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.

Re: Switched Mode Power Supply

Postby Finn Hammer » Tue Jul 18, 2017 6:20 pm

Niels,

I have been following this thread from the start, and I feel it is time to chip in.
My background is in the special full bridge converter that you will find in a DualResonantSolidStateTeslaCoil, where I have come up with a couple of engineering "firsts", as well as produced several units for museum, art and television shows. I have also built a handfull of the most common DC-DC converter topologies.
1. About driving mosfets with an Arduino:
Take a look at the datasheet of a IRFP250
https://www.google.dk/url?sa=t&rct=j&q= ... uw&cad=rja
on page 4 you will see this diagram:
Udklip.JPG


This diagram shows how the Drain-Source voltage relates to the charge you transfer to it (the gate). As you will see, the voltage stops rising momentarily when it reaches 6.5 volts, and this horisontal line on the graph is called the miller plateau, this is where the mosfet starts to turn on. You have to drive the gate higher to turn the mosfet properly on, and you can not do this with the output of an arduino, which only puts out 5 volt.
The arduino can also not deliver enough current to turn the mosfet on properly, and that is why there are gate driver chips. You need a totem pole driver with a bootstrap capacitor to also turn the top mosfet on.
2. Topology
There is no need to invent a new topology, as the one you show in your last post. Read a good application note on topologies
https://www.google.dk/url?sa=t&rct=j&q= ... SA&cad=rja

and choose one that is suitable for a high power supply. You will see that most of the topologies stop at around 500W, and from there, there are only the resonant converters that come through. My own pet is the series resonant converter, because it is easy to build and understand, it supports zero current switching, and with that, a lot is gained. If you have ever tried to design snubber circuits for a hardswitching converter you will follow me here. Zero voltage switching is even better, but then you have to build a LLC converter, which is harder to construct.
As a note to topology: My advice, ditch the arduino and get a proper dedicated converter chip that is designed to do just that job. A good old fasioned chip for an entry level converter is the TL494, but there are newer ones also.
3.Layout
When it comes to building power electronics, proper layout practices are key, and along that note, I will encourage you to post actual pictures of your work, because in that case we will probably be able to point at obvious mistakes.
These are the most obvious points I will point at, you have a lot of gumption and that is commendable, and needed, if you are going to see this project to the end with success. The most important advice I can give you is: read the application notes from TI, Linear, etc. They have covered this subject in great detail, and often the notes are written by authors who have exceptionally great command of written language, so that it is well suited for even a lay person studying on his own.
I wish you good luck on this journey, there are a lot of frustrating moments ahead, but there is also a lot of satisfaction to be gained.

Cheers, Finn Hammer
User avatar
Finn Hammer
 
Posts: 29
Joined: Sat Mar 05, 2016 12:21 pm
Real name: Finn Hammer

Re: Switched Mode Power Supply

Postby Niels Geerits » Fri Jul 21, 2017 10:18 am

Finn,

Thanks for your thoughts and input!

1. I actually do not drive the power MOSFETs directly with the arduino. I use the Arduino to switch a smaller MOSFET (e.g. BS170), which acts as an amplifier for the power FETs (IRFP360, 250 and IRF9640). The drains of these smaller FETs have a 12V Input and are hooked to the gate of the power MOSFETs and a 47 Ohm resistor which limits the current to around 250 mA (this current seems high enough to limit switching heating). You certainly do have a Point, first I was trying to drive the power FET directly with the arduino and this did not work very well. Also before I used the BS170 as amplifier FET I was using IRF540s which should turn on somewhere at around 5V, however this was causing shorts (5 A current with no on the secondary) even at low duty cycles. The BS170s seems to turn on at 5V though.

2. Thank you for the report it is very helpful! I am not sure I know what you mean when you say don't invent your own topology? I think the one I showed here is the half bridge topology, I just added a simple and cheap method to operate the top switch. I have headr of the resonant converters but have not looked into it very much. I think I would need better instrumentation to build one since I would need to know my primary and secondary inductance very precisely, so that I can select the right capacitors for my frequency (or select the right frequency for my capacitor). But it does seem like an interesting project.

3. Thanks I will try and post pictures some time next week. Once I get everything from the tabletop into a box.
Niels Geerits
 
Posts: 27
Joined: Tue Jun 13, 2017 10:25 pm
Real name: Niels Geerits

Re: Switched Mode Power Supply

Postby Rich Feldman » Mon Jul 24, 2017 2:27 pm

>> I just added a simple and cheap method to operate the top switch.

