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

1 2 3 next

Move Thread

LAN_403

Finn Hammer

Sat Dec 21 2013, 10:55PM

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

For power supplies in the 100W+ range, the Buck current fed pushpull converter has been advertised as the ultimate choice, inherently shortcircuit able, isolated and, well, sweet! Since I want a 200kV @ 1mA Greinacher, this is the type of supply I want to build.
I am basing my design on this:

,d.bGE The UC3827-1 combined buck and pushpull controller.
I have little experience with these things, so to get started, I mounted the troller on the breadboard, passified the error amplifiers and fed a DC voltage into pin 6, one of the pins on the PWM comparator. The other pin sees the ramp voltage on the oscillator timing capacitor, and I did this in order to investigate and get a feeling for, how the chip adjusts the pwm, and what makes it do so.

It would appear that a rising voltage on pin6 causes an increase in duty cycle and this observation leads me to conclude that the voltage error amplifier should be inverting. In that case, when the output voltage rises, the voltage error amplifier will return a decreasing controll voltage, in turn instructing the controller to back off the duty cycle.
Next step will be to build the Buck stage, get current feedback to work, then the pushpull stage and finally the voltage loop. This will take some time.

This is how I intend to learn, step by step, blowing chips as I go, but finally ending up with an understanding of the function of the supply.
It is a creepy thing to do, I know, strapping a chip to the breadboard without prior experience. Not unlike if Gollum caught a Mermaid and kept her in a cage on the beach: "We's really likes to get to knows you, sssch" You get the picture, death and havoc.

One problem I have encountered is this:
I cannot for the love of god get the Buck stage gate driver to function as a floating supply. If i tie the source pin to ground, the gate signal looks good and as it should: (where cyan is timing capacitor ramp, Magenta is fake voltage feedback, yellow is Buck gate signal and green is one PushPull gate signal

However, if I remove the wire from source to ground, and float the supply, this is the gate signal: yellow again

I have a diff probe on the gate signal.

I wish someone could recognize the fault, because it is a flaw. Not that it probably will mean a lot, I have some LM5104 sync. buck drivers around, and one of these can interface to the Buck stage and make it sweeter, so that is the way I want to go.

Cheers, Finn Hammer

Wastrel

Sun Dec 22 2013, 12:41PM

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

I think you may have referenced the wrong link when talking about the chip. Please show your schematic so there is no confusion.

Finn Hammer

Sun Dec 22 2013, 05:37PM

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

Wastrel wrote ...

I think you may have referenced the wrong link when talking about the chip. Please show your schematic so there is no confusion.

Wastrel, all,

the link to the datasheet is here:

and here is a grab of the internals of the chip:

As you see, the driver for the buck transistor is supposed to be floating.
Here you see how I hooked the chip up, when the strange fault shows up:

As soon as I attach terminal 3, SRC, to ground, the gate signal behaves well, as shown in the videos in the first post, but since the gate driver is not floating anymore, it would need a buffer to function.

Why not just buffer it, I want to do syncronous rectification anyway.
So I grabbed a surplus gate driver from my Thumper DRSSTC, and hooked it up as a syncronous buck, therefore using the same switches and gatedrivers as I intend to use in the final build.

This worked really well. With 15000uF in and out, a 450uH inductor I was soon putting out from 500mV to 13V into a 9.4ohm resistor ballast, thus processing 18W. The only thing that got hot was the resistor ballast.
I tried to change frequency, the buck functioned from 20 to 200 kHz. Here you see it at 40kHz, and a 300mA buck ripple current, the upper green trace.

This is fun! I never thought Buck converters were easy to make, but then the real challenge comes when the controller feedback loops are going into the equation.
And by the way, I tried to inject DC into the current sense amplifier output, pin 7, and right enough, at 3 volts, this causes the controller to stop cycling the switches. This makes it straightforward to calculate a combination of current sense resistor and current sense amplifier input and feedback resistors.
SInce things appear to function as advertised, i have ordered smd versions of the controller, and inted to design and etch a mk1 pcb, with proper ground planes, to get closer to a fully functioning supply.

Cheers, Finn Hammer

Wastrel

Wed Dec 25 2013, 06:25PM

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

After some reading and some thinking I'm quite surprised removing the SRC connection from GND works at all. Without that where is the driver supposed to sink current to? To have a floating output you need floating power between V+ and SRC. At least I assume that is how this is supposed to work.

Finn Hammer

Wed Dec 25 2013, 09:48PM

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

Wastrel wrote ...

After some reading and some thinking I'm quite surprised removing the SRC connection from GND works at all. Without that where is the driver supposed to sink current to? To have a floating output you need floating power between V+ and SRC. At least I assume that is how this is supposed to work.

I was expecting it to function like the high side/low side drivers I usually use, which have a bootstrap capacitor and supply diode from V+, just like this chip has it.
But well, like I said, never mind, I will just use a syncronous buck setup and ground SCR pin.

I have been working on the Schematic and board layout, and I think it is about finished.
Setting it up as a sync- buck, and going from push-pull to full bridge is my part, the controll feedback scheme is from the app. note.

Going to be interesting to see, how hard it is to get the converter working with feedback.
The first hurdles have been overcome, though.
Selecting the input caps against a 2.3A ripple current and size restrictions was not so hard, I landed on PANASONIC EEVFK2A470Q 47uF 100V 0.5A ripple. A bit under spec., ripple current wise, but more than enough capacitance wise.
Finding a suitable inductor for the buck stage was a bit more challenging. I settled for a quad array of coilcraft - MSS1210-474KEB - INDUCTOR, 470UH,1.7A parts, to handle the peak current of 3.3A and still have enough inductance to perhaps lower the frequency a bit.
The final part I found was the regulator for the 15V household supply. With up to 90 Volts in, I need a part like the TEXAS INSTRUMENTS - TL783CKTTR - REG, HI-VOLT, ADJ, D2PAK-3.

Cheers, Finn Hammer

Wastrel

Wed Dec 25 2013, 10:29PM

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

Why ground SRC, why not connect it to the buck source pin?

Finn Hammer

Thu Dec 26 2013, 12:37PM

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

Wastrel wrote ...

Why ground SRC, why not connect it to the buck source pin?

I thought I had made it clear, that if I connected SRC to the buck transistor source pin, then the gate voltage sagged at low duty cucle signals.
Anyway, that is what you see on this video:

-So that would put me back to square one.

Cheers, Finn Hammer

Sulaiman

Thu Dec 26 2013, 03:02PM

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

your test circuit has no current path for gate drive as shown, try;
9v battery, + to pin1, - to pin3
load a few nf in parallel with 1kOhm pin2 to pin3
rest of circuitry as existing

pins 1,2,3 will be floating relative to power ground

you could (differentially) 'scope pin2 to pin3 to observe gate drive waveform?
maybe vary the voltage on pin1 relative to power ground
with an external supply to check that the drive is truly 'floating' and can be at a higher potential than Vcc

Vcc < 20v , Vpin1 <72v

Finn Hammer

Sat Dec 28 2013, 08:29PM

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

Sulaiman wrote ...

your test circuit has no current path for gate drive as shown, try;
9v battery, + to pin1, - to pin3
load a few nf in parallel with 1kOhm pin2 to pin3
rest of circuitry as existing

pins 1,2,3 will be floating relative to power ground

you could (differentially) 'scope pin2 to pin3 to observe gate drive waveform?
maybe vary the voltage on pin1 relative to power ground
with an external supply to check that the drive is truly 'floating' and can be at a higher potential than Vcc

Vcc < 20v , Vpin1 <72v

Thanks for pointing the current path out to me, I missed it, actually thinking that the SRC pin got grounded internally, when the gate pin went low. In sync. buck configuration, there will be a turned on low side switch to supply a path to ground, in the non-sync version, a forward biased diode will do the same thing. I missed that forward bias......

I got the prototype Mk1. board done today, it turned out right nice:
The narrow traces are 12mill wide, so no reason for complaint here.

Power should not be unleashed on less than 3oz (105Âµm) of copper.....

Cheers, Finn Hammer

Thomas W

Sun Dec 29 2013, 01:08AM

Registered Member #3324 Joined: Sun Oct 17 2010, 06:57PM
Location:
Posts: 1276

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

1 2 3 next

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