Winding Switchmode Transformers

[Chryseus]

Hi
I've decided it's time that I learned how to wind and design my own switching power supplies, I've already designed and built various non-isolated converters such as buck & boost but have never really been satisfied with their performance at higher power levels, the information I've found so far on winding your own transformer seems to vary greatly from source to source so I'm not even sure what information is correct.

Specifically I am trying to calculate the minimum number of primary turns required, this is the equation I got from this Infineon design note.

This all makes perfect sense except for two points.
Should Bsat be in tesla or gauss ?
The example value they give for Ae is for the ETD34 core, they show this value as 97.1e-6 however when I looked at the datasheet for the ETD34 the sheet says Ae = 97.1mm squared, correct me if I'm wrong but isn't that 97.1e-3 ?

Also does anyone know a good comparison of isolated driver topologies ?
Cheers.

[Macbeth]

I have no idea how to answer your specific question, but I have been learning how to machine sew recently and today instantly hit on the idea of using the spool winding mechanism to make inductors and transformers really quickly and neatly, using enamelled copper wire instead of thread! Anyone tried this?

[sigxcpu]

97.1 mm => 97.1e-3 m
97.1 mm^2 => 91.1e-6 m^2

[Chryseus]

97.1 mm => 97.1e-3 m
97.1 mm^2 => 91.1e-6 m^2

I'm not exactly sure how you came to make the answer 3 orders of magnitude lower unless there is something important I'm missing 

[ElektroQuark]

You need to use m not mm.

[tonyarkles]

1mm = 1e-3 m
1mm^2 = (1mm x 1mm) = (1e-3 m x 1e-3 m) = 1e-6 m^2

[Chryseus]

I still think I'm missing something.
0.0971 * 0.0971 = 9.42e-3
The answer should be 97.1e-6

If I put in google 97.1mm squared it gives the correct answer, I'm not exactly sure what it is doing different to me.
Putting any variation of 0.0971^2 in a calculator gives the wrong answer.

[AndreasF]

97.1 mm => 97.1e-3 m
97.1 mm^2 => 91.1e-6 m^2

Nope:

(97.1mm)^2 = (97.1*10e-3m)*(97.1*10e-3) = (97.1*97.1)*(10e-6) = 9428.41*10e-6 = 9.42841*10e-3

[Neverther]

The unit given is alredy in squaremillimeters, you need to note this fact.
The unit used in calulations is squaremeter, without prefixes as with all SI units (well almost all).

For example
1mm side rectangle = 0.001m rectangle
The areas are
(1mm)^2 = 1 mm^2
(0.001m)^2 = 1*10^-6 m^2

I think thats simple enough....

[Chryseus]

The unit given is alredy in squaremillimeters, you need to note this fact.
The unit used in calulations is squaremeter, without prefixes as with all SI units (well almost all).

For example
1mm side rectangle = 0.001m rectangle
The areas are
(1mm)^2 = 1 mm^2
(0.001m)^2 = 1*10^-6 m^2

I think thats simple enough....

I think I understand it now.
Thanks 

[Harvs]

When I first started doing this a few years ago it all seemed very difficult as well.  The formula's seemed to change from one book to the next, certain cores were very difficult to get info on, then there was the differences between core materials, gapping etc.

However, all I can recommend is get some cores of different types, grabs some enameled wired and start winding and testing some.  You'll probably kill the odd circuit in the process, but it will demystify the whole thing. I can now look at an ETDxx and have a rough idea how many turns with what gap is going to give me what inductance and saturations current.  When you know that sort of thing, the formulas make a lot more sense.

[T3sl4co1l]

http://webpages.charter.net/dawill/tmoranwms/Elec_Magnetics.html

I should probably rewrite this some time, clean it up, and finish the inductor section.  What's there should still be a little useful I think.

Tim

[diyaudio]

Hi
I've decided it's time that I learned how to wind and design my own switching power supplies, I've already designed and built various non-isolated converters such as buck & boost but have never really been satisfied with their performance at higher power levels, the information I've found so far on winding your own transformer seems to vary greatly from source to source so I'm not even sure what information is correct.

Specifically I am trying to calculate the minimum number of primary turns required, this is the equation I got from this Infineon design note.

This all makes perfect sense except for two points.
Should Bsat be in tesla or gauss ?
The example value they give for Ae is for the ETD34 core, they show this value as 97.1e-6 however when I looked at the datasheet for the ETD34 the sheet says Ae = 97.1mm squared, correct me if I'm wrong but isn't that 97.1e-3 ?

Also does anyone know a good comparison of isolated driver topologies ?
Cheers.

This is really complicated topic. You will never find a standard equation, the results varies with geometry, magnetic material and Fsw ( determined by the material )

Join up here www.diysmps.com and see what the guys have done
 

[Chryseus]

Thanks for the information guys.
I'm going to order up a bunch of cores and do some experimentation. 

[mos6502]

I can highly recommend this site:

http://schmidt-walter.eit.h-da.de/smpshome/

You can enter your parameters and it will calculate winding data for all kinds of topologies. It also gives you a list of suitable cores.

[Richard Head]

I learned about switchmode power supplies and magnetics design from the excellent Philips application notes that they published about 25 years ago.
There is also a very good book by Snelling that is really the magnetics bible. I think it's out of print now though. Also "Applied magnetics" by JK Watson is recommended for more advanced study, but a bit theoretical.
Also, get a ferrites databook if possible. I use the Siemens (Epcos) and Philips ones. The books are easier to use than messing around with pdf's on a PC.
Only once you make a power supply and measure the effects of leakage inductance/inter-winding capacitance, high flux density etc will you begin to appreciate the design trade-offs required in magnetics design. The heart of a SMPS is the transformer. A lot of the circuitry in a SMPS is based around the transformer characteristics.

Dick

[IanB]

This all makes perfect sense except for two points.
Should Bsat be in tesla or gauss ?
The example value they give for Ae is for the ETD34 core, they show this value as 97.1e-6 however when I looked at the datasheet for the ETD34 the sheet says Ae = 97.1mm squared, correct me if I'm wrong but isn't that 97.1e-3 ?

In the design note they say that B_sat is in Tesla. However, even if the information is not available you can work such things out by dimensional analysis.



The whole expression gives a number of turns, which is dimensionless. Therefore the expression:

Vi_MAX · D_MAX · (1 / fs)

must have the same units of measure as the expression:

B_sat · A_e

On the top we have units of:

V · s

On the bottom we have units of:

T · m² = V s m^-2 · m² = V · s

Since we have V · s on the top and bottom of the formula, we know the unit of B_max should be Tesla.

[Conrad Hoffman]

There was a Ferroxcube binder for transformer design several decades ago, with nomographs to figure everything out. I started with that and it made things at least somewhat clear. After that was an early book by Billings (I think) that was sort of hit or miss in its chapter organization, but had a huge amount of practical info. Finally I got the Transformer and Inductor Design Handbook by McLyman. One reference just didn't do it for me, but by comparing explanations and formulas from all three, plus the 'net, I was able to design pretty good supplies. A caution- my interest was high voltage supplies up to about 1500 VDC, and a lot of the common design recommendations that apply to low voltage supplies aren't valid for high voltage supplies!

[Fank1]

Try this site:
It will make it a LOT easier.
http://schmidt-walter.eit.h-da.de/smps_e/smps_e.html#smps

[PChi]

97.1 mm squared = 97.1 E-6 m squared.
Also I would oprefer to use the switch on time (Ton) instead of Duty cycle (Dmax) / Switching frequency (fs). The formula is assuming that the flux is zero at the beginning of each cycle.

[Jay_Diddy_B]

Hi,

The formula that was given by the original poster has its roots in Faraday's Law of Induction:

V=N dPhi/dt

where:

V= voltage
N = Number of turns
Phi (should be Greek 'Phi') = flux density
t=time

The total flux Phi = flux density x Area

Flux density is B

Since the area of the core doesn't change with time.

We get V=NA dB/dt

The Si units are:

Volts
Turns (no units)
A in square metres.
B flux density in Tesla
t in seconds.

Ae is different than Amin

Ae is the effective Area of the core
Amin in the minimum area of area of the core.

But, in most core geometry Ae is approximately equal to Amin.

The formula that was given by the original poster gives the minimum number of turns to for a given peak flux.

How do you choose B?

Every time you magnetize and de-magnetize the core you travel around a B-H loop. The core losses are proportional to the area enclosed within the B-H multiplied by the frequency that you go around the loop.

The core loss is given in the ferrite material datasheet. The units may be kW/m² for given peak-peak flux change.

Since the area inside the B-H loop changes approximately with B² then the core loss is approximately proportional to B².

So we have the core loss

Core loss = k x core volume x Fsw x B²

where k is a constant based on the material properties and Fsw = switching frequency.

Consider the power supply topology

Some power supplies, like discontinuous mode flytback converters, the flux density goes from 0 to Bpeak each cycle. In other topologies the flux may go from -Bpeak to +Bpeak (Full-bridge). So you have to understand the flux waveform when considering what value of Bpeak to use in the calculation.

Once you have determined Bpeak you can then calculate the number of turns.

Once the turns have been calculated, the core chosen, the construction of the transformer is another topic in its own right.


If you post you circuit and your requirements, I might be able to help with a transformer design.

Regards,

Jay_Diddy_B