Volts per turn question. / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Volts per turn question.

previous 1 2 3 4 next

Move Thread

LAN_403

Ash Small

Sun Dec 04 2011, 08:34PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Do you have any figures for Bmax for any cores with similar characteristics to N27 cores, Steve?

EDIT: The datasheet gives a figure of 500 mT.

If I put this into the above equation, I get Npri = 240/500 = 0.48, which makes no sense at all.

What have I done wrong? Do I need to multiply by 1000, maybe? or is it something else?

Steve Conner

Sun Dec 04 2011, 09:40PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Well, the core datasheet that you posted earlier gives an example operating point. Something like 200mT, 16kHz, at which point the core loss is 22W for the UI set and 30W for the UU set. The core loss depends on temperature which is why they specify 100 degrees, but the implication is that that is a reasonable operating temperature too.

200mT is 2000 gauss, and that is kind of in the ballpark. I've seen 300mT quoted elsewhere as a typical saturation flux density for power ferrite.

You can only run ferrite near saturation at low frequency. You need to decrease the Bmax as you go to higher frequencies, because the core loss increases with frequency. If you went to 160kHz,
maybe the core could only stand 200 gauss instead of 2000. So the volts per turn figure you calculate is about right for any frequency above 16kHz.

Edit: 1mT is 10 gauss. Also what units does the formula expect cross-sectional area in? Could be square centimetres, inches, metres, whatever. I'd expect about 10V per turn from this core, but this is a real hand waving guess!

Ash Small

Sun Dec 04 2011, 09:51PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

I'm still none the wiser. Which figure is about right for volts per turn?

Steve Conner

Sun Dec 04 2011, 10:10PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Like I said, about 10 volts per turn, maybe a little more, for a core this size in continuous operation. I've seen people claim 25 or 50, but that was with the biggest cores they could get anywhere, and probably only for a minute or two.

Marko

Sun Dec 04 2011, 10:29PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Hm, looks like one pretty damn big core to me, with nearly 9 square centimeters surface area... for comparison, I've ran a 4sq cm ferrite toroid at about 15V/turn at 100kHz with little complaint. So I think it's fairly safe for this core to go up to 25 (of course, if frequenccan be high enough toa void saturation).

The most cruel experiments on V/turn limits were those I've done on PC PSU transformers. They seem to have cross-section area of about 1cm^2 or less.
I've driven one of them in reverse far over it's ratings - 600V sinewave peak out instead of usual 150 square it sees on the primary in normal operation. It could drive bunches of fluorescent tubes and like... though the core would get untouchably hot very fast.

I've also noticed that one outer leg of the transformer core heats up significantly more than the other (and windings were inclined a bit towards one of legs; not sure which).

If I recall right those transformers tended to have 50..60 primary turns. If it was 50 then this would put the outer limit to V/turn per cm^2 to about 8 which is pretty insane.

If you're really keen on getting 50kV AC from a transformer, I think it'd actually be recommended to push the V/turn on your transformer as much as you can, because less secondary windings will allow for less parasitic C and less trouble. You can let the core run hotter if your whole system is under oil (which it really should be).

You can do some preliminary tests by energizing the core by an inverter at your target frequency and V/turn with only primary coils on it, and see how it behaves.

Marko

jpsmith123

Sun Dec 04 2011, 10:46PM

Registered Member #1321 Joined: Sat Feb 16 2008, 03:22AM
Location:
Posts: 843

You have to start by choosing a peak flux density (in the core) and a frequency that you want to run at. In most cases, core loss will be the main concern, but in the case of a high voltage transformer, high frequency effects in the secondary may limit your options. (I was just reading about a HFHV transformer that failed apparently due to I^2*R heating in the secondary, due to a high circulating (reactive) current.

In my experience it's sort of an iterative process, starting from a semi-arbitrary point and looking to see it you have enough room for the secondary, for example then maybe adjusting something, etc.

Being that most modern ferrites (intended for power conversion) have similar characteristics, you can almost say without even looking at the data sheet (for the particular material in question) 25 khz @ Bpeak = 0.2T would be a reasonable starting point.

BTW I think I remember posting a loss curve for that N27 material (and also ferroxcube 3C90) a while back. IIRC your material is somewhat more lossy than 3C90.

You would use the formula: N = V/[4*f*A*B]; where N is the number of primary turns; V is the DC bus voltage in volts that your bridge would apply to the primary; f is drive frequenct in Hz; A is the core cross sectional area in square meters; and B is the peak flux density in Tesla. The numerical constant ("4" in this case for a 50% duty cycle square wave) is waveform dependent. (For a sine wave it would be [2*pi]/[2^1/2] = 4.443).

Anyway for the case of 25khz and 0.2T, assuming a core cross sectional area of 6.45exp-4 (1 square inch) using the above equation setting N=1 and solving for V, you get about 13 volts per turn for example.

Ash Small

Sun Dec 04 2011, 10:48PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Thanks for the pointers, Steve.

So I may as well just use rectified 240 V mains, with ~30 turn primary and ~5000 turn secondary.

(At least it's a starting point.)

EDIT: OK, Dr. Pork says I can get away with a 2000 turn secondary. I'll take these as the upper and lower limits, unless I recieve any more input on the subject. JPSmith gives a figure in between.....maybe 3,500 would be a good starting place?

jpsmith123

Sun Dec 04 2011, 10:57PM

Registered Member #1321 Joined: Sat Feb 16 2008, 03:22AM
Location:
Posts: 843

What about if you took your transformer AC output and ran it into an L network (a series resonant circuit) to boost the voltage that way?

Marko

Mon Dec 05 2011, 03:19AM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Ash Small wrote ...

Thanks for the pointers, Steve.

So I may as well just use rectified 240 V mains, with ~30 turn primary and ~5000 turn secondary.

(At least it's a starting point.)

EDIT: OK, Dr. Pork says I can get away with a 2000 turn secondary. I'll take these as the upper and lower limits, unless I recieve any more input on the subject. JPSmith gives a figure in between.....maybe 3,500 would be a good starting place?

I wouldn't mess with secondaries until I was sure how many turns I exactly need (or you could, by making a few of those pancake windings and then adding more later if necessary).

Drive the core alone at first, and progressively increase V/turn over time and monitor how hot it gets over longer periods.

Note that all I've taken into account are eddy current losses (which are the main killer here); you need to make sure your core won't saturate at chosen V/turn by choosing a right frequency.

Marko

IamSmooth

Mon Dec 05 2011, 03:55PM

Registered Member #190 Joined: Fri Feb 17 2006, 12:00AM
Location:
Posts: 1567

I know I'm a little late on this thread, but if you look here I had derived the formula for transformers and added it to the wiki site under equations. Maybe it will be helpful for some.

previous 1 2 3 4 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob