Switched Mode Power Supply

This forum is for specialized infomation important to the construction and safe operation of the high voltage electrical supplies and related circuitry needed for fusor operation.
User avatar
Rich Feldman
Posts: 968
Joined: Mon Dec 21, 2009 11:59 pm
Real name: Rich Feldman
Location: Santa Clara County, CA, USA

Re: Switched Mode Power Supply

Post by Rich Feldman » Tue Jul 11, 2017 5:18 pm

What happens if you reconfigure the voltage multipler
so there's a path for DC current in the secondary winding?
For example, from http://www.oldtellys.co.uk/otltbeht.html
ltbeht8.jpg
ltbeht8.jpg (18.74 KiB) Viewed 757 times
Don't be led astray by the many DIY circuits that use "flybacks" as regular transformers, with drivers that involve more than one switch.

Your HV measurement methods sound fine, if you respect the voltage rating of the high-ohm ( 2 MΩ ) resistor.
If you are using a computer keyboard with a calculator-style number key array, what happens if you type 234 (over there) while the Alt key is depressed? Ω
Richard Feldman

Niels Geerits
Posts: 56
Joined: Tue Jun 13, 2017 10:25 pm
Real name: Niels Geerits

Re: Switched Mode Power Supply

Post by Niels Geerits » Wed Jul 12, 2017 8:52 pm

Took apart my transformer and rewound the primary and secondary. I now have 10 primary turns and 37 secondary turns. I checked for the DC current component you mentioned Rich. I did this by putting a large value resistor across the secondary and measuring the DC current through it. I was unable to measure anything. I think there may be some confusion. I am using a driver that one would use with a flyback transformer, however my transformer is not a flyback. It is a toroidal ferrite transformer. Maybe "forward converter" is a more adequate name (though to be clear I do not have an extra inductor. The circuit I posted here is the circuit I am using).

I am getting an issue that my output voltage suddenly drops once I get above a certain input voltage. I believe this is due to breakdown occuring somewhere which causes a short circuit causing the voltage to drop (this happens somewhere in the CW Multiplier, probably soldered it too compact on the perfboard).

Also for 234 on the numpad I get Û

User avatar
Rich Feldman
Posts: 968
Joined: Mon Dec 21, 2009 11:59 pm
Real name: Rich Feldman
Location: Santa Clara County, CA, USA

Re: Switched Mode Power Supply

Post by Rich Feldman » Thu Jul 13, 2017 1:35 am

Well Niels, with the "flyback" drive circuit, you know there's DC current in the primary winding. How many ampere-turns can your toroid take without saturating? We can help you figure that out, if you post the core's dimensions and material properties or inductance index (usually given as some inductance value for some number of turns). Or we can review the calculations you made before you chose a toroid and bothered to wind wire on it.

In a regular transformer, the ideal winding inductance is infinite, and the ideal energy storage is zero. Instant by instant, the power going in at the primary terminals matches the power going out at the secondary terminals. Conversely, finite inductance and energy storage are essential in the magnetic component of a flyback converter. When the primary switch is "on", power goes in and accumulates as stored energy ( L * i^2 / 2 ). When the primary switch turns off, a secondary winding diode turns on. The stored "magnetic" energy goes out as power (volts * amps) through the secondary terminals.

CRT "flybacks" (for horizontal deflection & anode power) generally have an air gap in the ferrite core, at the joint hidden by the winding subassembly. Air gap greatly reduces the winding inductances and greatly increases the ampere-turns limit. Net result is a useful energy storage capacity (with most of the magnetic energy concentrated within the air gap). Became pretty universal before 1960.

If you want to use flyback converter topology with a ferrite toroid core, I think you should learn how to understand and calculate details that you've avoided so far. That knowledge will serve you better than having an oscilloscope. The real world is analog! :-)
Richard Feldman

Niels Geerits
Posts: 56
Joined: Tue Jun 13, 2017 10:25 pm
Real name: Niels Geerits

Re: Switched Mode Power Supply

Post by Niels Geerits » Thu Jul 13, 2017 7:35 pm

Darn it. Happened again got logged out when I had typed my response.

