PDA

View Full Version : OT any transformer experts here?

darryl
01-07-2012, 08:03 PM
I expect a few here at least will know about this-

I have not been able to find much about the relationship between transformer core size and power capability. I'm talking about the typical EI core steel lamination transformer core construction.

I have a formula that suggests that multiplying the core area (in sq cm) by itself will give a good representation of the power capability of the transformer. In other words, a transformer with a core 2.5 cm wide and stacked 6 cm high gives an area of 15 sq cm, which should be good for about 225 watts. I can find examples of power transformers from audio amplifiers which are quite close in core area, but are a factor of two apart in total weight. Offhand, the heavier of the two would seem to be capable of more power, but not according to the formula. Magnetic path length is longer on the larger transformer and winding area is larger, though the same distance around the core.

How does magnetic path length enter into the equation? I understand that a longer center leg gives more length to wind the wire onto, so you could decrease losses by using a larger gauge of wire, or by keeping more of the wire closer to the core for a shorter length overall, thus reducing copper losses. Also, I have read that the optimum height of the stack of laminations is equal to the width of the center leg. But if you double the height of the stack, for instance, you are doubling the amount of core area, doubling the power capability, but only increasing the length of the wire by 50%. At the same time you are reducing the number of turns required, so it would actually seem like you would be improving the efficiency by making the height of the lam stack more than the width of the center leg.

Turns per volt seems to be closely related to core area, but how does magnetic path length relate to turns ratio?

I understand that transformers are designed differently depending on use- intermittent use under full load gets you maximum power for core area, continuous use under mainly part power lets you minimize the magnetizing current for lower operating temperature, but costs in the ultimate power producing capability, etc. My questions relate to the latter type of use.

I'm leaning towards mating the lamination stacks from two or three transformers into one stack, thus making one higher powered transformer of the same low profile.

Black_Moons
01-07-2012, 08:15 PM
Most of the transformers iv seen have all been pertty much the same aspect ratio, Likey theres an ideal aspect ratio for turns/core volume per length of copper...

I'll also note that older transformers had WAYY more iron/copper to em for a given wattage, I suspect its mainly they just wanted em to be more efficent, never get close to saturation and had money to spare.. also copper was wayyy cheaper back then.

If you need serious wattage, Why not just take the MOT (Microwave oven transformer) outta a microwave, cut through the 2kv secondary (BEWARE THE CAPACITOR in the microwave in series with the HV, it may still be charged with a LETHAL charge) and rewind onto that?

If you need low profile and high power, Look into toroidal transformers.

dp
01-07-2012, 08:31 PM
You might give this a try, or search for other transformer design calculators.

http://www.electronicecircuits.com/electronic-software/transformer-calculation-software

darryl
01-07-2012, 08:43 PM
I have several MOT's, some of which I have already partially prepared for other uses. One of the problems with them is that they are seam welded in a few spots, which pretty much precludes taking them apart. You would then have to wind the turns through the core, which is time-consuming, and risks damaging the insulation on the wire. You can do a fairly nice job of it, but what a pia.

Another problem with them is that they are wound for a fairly high flux density- not a problem if the load is high all the time, but it does have a high idle current and would easily overheat if left on for any real length of time. You would have to add turns to the primary to reduce the magnetizing current- or cut all the wire off and start fresh. Seems the best use of a MOT is for a spot welder or similar high current, low voltage output and intermittent operation. It's less demanding to wind on relatively few turns through the core, but if you need to place say 120 turns for a primary, then possibly two bifilar winds of say 50 turns for the secondary, you're in for a frustrating time.

