Power Toroid Wdc vs rated VA and acceptable temperature rise

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
16,152
Location
Norfolk - UK
When designing a linear power supply it is well known that you usually calculate the ac current rating of the winding as some factor times the dc current. Part of the reason is that the peak dc output is 1.414 times the ac winding voltage so the current needs to be at least 1.414 times the dc current because power in must be at least as much as power out. The current multiplier is usually 1.6 for a full wave bridge. See Sowter info here:

http://www.sowter.co.uk/rectifier-transformer-calculation.php

One design question is how big to make the reservoir capacitor. For low ripple bigger is better. This also makes regulation easier because it tends to reduce the voltage drop across the regulator and hence its dissipation. In a 12V 2A supply I use a 22000uF capacitor which gives a peak to peak ripple of less than 1V.

Recently however, I have discovered that designing using these formulae tends to lead to quite significant temperature rises is the transformer which begs the question what is an acceptable temperature rise in a toroidal mains transformer. I then came across this data sheet at Farnell:

http://www.farnell.com/datasheets/2200017.pdf?_ga=2.243874893.939696366.1502612378-493727691.1475222435

This is unusual in that not only does it separately  list iron and copper losses but ait lso specifies a temperature rise which for a typical 50VA toroid is a whopping 50 degrees Celsius and this assumes an ambient of 40 degrees. It is not clear if this is core temperature rise or the rise for the transformer as a whole but wither way it implies significant temperature rises are not unusual.

Then most recently I came across this piece by a transformer manufacturer.

http://www.precision-inc.com/power-toroid-p-1296-l-en.html

wguch gives a grapgh of dc output power versus rated VA of the transformer for various core sizes and dc voltages.  The dc applications section begins by saying " A transformer VA rating must always exceed the WDC rating due to the equivalent circuit series resistance, rectifier voltage drop and high peak currents during the charging of output capacitor"

This is the first time I have seen mention high peak capacitor charging currents in a transformer manufacturer's literature. The really interesting part is that for low voltages like 12V, you should design for a dc output of about  55% of rated VA which gi gives a current multiplication factor of 1.8 rather than the usual 1.6.

I have experimented with one transformer that at present is giving what I consider to be an unacceptable temperature rise. It has winding for three different supply voltages. By using just the one that consumes about 2/3rds of the power, the temperature rise is quite modest - just warm to the touch.

All this is leading me to the conclusion that, for a conservative design, it might be best to add up all the dc powers, double it and call that the VA rating of the transformer.

I am no expert on transformers so any insights would be welcome.

Cheers

Ian

 
ruffrecords said:
When designing a linear power supply it is well known that you usually calculate the ac current rating of the winding as some factor times the dc current. Part of the reason is that the peak dc output is 1.414 times the ac winding voltage so the current needs to be at least 1.414 times the dc current because power in must be at least as much as power out. The current multiplier is usually 1.6 for a full wave bridge. See Sowter info here:

http://www.sowter.co.uk/rectifier-transformer-calculation.php
I seriously doubt these formulae, in particular the factor 1.11 that seems to come out of nowhere.
Simulations in LTspice show that there is no such factor, as well regarding average and rms values than instantaneous. Indeed the differences due to the number of diode voltage drops are correct, but assuming consistency in use of materials, resulting in a CT winding having a total resistance of twice that of a single winding, the net result is just about status quo.


One design question is how big to make the reservoir capacitor. For low ripple bigger is better. This also makes regulation easier because it tends to reduce the voltage drop across the regulator and hence its dissipation. In a 12V 2A supply I use a 22000uF capacitor which gives a peak to peak ripple of less than 1V.
Indeed. According to Q=C.V, dQ/dt=C.dV/dt, hence I=C.dV/dT or dV=1/C.dT At 50Hz, dT=10mS and for C=10mF (10 000 uF), dV=1V


Recently however, I have discovered that designing using these formulae tends to lead to quite significant temperature rises is the transformer which begs the question what is an acceptable temperature rise in a toroidal mains transformer. I then came across this data sheet at Farnell:

http://www.farnell.com/datasheets/2200017.pdf?_ga=2.243874893.939696366.1502612378-493727691.1475222435
These are also somewhat questionable, since their calculations seem to be based on resistive loads at rated power, which is not the way most of these transformers are used most of the time.