Its output impedance is almost 1000 ohms! How many mA are there to charge, and to discharge, the P-FET gate?
The charging current, and ultimate Vgs level, are proportional to the bus voltage. For what voltage did you design it?

When are you going to get an analog circuit simulator program and learn how to drive it? There are good ones that can be downloaded and used, honestly, for free. One popular one is SwitcherCAD, from Linear Technology, often informally (and later officially) called LTspice. We'll have to see what happens now that the LT company is part of ADI.

SPICE is but one limb of an evolutionary tree; another is MSINC. https://www.allaboutcircuits.com/textbo ... -of-spice/ gives some history and credit. SwitcherCAD was developed to be good with magnetic components, power switches, and feedback regulation in SMPS circuits.
Richard Feldman
User avatar
Rich Feldman
 
Posts: 861
Joined: Mon Dec 21, 2009 11:59 pm
Location: Santa Clara County, CA, USA
Real name: Rich Feldman

Re: Switched Mode Power Supply

Postby Niels Geerits » Sat Jul 29, 2017 1:00 pm

Rich the P Fet will turn on at 100V input this corresponds to 100mA roughly. It is designed to work up to 350V so roughly 350mA at the gate. Perhaps you recall that the n channel fet is getting roughly 250mA to charge the gate. Until now I have had no issues.

The reason I haven't done any simulations yet is because I assumed simulators would cost a lot of money so I didn't bother to look so this is good to know. I will try it out.

I setup my new design and wound my transformer for "full scale operations" 10 primary turns and a bit over 230 secondary turns. I also soldered my quadrupler and made some arcs yesterday. it pulled about 2 amps in the primary while outputing roughly -20kV (this is based on the arc size still need to build a suitable voltage divider for this voltage).
Niels Geerits
 
Posts: 27
Joined: Tue Jun 13, 2017 10:25 pm
Real name: Niels Geerits

Re: Switched Mode Power Supply

Postby Niels Geerits » Mon Jul 31, 2017 1:28 pm

As promised some videos and pictures of the supply. First the complete circuit diagram (probably more valuable than pictures of the actual supply):

Image

Now a picture of the actual supply. The switcher part:

Image

It is hard to see what is what if you didn't build this due to the cable jungle (before switching on I always make sure conducting surfaces that should not be touching are not touching). Upper left corner low side switch on a massive heat sink, next component to the right is the 3.3mF cap, to the right of that you will find the high side switch on a not so massive heat sink. In the lower right corner you find the FET I use to switch the high side switch. The reason for different heat sink sizes is simply because I am trying to use what I have laying around (not that much). Just to the left of that FET you will see the diode bridge rectifier. To the left of that on the perfboard you will find the two caps which create the common terminal, two 120 ohm resistors in paralell forming effectivly 1 60 ohm resistor, followed by two 50W 560 ohm resistors in series. On the perfboard you will also find a few leads going to various drains, sources and gates of different MOSFETs. I also have a few ground leads and 1 "common" lead to which one terminal of the transformer is attached. On top of the perfboard you will also find an Arduino UNO and a breadboard. The breadboard has two 47 ohm resistors and 2 bs170 FETs plus the leads of the 12V power supply go here. This setup is used to generate a 12V 250mA PWM signal for the two N-FET gates. You will notice I use soldered wire, crocodile wires and thin wire for the breadboard. The soldered wire is used where there will be 1A or more, the other wires are used where I have lower currents. You will notice that some places that will see more than 1A still have crocodile wire. Don't worry this will be replaced.

The transformer and multiplier (not in oil yet):

Image

You will notice that the secondary of the transformer has mutliple layers. The upper layers are thin insulation while the lower layers are thick insulation. The area with a lot of tape is the solder point. Below the tape are 2 more layers of tape and 2 layers of shrink insulation (not sure if this is the correct term, it is insulation that shrinks when heated, pretty standard). You will also notice that there the blue secondary wire is never on top of the primary. This is to avoid breakdown and a short. You may also notice that the winding pattern of the blue secondary wire is less regular. This is a result of me trying to avoid gaps in the lower layers. If the blue secondary wire were to touch the transformer core anywhere there would probably be breakdown followed by a short.
The multiplier is soldered to a perfboard and connected to the transformer via some aligator clips. I will solder some wire to connect these. I will also solder some 50kV wire to the output and common of the multiplier (at the point where the wire leaves the oil).