So here are my datasheets: http://www.farnell.com/datasheets/15958 ... 1499594956 and http://www.ferroxcube.com/FerroxcubeCor ... t/3c90.pdf (can somebody tell me what Sigma I/A is?)

Yesterday I actually measured the DC current in the Primary to calculate the inductance. At 50kHz and 35V I measured an average current of 66.7mA. From this I concluded that the primary inductance equals 1.3mH. From this I calculated that the DC current at 350V and 50kHz would be something like 337mA. At 10 Primary turns I have 3.37 Ampere Turns. Using the datasheet we find the toroid core is 0.26m Long. Therefore the field equals about 13 A/m, this is far from saturation if I read the datasheet correctly. Also some of this may not be optimal because I did a poor job on my primary windings (some air gaps) because the wire is not ideal for winding (secondary looks a lot better)

As a side Project today I made a sound card silly scope which despite its sampling frequency of 96kHz can only measure signals up to 20 kHz so I think it samples at around 40 kHz. So I lowered the frequency of my driver and hooked up the scope to my seconday windings and played around with the Input voltage a bit. I did not get any saturation this time. I don't recommend anyone try this unless they want a headache from all the high frequency ringing.

I don't think my setup could ever deliver the power needed for fusion in flyback mode. This is why I eventually want to look into push pull and or the half bridge topology. The problem with push pull is the high voltage stress on the switches.

User avatar
Rich Feldman
Posts: 968
Joined: Mon Dec 21, 2009 11:59 pm
Real name: Rich Feldman
Location: Santa Clara County, CA, USA

Re: Switched Mode Power Supply

Post by Rich Feldman » Fri Jul 14, 2017 12:20 am

Love it, Niels. I'm pretty addicted to the stuff, and learned some things from working on answers for you.

On the toroid datasheet, that "Sigma I/A" value is in the magnetical dimensions section. The I is actually a l (lowercase letter L), like the one in the symbol for effective length (l sub e: 255 mm). Unless the Polish word for length begins with the letter between h and j. We see that those sigma term values, and their unit dimensions, match the quotients l/A and l/A^2 of the magnetic length and area parameters. Perhaps they are convenient criteria for some core selection process.

I did an exercise to get the toroid's inductance index, Al, from that Sigma l/A and the initial permeability. In your 3C90 material datasheet, the typical ui value is given as 2300. In absolute terms, that's 0.0029 henries per meter. Dividing by your toroid's le, and multiplying by its Ae, we get 5041 nanohenries [per turn squared], just like the toroid datasheet says!

The primary winding inductance you got, 1.3 mH, sounds perfectly reasonable. That would make the Al value 13000 nH, and relative permeability u = 5930, at the operating point of your measurement. 3C90 datasheet gives u = 5500 at Bmax = 200 mT and t = 100 °C.

I have issues with your extrapolation from 35 to 350 volts, and with your saturation margin numbers. No time to write it now.
Richard Feldman

User avatar
Rich Feldman
Posts: 968
Joined: Mon Dec 21, 2009 11:59 pm
Real name: Rich Feldman
Location: Santa Clara County, CA, USA

Re: Switched Mode Power Supply

Post by Rich Feldman » Fri Jul 14, 2017 1:40 am

OK, let's make this fast.
The details you gave about the 1.3-mH measurement, operating at 35 V, suggest a primary current waveform that looks like:
niels_flyback.PNG
niels_flyback.PNG (5.57 KiB) Viewed 675 times
To get DC current of 66.7 mA, the duty cycle needs to be about 50%, with peak current about 4 times greater than average. Have I got that straight? If so, then your Bmax is already up to around 80 mT.

What do you think the secondary current waveform looks like?

How come your expected current at 350 V isn't ten times greater than that at 35 V?

I like your "silly scope" side project.
Suppose you can build a fast-acting voltage sample-and-hold circuit.
Drive it with your Arduino at 49 kHz while the circuit under test is running at 50 kHz.
The output waveform will look like the input, slowed down by a factor of 50. At 1 kHz, your sound card ought to be able to do a pretty high fidelity waveform capture.

