Making a DCDC converter

mxlemming

100 kW
Joined
Jul 17, 2020
Messages
1,117
I'm really not sure why I'm doing this so it'll get kicked to the wayside first of all my projects.

Inspired by zombiess DCDC thing, I decided to quickly lay out a DCDC controlled by an stm32 (literally the cheapest stm32 made) and an ir2184 gate driver with external FETs. I got 5 of them from jlcpcb fully built for 60usd.

The concept is to use the 12V output to feed the gate driver and MCU via a regulator, with a push to start few kohm resistor and 12V Zener diode to kick it off.

IMG-20211114-WA0000.jpeg

Well here it is, 20x40mm and might one day be capable of 4A.

So far I was initially very pleased that it programmed and can read all the voltages and currents and do pwm and hit all the interrupts... then I found the kick start circuit can barely power the MCU and realized that the gate driver cannot be used in discontinuous conduction mode because it needs time to charge the bootstrap cap.

This thing is going to be a pain in the arse. Why am I doing this.
 
After ~1.5 years, I got some time to finish this, spent about 6 hours this afternoon. Total time on this I think is about 24h effort between design, programming, debugging...

The original hardware works, with a minor mod to the starting resistor.
I ran it at 1.2A for about half an hour. 12V 1.2A = 14.4W... PSU flips between 15.99 and 16.28W so... call it 88% efficient/
WhatsApp Image 2023-01-25 at 17.21.53.jpeg
The main heating, surprisingly, is the bootstrap diode. I messed this up and used a 1n4007 2 years ago, need to swap it for a Schottky, and all that heat will probably go away. High side MOS seems to be next hottest, surprising since it is running at about 35% duty.
WhatsApp Image 2023-01-25 at 17.36.31.jpeg
The switching is fine, and the regulator keeps reasonable accuracy. It is a really crude PI controller, hacked for M0+ integer only core, with an extra loop for overcurrent protection and feedforward regulation limits.
WhatsApp Image 2023-01-25 at 17.48.44 (1).jpeg
Switching is really perfect at 1.2A. Nothing to see here, except that the dead time is waaaaay too long. Could have 1/4 the dead time.
WhatsApp Image 2023-01-25 at 17.48.44.jpeg
Regulation with 43V input is 300mV ripple. This is OK, but it would probably benefit from a bit of a higher switching frequency. Maybe I'll up it to 200kHz with the schottky diodes.
WhatsApp Image 2023-01-25 at 17.48.45.jpeg
No zingy stuff on the output. Absolutely clean. I am quite surprised by this to be honest.

Not been brave enough to use it for anything other than powering a heatsink yet... Think I'll run it up to 84+V and see how I feel.

Moderate success. Woo.

https://github.com/davidmolony/HVDCDC
I really don't advise you build this. You can if you want, but you can buy perfectly reasonable ones from mouser for 20$.
 
Ran this up to 90V today, switching still clean, but some zinginess appeared (edit, that was me having a ground clip fall off, it's totally clean).

DCDC 90V in 1.2A.png
DCDC 90V single pulse.png
DCDC ripple at 90V.png
Ripple increased, but not too bad. Higher frequency would be nice.

All good basically, surprisingly OK. Wonder how high it COULD go, but not going to push my luck for now.
 
Hi Sir, hoping Ur having a nice weekend :thumb:
There is a lot effort into that DC-DC module development!! Congratulations!
Do you plan to resell modules as a final product?

Last post U said limit to 90V, thats great for battery variable charging applications

Is possible to run module output from 42V to 72V and CC adjsutable?
two potentiometers to control module output, one for voltage adjust CV, and other for CC constant current adjust

How much amps draw max at 42V? and how much Amps draw at 72V setting?
step down topology
 
Awesome work 👍.
A decent DCDC is a valuable thing.
I really need someone to produce a compact (like, pack of smokes size) DCDC capable of stepping 700 V DC down to 12 V. Seems most stuff is designed for trams and railways, and weighs about half a ton.
 
Awesome work 👍.
A decent DCDC is a valuable thing.
I really need someone to produce a compact (like, pack of smokes size) DCDC capable of stepping 700 V DC down to 12 V. Seems most stuff is designed for trams and railways, and weighs about half a ton.
Cheers, it really is quite hacked together at the moment, I did not ever think people would be that interested.

700VDC step down is not an easy feat I think.
1) It will need a transformer to get the voltage ratio.
2) With inductors/buck topology they will probably end up being very bulky to handle the voltage and the current, and the on time of the buck will get very very short.
3) Startup becomes an even bigger hassle since the linear regulators you can use for jump starting a 100V buck get very big and lossy for 700V.
Not one I would jump into readily (would need a 700V power source as well). I could do it... It's within my ability, it's just not a case of a few evenings like this DCDC.

Hi Sir, hoping Ur having a nice weekend :thumb:
There is a lot effort into that DC-DC module development!! Congratulations!
Do you plan to resell modules as a final product?

Last post U said limit to 90V, thats great for battery variable charging applications

Is possible to run module output from 42V to 72V and CC adjsutable?
two potentiometers to control module output, one for voltage adjust CV, and other for CC constant current adjust

How much amps draw max at 42V? and how much Amps draw at 72V setting?
step down topology
No plan to sell it, the files are online if you want. It is a curiosity rather than a useful thing for me.

You could easily reprogram the MCU to run CC and CV, but you might end up needing two of them - one to power the logic side of the second and the second to do your battery charging.

This is a Buck topology, and the amps are related to the duty ratio. 1A output at 12V means you need 0.25A at 48V, 0.133A at 90V (plus a bit for inefficiency).

Again, making a battery charger is something I COULD do, but I am unlikely to because I have not got the time to do it recreationally.

If I was making a battery charger, I would make a boost converter, since that is a lot easier. Could run from a 48V off the shelf PSU and step it up with current and voltage regulation. It's a much much easier problem.
 
Back
Top