Forums

transformer thermals

Started by Unknown August 4, 2020
In article <5h5kifpi9csk1ps3r90q0ht7srn3qf1ucj@4ax.com>, 
jlarkin@highlandsniptechnology.com says...
> > We can sense the primary current of this transformer, with a shunt and > an isolated delta-sigma ADC. The FPGA squares the samples and filters, > so we can pick that up and square root to get RMS current. The tranny > is rated for 240 VA, which would be 9.6 amps RMS in the primary. > > So I ran it for a few hours with 10 amps DC in the primary. Temp rise > was about 26C in free air. I think people design transformers for > equal copper loss in the primary and secondary, so temp rise would > double when loaded in the system. But wafting a little air over it > cuts the rise in half or so, so I'm back to something like 25 c rise > in real life, where there will be lots of air. So we'll set the > software shutdown at 12 amps maybe. > > https://www.dropbox.com/s/ylm8dc1e14dwv7y/P900_Xfmr_Thermal.jpg?raw=1 > > https://www.dropbox.com/s/5b55ybfoq2pkp9v/P900_Xfmr_Thermal_2.jpg?raw=1
That reminds me of when I was a schoolboy my Dad got me inolved in rewinding a transformer to get 20V AC from 240V mains. I don't remember what the original secondary rating was. This was ostensibly to run some 20V decorative candle bulbs for pretend safe candles for the upcoming Christmas. Only when presents were opened I discovered that the real reason was to power a 20V Meccano electric motor (which I still have ~65 years later.) Anyway, the relevance was that we put back only ~ half of the laminations before trying it so see whether our calculation of the turns-per-volt was correct. When we got 20V AC I went rushing off to tell my Mum of our success (I was only junior!). Only to be called back by my Dad shouting that the transformer was smoking... After a pow-wow, we decided that maybe all the laminations were needed to keep the losses within bounds, and finished re-building the core. That transfomer gave years of service subsequently, and was not the only one I built or rebuilt. I even used my Meccano to build a coil-winding machine! Mike.
On Wed, 05 Aug 2020 13:21:31 -0700, boB <boB@K7IQ.com> wrote:

>On Tue, 04 Aug 2020 19:35:39 -0700, jlarkin@highlandsniptechnology.com >wrote: > >> >>We can sense the primary current of this transformer, with a shunt and >>an isolated delta-sigma ADC. The FPGA squares the samples and filters, >>so we can pick that up and square root to get RMS current. The tranny >>is rated for 240 VA, which would be 9.6 amps RMS in the primary. >> >>So I ran it for a few hours with 10 amps DC in the primary. Temp rise >>was about 26C in free air. I think people design transformers for >>equal copper loss in the primary and secondary, so temp rise would >>double when loaded in the system. But wafting a little air over it >>cuts the rise in half or so, so I'm back to something like 25 c rise >>in real life, where there will be lots of air. So we'll set the >>software shutdown at 12 amps maybe. >> >>https://www.dropbox.com/s/ylm8dc1e14dwv7y/P900_Xfmr_Thermal.jpg?raw=1 >> >>https://www.dropbox.com/s/5b55ybfoq2pkp9v/P900_Xfmr_Thermal_2.jpg?raw=1 > > > >Why aren't you measuring this transformer using AC ?
Because there's a 10-amp DC supply on the bench.
> > It is going to be hotter with AC.
Hardly any.
> >DC will certainly give you a best case reference though. >
jlarkin@highlandsniptechnology.com wrote:

> I think people design transformers for > equal copper loss in the primary and secondary
BTW, what is the source of this and similar rules of thumb (e.g. equal copper and core losses)? Logic says that one should always design for minimal total losses, given the economic constraints. Best regards, Piotr
boB wrote:

----------
> > Why aren't you measuring this transformer using AC ? > > It is going to be hotter with AC. > > DC will certainly give you a best case reference though. >
** The standard test involves shorting the secondary with an amp meter and applying enough AC volts to the primary to get the rated current. This will, after a time, establish the temp rise and copper losses. DC only heats one winding and needs an expensive supply. JL likes to do things the hard way. .... Phil
On Wed, 5 Aug 2020 23:32:02 +0200, Piotr Wyderski
<peter.pan@neverland.mil> wrote:

>jlarkin@highlandsniptechnology.com wrote: > >> I think people design transformers for >> equal copper loss in the primary and secondary > >BTW, what is the source of this and similar rules of thumb (e.g. equal >copper and core losses)? Logic says that one should always design for >minimal total losses, given the economic constraints. > > Best regards, Piotr
I don't think transformers are usually designed for equal core and copper losses. This one runs cold at full AC voltage but no load. They are designed for equal primary and secondary copper losses at full load, I think. Cooling depends on surface area. You get more surface area by adding more copper, and that relationship is not linear (the volume-surface thing, like mice and elephants) so core loss might require a lot of expensive copper.
On Wed, 5 Aug 2020 16:33:48 -0700 (PDT), Phil Allison
<pallison49@gmail.com> wrote:

