more than 50% efficiency, given that SSamps are about 75% efficient ... and donīt have to waste power in heaters, screens or plain saturation voltage (about 60V) or even pure waste in high idle current "just to sound better"

Sec 2: 6,3 VCT @ 7,5A <-- 48VA IF it adds up to your hungry filament needs, no complaints. But ... does it?

Sec3: 98 V @ 150mA <-- 15VA No big deal.

Sec 4: 12,6 V @ 1A <-- 12VA Same.

Losses are more into copper DCR than actual magnetic ones, the coil under constant load heats up more (and faster) than core itself.

As RG said, coil temperature can be accurately measured by measuring copper resistance increase ... which is huge.

As of load testing, resistors are not a real world load ... since you amp power supply is not a resistor by far.

IF you want realistic load testing, load filament winding with resistors to full rated current with resistors, that oneīs easy, BUT for the HV winding, ONLY realistic load is to connect it to a diode bridge, proper capacitors, and load resulting DC with resistors pulling same current as amplifier at full blast.

Only other way is to actually build the amplifier and burn it in. Not kidding.

[nevetslab]

IF you want realistic load testing, load filament winding with resistors to full rated current with resistors, that oneīs easy, BUT for the HV winding, ONLY realistic load is to connect it to a diode bridge, proper capacitors, and load resulting DC with resistors pulling same current as amplifier at full blast.

Only other way is to actually build the amplifier and burn it in. Not kidding.

Or, in lieu of the planned amplifier, substitute that with a similar amplifier that would be running the same planned power level. I'd set up the connections of your Transformer-Under-Test with quick Disconnects, or even a well-isolated terminal strip so you can get the Kelvin 4-wire test clips onto the multiple pairs of windings. Obviously with AC mains disconnected. That's how I had characterized prototype power transformers for new products in the engineering lab. I also had the prototype xfmrs wound with K-type thermocouples to add to the temperature rise trend, but relied on the change of resistance of the copper in all primary/secondary windings. Also tracked the ambient temperature in the localized test zone, as it changes over time from the radiated heat, and is in the equation.

[catalin gramada]

Finally my PT was coming. German quality made. EI 114 / 50mm stack. At first sight it looks acceptable built, but two remarks: the lamination looks thick, actually twice thicker than regular Hammond use,for instance, and the end bells are huge, there are more than twice deep than is necessary, which could be a problem for room accomodation on some chassis. Pictures below:





And my power jig: 610 ohm/240w and 0.9ohm/60w




I didn't found my temp probe for my Fluke so have to improvise something else

[Helmholtz]

From the pictures it looks as if the core is stacked with alternating lamination pairs. Nothing wrong with that. What is the brand/ manufacturer?

[catalin gramada]

From the pictures it looks as if the core is stacked with alternating lamination pairs. Nothing wrong with that. What is the brand/ manufacturer?

Hi, It is a Tube Town transformer and think is one they made it "in house". It did not looks like Chinese rebranded one but more like craft one instead. Just judging by wires/leads they used stuff for professional electrotechnic application...just an example, It have a touch of professional electrotechnic workshop, if you know what it means...So I believe is a well built transformer judging by how it looks like. ...I have a problem with its specifications instead...

[Helmholtz]

The wire resistance method gives an equivalent average wire temperature along its length. For comparing to the (to be known) insulation class temperature limit, the internal hot-spot temperature needs to be known. Is there an established method to derive/estimate the hot spot temperature from the average for a given transformer size?

I remember that we had to provide the certification lab with transformer samples having embedded thermo couples to get the safety approval.

[Helmholtz]

Hi, It is a Tube Town transformer and think is one they made it "in house". It did not looks like Chinese rebranded one but more like craft one instead. Just judging by wires/leads they used stuff for professional electrotechnic application...just an example, It have a touch of professional electrotechnic workshop, if you know what it means...So I believe is a well built transformer judging by how it looks like. ...I have a problem with its specifications instead...

The Tube Town specifications add up to 288.15W secondary power. I think that's feasible, especially with grain-oriented (as advertized) laminations and over-sized coil former. It is clearly advertized as a replacement for a 100W Marshall - not for a 200W 4xKT88 Major.

[catalin gramada]

The Tube Town specifications add up to 288.15W secondary power. I think that's feasible, especially with grain-oriented (as advertized) laminations and over-sized coil former. It is clearly advertized as a replacement for a 100W Marshall - not for a 200W 4xKT88 Major.