Dp, thanks for that link. I'll check it out- don't think I've seen that one.

macona
01-07-2012, 08:52 PM
Someone posted a formula (Want to say evan) a while back that allowed you to estimate a transformer's capacity from the core size. Dig though the archives and you might find it.

alanganes
01-07-2012, 09:00 PM
The power handling is primarily related to the core area, the area of the center leg of the core. Path length does not really come into it. Mag path length does not affect the turn/volt ratio. The formula is this:

Turns = volts/[4.44*B*A*F] where:

B = flux density
A = core area (center leg for "EI" laminations)
F = frequency

The considerations for intermittent vs. continuous duty are primarily ones of heat removal. The major contributor to to losses is generally IR loss in the windings and not magnetizing current unless you are talking about pretty small transformers.

darryl
01-07-2012, 09:57 PM

Just used the software program to give some figures- an MOT that came from a 600 watt oven is good for 380 watts, and the turns per volt is also about 50% higher than original. It does appear that the calculations are based on a fairly conservative design, which probably allows for continuous operation of the transformer without too high of a temperature rise. One thing I won't know is whether the steel in the transformer is good for a higher flux density than the program might be taking into account- the only way I can know this is to manually experiment with the core to see how high I can raise the input voltage before the core heats to an unacceptable degree. That's a test I can set up easily using my variac.

I can also monitor the input current as I raise the voltage- at some point the current will begin to rise more quickly than the voltage, and I'll call that the saturation point. The final design would be set for a somewhat lower idle current than that.

Once I find a suitable maximum input voltage, I can check the turns per volt ratio by adding say 10 turns around the core and checking the output voltage from that. Then I'll compare that to what the software program says for that same core.

Bruce Griffing
01-07-2012, 10:10 PM
A couple of points. MOT's are designed for intermittent use and are normally fan cooled. Using the same core for another application should, in most cases, be at least half the power level. The design limits for a transformer are limited by core saturation and total power dissipation. The core area factor is all about the saturation limitation. But you must also pay attention to resistive, eddy current and core power dissipation as well.

This link is a good start at an explanation ----

http://ludens.cl/Electron/Magnet.html

J Tiers
01-07-2012, 11:53 PM
In the most general sense, the size of transformer has NO relation to power capability...... but that is if frequency is allowed to vary. Even then, the core size is almost irrelevant to power..... The flux in the core is pretty constant at any power level, and in fact REDUCES slightly as current increases.

it is fairly constant because the secondary current produces a flux opposing that of the primary current, balancing out to leave the magnetizing flux as a remainder.

it reduces with current because some of the primary voltage is dropped in wire resistance, and is no longer "available" to drive current through the primary inductance.

The main limitation is heating of the wire.... With excessively high current the wire IR losses get to be too high, and the output voltage is no longer usefully close to the open circuit voltage. So the transformer heats up and puts out a low voltage.

To avoid that. you need to use bigger wire.... but to fit that, you must increase the winding "window" area... the open space for wire. if you do not, you cannot fit enough turns in to get an inductance high enough to keep the transformer from saturating the core.

if you increase the "window" area, you increase the magnetic path length.... that reduces the inductance per turn, and gets you back in trouble with magnetizing current.... So you have to increase the core cross-section to get the inductance per turn back to a usable level.

That's how the transformer size gets related to power FOR ANY GIVEN FREQUENCY.

But, at higher frequencies, you need less inductance to hold down the magnetizing current. That means you can use fewer turns, and a smaller core with smaller winding window....

I have not been able to find much about the relationship between transformer core size and power capability. I'm talking about the typical EI core steel lamination transformer core construction.

I have a formula that suggests that multiplying the core area (in sq cm) by itself will give a good representation of the power capability of the transformer. In other words, a transformer with a core 2.5 cm wide and stacked 6 cm high gives an area of 15 sq cm, which should be good for about 225 watts.

I understand that a longer center leg gives more length to wind the wire onto, so you could decrease losses by using a larger gauge of wire, or by keeping more of the wire closer to the core for a shorter length overall, thus reducing copper losses. Also, I have read that the optimum height of the stack of laminations is equal to the width of the center leg. But if you double the height of the stack, for instance, you are doubling the amount of core area, doubling the power capability, but only increasing the length of the wire by 50%. At the same time you are reducing the number of turns required, so it would actually seem like you would be improving the efficiency by making the height of the lam stack more than the width of the center leg.

Turns per volt seems to be closely related to core area, but how does magnetic path length relate to turns ratio?

Longer path length means less inductance per turn*, so you need a bit more in turns to get the same inductance. Not a big deal for most sensible sized transformers.

Long thin coils tend to be better at higher frequencies to reduce losses which increase rapidly with the number of layers of winding. At 60 Hz it is not such a big deal how many layers of winding there are.

When you stack a lot of laminations to get a long that is far past the normal square cross-section, the transformer becomes less efficient. I used to call them 'caterpillars".... vendors would offer them up as alternatives to toroids for thin spaces, but they were a distant 3rd in usability...

The added wire is very significant in terms of "regulation", the voltage drop at full power. Also that construction is horrible for heat dissipation, although it tends to take a while to heat up. With so much wire "buried" away from cooling air, AND the added resistance of more wire adding even more heating, the net effect is a double whammy of bad heat removal....

You don't want to compare it against a smaller transformer, you want to compare it against an "optimized" part of the same capability.... with generally same total cross section of core, etc. That shows up the inefficiencies.

* thought experiment on inductance per turn.....
Imagine a core of generally optimized design, another one with a core having a path length of 10 metres but same cross-section, and a third with only a short section of core sticking out perhaps 20mm at each end of the coil, not "closed".

Obviously the first will have more inductance per turn than the third with its "open" core. Most of the path length of the third unit will be in air, with permeability of approximately 1.

Now, the second unit, with a 10 metre path length, is a lot closer to the third than the first. The long path has a higher magnetic reluctance, "absorbing" more ampere turns to drive the same flux through the core, and the only way to reduce that for the same path length is to increase the cross section.

darryl
01-08-2012, 01:33 AM
Ok. I'm starting to get a better handle on this now. Looking at the stack length compared to the leg width, I see that if you make one loop around the leg, then expand that loop into a circle, the volume of the circle would be much greater than what is enclosed within it when it's a long rectangle. Same length of wire in either case- less enclosed iron in the long rectangle than in the circle. A square core cross section would be closer to optimum then, and you would only deviate from that if you had physical size constraints, and then not by much.

It would seem that if the total transformer structure was more or less cubic, that would represent a close optimum relationship between all important factors. Fair enough.

I got through the formulas well enough to deduce my own equation, which in assuming a fixed frequency of 60 hz, and flux density factor for the steel of 1.2, boils down to this- turns per volt = 31/ core area in sq centimeters. Using this formula, the turns per volt that I'm measuring on two different transformers works out very closely. One measures at 3 turns per volt, while the formula says it should be 2.9- the other measures at 3.2 turns per volt, while the formula says it should be 3.17. Funny thing is, one of these runs with very little temperature rise, even when on overnight- the other is quite warm after only an hour of running. It's the core where the heat is being generated, not the windings. The laminations are the same thickness. One material is lighter in color than the other- it's the lighter one that's getting warm. Maybe there's less insulation, leading to higher eddy current losses, or maybe one has a different steel than the other.

Well, I've got some answers, and more questions at the same time- what else is new :) Thanks for helping me out.

Evan
01-08-2012, 02:23 AM
I'm leaning towards mating the lamination stacks from two or three transformers into one stack, thus making one higher powered transformer of the same low profile.

When you start messing around with iron core transformers you will find that making your own will often result in very inefficient transformers. The problem is the iron laminations. Recyling old ones will result in poor stacking, shorted laminations and work hardening from handling. Even the slightest amount of work hardening reduces efficiency dramatically.

The best laminations these days are laser cut because it produces zero mechanical stress on the iron and the laminations stay dead soft.

The Artful Bodger
01-08-2012, 03:21 AM
Does burning the varnish off the old laminations leave them in a soft state?

darryl
01-08-2012, 04:49 AM
Evan, I've never had that problem, but then I'm very careful with the pieces since I am aware of some of the problems. I've never heard of work hardening in laminations, though.

I've just re-built two transformers, stripping off the secondaries and winding on the custom ones I needed. They seem to be fine. These are the ones that run cold, which is good because they may end up being powered for days at a time. I do have four laminations left over that I couldn't fit back on- that's two per transformer. I'm not worried about it. They don't buzz, even though I haven't added any shellac. I suspect they will be fine.

Now I need two more- these have to be good for about 300 watts, though they won't be operated continuously at that level. The cores I wanted to use are too small, so I need to look in the dark recesses of the workshop to find a pair that would be suitable. All I'm finding so far are these MOT's.