>boB wrote: > >---------- >> >> Why aren't you measuring this transformer using AC ? >> >> It is going to be hotter with AC. >> >> DC will certainly give you a best case reference though. >> > >** The standard test involves shorting the secondary with an amp meter and applying enough AC volts to the primary to get the rated current. > >This will, after a time, establish the temp rise and copper losses. > >DC only heats one winding and needs an expensive supply. > >JL likes to do things the hard way. > > >.... Phil
The 10-amp DC supply was right there on the bench, plugged in even. We may have an old Variac somewhere in the basement, but I haven't seen it in years. The DC measurement told me what I wanted to know. It wasn't hard. The fan has a USB cable for power, and I didn't have a USB power supply, but the oscilloscope has a memory stick connector on the front, so I used it to run the fan. The scope was also right there, plugged in too. https://www.amazon.com/gp/product/B00080G0BK/ref=ppx_yo_dt_b_search_asin_image?ie=UTF8&psc=1 Really cool gadget.
 Piotr Wyderski wrote:

===============================
> > > > I think people design transformers for > > equal copper loss in the primary and secondary > > BTW, what is the source of this and similar rules of thumb (e.g. equal > copper and core losses)? Logic says that one should always design for > minimal total losses, given the economic constraints. >
** The source is simple calculus that finds the minimum or maximum in a curve. It's also kinda obvious that if the primary and secondary run at different temps then you have one heating the other. Same goes for core and windings. The issue with tape wound toriodals is their very sharp saturation curves - forcing the designer to keep away from that condition. Hence their low contribution to heat from the core. Commercial toroidals are made using a clever machine passing wire through the centre hole - this sets a limit on the amount of copper that can be used. I know of one amplifier designer who at my suggestion had the winder use his machine for the primary and wind the secondary by hand in order to fill the hole almost completely. This almost doubled the amount of copper used. Along with a bit of fan cooling, the result was a 3kW rated amplfier that used a 1kW size transformer core. ..... Phil
On Wednesday, August 5, 2020 at 1:45:36 PM UTC-7, John Larkin wrote:
> On Wed, 05 Aug 2020 13:21:31 -0700, boB <boB@K7IQ.com> wrote: > > >On Tue, 04 Aug 2020 19:35:39 -0700, jlarkin@highlandsniptechnology.com > >wrote: > > > >> > >>We can sense the primary current of this transformer, with a shunt and > >>an isolated delta-sigma ADC.... > >>So I ran it for a few hours with 10 amps DC in the primary. Temp rise > >>was about 26C in free air.
> >Why aren't you measuring this transformer using AC ? > > Because there's a 10-amp DC supply on the bench. > > > > > It is going to be hotter with AC. > > Hardly any.
Iron lossses due to eddy currents and hysteresis aren't insignificant. It this for variable frequency use? Those iron losses are frequency-dependent. it might take a bunch of testing to characterize. Early switchmode power supplies occasionally heatsinked the transformer cores...
whit3rd is Witless wrote:

-------------------------
> > > > It is going to be hotter with AC. > > > > Hardly any. > > Iron lossses due to eddy currents and hysteresis aren't insignificant. >
** They are with GOSS tape wound into a toroid.
> It this for variable frequency use? > Those iron losses are frequency-dependent. > it might take a bunch of testing to characterize.
** No it wont. JL's tranny is rated for 50/60Hz and will work fine with lower losses at any higher frequency. Core magnetisation goes DOWN with rising frequency for a given primary voltage. Never seen an audio output tranny in your entire life ?
> Early switchmode power supplies occasionally heatsinked the transformer cores...
** So what ? .... Phil
On Wed, 5 Aug 2020 18:01:49 -0700 (PDT), whit3rd <whit3rd@gmail.com>
wrote:

>On Wednesday, August 5, 2020 at 1:45:36 PM UTC-7, John Larkin wrote: >> On Wed, 05 Aug 2020 13:21:31 -0700, boB <boB@K7IQ.com> wrote: >> >> >On Tue, 04 Aug 2020 19:35:39 -0700, jlarkin@highlandsniptechnology.com >> >wrote: >> > >> >> >> >>We can sense the primary current of this transformer, with a shunt and >> >>an isolated delta-sigma ADC.... >> >>So I ran it for a few hours with 10 amps DC in the primary. Temp rise >> >>was about 26C in free air. > >> >Why aren't you measuring this transformer using AC ? >> >> Because there's a 10-amp DC supply on the bench. >> >> > >> > It is going to be hotter with AC. >> >> Hardly any. > >Iron lossses due to eddy currents and hysteresis aren't insignificant. It this >for variable frequency use? Those iron losses are frequency-dependent. >it might take a bunch of testing to characterize.
200 Hz to 4K. This simulates a 6-pole PM alternator on a jet engine. Primary current is very low when the transformer is unloaded. It's a good transformer. -- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard