Building my first 12v smps...

Tolik

Member
2007-03-24 6:31 am
Hi all,

After looking at all nice projects here, I started to build my own 12V SMPS for my SKA modules. After prototype was assembled, I
noticed that while SMPS not loaded, primery`s waveforms are pretty rectangular:), but some strange forms of voltage on primery windings while simple resestive load connected. The output voltage drop from 60 to 30 V under 200-240W load :confused:. This is a kind of rigning. Why it occurs ? Is this caused by flux imbalance, bad pcb layout, or core saturation ?
I captured some screenshots from my scope to illustrate problem.
Please help me to determinate what`s wrong here ?
My setup:
Core - Amidon FT140-A-F (Area of core 0.968cm^2, A window 4.155 cm^2, Mean Lenth 9.252cm)
Primery - 2X5 turns 3X18AWG
Secondary - 2X28 turns 18AWG
F=50000Hz
MOSFET`s - STP80NF06 2+2
Diodes MUR1520G 15A 200V 30ns
Capacitors : input - some cheap caps 6X 4700 uf 16V, output - currently 27000uf.
 

Attachments

  • psu.zip
    14 KB · Views: 470

Tolik

Member
2007-03-24 6:31 am
First - thank you so much for trying to help !

The shematic attached is for power stage only. Please notify me if you wont to see the cshematic of control stage, or pcb images.
I think about several possible reasons of my problem:
Core too small for 350...400W ?
Area of windings (total cross section of wires) too small ?
Bad quality of main primery capacitors (6X 4700uf 16V are no brand, but 4X 1500uf 16V are the best motherboard caps from Nichicon)
Bad PCB layout ?
MOSFET`s are connected to SG3525 outputs directly ? (now I ordered UCC27324 drivers to improve situation)

What you think about it ? Any you suggestions will be much appreciated.
 

Attachments

  • power_st.zip
    16.3 KB · Views: 350

ifrythings

Member
2005-10-24 2:07 am
switchmodepower said:
Can you post control schematic also?
Are you running a push-pull in voltage mode? If so, that's a no no- how do you handle flux imbalance?

He's running it un-regulated, so thats not a big deal then.

When you have core saturation, you will see (and maybe hear the transformer) very large spikes on the leading egde of the waveform.

Try a larger value resistor for your mosfet gates 10 -22 ohms and see if that helps
 

Tolik

Member
2007-03-24 6:31 am
Thank you all for trying to help !

Hello guys,

As ifrythings told, currently control stage is un-regulated for simplifyng of test procedures - I can play around Fsw, dead time, & duty cycle. Generally was SG3525 selected because of low price - I don`t afraid to burn some chips during tests.
In final version of my SMPS I`m planing to use current mode operation. What will be better UC2825 or UC2856N for this purpose ?

Iifrythings,
The transformer not produsing neither noice nor heat, it running at may be 5..10C* above room temperature, but spikes are up to 50v:bawling: as seen on screenshots.

Switchmodepower,
Absolutley agree with you, disadvantages of voltage mode control for push pull topology are well known. Just I`ll fix the power stage, I`ll start to design current mode control block based on one from the chips specifed above. I wont to make powerful & relable, like military grade SMPS, available for DIY community.:xfingers:

Luka,
Now I using 12V 16A rated output from old 400W PC`s SMPS because I don`t have another power source. Is it able cause such problems ? Is simple mains transformer (250-500W) based power supply will be better here ?

To all,
I`m able to work around this my project at weekends only due to my very busy job- 12..14 hours per day, so results/updates of testing on prototype I`ll be able to post mainly at fridays - saturdays.

Chematic of control attached-
 

Attachments

  • schematic prints.pdf
    11.5 KB · Views: 423
Hi

Looking at your gate picture, the one @ 200w load, you have only ~8v on gates. How did that happen? I mean I have supply for SG connected to input of smps, so if input is 12V, gates signal will be just little bit lower, no where few volts lower.

Oscilations that you have on drains @ 200w are I think just after turn ON of fets, leakage inductance creates those spikes, use of filter will make them much more smaller.

You should put filter on seconday too, for even better performance once you even get it. I would use freq. in 20-35kHz range, this will help all spikes, oscilation,.. you have.

Use of unregulated, or regulated, just to keep low and no load voltage to what you want, but unregulated will perfor better then you think...

If you could explain how did you wind your trafo and how many of what wires will maybe help too.


Luka, now I using 12V 16A rated output from old 400W PC`s SMPS because I don`t have another power source. Is it able cause such problems ? Is simple mains transformer (250-500W) based power supply will be better here ?
It doesn't matter what you use, but your 12v 16A supply probably isn't enough, since whis is only 192w IF voltage is really 12v then and IF current is really 16A... with this supply you will be able to test probably no more then 170w of load to smps... Check your input voltage and at the same time output...you will see that transformer voltage ratio won't change much
 

ifrythings

Member
2005-10-24 2:07 am
Re: Thank you all for trying to help !

Tolik said:


Now I using 12V 16A rated output from old 400W PC`s SMPS because I don`t have another power source. Is it able cause such problems ? Is simple mains transformer (250-500W) based power supply will be better here ?

The problem with PC power supplies is that the 12V line is loosly regulated, the 5V is the main regulated line. I've seen the 12V get pulled to ~10V before the UVLO kicked in.

My main guesses right now would be

1) What is the input voltageage at full load (As Luka stated)
2) Hows your transformer wound (As Luka stated agin)
3) Do you have thick short traces from the mosfets to the transformer?