This is unusual in that not only does it separately  list iron and copper losses but ait lso specifies a temperature rise which for a typical 50VA toroid is a whopping 50 degrees Celsius and this assumes an ambient of 40 degrees. It is not clear if this is core temperature rise or the rise for the transformer as a whole but wither way it implies significant temperature rises are not unusual.
It seems to me pretty consitent with reality. And I wouldn't really be concerned with an electronic component sitting at an even higher temperature. Transistor junctions commonly sit at 80°C.


Then most recently I came across this piece by a transformer manufacturer.

http://www.precision-inc.com/power-toroid-p-1296-l-en.html

wguch gives a grapgh of dc output power versus rated VA of the transformer for various core sizes and dc voltages.  The dc applications section begins by saying " A transformer VA rating must always exceed the WDC rating due to the equivalent circuit series resistance, rectifier voltage drop and high peak currents during the charging of output capacitor"

This is the first time I have seen mention high peak capacitor charging currents in a transformer manufacturer's literature. 
It's often neglected but real. Voltage drop due to resistive and inductive parasitics impair voltage regulation significantly because the charging currents are extremely brief and intense.


The really interesting part is that for low voltages like 12V, you should design for a dc output of about  55% of rated VA which gi gives a current multiplication factor of 1.8 rather than the usual 1.6.
OTOH, I seriously doubt the graphs that show considerable reduction in available DC power for full-wave bridge. These lack explanation of how tests were conducted. They don't make sense to me as they are. As to the variants with a choke at the input, it's all so dependant on the value of the inductor that it's impossible to characterize the circuit with a single value. Unfortunately, these "formulae" continue to reappear on the internet like they were cast in bronze.


All this is leading me to the conclusion that, for a conservative design, it might be best to add up all the dc powers, double it and call that the VA rating of the transformer.
How much conservative is conservative? The manufacturer has probably dialled in a factor 1.2-1.3, so you're adding your own comfort zone. I have consistently used thousands of toroidal xfmrs with a margin of about 1.3 and never had a single issue with temperature or performance.
 
> for a conservative design, it might be best to add up all the dc powers, double it and call that the VA rating of the transformer.

Agree.

Ignoring diode losses, you could use 1.6X for temp rise and 1.8X for sag. But because standard parts go in 1.25+X steps, such exactitude is not needed. And you never complain about a bit too much, while a bit short is a disaster (buy and ship a new PT, use your chassis-stretcher to make it fit....). 

> a whopping 50 degrees Celsius

This is a quite normal rating. "Class A". You can design for 125C rise, but the materials are more expensive; also the copper-loss rises.

http://solarprofessional.com/sites/default/files/articles/ajax/docs/7_SP8_2_pg18_Kondrashov.jpg

> an unacceptable temperature rise

You may consider this like the power rating of your car. Mine claims 150HP out of 2.4L with a radiator little bigger than a magazine. I go to the truck shop and their 150HP is a 5L with a rad bigger than a spread newspaper. My car will do 150HP but (if I find a way to do that all-day) it will burn-up soon. The truck will do 100-150HP all the way up Wyoming or the Alps, every day for years. Also note that it may be hours to reach full rise. A gigging guitar-amp may not get full hot before the joint closes- car-class is OK. Studio gear may run 12 hour shifts or even be left on 24/7, you want the truck. 
 
abbey road d enfer said:
I seriously doubt these formulae, in particular the factor 1.11 that seems to come out of nowhere.
Simulations in LTspice show that there is no such factor, as well regarding average and rms values than instantaneous. Indeed the differences due to the number of diode voltage drops are correct, but assuming consistency in use of materials, resulting in a CT winding having a total resistance of twice that of a single winding, the net result is just about status quo.

