On Vintage Power Transformer Operating Temperature

BinaryMike

Pelagic EE
There have been many threads on this topic, so when I found an authoritative treatment in the Radio Designer's Handbook (Classic Edition), I decided it was worth quoting here verbatim:

Temperature Rise

This is dependent on the cooling area, the total loss and the ratio of iron to copper loss. To avoid deterioration of the insulating materials, it is necessary to limit the working temperature to 105°C for Class A insulation which includes paper, cotton, silk, varnish and wire enamel. The temperature rise in the winding, as measured by the change in resistance method, will be about 10°C lower than the maximum (hot spot) temperature. Thus with an ambient temperature of 40°C (104°F) plus a margin of 10°C for the difference between measured and hot spot temperature, it will be seen that the maximum permissible rise is 55°C, as measured by the change of resistance.

It is common practice to allow 10°C margin for change in line voltage, frequency, or operation in situations with restricted ventilation. Thus 45°C is generally accepted as the maximum permitted rise above ambient when measured by resistance change.

The temperature difference between winding and core varies between 10° and 20°C according to the distribution of losses. This means that even with an ambient temperature of 25°C (77°F), the core temperature may be 60°C (140°F). This will feel quite hot to the touch, although the internal temperature may well be under the permitted maximum. Measurement of the core temperature may be made with a spirit thermometer if good contact is maintained between the core and the thermometer bulb.

Langford-Smith goes on to discuss the winding resistance measurement method in detail, but his example assumes high enough winding resistance that standard two-wire ohmmeters are adequate. Be aware that low resistance windings such as heater windings generally require the use of four-wire ohmmeters for sufficient accuracy and repeatability. Modern IR thermometers are convenient for core temperature measurements.
 
For the sake of completeness, here's the relevant formula for determining winding temperature rise via DC resistance measurement:

Tr = (Rhot - Rambient) / Rambient * 0.00393

For example, given a winding that measures 10R cold and 12R hot, its temperature rise is:

Tr = 12 -10 / 10 * 0.00393 = 50.9°C with estimated hot spot of 61°C. To see how this plays out, add your maximum expected ambient operating temperature for this transformer, say 40°C in the cabinet, in Summer, and a 10°C safety margin, to get 111°C. In this case, we can expect winding insulation rated at 105°C to deteriorate when operating conditions are stressful.

To boil this down a bit, we can say that a 20% rise in winding DC resistance is about the limit for a vintage transformer, regardless of core temperature. And that's with all windings loaded per normal operation.
 
There have been many threads on this topic, so when I found an authoritative treatment in the Radio Designer's Handbook (Classic Edition), I decided it was worth quoting here verbatim:..
BinaryMike, thank you for this posting. :thumbsup:
 
This thread should be bookmarked. Seems like once a month or so there is a question about how hot is too hot for a tube amp power transformer.
 
It's the fourth edition, originally of 1953, revised 1967, reprinted 1997 and 1999.
So up to at least 1999, this is still current. All my tube amps. p.trans. get hot. to about 70 deg.C. on the core. Yet, they run fine year after year and are 60 years old. So why worry so much. Electronics and price point drive the transformer specs. To have more thermal headroom, it takes more steel. More steel is more expensive to buy and to ship in the final product. So, it's always a compromise.
 
So up to at least 1999, this is still current. All my tube amps. p.trans. get hot. to about 70 deg.C. on the core. Yet, they run fine year after year and are 60 years old. So why worry so much.

I'm specifically not stating you are incorrect, because you aren't, just that heat has a price and you may not have yet seen it.

Transformers in audio, like those for most applications, are designed to meet a target point. That point is fuzzy and is sometimes barely surpassed or just missed shy of reliability. Transformer heating often indicates a problem with marginal windings or overloaded cores. Witness the power transformer failures plaguing the Heathkit Williamson amplifiers. Metal more rapidly oxidizes and migrates with temperature, and a minor flaw, such as occurs during wire draw, can cause failure.

Transformers sometimes are overloaded, such as when designers use a transformer specified for a full-wave rectifier with a half-wave rectifier. The form factor is often more than enough to push a transformer into overheating which shortens its lifespan.

For a consumer device which is not often used (see below) the heating never becomes an issue. But that doesn't mean it isn't a potential problem.

To have more thermal headroom, it takes more steel. More steel is more expensive to buy and to ship in the final product. So, it's always a compromise.

Yes. But it is not merely steel:
(1) The alloys used for lamination have different thermal conductivity. Some are rock-bottom inexpensive. Others require approval from a vice-president.

(2) The physical construction of the core affects heat transfer. A modern microwave oven transformer is made from inexpensive steel stacked and strip welded to hold it together because screws are too expensive for this application. Thermal conductivity between the laminations may not be what it could be for a better design. But overheating in this application isn't critical, because the duty cycle is low, the product must meet a low retail price point, and lifespan is generally not a serious issue as long as it is long enough, a fuzzy term defined by returns and damage to the brand name.​
 
Back
Top Bottom