Clearly understood , less than didn't bought this PT as an replacement item but for its electrical specifications which should be in conformity with what is state. If it is proper suited for one or another application this is another story. In my understanding it should work on its rated specs, except the fact of overloading somehow.
I made some custom powertransformers in a past at one workshop in my town. I don't remember to be asked from one technicians what I need it for? I specified my volts/currents demands an they dimensioned and made it by request. It is a first time I heard about this nuances like ..ok it is good enough if you use it like this or that....It is a simple mains transformer , for Christ, should no matter if I supplied a bulb , a power resistor or a power amp. As time I don't exceed its specs should be able to perform as same in terms of V/A
If it will not work into a Marshall Major 200w will be because its voltages and current ratings didn't recommend it for those application and not because this transformer it was recommended by seller as an 100w Marshall replacement....just my 2c

[catalin gramada]

Allright, I have some preliminary measurements. With 230V on the mains and 7A draw from heaters winding I got 366V no load/ 6.12V heaters, and with 610ohm load for 580mA get 352,5 V but heaters drop to 6 V. This is a damn good regulation, less than 5 per cent.Pretty unhappy by heather voltage but think I can fix it if have room to wind one or two turns over insulation. Further tests will follow. Thank you guys!

I will let it toast for a while but still not have a proper temp probe to determine the moment when comes in balance with ambient so it have no sense to let it toast for hours without to be able to have a conclusion

...and to not forgot: It is absolutely quiet. I cannot heard it at all

[trobbins]

Yeh I'd imagine a type test may somehow include embedded sensors, or a confident margin is used based on previous type tests of similar parts.

With respect to 'full spec load' testing, you still end up having to assume (or make an educated guess) the designer's insulation system rating, and also rationalising your own operating margins, such as for environment ambient temp, and the 'average' music power you could be operating at given that would be more than idle power but less than continuous sinewave testing for x hours.

If you don't want to include a rectifier and filter in to your test jig, you can use PSUD2 to estimate the rms winding current for a given DC output loading, based on your rectifier and filter part choice.

[catalin gramada]

Yeh I'd imagine a type test may somehow include embedded sensors, or a confident margin is used based on previous type tests of similar parts.

With respect to 'full spec load' testing, you still end up having to assume (or make an educated guess) the designer's insulation system rating, and also rationalising your own operating margins, such as for environment ambient temp, and the 'average' music power you could be operating at given that would be more than idle power but less than continuous sinewave testing for x hours.

If you don't want to include a rectifier and filter in to your test jig, you can use PSUD2 to estimate the rms winding current for a given DC output loading, based on your rectifier and filter part choice.

Thanks for You advices. I will but have not conditions for the moment. Further tests will follow.
Btw. It is from almost one hour under loads and core temp raised insignificant, max 5 Celsius degree. It is same cold. But this is something relative.I still not have a temperature probe to perform a test.
Till this point I.m happy with what I get.

[J M Fahey]

Cool results so far, but please remember:

It is a simple mains transformer , for Christ, should no matter if I supplied a bulb , a power resistor or a power amp.

is absolutely not realistic

Filaments and resistors pull sinusoidal current from a sinusoidal voltage, and thatīs what you are measuring so far, but diodes feeding a *loaded* capacitor pass narrow peaks of very high current , absolutely different from the first case.

RMS Loss/dissipation/Temperature rise (all related to each other) depend on the square of current.

Consider a simple 50% duty ratio compared to continuous/full cycle load (resistive).

Not mentioning any particular resistor, since value is same in all examples:

* 1 A during 1 second (resistive load) means "X" power

* 2A during 0.5 second means same average but twice RMS power

* now consider that in typical amplifier power supplies capacitor charging duty cycle is around 10% to at most 20% (under very heavy load) and youīll see that loss situation is serious.

From experience: 5% regulation with resistive load, and the only parameter manufacturers supply, and they swear by it because "I measured it at the factory test bench" translates to 15% , often 20% drop in amplifier voltage rails (rectified and capacitor filtered DC).

Of course, hobbyists and diyers donīt care about cost and often overspec what they use, so sometimes they get better numbers, but thatīs a particular case, not widespread.

So in a nutshell: in principle I expect youīll find it a VERY GOOD 100W Marshall transformer, that does not mean it will feed a Major, not even halfheartedly

[catalin gramada]

Thanks JMF. So, to understand the application (meant a bridge/capacitor supply which should provide a certain DC current onto a load) determine an real operating current which is different from rms one. The abilities of a PT to sustain this type of current is a matter of transformer characteristics to can manage it in respect with "thermal stress". From this point a insulation classification became very important for transformer abilities.Over that should be an equivalent model which can translate this current regime (taking in consideration the load and capacitor battery)into a rms current one. I.m far out of math to can calculate but now all suggestions started to have more sense for me. The application (power supply) is not a rms one but have a rms equivalent which should be related by transformer characteristics. A true test should include the real PS application
In this case measuring the rms current between transformer and rectifier and raised progressive the load in DC side monitoring permanently the tranny temperature till stabilization point should be pretty close, please ?



I can modify the value of load in DC side rising the current from transformer till 600 mA ( supposed my true rms dmm integrated the type of currents by PS application )
After stabilization I can move the measurement after rectifier to see how much resources have in DC side
My question for now is if my dmm is realy able to show true rms value no matter by current operation conditions? Or is just a rms when operate only with sinusoidal current ?

For my understanding is more simple to conclude: a 300mA bridge power supply application request a 600mA RMS power transformer...for instance. But if rms value is only derivated from sinusoidal currents and not from whatever peaks current operations , things become more complicated to get it