Indeed. According to Q=C.V, dQ/dt=C.dV/dt, hence I=C.dV/dT or dV=1/C.dT At 50Hz, dT=10mS and for C=10mF (10 000 uF), dV=1V

dV = i/C.dT not 1/C.dT;I assume this is a typo but for i =2A it gives 20000uF for dV = 1V
How much conservative is conservative? The manufacturer has probably dialled in a factor 1.2-1.3, so you're adding your own comfort zone. I have consistently used thousands of toroidal xfmrs with a margin of about 1.3 and never had a single issue with temperature or performance.

That is indeed the point. Right now I am specifying transformers according to the 'standard' rules and unlike you I am getting very hot transformers. How hot I will be able to say as soon as my IR temperature meter arrives but it is certainly too hot to touch and much hotter than I  have experienced before.

Cheers

Ian
 
ruffrecords said:
dV = i/C.dT not 1/C.dT;I assume this is a typo but for i =2A it gives 20000uF for dV = 1V
Correct. My intention was to demonstrate that for each amp you need 10 000 uF for 1V ripple. I failed miserably for lack of an explanatory line.

  Right now I am specifying transformers according to the 'standard' rules and unlike you I am getting very hot transformers. How hot I will be able to say as soon as my IR temperature meter arrives but it is certainly too hot to touch and much hotter than I  have experienced before.
Is it something you experience with toroidals and didn't with conventional xfmrs?
 
abbey road d enfer said:
Correct. My intention was to demonstrate that for each amp you need 10 000 uF for 1V ripple. I failed miserably for lack of an explanatory line.
Is it something you experience with toroidals and didn't with conventional xfmrs?

No, I have used toroidal transformers successfully in the past. It it just this one combination that seems to be giving me trouble. Secondaries are:

250VAC @ 100mA for 60mA dc HT supply (simplr CRCRC filter) = 25VA
50VAC @ 60mA for Phantom power for for mic pres (TL783 regulator) = 3VA
12VAC  @ 3A for 1.8 amp dc heaters (LTR1084 regulator) = 36VA

Total VA = 63VA

Running just the heaters or just HT plus phantom and the transformer runs cool. Running all three and it really gets hot. I have two made by two different manufacturers. One gets rather hot but possibly OK in the light of what has been said already. The other gets very very not.

Something is happening in the core because I look at the preamp output noise floor and I can see the mains harmonics in it. They are not huge - worst case -80dBu at 60dB gain but the spectrum is odd. There is a small fundamental, absolutely no second harmonic  and then a 3rd harmonic higher than the fundamental ( and a lot of higher harmonics of a similar amplitude).

The lack of second harmonic implies it is not sourced from the rectified signal. At the moment I am running two transformers. One runs the heaters alone and the other runs the HT and phantom. Both transformers are barely warm but the strange harmonic structure is still there. If I turn off the heater transformer the fundamental and all harmonics instantly disappear. Just for fun, tomorrow and am going to try ac heaters.

Cheers

Ian
 
Phantom power winding indicates that you're planning on using this in a high-gain scenario - so you will want this transformer as electromagnetically quiet as possible.

For that, you'll want to run the core at only 70-80% of it's rated gauss (what you do is that you calculate the primary as if it was made for a significantly higher voltage, e.g. 290V in stead of 230) - this keeps it away from core saturation and thus discortion.

as a side benefit, core will heat less

If you're looking into manufacturers, I can recommend http://toroidy.pl - talk to Tomasz, he's one of the few that actually understands people like us :)

Jakob E.
 
ruffrecords said:
Total VA = 63VA

Running just the heaters or just HT plus phantom and the transformer runs cool. Running all three and it really gets hot. I have two made by two different manufacturers. One gets rather hot but possibly OK in the light of what has been said already. The other gets very very not.
What size did they come with? 63VA is at the limit between 60 and 80VA types. I guess the manufacturers chose the smaller (cheaper) one.
Also some manufacturers choose to run their cores at high induction, even if it's not strictly needed for the application. They have one typical primary for each core, that is rated for the highest power (and highest induction), even if the actual power need is less. You have to tell the winder to use lower induction because it's in a sensitive application. If they don't understand this, find another. Increasing the number of turns by 20% generally produces a drastic reduction of stray field. In some cases, that will imply using a larger core.
 
abbey road d enfer said:
What size did they come with? 63VA is at the limit between 60 and 80VA types. I guess the manufacturers chose the smaller (cheaper) one.
Judging by the dimensions ( over 90 mm dia and nearly 40 mm high) they are 80VA types.