Weston Bye
01-08-2012, 08:33 AM
Be aware also that the surface of the laminations is important also. Generally they are oxidized in some manner to produce an insulating layer that helps reduce current paths for localized eddy currents from one lamination to the next, preventing current losses and core heating. The oxide layer is not a particularly good insulator, just good enough, and cheap.

Duffy
01-08-2012, 11:40 AM
Darryl, I am in the process of a) dismantling a large, (about 200lbs,) transformer, and b) rewinding the secondaries of a couple of magnetron transformers.
The large one is not an E-I design, but rather a rectangle. The laminations are 0.014", dead soft and measure 3X5 and 3X11. If you want a few pounds, speak.
As far as Xfmr capacity, my reasoning is that if the unit drove a 1200 watt magnetron, then it probably is good for about 1.5 Kw, since it is really only a class C amplifier circuit.
I measured one of mine by wrapping 10 turns around the secondary window, and got 11.2V which I guess is a "good enough" approximation of one turn per volt. Since these are ALL 110 volt appliances, it follows that they all have a similar number of turns on the primary. I have observed that the newer models have copper coloured aluminum wire. Also, the old ones, (1981 vintage,) are A LOT larger for the power output. I guess the designers were still dumbing down, (optimizing,) the design to minimize cost.
All I am trying to do is produce a couple of workable 12V-20V power transformers to drive salvaged drill motors.

01-08-2012, 12:17 PM
Not sure why you are worrying about VA when intended to run some low voltage hand tools and the transformer weighs 200LB'S!!!!.
BTW, In new designs I use Toroidal type, very easy to modify or add a small overwind etc.
Max.

kf2qd
01-08-2012, 01:54 PM
Core size is also influenced by frequency. At higher frequencies the core can be smaller. Part of the reason that the military uses 400Hz in aircraft and ships.

Different frequencies are also used in foundries for melting different materials in induction furnaces. 60 Hz for iron, 400Hz for aluminum. Not exactly applicable to the topic, but thrown in for free...

The Artful Bodger
01-08-2012, 02:50 PM
Core size is also influenced by frequency. At higher frequencies the core can be smaller....

This is true and causes us the occasional problems in our 50Hz domain when US made transformers although rated for 50Hz run hotter than one might like and die a sudden death if the supply frequency sags (which it only does when on generator supply).

Perhaps if you are in a 60Hz area and a 50Hz tranformer comes into your hands that would be a fine candidate for experimenting having more leeway in the form of the extra iron.

Duffy
01-08-2012, 02:57 PM
Maxheadroom, you misunderstood. The transformer is from an impressed current cathodic protection system, removed from a building. It put out about 40 amps at up to 40 volts, through a selenium bridge rectifier. I could not figure how to adapt it for any useful purpose and it sure was overkill for my purposes, AND it was over 50 years old. I know, 50 years is nothing hell, I am over 50!:D

dp
01-08-2012, 03:43 PM
Many years ago I made a scratch-built high power audio amplifier. I needed +-68vdc for the power output stage, and +-12vdc for the rest of the circuit. I don't recall the current required. But I calculated the winding ratios needed to convert 110AC to the values needed (two secondaries), then calculated wire gauge necessary for the current.

Now I knew the window size of the IE transformer and in essence, the core size. I rummaged around my stash of parts, found an old TV transformer, cut off the old copper, and rewound it to my design. It worked perfectly.

I think that was in 1984 or so. I never finished that amplifier because I could not find one of the components in the parts list - I still believe that parts list was in error. The chassis is in my shop to this day awaiting my reborn interest in it. I have a 100W Fender amp now, though, so except for curiosity...

darryl
01-08-2012, 06:07 PM
I kind of got spirited away last night. The girls wanted to go dancing, so I didn't get home till after 1 am. Too late to play with transformers, so here I am today, after having slept in quite late, crawling around on the basement floor looking for hunks of iron.

I have a pair of transformers I can use, though they are a little larger than needed. That will be fine- if I need to reduce the idle current I can add a little winding to put in series with the primary.

