SMPS for gainclone

[DT]

Hi to all,

I am making this SMPS (figure 9.) for car sub amplifier. My first block - control cirtuitry is completed and tested. But i have some questions.

I am using not SG3525A but SG2525A. In the datasheet i found that they are absolutely identical, except operating temperature for SG2525 is -25 to 85 and for SG3525 from 0 to 70. But i am not sure about the output logic. Datasheet says - "The
SG3525A output stage features NOR logic, giving a
LOW output for an OFF state" Maybe SG2525 outputs operates different? Must i get exactly 3525 chip for this project?

Chip operates from 12V source and consumes 16mA of current. Slow start works very nice. After power turning on, pulses are slowly groving to its normal size in about 2-3 sec.

I watched with osciloscope at the outputs. The High output is 8us and the LOW output is 10us. So if i am right my oscilator works at about 55kHz. But i am afraid about this difference in duration of LOW and HIGH outputs.
In the project description there is 12k ohm resitors , mine is 11.9k ohms. Maybe this caused such problem? Or anything else?

I am absolutely not strong in this stuff, so your help will be really appreciated.

Thanks to all , and sorry for my bad english.

[sivan_and]

Both chips are interchangable....but for the operating temp.
THe difference in pulse width is due to dead band,Dead band can be minimal if using MOSFETS ....by reducing Rd between Ct pin and Rd pin ie Pin 5 and Pin 7,........Try 47 Ohms..
PS: Not seen the circuit u mentioned.
Good Luck
Sivanand

[DT]

the circuit:



Thanks for suggestion. I will try smaller value of R8.

[DT]

So i have just changet the resistor for smaller one, and now the dead time is shorter a bit.
I connected switching transistors ant transformer and this is how looks my secondary windings waveform:

wave form

main pulse is about 40V but that spikes up to 120V looks to me not very nice

I measured with no output diodes and capacitors and no load. And wires between transistors and transformer was long (~20cm) and thin. Maybe this causes such result?
Where can i look at right smps waveforms?

Thanks again.

[sivan_and]

If therez too much spike...reduce stray inductance in the circuit,ie the lead connecting mosfet drain to tranf should be minimum,not more than few cm,practically posible,If using PCB try few parallel tracks 3#.
R15 and C15 should be there,will reduce overshoot!!!
GOODLUCK!!
Sivanand

[DT]

so, i am back again

I just have rewinded my transformer, now on two glued ferite rings, each inner diameter 29mm , outer diam. 45mm and cros-section 0.65 cm^2. Three enamelled copper 1.5mm diameter wires in parallel form primary - four turns - center tap - and another four turns in the same direction. Then the secondary over the primary in the opposite direction, with two 1.5mm diameter wires in parallel - eight turns -center tap- another eight turns in the same direction.

I connected my setup again - control circuitry with BD139-140 , switching mosfets IRF540 and transformer. Suply voltage 14V. Current from suply is about 0.2 A. Is this normal?
And after 10 min. transformer becomes markedly warm. Not hot, i can hold it easy with fingers, but warm. Must it be warm without any load, even without diodes and capacitors ?
I am sligtly worried about this, maybe i must rewound my transformer in different way?

Thanks for help.

[TwoSpoons]

Check your primary current waveform and make sure you're not saturating the core.

[N-Channel]

DT,

WOW! That's alot of ringing. I'm glad you're using the IRF540s, because at 100V(ds), you will need it. As for core saturation (presumably from MOSFET imbalance), as long as the MOSFETS came from the same production batch, you should be OK. This is indicative if the date codes on the individual MOSFETs match. Better yet, try hand-matching them. Set up a test rig, where you measure the V(gs), as the I(d) goes up. Measure these curves to within 0.5mA (500uA), and you should be good.

The SG2525 is the industrial version of the SG3525. If you're fortunate to get any SG1525s, you have the military version. These have storage temperatures ranging from -65 Centigrade to +150 Centigrade, and an operating temperatures ranging from -55 centigrade to +150 Centigrade. Great for those Baltic winters!

Anyway, you should not measure the MOSFET wavefroms with the outputs unloaded. SMPSs don't like to run without even a minimum load, because the very narrow pulses at light (or no) loads will produce alot of spike-like ringing. Loading the outputs with +/-100mA of current should do the trick. If this doesn't work, try playing with the RC snubber component values across the transformer's primaries.

Are you regulating the outputs, or are you running at max duty cycle (minus the deadtime)? Also, you could try putting a common-mode choke (wound on a powdered-iron toroid core) in line with the outputs for smoother filtering.

Please let me know, as I have made a Switcher like this, only with output regulation, and it works very well. And, if I could figure out how to compress the pic file small enough, I would post it here. The power supply I made comes from Audio Amateur magazine from the last issue of 1989 and the 1st issue of 1990 (it was a two-part article), out of Randy Vikan's 75W + 75W MOSFET car amp.

Best of luck,

Steve

[TwoSpoons]

Just occurred to me 1.5mm is too fat at 55kHz. Can't remember skin depth equation, but I'd say don't bother with bigger than 0.5mm dia. wire. Use about 12 x 0.5mm instead, woven or twisted, like Litz wire.

[DT]

very thanks to all for answering, i am just beginner in this stuff, so your help is very appreciable.

I will try to make another transformer with few thinner wires. But dont know how to check primary current waveform , and how to realize that core is saturating or not
To my this is a little painy, because i dont own an osciloskope, i can just borrow it. Maybe one can explain to me how to check saturation?

I even dont know the procedure of hand matching the MOSFETS ...

There is no SG1525 in our local stores, but 2525 i think must work pretty well too.
My SG2525 operates without regulation, exactly like here, just in the place of R3-R4, R5-R6 22Om i am using BD139-140 with 10Om and 2.7Om resistors to switch MOSFETS. With sthese small transistors there is less ringing in square waveform than with only SG2525 outputs.
I will try put some load on output and then measure all again.

Steve, to make the pic file smaller you can use any graphic program and resize it. Just in case of Windows Paint : Opent the file, then press CTRL+W (Menu Image-Stretch/Skew) then in the Stretch in horizontal and vertical positions you can change the value of 100% to smaller one. And after saving this you get smaller file.