Finally a video of an arc I made with this supply:

https://youtu.be/u6_XeROXMhY

EDIT: The arduino code I wrote (you do not need pin 7 or the serial begin. I had that in for testing purposes):

Image
Niels Geerits
 
Posts: 27
Joined: Tue Jun 13, 2017 10:25 pm
Real name: Niels Geerits

Re: Switched Mode Power Supply

Postby John Futter » Tue Aug 01, 2017 7:03 am

Niels
The multiplier looks good
The rats nest in the plastic box is sure to please the purveyors of silicon devices
Even at these low frequencies stray inductance and capacitance are destroyers as your circuit goes into parasitc oscillation.
What you do should look like your circuit diagram, print off a copy and lay each bit over its diagram counterpart. That now is the finished size and look of what should be built
Ie no wire longer than it has to be especially the gate drain and source wires of your active devices
John Futter
 
Posts: 1168
Joined: Thu Apr 22, 2004 2:29 am
Real name:

Re: Switched Mode Power Supply

Postby Niels Geerits » Wed Aug 02, 2017 7:59 pm

John,
Thanks. I did not know that stray inductance and capcitance could be that destructive. I thought it would only matter for applications where the form of the signal is absolutely critical. I will work on making it cleaner.

The past few days I distilled and then froze some water in an attempt to make a high value high power resistor for the HV supply and it worked! I also used a small bit of hair wax (parrafin) which almost instantly melted. Tomorrow my 100 10 M ohm resistors shall arrive so I can do some accurate voltage measuring.

I also ran into some trouble. I took an old light bulb and put the negative HV terminal on one of the light bulbs terminals (hoping to ignite a plasma as they have a vacuum inside AFAIK) and put the reference terminal on the glass of the bulb. No plasma but I did see a thin purple beam (basicaly an arc) in the bulb. Eventually it arced through the air instead of the light bulb (huge arc btw). At that point the primary drew more than 5 A (at 200V variac). After that my transformer made a strange noise which can best be compared with putting a small screw in a glass cup and shaking it around. I assumed that this meant I had a short somewhere. The primary current draw made it look like I had a shorted secondary. So I removed most layers of the secondary and found that there were some holes in the tape around the primary. I replaced the primary rewound most of the secondary and now it seems like it is working again.

Now I wonder did the primary break due to high voltage arcing from the secondary or did that current draw >5A break the primary insulation?
Niels Geerits
 
Posts: 27
Joined: Tue Jun 13, 2017 10:25 pm
Real name: Niels Geerits

Re: Switched Mode Power Supply

Postby Niels Geerits » Sun Aug 06, 2017 12:30 pm

I succesfully replaced the broken components and rewound the transformer with some extra plastic sheets in between layers for extra protection. I kept the wires short as recommended and added a few extra components in the arduino subsection to limit current (plus some pull down resistors on the bs170s). I also placed 100 10 M ohm resistors and 1 100 k ohm resistor in series behind the multiplier. I measured 2.5V over the 100k resistor with the variac at 30% of max power. With a frozen distilled water load the supply pulls 4mA and stays stable at -25kV.

Here some pics of the supply:

1st the supply without the arduino inside:

Image

The remaining cable jumble goes to the rectifier and variac.

2nd the supply with the arduino. The Bs170s, current limiting resistors and 12V 1A supply are hooked to the breadboard:

Image

And a final note I am getting an oscilloscope to look at the waveforms in my circuit. I want to know if there is anything that might lead to instability.
Niels Geerits
 
Posts: 27
Joined: Tue Jun 13, 2017 10:25 pm
Real name: Niels Geerits

Re: Switched Mode Power Supply

Postby Niels Geerits » Sun Aug 13, 2017 1:12 pm

I used my oscilloscope to probe the circuit. I used both channels and used the difference math function, because my circuit is earth referenced. All gate source signals look clean with no ringing, same goes for the drain source voltages except for the P-FET I found some ringing there during turn off. Here is a picture of the waveform (note its inverted):

scope.jpg


I am thinking about using some zener diodes and a resistor to clip the ringing at 400V (FET is rated for 500V) it seems like the simplest adjustment to protect the FET.

The pulse train you see after turn off is from the transformer as far as I know (this was also present in simulations)
Niels Geerits
 
Posts: 27
Joined: Tue Jun 13, 2017 10:25 pm
Real name: Niels Geerits

Previous

Return to High Voltage - Fusor Input Power (& FAQs)

Who is online

Users browsing this forum: No registered users and 2 guests