Cheers
 
I guess one major difference between a toroid and a regular Ei transformer is that an Ei can quite effectively loose heat via conduction to the chassis its bolted too ,a toroid cant do this .
The old style Ei units usually had an interwinding screen to prevent capacitive coupling between windings ,im not sure if thats even possible or applicable with a toroid . I have only one tube mic power supply using a toroid ,I made a few mods to it ,including bypasing all the lytics with foils ,useing schottky diodes , and moving the toroid to the far corner of the enclosure. It definately improved the immunity to mains garbage ,things like switch on spikes from flourescent lighting were more or less completely masked by circuit noise after the mods were done . I wonder how those R core transformers out of china work in tube power supplies,there seems to be quite a range of voltages available in sizes 30 to 100 va .Of course it makes alot of sense to try and find local suppliers of components where ever possible ,keeps the profit and the jobs at home ,rather than supporting whats most likely little better than slave labour in chinese factories.
 
Just for comparison ,I made a preamp with an EL84 , I reused an old philips radio transformer for ht,6.3v winding supplying an Ez80 rectifier. Then for the heater of the El84 I used a 2x6 volt 12 va EI transformer (series connected ), I initially had 20,000 uf in two caps directly after the bridge,then a 317 To3 regulator down to 6.3volts dc.Transformer did run a bit hot and so did the regulator ,so I put in a 2 ohm resistor between the two caps and a common mode choke after the regulator  . I guess this made life a bit easier for the transformer and shaved a couple of volts off what the reg had to drop  . 780 ma/6.3v is the rating of the el84 ,Im a bit surprised that now my transformer runs warm but not hot . 12 volts at  1 amp delivering 780ma at 6.3 volts ,its maybe a bit under spec'd ,but it all works away grand now.
 
Tubetec said:
I guess one major difference between a toroid and a regular Ei transformer is that an Ei can quite effectively loose heat via conduction to the chassis its bolted too ,a toroid cant do this .
That is true but it's not all? Since toroidals make better use of iron and copper for a given power rating, the mass is actually lower. Since winders use the same design objective in terms of relative  loss, the temperature is higher with toroidals. In addition, the part that is accessible to the hand is the winding, which is subject to more losses than iron, thus increasing the perception of temperature.


The old style Ei units usually had an interwinding screen to prevent capacitive coupling between windings ,im not sure if thats even possible or applicable with a toroid .
Toroidals with electrostatic screens are available. Winders do not like them much because they don't lend themselves easily to automation.


I wonder how those R core transformers out of china work in tube power supplies,there seems to be quite a range of voltages available in sizes 30 to 100 va .
The big advantage of R-core xfmrs is they are usually wound with completely separate primary and secondaries, one on each jamb, which minimizes capacitive coupling. They also make a better use of iron compared to EI core, but not as good as toroids. It's a different optimization.
 
note that toroid power transformers usually have a thermal switch taped to the outside winding,

transformer engineers hate thermal, it is the biggest chapter in the book,

it is not bad for small transformers, but if you are designing for Hoover Dam, you do not get a second chance,
 
abbey road d enfer said:
...... The big advantage of R-core xfmrs is they are usually wound with completely separate primary and secondaries, one on each jamb.....

Are you sure? R-core transformers are usually wounded using balanced windings to  achieve low leakage. I newer saw separately wounded  primary and secondaries on R and C cores.
 
ruffrecords said:
Judging by the dimensions ( over 90 mm dia and nearly 40 mm high) they are 80VA types.

