My TSC International Ferrite Shipment. (1kVA Txfmr) / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

My TSC International Ferrite Shipment. (1kVA Txfmr)

1 2 next

Move Thread

LAN_403

Patrick

Sat Apr 23 2011, 01:24AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

These were ordered as part of a project I plan to post here on the forum eventually.

I want to see if its possible to make a 10 Megawatt Ferrite transformer which in volume is about or less than 1 meter cubed!
(I am not sure at all that this is even possible) First using math, then second, building a scaled down example as a technology demonstrator/proof of concept device.

The intent for this thread is to identify some things regarding TSC international Ferrite, like these shipping details that were unexpected. Also technical matters related to ferrite power and heating.

Is there a corralation between the 16 items on paper and there placement of the 32 pcs on the shrink wrapped board?

EDIT: Title changed.

Sulaiman

Sat Apr 23 2011, 06:35PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

I guess 32 pieces = 16 pairs/sets.
Each set/pair has had it's Al measured for you. 30uH per (turn.squared)

Core geometries are carefully chosen for various applications
(these core sets look optimised for using pcb tracks as windings
not a particularly good power/volume ratio)
For high power you may be better off using multiple smaller transformers rather than one large one due to cooling requirements.
Is that why you went for 16 sets?

Based on smps that I have serviced;

For high power density you need high frequency,
25 kHz was common, then c50 kHz, 100kHz, 200kHz, 500kHz and even >1MHz.
I have read manufacturers data for 4MHz and up!
Rectifier choice seems important, MLCC essential, good quality electrolytics (if used)
switching ic manufacturers often give design tools that include all parameters and give total footprint area etc . A good one is

Ash Small

Sat Apr 23 2011, 06:48PM

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

Maybe you could measure Al of the 6th pair and 7th pair, using a capacitor , sig. gen. and 'scope, and see if you can detect a difference?

(if I remember correctly, the Al is with one turn, if you can't detect a difference, try ten turns, or whatever, then work out Al from there)

This should at least give you an indication of whether they are layed out on the board in the same order as in the list.

EDIT: and a resistor.

jpsmith123

Sat Apr 23 2011, 07:10PM

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

What exactly are you going to do with all those cores?

Steve Conner

Sun Apr 24 2011, 08:14AM

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

If he's anything like me, put them on a shelf for the next 10 years.

If you want to design a 10 megawatt transformer, the two things that will bite you in the ass are:

Surface area to volume ratio of the core. (heat generation is proportional to volume, heat dissipation to surface area.)

Skin effect in the conductors. (Doubling the cross-sectional area doesn't halve the AC resistance.)

These two effects mean that you can't just scale up a smaller design, you will have to get innovative.

Oh, and Al isn't measured with a one-turn winding, because the inductance is so small that stray inductance of the wiring gives an inaccurate result. Use a 10 turn winding and divide the answer by 100.

teravolt

Sun Apr 24 2011, 02:27PM

Registered Member #195 Joined: Fri Feb 17 2006, 08:27PM
Location: Berkeley, ca.
Posts: 1111

Are you shure you have not done to much partying up there in Chico 10Mw is a lot of ferites. Maby you ment 10Kw

Patrick

Sun Apr 24 2011, 06:14PM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

I should say that these modules will be immersed in oil, activley pumped, with an external radiator.
Also the h-Bridge, regulation, feedback, CCM IC, series resistor, and transformers will all be integratated into a single PC board, in oil. Also, the physical/mechanics and stucture of all this worries me more than the thermal matters. If all this cant be made to work, then id rather fail now... early, then fail after years of work and money. This is a part of a larger projet, I scheduled about 2 years to develop this idea (Unless I see that it cant be done sooner).

Sulaiman wrote ...

I guess 32 pieces = 16 pairs/sets.
Each set/pair has had it's Al measured for you. 30uH per (turn.squared)

Yes, I realise this. I was hoping it averaged out to 12 uH per turn, since there test coil was 2 turns.
0.0301mH avg / 2 = 150.5uH, sqrt(150.50) = 12.2uH per turn or did I do the math wrong?

Sulaiman wrote ...

Core geometries are carefully chosen for various applications
(these core sets look optimised for using pcb tracks as windings
not a particularly good power/volume ratio)

Yes they are carefully chosen. These are optimized for the planar circuit pattern coils, but they will have normal wire primaries. No matter how I did the math, the planars always came out on top, Power, temp, High freq, few turns, High V per T. Power per volume came out favorably too.

