Switching 12V power supply PCB

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
current sink capabilities

Hi all,


Smarmydog, why are you using the 3906's?, they can only sink 200mA, while the outputs of the sg3525 can sink 500mA. Do you really get a speed up at the mosfet turn-off placing that transitors instead running them directly from the controller?
 
Good eye J.Carlos!:magnify: The transistor in the actual circuit is an On Semi MPS6652. I grabbed the 3906 part from a Protel library and didn’t change the part value. I’m with Alme – the magnetics design is where the real challenge is. Hopefully after I nail that I won’t need any special gate drive stuff.

-Casey Walsh
 
I grabbed the 3906 part from a Protel library and didn’t change the part value

That's what I was thinking

By the way, did you notice an amplitud increment on the "noise" at the output voltage with that fastest turn-off time?, many car audio amplifiers that I have seen uses relatively long times to turn-on and turn-off their mosfets, maybe 'cause that's the way they meet EMC.


Greetings from México

Juan Carlos.
 
Got my design now fellas. Going with a SG3525, with 3 IRFZ44 per rail, using the two inch toriod core from my blown amp (yes, its ferrite, 5 turns for each primary, and a +/- 60V output), got highspeed diodes from blown amp that still work(25nsec), and capacitors(about 10000uf on the input and 4500uf on each output rail), using a frequency of 37khz with the totem pole transistor out puts off the pwm and 10ohm resistors going into each... will get results back to you in a few weeks, maybe even with a few pictures...

two quick questions, input choke value(thinkin 100uH), and i read earlier in this thread that someone shorted the delay to reduce dead time to a minimum, is this advised???

michael
 
Michael, I think such a choke 100uH, which will accept 50A (peak even more?) current without saturation, is quite large. Usually they use a choke in battery wire of up to 10uH to improve EMC. You will need its wire cross-section not less than 5sq.mm.
If you use some 2A transistor followers at PWM outputs, it's possible to decrease deadtime to its minimum. Without them, keep additional discharge resistor of say 100 ohm.
Also you can try to decrease switching frequency to some 27kHz (if there's really 2-inch core, you'll get slightly higher efficiency).
And report about your progression :)
 
Hi Everyone!,,

I am actually working on a 500W SMPS for a Tripath Class D Amplifier. The PCB layout is complete and I am awaiting the PCB from the manufacturer.

I am using a PSoC microcontroller to create a dual PWM with deadband at 200KHz. These signals are going into non-inverting 6A MOSFET drivers. The drivers switch the MOSFET through 4.7 Ohm gate resistors. The MOSFETs then switch on a custom wound toroidal transformer. Owen Columbus (diy member opc) painstakingly wound the transformer, which is composed of 22 primary turns and around 80 secondary turns, it's beautiful!

I have a bank of capacitors 26400uF 25V on the input a total of 40mH of power inductors and 5440uF capacitors for the output filter. The output is rectified with ultra high speed switching diodes. The DC voltage on both rails is monitored via a zener diode based feedback into an opto isolator. The output of the optoisolator is fed back into the microcontroller's ADC input. The PWM's period is then adjusted accordingly when the voltage starts to exceed the limit.

The problem with the design is the expense! The capacitors alone cost 50$, which all are surfacemount electrolytics. I have some pretty crazy via work and wire to solder to unmasked traces for it to handle the current..

I will test the design in roughly two weeks and will post the PCB here when complete with full part list and program. I can probably program some microcontrollers if someone is interested. I guess we will see how the SMPS performs first!

Steve
 
follow up

sorry, no pictures

well, i did it. i am succesfully running a 700 watt mono class d amp of the smps i built without any problems. the transformer gets a little warm, but everything else is great. my base circuit is from ESP's smps project. i decided to just go with what was proven to work except i ran 4 mosfets a side on a big self wound two inch ferrite toroid(16 primary 80 secondary). this thing could deliver over 1000watts continuously. it is not feedback regulated so i do encounter some dip in voltage over peak power demand, but never more than 3 volts, its truly amazing. all of the circuits are built on vero board(the cheap radioshack stuff), and the mosfets and diodes are hardwired with the transformer because of current demand. at 700 watts continuous i'm only pulling 75 amps so thats about 75 to 80 percent efficiency throught the amp. its in a huge case though. the ESP project is a great base for any where between 300 to 1000 watts, thanks a bunch for the help. feel free to ask any questions about my project.

michael
 

opc

Member
Joined 2004
Paid Member
Eva,

There are some concepts that are pretty sound with SMPS design. Honestly, there isn't much you can screw up with the simple typology used. I am basically switching the primary via a center tap and rectifying the output. I'm not trying anything new and groundbreaking. Also, it's quite hard to prototype something accurately that is such high power... I'd want it to run at its expected capacity.

I have been pretty lucky with these things in the past and hopefully the trend will continue. Or maybe I will get a nice wakeup call and do the best prototyping job possible.

Thanks 'burnedfingers'! You gotta love that attitude! At least if it doesn't work, you can get a good show!

Steve (whoops, on OPC's account :( )
 

opc

Member
Joined 2004
Paid Member
kysoftdrink,

How did you measure this 1000W continuous power from your switching power supply? I am skeptical of core saturation.. I am told that these cores can only produce around 500-600W max., then they saturate :(

I have only placed an upper limit on mine to be around 500W due to core saturation. This would be the only limitation up to that power lower with my design..

STeve (opc account, whoops :) )
 
opc :

Core saturation is not related to output current at all, it's related to input voltage and dead time


Also :

I am using a PSoC microcontroller to create a dual PWM with deadband at 200KHz. These signals are going into non-inverting 6A MOSFET drivers. The drivers switch the MOSFET through 4.7 Ohm gate resistors. The MOSFETs then switch on a custom wound toroidal transformer. Owen Columbus (diy member opc) painstakingly wound the transformer, which is composed of 22 primary turns and around 80 secondary turns, it's beautiful!

I have a bank of capacitors 26400uF 25V on the input a total of 40mH of power inductors and 5440uF capacitors for the output filter. The output is rectified with ultra high speed switching diodes. The DC voltage on both rails is monitored via a zener diode based feedback into an opto isolator. The output of the optoisolator is fed back into the microcontroller's ADC input. The PWM's period is then adjusted accordingly when the voltage starts to exceed the limit.

I was talking about this. It's quite an ambitious project, particularly the stabilisation of such a control loop with dual outputs in the digital domain by means of a DSP
 
well, the toriod came from a blown amp that put out 1200 continuous. so i was planning that if it could put out 1200+ watts continuous, it should do 1000 on a custom design with no problems. but i was able to get the dead time switching frequency of the old amp from the chip design and various resistors and capacitors and i just ran my design on the same frequency and dead time to save trobleshooting. so large toroid with the same switching frequency and dead time putting out 1000 continuous when it is capable of putting out 1200 seemed alright in theory. when put to the test, it performed marvilously.

i recomend, if u have the money, find a high powered amp that uses a single toroid and is blown, and use the toriod, switching frequency and dead time from that amp since it was designed by profesionals.

michael
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.