Or drive this fast-acting voltage sample-and-hold with your Arduino at 50 kHz, and a programmable phase with respect to your SMPS under test. Like using a stroboscope to "freeze" the view of a rotating part. Take readings manually with a voltmeter, at whatever phases you choose. Or use the Arduino to read an analog to digital converter, which can be as slow as you want.
Richard Feldman

Niels Geerits
Posts: 56
Joined: Tue Jun 13, 2017 10:25 pm
Real name: Niels Geerits

Re: Switched Mode Power Supply

Post by Niels Geerits » Fri Jul 14, 2017 7:27 pm

Glad you like it too Rich, you have been very helpful. Thanks for clarifying my question about the datasheet and the example calculation!

Yes that is the waveform I was using in my calculations (no reverse current). You are correct the peak current should be 4 times higher than the average current. Using the 0.0029 H/m you found that means the B_max=66.7mA*4*10Turns*0.0029H/m=150mT! (this explains why saturation is happening so soon). This could be reduced by increasing the number or primary turns, because more primary turns increases the inductance (goes with turns squared) which decreases the current, the amp turns goes linearly with the number of turns, so B_max will drop linearly with the number of turns.

I am a bit confused though: imagine that we sent an AC into the transformer (+35V to -35V square wave). If we look at the current like we just did we fill find that B_max=80mT, however if we use the universal EMF equation for transformers B_max=V_rms/(4.44*f*a*N)=35/(4.44*5*10^4*4.45*10^-4*10)=35mT. And if we drive the transformer with a square wave of 35V to 0V we get a lower RMS voltage which would give a lower B_max.

As for the secondary waveform: in the first phase as the magnetic field builds up the (shorted) secondary should counteract the magnetic field. So first we will see a current in the opposite direction: I_s=-I_p/T, where T is the turns ratio. Then in the second Phase the magnetic field collapses so the current in the secondary reverses its direction instantly. We will see I_s=I_max-k*t, where k is V/L like in the first Phase.

Also you are right the expected current at 350V is 10 times higher than at 35V. I divided by two somewhere in my calculation, my mistake.

I though of a similar way to bypas the frequency limit. I can use a pair of MOSFETS the arduino and the silly scope to do this. I just use the probe I have been using but add some MOSFETS to it which are turned on and off with the arduino at say 1 kHz with a low duty cycle and then using the arduino slowly shift the turn on time. This only works for periodic signals, but that isn't a problem since my Signal is periodic. I will add a drawing in paint to illustrate what I am doing. The red lines are samples which go to the sound Card and can be sampled at 48 kHz.

Image

Also today I took my rectifier from the bread board and soldered it making my Setup a bit cleaner. In addition I added an amp meter in the primary circuit, when I turned it on my IRFP360 MOSFET broke. I replaced it with a IRFP250N, which also broke. I checked the circuit for any shorts, checked wether or not the arduino was creating the correct Signal etc. couldn't find anything. So I put in a new IRFP250N with an extra 100 ohm 50 W resistor on the drain. This time everything worked. However once I removed the resistor the MOSFET broke again. Which is strange because my IRFP360 was working all week without a resistor in place (just the coil). I am also using the 15nF cap to filter the HV feedback. Do you know what is going on here?

I should add I use two power supplies: one is a variac connected to the rectifier and my large 3mF cap, which powers the circuit. The other is a 12V 1A supply to power the switcher and the arduino. I make sure all ground are connected to a common terminal (the ground terminal on my cap) so that nothing floats away. So Arduino, 12V PS and large value cap are all connected.

User avatar
Rich Feldman
Posts: 968
Joined: Mon Dec 21, 2009 11:59 pm
Real name: Rich Feldman
Location: Santa Clara County, CA, USA

Re: Switched Mode Power Supply

Post by Rich Feldman » Mon Jul 17, 2017 10:19 pm

I'm giving this a rest for a week. In the meantime, you could find out whether that formula with the 4.44 term is for sine waves (like I warned you about) or square waves. The square and sine coefficients differ by some factor like 2/pi; flyback formula has yet another value in that place. It should be clear if you work it from the laws named for Ampere and Faraday.

In fact for volts-per-turn at some frequency and Bmax, Faraday's law is sufficient. The core needs enough flux swing to match the voltage. The current to get it there (a.k.a. magnetizing current or no-load primary current) is ordinarily very small compared to the operating current, and isn't critical for design choices. Except in flybacks.

For both regular transformers and flybacks, increasing the turns count has the bad effect of increasing the winding resistance and "copper loss" heating. Compensated by using thicker wire, if there's room. If the hole in the core isn't mostly filled with primary and secondary windings, you are wasting core material, and leaving some power efficiency on the table.
Richard Feldman

User avatar
Richard Hull
Site Admin
Posts: 10842
Joined: Fri Jun 15, 2001 1:44 pm
Real name: Richard Hull

Re: Switched Mode Power Supply

Post by Richard Hull » Tue Jul 18, 2017 12:25 am

Power efficiency?

I am a heathen. The very least part of a power supply design I would consider here is power efficiency. The rule is git 'er done!

Why not worry about power efficiency, you might ask....

1. Do you get a significant reward for power efficiency?
2. Are you building this for yourself or for your employer who "wants it all"... small, lightweight, efficient and pleasing to the eye?
3. For some reason is your power company unwilling to supply you with any amount of power for a fixed amount of money?

point number one....

The average working fusion at 2 million fusions per second can easily demand 40kv at 12ma. 480 watts
I pay about 12.4 cents per KWH here and this figure includes all taxes, fuel price adjustment additions, transmission costs, etc.

So....

a 100% efficient supply would cost me about 6 cents per hour to run my fusor at full bore, max power.
a 40% efficient supply would leap the per hour run rate up to a wallet busting 15 cents! Ouch! That stings
A rotten 25% efficient supply would break the bank at almost 25 cents per hour! Oh mercy!

Based on about 15 hours operation at full power, per year of fusor IV, my yearly fusor power bill with my 50% efficient supply is under $2.00
I don't know about you, but I pay almost double my yearly fusor power bill buying a whopper with cheese, and this does not include a drink or fries to go with it.
I have the fusor in my hosehold budget as the wife gets mad spending two bucks a year on fusor operation. She' pushing me to make a more efficient supply to save money.

Point number two

No one is paying you the big bucks to make your supply efficient, light small or nice looking. In addition efficiency is normally a real issue for portable or battery powered gear where every watt is dear and costly.

It seems stupid to try and get 90% efficiency in a home built supply to run a fusion device that is 0.00000001% efficient at doing fusion. The fusor, in all sincerity, is a great space heater when operated in a chilly winter lab. It heats the lab not by fusion energy but by wasted electricity from the wall outlet.

Point number three

Power is cheap when wall power is used at the level of energy and net run power that the average fusioneer will require. The power companies across the land will back you play for pocket change.

I would be far more sympathetic with a 100 fold increase in fusor efficiency to a 0.000001% efficient fusion device. Of course, with a D-D fusor, at amateur level voltages, this goal seems impossible. Thus far, no one has even advanced the fusor to a ten fold increase over the big, "normalized", goal of 2 million fusions per second.


Richard Hull
Progress may have been a good thing once, but it just went on too long. - Yogi Berra
Fusion is the energy of the future....and it always will be
Retired now...Doing only what I want and not what I should...every day is a saturday.

Niels Geerits
Posts: 56
Joined: Tue Jun 13, 2017 10:25 pm
Real name: Niels Geerits

Re: Switched Mode Power Supply

Post by Niels Geerits » Tue Jul 18, 2017 7:50 am

I found answers to most of the questions I asked this weekend. First of all I think the MOSFETs were breaking due to the high voltage spike when the switch turned off, so I actually changed my switching topology to a "half bridge" schematic below:

Image

As you can tell this circuit only turns on above a certain Input voltage, when V_gs on the P-MOSFET gets large enough.

I also derived the not so universal EMF equation for a square wave signal: V=N*A*B_max*f/D (V voltage, N turns, A core area, f frequency, D duty cycle).

I don't care about effeciency that much as long as it is above 50% (my variac can deliver 1kVA). I also do not want to lose too much energy in my MOSFETs or core simply because I do not want them to overheat.

The turn count should stay as low as possible because I only have 10 meters of HV wire, however if I use all of that wire it will cover the entire core, so I could wind a second layer using standard wire. There is no way the top layer could cause breakdown through the bottom layer into the primary coil

Post Reply