It would be nice if I didn't have to disassemble the cores at all. Usually, when I want a low voltage, high current secondary, I cut a few strips from copper sheet, insulate one side of the strip with scotch tape, then just pass that through the slots. Each of these becomes one 'stack', or disc, and then I wind on another one beside that. I try to shoot for four stacks side by side, since that allows me to wire it up as a center-tapped winding. With care, I can get a decent fill ratio and low copper losses, and it's a lot easier to wind a flat strip fairly tightly than it is to pass heavy gauge wire through the core and end up with a tight winding. If I've scotch taped the inside of the copper strip, there's little chance of that becoming damaged while feeding it through the core slots.

Cutting out the copper strips from sheet is problematic, though. Without a custom machine to cut the strips equal width and roll off the burrs, it's just another time-consuming part of the total project. And if the copper strips are relatively thin, the scotch tape takes up a significant portion of the available winding space. An ideal width for the strip is something just less than the width of the scotch tape, so there are some limitations between thickness and the effective gauge of the strip. Sometimes it works out ok.

Just did some checking- the scotch tape I use measures .002 thickness- the copper I have on hand measures .006. 25% of the winding space is taken up by the insulation. A strip of this copper just under an inch wide is roughly equivalent to 10 gauge wire. Of course, when you flat-wind this, there is no air space between windings as there is when you lay round wire side by side, so this isn't really a bad situation.

I watched one video where is shows distribution transformers being made. It shows a flat strip being wound as one of the windings, probably the secondary. It's pretty much the width of the core. Nice that it leaves a very flat surface upon which to wind the higher voltage windings. Then I see the core being inserted- this was the interesting part, as it looks as if the core is U shaped pieces being inserted from either side of the winding bobbin, and it looked like they were thin. This would of course mean that several individual pieces would be nested together before the core is complete. There must be a way that they make a good edge to edge contact in order to maintain a high density flux path through the core. It seems no small feat to be able to achieve this for every piece of the core that is inserted, while keeping all these steel strips in contact as one 'lump'.

The other way to do this, and the one that interests me the most, is to have this steel strip as a length, which you then wind through the bobbin. Normally, you would wind two strips through the core, making two doughnuts. Essentially then, you have two torroids passing through the winding bobbin, and it should end up being very efficient. Not only electrically, but mechanically as well, since you can avoid having to pass miles of wire through the hole in a doughnut. It's the procurement of this steel strip that is problematic for me- where would you go to find a hobbyist quantity of suitable steel of a suitable width?

Bruce Griffing
01-08-2012, 06:42 PM
If you are interested in the most efficient way to build special transformers look into cut cores. You wind the primary on one bobbin and the secondary on another. You can take them apart and try alternatives. The only downside of this approach is that you have to buy the cores rather than recover them from scrap.

Evan
01-08-2012, 07:01 PM
Does burning the varnish off the old laminations leave them in a soft state?

I don't know. I'm not up on the annealing procedures for silicon iron, which is what transformer laminations are made from.

darryl
01-08-2012, 08:08 PM
I've only done this once, but it could happen more often in the future- I had all my laminations removed from the old transformer, gave each one a pass over sandpaper laid on a flat surface, blew them off, then laid them all out closely and gave them a coat of lacquer. We had the spray booth operating, so it wasn't a big deal for me to go in there on a weekend and do this. I gave each side one coat.

I should have bothered to measure the thickness before and after, but I didn't. We had lots of laminate laying around, so I made up my own bobbin from that. We also had lots of backing laying around, so I cut some of that into 2 inch squares to mix epoxy on. The tricky part was holding the bobbin pieces in good square relationship while the epoxy cured- the hardest part was cutting a nice and neat rectangular hole in two of the pieces where the core would pass through. Laminate tends to chip and crack, etc. A laser or water jet cutter would sure be nice for some of this work. Next time I build my own bobbin, I'm going to make it in two parts so there are no enclosed holes to cut.

I've just finished completely destroying the secondary windings on one of the last two transformers I need to wind. I've measured the turns per volt and now need to figure out my best way to cut this copper sheet into strips. I'll need to use three strips soldered up in series for each half of the secondary winding. One thing about using flat strip- it's super easy to add breakout taps to the winding- just lay another piece of strip across and solder. Makes a very low profile junction. At the beginning and ends, I just make a fold at 45 degrees and there's a connection tab.