Sulaiman wrote ...

For high power you may be better off using multiple smaller transformers rather than one large one due to cooling requirements.
Is that why you went for 16 sets?

Yes, I plan to use modules, two transformers in series per module (CT like an NST) for V, then many modules in parallel power/current, this also improves reliability. 100 modules per 1Mw.

Sulaiman wrote ...

Based on smps that I have serviced;

For high power density you need high frequency,
25 kHz was common, then c50 kHz, 100kHz, 200kHz, 500kHz and even >1MHz.
I have read manufacturers data for 4MHz and up!

I am thinking initially 50-70kHz, later generations maybe higher in freq. I seem to see a corralation in the math, with increasing Freq I see more power out of the same volume of ferrite. Unless I am in error? Higher freq than 200kHz, in large supplies seem difficult, ive only seen small power supplies use 400kHz and only a single small Tesla Coil use 4MHz. 400kHz to 4Mhz seems to me to be off limits for large SMPS's, but maybe some one will prove me wrong.

Sulaiman wrote ...

Rectifier choice seems important, MLCC essential, good quality electrolytics (if used)
switching ic manufacturers often give design tools that include all parameters and give total footprint area etc . A good one is

Yes on rectifiers, caps and resistive feedback. I will be using current mode regulation, each pulse will be current measured and limited. also current mode control is easier to parallel anyway, for safety as in one module failure not becoming contagious, and that module should have the ability to be isolated. Also from a practical view voltage mode can become problematic, with load shedding form some PS's to others, again causing contagious failure. Human safety and maintainability is an issue as well.

teravolt wrote ...

Are you shure you have not done to much partying up there in Chico 10Mw is a lot of ferites. Maby you ment 10Kw

Perfectly sober, of clear mind. Now do you see why I need the HV oscilloscope probes? Perhaps I Should have said 1kW transformer, since I will be using modules to make up the 100kW, 1MW, 10MW types in that order. First using math, then the scaled down devices, finally the all up full versions.

If EE PH.d. 's were so easy to get, I guess everyone would have one.

EDIT: Yep title changed. Less hype more fact is better anyway.

Patrick

Sun Apr 24 2011, 09:14PM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

Dam it, i double posted, sorry my bad.

Ignore the 100kV, 3kVA label. They were just for show, these containers worked well for one experiment so Ill repurpose them for this one. I CNC'ed them myself, there PE, PVC and aluminum.

Sulaiman

Mon Apr 25 2011, 10:48AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

Ah! didn't see it was for 2 turns,
Al = 30 x (1/2)^2 = 30 x 1/4 = 7.5 uh/turn2

I think that the space-saving exercise should concentrate on
switching transistor and rectifier diode dissipation
I have not come across a burned-out smps transformer yet, but many switches and rectifiers.
Seems manufacturers specs are conservative for magnetics and marginal for semiconductors.

My favourite equation; N.I = B.(Ae/Al) to check amp.turns vs flux density(T) given Ae and Al
e.g. Your core Ae = 0.25" x 0.799" = 0.2 sq. in. = 129 x 10E-6 sq.m
at 100 mT (1000 gauss, quite high) flux density,
N.I = 0.1 x (129/7.5) = 1.781 Amp.Turns .... OOPS!
maybe not the best cores for a flyback smps?
You will need to add airgaps, I suspect.
P.S. the equation is from, L=N.phi/I, L=Al.Nsquared, phi=B.Ae
P.P.S. since you will have to achieve high efficiency conversion to get high power density
you could consider progressive 'switching-in' of modules so that all but one operate at peak efficiency.
You will still have to design for worst case but operational efficiency (cost) would be optimal.
A daisy-chain hardware solution could be as simple as
I/P = Enable = Previous output (first module = permanent enable)
O/P = At full efficiency / limit / help!
One led per module would make a great peak-power meter
Good luck with stability calculations!

also, I only try to repair, I'm not a designer, so treat info. with suspicion!

Patrick

Sun May 15 2011, 06:27AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

I will be using this IC, Pulse by pulse current limit, parallelability, and load-shed resistence.
]uc3846.pdf[/file]

Or this one its cheaper, though i need 2 ouputs, i guess for this ill have to have external loigc divide be two? (for an H-bridge)
]fan7554.pdf[/file] (cant be paralled either)

Unless someone else can suggest better/other current mode controllers, that cost less then $7 for each IC.

1 2 next

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