A Buck Current-fed Push-Pull Converter / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

A Buck Current-fed Push-Pull Converter

previous 1 2 3 next

Move Thread

LAN_403

Sigurthr

Sun Dec 29 2013, 03:23AM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

That is one beautiful etched board! Really loving watching you work on this project. Love seeing people take on projects above my ability level!

Linas

Sun Dec 29 2013, 10:59AM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

Are you using UV method for pcb ?

Somehow my mask leaks UV and i get texture on copper. Where do you get your mask ?

Finn Hammer

Sun Dec 29 2013, 05:07PM

Registered Member #205 Joined: Sat Feb 18 2006, 11:59AM
Location: SkÃ¸rping, Denmark
Posts: 741

Linas wrote ...

Are you using UV method for pcb ?

Somehow my mask leaks UV and i get texture on copper. Where do you get your mask ?

Linas,

Yes, I use transparencies printed on OKI C3100 Laser printer.
A bit more than 2 minutes exposure, and developed with drain cleaner in warm water. Warm water and a squirt of drain cleaner.
Etch in hydrocloric acid, 30%, and about same amount of peroxide 10%. The latter because I cannot get 35% peroxide due to anti terror legislation. Doh!

__=|(:3)-|--{__ wrote ...

Looks lovely, why does it have a shine to it like its gold plated?

Due to the light, which was low, I cranked the exposure to 3200 ISO, and still it got a bit dark. Also the photoresist coating was still on it.

Cheers, Finn Hammer

Shrad

Sun Dec 29 2013, 05:54PM

Registered Member #3215 Joined: Sun Sept 19 2010, 08:42PM
Location:
Posts: 780

for the peroxyde, you should ask your pharmacist about it and show him your boards, they usually get their 30MOL peroxyde in 1L bottles...

mine now accepts to sell me some 1L bottles, but we are in Belgium... they might be less strict on those things... I once received a box of 20kg of KNO3 from Holland, and they were packed in 1kg ziplock bags, and nothing got opened or checked...

Wastrel

Mon Dec 30 2013, 12:20AM

Registered Member #4095 Joined: Thu Sept 15 2011, 03:19PM
Location: England.
Posts: 122

It's a new European law intended to restrict the precursors of explosives. It makes selling more than 10% peroxide to an individual a crime, among other things. If you register as a company the law does not apply. I don't think it's been implemented here yet.

Finn Hammer

Fri Jan 03 2014, 11:43PM

Registered Member #205 Joined: Sat Feb 18 2006, 11:59AM
Location: SkÃ¸rping, Denmark
Posts: 741

All,

The first PCB that I so proudly presented as prototype board Mk1 was an utter disaster. I use Eagle for layout and schematics, and without reflection, just used the footprint of the first mosfet that came in a Dpak.
To my complete embarrasment, and after much debugging, I could finally conclude that it was in fact a type P mosfet, so current didnÂ´t flow quite as expected.
Well, all is not so bad, I had time to think more about what I wanted from a prototyping board, and this is what I came up with.

The whole power section of the board is as I expect it should be, although I will no doubt run into thermal problems when I start to crank the power up.

The controll section, on the other hand, is fully configurable:

Using cut up Dill sockets (the good ones with turned pins)

I created accesspoints to all 3 pins of both the voltage and the current error amplifiers. The internal voltage reference is brought out as well together with a couple of ground connections. The 2 inputs to the current sense amplifier are tied to the current sense resistor at the back of the board, and the feedback resistor to set amplification is the blue one hovering above the LM3827-1. This resistor can be changed in *a jiffy*, so the cycle by cycle current limit can be dialled in to any value desired. The one potentiometer adjusts the controller frequency, within a range from 10kHz to 200Khz. The other potentiometer is part of a voltage divider that I presently use to feed pin 6 of the chip for duty cycle adjustment. Apart from this, I sprinkled the board generously with various test points and ground connections for scope grounding clips.

The last picture shows that every chip on board has its own 20ÂµF bypass cap apart from the 1ÂµF ceramic up on top, It shows the separation between Analog power ground, and the digital ground which is attached to analog at the far side of the current sensing resistor, as I beleive it should. Hopefully there are no current loops worth mentioning, since every current path on the top side is dublicated on the bottom, and the power does not cross over the digital area.

At present, all 3 sets of switches are working, the controller is oscillating and putting out timing signals for the switches.
The signals for the bridge switches have a 100nS overlap, as expected, and the signal for the buck switches reacts to duty cycle adjustment via pin6.

So things are looking good right now. Tomorrow, I will install the inductors and start to process some power at low levels.
Until then,

Cheers, Finn Hammer

hvguy

Mon Jan 13 2014, 07:06AM

Registered Member #289 Joined: Mon Mar 06 2006, 10:45AM
Location: Conroe, TX
Posts: 154

Hi Finn, I've been using this topology for ~10 years to drive all types power supplies with good results. There is a lot of good information regarding it in Abraham Pressmans book switching power supply design. The buck transistor does dissipate a significant portion of the total loss though, especially at higher frequency and power. Also, depending on switch timing, a spike may appear across the bridge rails at turn off since there is no C to dampen it. At lower powers I let the fets clamp and dissipate it. At the higher powers a snubber may be worth implementing. GDTs work well for higher input voltage if a dc restore circuit is used. It's cool to see this show up, I don't see it very often. I hope it works out for you.

Finn Hammer

Mon Jan 13 2014, 03:54PM

Registered Member #205 Joined: Sat Feb 18 2006, 11:59AM
Location: SkÃ¸rping, Denmark
Posts: 741

Thanks, Aron.

The snubbers are what I am focussing on right now, and Pressman is my hero!

I have had the controller running for some time, only with a variable voltage into pwminput, no feedback yet.
The buck behaves well, but perhaps because my components were a bit too low voltage types, I blew them regularly, untill I got a snubber on the switch node.
I should probably just have made a buck controller to start with.

Anyway, I have loaded the output of the buck stage with both a cap/resistor and just a resistor.
The latter seems like a good idea, since it allows me to monitor the inductor current directly at the output.

When I start to load both push and pull switches with resistors, everything is fine, I get the switch overlap that is needed and the waveforms check out.

When I put the transformer on, all I see is horrid ringing during the main buck switch *on* period, and this ringing takes out the controller.

I suppose it could be the dreaded leakage spike, but I really donÂ´t have a clue right now.
I am reconfiguring the board to utilise high voltage parts, and only single switches, to better understand what is the problem, and this board has generous space to play with snubbers.
I am also changing to a buck inductor that has less tendency to drop into saturation.

Interesting and hard.

Cheers, Finn Hammer

Finn Hammer

Tue Jan 14 2014, 11:09PM

Registered Member #205 Joined: Sat Feb 18 2006, 11:59AM
Location: SkÃ¸rping, Denmark
Posts: 741

I put some pride in making proper pcbÂ´s, even during the design phase, undoubtedly not just due to my ability to produce them myself. I like the process of laying them out in eagle as well as the chemistry. I hate to work with birds nests and stripboard.
And I love these pcbÂ´s, even affectionately calling them names.
Like these last ones, I call them "Buckingham Palace", and the royal quarters are under constant and heavy remodelling.

This last iteration will bring me success, I can feel it!

First of all, I have independent household power to the controller and drivers, separated from the adjustable power to the buck section. The household supply is one of those silly cheap boost converters that China made available to us.
There is a 600V high side gatedriver, the FAN73711, for the buck switch, which is now a IFRP460. Rugged. The push pull stage will also be equipped with 460'es, driven by UCC37321'es.
The buck inductor is a rewound Coilcraft, @386uH. This choke is wound on an open core, has plenty of airgap, and will not saturate before 15A

Last. but not least, active switches and inductor are mounted in screw terminals, and can therefore be replaced without soldering. An old trick from the SSTC days of yore.

The first task was to bring voltage slowly up on the buck, and monitor the switch node for ringing.
Compared to the predecessor, this board has longer leads to everything, meaning higher inductances, and the output capacitances are bigger too, so the ringing frequency turned out to 62.5mHz instead of 139mHz in the previous board.

There is a fancy way to design the RC snubber which goes like this: Add a cap from switch node to ground, one that will half the ringing frequency. halving the frequency means the capacitance has been quadrubled, so the original capacitance is one third of the added capacitance. From this you can work out the inductance that causes the ringing, and with these 2 values, calculate the impedance of the ringing circuit. Use this value as snubbing resistor together with the cap that cut the frequency in half. This way you get Q=1 damping.

You can catch the description here:

,d.bGE

All very good, but I did not have a capacitor small enough to just cut the ringing frequency in half, so I used the snubber from the previous board, a 2.2nF/5ohms combo, and ended up with this:

Not bad, even if I must say so myself, I am satisfied with this result, so will leave it at that. For now.

Final picture shows switch node voltage, with inductor current, I know I've said it before, but it is looking sweet.

The buck frequency is 100kHz.

Cheers, Finn Hammer

DÃ©dÃ©!

Wed Jan 15 2014, 01:27PM

Registered Member #4932 Joined: Thu May 17 2012, 01:42PM
Location:
Posts: 59

Linas wrote ...

Are you using UV method for pcb ?

Somehow my mask leaks UV and i get texture on copper. Where do you get your mask ?

Linas, you can also print several (about 3) transparencies and tape them on top of each other. That's what I do and it works better than you would think.

__=|(:3)-|--{__ wrote ...

Looks lovely, why does it have a shine to it like its gold plated?

Due to the light, which was low, I cranked the exposure to 3200 ISO, and still it got a bit dark. Also the photoresist coating was still on it.

Cheers, Finn Hammer
[/quote1389791724]

I think it's just because the photoresist coating was still on it. I don't think lighting has anything to do with it.

previous 1 2 3 next

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