Cheers

I just purchased an 80VA standard 2 x 12V secondary toroid and it is definitely bigger than the custom toroids I have. The very good news is that using just one sec of this transformer for the heater supply ensures both it and the one providing HT and phantom power run barely warm. Another very nice benefit is that the stray radiation from this under utilised transformer is very low which has improved my mic pre EIN figures by a couple of dB.

As part of this exercise I also decided to try a compact SMPSU as a heater supply. The lunch box needs about 1.8A at 12V for the heaters which is less than 22W. I decided to use a 50W SMPSU which is capable of 4.2A at 12V in order to account in part for inrush current into cold tube heaters. In selecting a SMPSU I discovered a new (to me) method of current limiting in SMPSUs called hiccup mode. It does not have the power dissipation problem of a conventional current limit, nor the lack of current of foldback current limiting. It is a kind of constant power limit so it seems ideally suited to tube heaters. So it was with some trepidation that I connected it up to the heaters of the lunchbox and turned it on. I was pleasantly surprised to see the tubes fired up in a couple of seconds. I then turned on the HT/phantom and looked at the mic pre output noise spectrum. I was astonished to find it was the cleanest totally hum component free noise floor I have ever seen in any of my designs. No splatter from the SMPSU, no noise getting in via the heater circuits and after running for an hour the little thing was  still cool. Here is the one I used:

http://uk.farnell.com/webapp/wcs/stores/servlet/ProductDisplay?catalogId=10001&langId=44&urlRequestType=Base&partNumber=2363864&storeId=10151

I need to do some more tests but I think I will be using SMPSU heater supplies from now on.

Cheers

Ian
 
Hey Ian, I went trough this not to long ago.  It was a toroid for a large tube compressor. Sixteen 6BA6's plus +/- 18V for the sidechain and various other things.  From working on tube guitar amps all my life, I tend to feel things out rather than do the math, as you probably already know:) I started out by doubling the heater DC current draw for the AC winding spec. Did the same with the +/- 18V windings, but for the HV I ran out of space so I could only go about 1.5 times the DC current draw or I would have had to move up to the next size.  The proto type I ordered worked well and didn't get to hot. Pretty warm but not to hot to touch.

I had the company send one more updated one with some of the voltages lowered as I wasn't sure how to calculate the regulator loss. But one revision wasn't bad. Basically I used CJ's method, "buy then adjust". The company I used was https://toroid.com/

They are in the USA so not much help to you but if anyone else is specifying a custom toroid, they are very helpful and prices are fair.

I also looked into the switch mode power supply option. I had good results as well with some off the shelf supplies I had lying around for the heaters. But when I looked into having custom units made with all the voltages I needed it was difficult to get small quantity's made.  Even considered designing my own, but gave that up after about 20 min of googling SMPS designs. Too much math for me but you might want to look into it.
 
moamps said:
Are you sure? R-core transformers are usually wounded using balanced windings to  achieve low leakage. I newer saw separately wounded  primary and secondaries on R and C cores.
That's an option when optimizing for low leakage, which is a concern for audio transformers, wher  minimizing leakage is essential for good HF response. OTOH, for power xfmrs, leakage is not a big issue, but capacitive leakage between primary and secondary is a concern for medical and audiophool applications.
 
bluebird said:
I also looked into the switch mode power supply option. I had good results as well with some off the shelf supplies I had lying around for the heaters. But when I looked into having custom units made with all the voltages I needed it was difficult to get small quantity's made.  Even considered designing my own, but gave that up after about 20 min of googling SMPS designs. Too much math for me but you might want to look into it.

I think SMPSUs come into their own where you need a regulated voltage at high current (like heaters) and then off the shelf product will do the job. For HT I never regulate because all my stuff is class A so you don't need it. A linear supply for phantom is dead easy with a TL783 so between them that takes care of all my power needs.

It is a different matter for class B or AB push-pull power amps where the current draw is signal dependent. A much harder nut to crack.

By the way, I forgot to mention, the little 12V 4A SMPSU for the heaters is pretty much the same volume as the 80VA toroid, about the same price and about 10% of the mass.

Cheers

Ian
 
Back
Top