P.S: try exparimenting with different gate resistor values too.
 

Tolik

Member
2007-03-24 6:31 am
Hi ,

I think that input voltage drops to about 8V. I`ll chek it exactly.

Transformer -
primary winded directly to core, 3wires 18AWG (about 1.08 mm diamiter) per each half. Wires not twisted. Both half windings, and all wires are placed wery close one to other.
Secondary plased ABOVE primery such way, that it fills all spaces between primary turns. Some of secondary turns plased above primary wires. Secondary contain 2 wires 18AWG.
Primary - 5+5 turns, secondary 28 turns.
I`ll try to make good quality pics.

PCB - I`ll post the images tomorrow.

I tried to go to 24-30 kHz, with almost same results. I`m preffering higher F sw @ lower Bmax to get more power from the trafo
 

Tolik

Member
2007-03-24 6:31 am
Hi Luka,

Did you make whole turn on core with each primary...

The distance between primary`s ends is about 18-19mm...
Is it can to cause a problem ? On you taste, how better to plase windings - secondary over primary, or otherwice - to wind secondary directly to toroid, and than place primary over it ?
Today I recived a haf pound of 63V 680uF Panasonic FC caps. These should be good for output filter . I think to place 3-4 per each +/- output.

Ifrythings,

PCB image attached. Top copper - red, bottom copper blue.
 

Attachments

  • test_board.zip
    30.3 KB · Views: 185
Hi

You wind primary first always, then secondary over it, even if on core next to primary... Space between start and end is not important, just that you try to make whole turn, probably not because of how you make board... I did some.

You have "jumper", more like just holes to connect upper and ground on other side...what I would do is remove them, and connect lower and upper gnd where cable comes on board, other then that is ok
 

Tolik

Member
2007-03-24 6:31 am
Hi all,

As was suggested above, I tried to reduce Fsw to 25KHz, add snubber across primarys, also I added small parallel inductor between recetifers to output caps ( now two Panasonic FC 63V 680Uf per side), primarys cross section was increased to 5x18AWG.
After redusing Fsw, & adding 22R+10nF snubber, SMPS started to work some better, spikes almost disappeared. Transformer became warm - 12..15C* above room temperature. Due to my calculations B now is about 4000 Gauss :hot: Is it not too many ? Do I need to operate at so high B value ?

Another improvment of output voltage was after two 1uf ceramic caps was placed across output filters. Even when I switched back to 50KHz, output voltage was almost clean.

Everything above is intresting, & some miss understood for me, because due to component readings, SMPS of 1000W Genesis monoblock, or SMPS designed by mr. Lumanluv (http://valveaudio.tripod.com) are running at 50..75 KHz just fine with simple filter components, even without snubbers .:confused:.

Unfortunaly at this moment my old PC SMPS, that was used as 12V source was blown :flame: , and I was :redhot: :headbash: :cuss: :cuss:... Yeah
So I`m going to build new simple conventional 12V power supply.
 

Attachments

  • some_better.zip
    3.5 KB · Views: 158
Just a random thought somewhat off topic. You mentioned wanting to use a computer SMPS, but can't because the 12V output is unregulated.

A resistive divider from +12V to ground with the tap connected to +5V will put the 12V output inside the feedback loop.

maybe a 150 Ohm from 12V to the divider and 50 Ohms to ground. This results in a load of 47mA on the 12V and 100mA through the 50 Ohms to ground. Since about half the current in the 50 Ohm resistor comes from the 12V output any change in 12V output will affect the 5V line thereby closing the loop. The 150 Ohm resistor dissipates .33 watts and the 50 Ohm dissipates .5 watt. Many other value combinations are possible.

I have no idea what the current ratings are for a computer supply, but I do suspect that most modern ones will probably have and regulate a 3.3V output.
 
Tolik-

Getting back to your SG3525 control schemo, three thoughts:

1) I like the adjustability of F(osc). If there is a noise issue at a particular frequency, you can simply move it off the band. Try replacing the 50kW pot with a 47kW in series with a 5.6kW, giving an adjustability of (+/-) 10%.

2) I also like the deadtime adjustability. Very cool.

3) I found an obscure App note from Silicon General (the original creators of the SG3525) on synch'ing their oscillators together. In this note (PS-7), it mentioned that it was good engineering practice to ground any unused (sync) pins to eliminate any false triggering.

Hope you get the desire to go back and revive the dead SMPS.

Steve
 

Tolik

Member
2007-03-24 6:31 am
Hello Steve,

Nice to see you again :)

3) I found an obscure App note from Silicon General (the original creators of the SG3525) on synch'ing their oscillators together. In this note (PS-7), it mentioned that it was good engineering practice to ground any unused (sync) pins to eliminate any false triggering.

OMG :eek: :eek: :eek: HOW I LOOSE IT ?! This is very important .
Thank you for good reminding, my stupid @ss always missing something essential :( .

1) I like the adjustability of F(osc). If there is a noise issue at a particular frequency, you can simply move it off the band. Try replacing the 50kW pot with a 47kW in series with a 5.6kW, giving an adjustability of (+/-) 10%.

Can you to discribe it more carefurly plz ? I tried to play with F80...25KHz, with almost same results.
Is it possible that problem caused by powerless 12V source ?

Hope you get the desire to go back and revive the dead SMPS.

I`m deraming about it.:) :smash: