12v 250w Car SMPS based off SG3525

[codex653]

Hey all! I'm wanting to build this car smps (see link below), but it's not going to be use in a car. In fact, it's probably just going to end up powering an audio amp I built. Mostly though, the entire point of this project is for me to learn more about some higher powered switching supplies building techniques (while at a relatively "safer"-) voltage level) before going on and building one that can be plugged directly into the wall.


This is the site where I am getting the schematic from: http://sound.westhost.com/project89.htm
And these are some quick links to the two schematics i'm looking at:

http://sound.westhost.com/p89-f9.gif
http://sound.westhost.com/p89-f1.gif

I prefer to combine the two so that there is the output voltage adjust circuit as well as leaving out the Q1-Q4 driver transistors as they (according to the site) are apparently not essential because the SG3525 can source/sink enough current to drive the mosfets.


Now here is where i'm running into some problems. I have been breadboarding the driver circuit from Figure 9 for two days now trying to make it work and it just won't budge! Here are some of the changes I made. Keep in mind that when I finish this thing, all the part values will be correct. Just for initial breadboarding purposes are some of the components changed. I am powering this right now through a variable bench supply @15vdc.

-IC is a UC3525AN...according to the datasheets, there is no difference between the SG and UC version besides who manufactures it, but it was worth mentioning just cause i didn't use the "exact" IC
-R7,9,10: They are all 15K instead of 12k..I couldn't find for the life of me any 12k's in my parts bin.
-R8: 25 ohm...I know it's lower but it will still put the dead time on the output
-C11: 47uF but that shouldn't matter as it's a start up timing capacitor
-I didn't want the remote turn on/off feature so i removed R11, R12, C18, Q1, and grounded the shutdown pin.



I double checked every single connection with my ohm meter when I placed the parts to make sure that things were connected correctly. Everything seemed perfect but when I applied power, no output showed on the oscilloscope for either out A or out B! I checked all my connections once again and yep...still nothing on the output. So i tried removing the ground connection on pin 10...same result. I tried a 15k pull up resistor to VCC on pin 10 and still no pulsing output. Then I thought it might be the chip that was broken so I replaced that too and tried the same things with pin 10 as before. Same result. ::SD I know this thing works as Rod Elliot does NOT post circuits on his website that he has not already tested himself. Does anybody have some suggestions for me?

Thanks,
Codex653

 

[codex653]

if anyone would mind helping that would be wonderful

 

[McMax]

Hi Codex,

this is not easy to analyze. From my point of view you MUST add the external drivers as done in the "f1" schematics: according to the 3525 data sheet the absolute maximum source/sink current on the driver section is 500mA but with 2x22ohm gate resistor (resulting in a 11ohm resistor) and 15V you are sourcing/sinking a peak of more than 1A (15V/11ohm) which is more the double the "recommended operating conditions (400mA source/sink).
Another small thing: in the same "f1" schematic you would need an anti-parallel diode on the BD139 driving the relay coil...

 

[codex653]

Darn...I was hoping to not have to use Q1-Q4. Oh well, it's only 4 more parts. As for the relay coil, I'm not wanting to have any sort of remote turn on/off thing at all. It serves no purpose for what I would use the power supply for and it's extra parts I don't need.

Hey also for anyone reading this, if you happen to have a MULTISIM model for the SG3525 it would make it alot simpler to figure out what is going on here
Until then, does anyone else have ideas of what could be going on??

 

[e1000]

If you have 5v on the ref pin (16) of the 3525 it is working, check if you have frequency from +12 to the gate of the mosfets, if there is 30k or so you probably have damaged or false transistors. Check this and tell me whats going on.

Saludos!

 

[codex653]

If you have 5v on the ref pin (16) of the 3525 it is working, check if you have frequency from +12 to the gate of the mosfets, if there is 30k or so you probably have damaged or false transistors. Check this and tell me whats going on.

Ahhh ok I will check that then when I have the time. I have some baseball games tonight and tomorrow that are going to go VERY late, so I might not get the chance to look at it until wednesday.

As for the mosfets, I do not have them hooked up yet. I have only breadboarded the control circuitry to make sure it would work before I tried connecting the rest of the circuit. As of right now there is no output on either out A or out B.

 

[giololucas]

Ahhh ok I will check that then when I have the time. I have some baseball games tonight and tomorrow that are going to go VERY late, so I might not get the chance to look at it until wednesday.

As for the mosfets, I do not have them hooked up yet. I have only breadboarded the control circuitry to make sure it would work before I tried connecting the rest of the circuit. As of right now there is no output on either out A or out B.

This site you mentioned it´s very good (rod elliot).
I built an amplifier with ideias of this one. If you want to check it, search for JBL based amplifier on this blog.
More than 4 years working with a 15 inch subwoofer on heavy load, and never failed.

Best regards.

 

[codex653]

I found time this morning before classes to check the breadboard! I guess it helps to take a second look at a project after sleeping on it, cause I fixed it! Now I have both outputs pulsing, it's time to start tweaking the circuit to get the duty cycle and frequency correct! I'll see if i can figure out how to post images of the waveforms from my scope...I'm not sure how to do that yet

Got a question for you all. The transformer I am planning on using came from a generic 300W ATX computer power supply, nothing fancy at all. What typical frequency do those transformers run at? Yeah I know that depends on the type of ferrite, power level, core size and a whole bunch of other factors but I'm just wondering so I can have a base frequency I can start testing my transformer around. What do you think? 30Khz? 50Khz?? maybe even 80Khz?

 

[Redwire]

I find these transformers are run from 40-70kHz. You can go to 100kHz but the losses go up drastically in the core, the windings (skin-effect unless you use Litz), the mosfet switching and lastly your secondary rectifier diodes. It's a case of speed kills.
30kHz-50kHz is a good start. If you are using the 4-transistor mosfet driver scheme, you can push it past 75kHz, but not with slower (non-Schottky) rectifier diodes. My $0.02

 

[codex653]

I find these transformers are run from 40-70kHz. You can go to 100kHz but the losses go up drastically in the core, the windings (skin-effect unless you use Litz), the mosfet switching and lastly your secondary rectifier diodes. It's a case of speed kills.
30kHz-50kHz is a good start. If you are using the 4-transistor mosfet driver scheme, you can push it past 75kHz, but not with slower (non-Schottky) rectifier diodes. My $0.02

Ok thanks Redwire! I have my breadboarded circuit at 32kHz right now, so I'll change around a few components to try and get close to 55kHz when I get the chance. In the mean time, I took a few pictures of the waveform from the circuit. I'll post em in just a min here.

 

[codex653]

Here are the images

The first one should be the waveform. It is 32kHz with 48.6% duty cycle. Ignore the "double" lines on the waveform. I had my scope on the "fit" setting so the two lines you see are just it flickering and the camera caught it. I'm still trying to figure out how to get the images directly from the scope to the computer so it will be a lot cleaner than this.

The second picture is of my initial breadboarding

 

[McMax]

you need to increase the caps and don't forget that you're running out of specs with the gate current!

 

[codex653]

you need to increase the caps and don't forget that you're running out of specs with the gate current!

i need to increase the caps on what?? I'm not sure I understand you

 

[codex653]

Switching Problems?

Ok so I decided to try switching some Mosfets through a direct drive from the SG3525 outputs. The load resistor was 15k and the mosfet was on the low side. You'll see I uploaded 3 pictures of some waveforms I saw on my scope with a couple of different types of mosfets. Keep in mind that each time I checked the Gate drive waveform for each mosfet and it did not degrade at all. I'll detail some things for each picture.

IRF9640: A bit different as it is a P-channel mosfet compared to the typical N-channel. The Drain is connected to ground, and there is a 15k resistor in series between the +15v rail and the Source pin. The Gate is being directly driven from the SG3525's "A" Output. I think it worked quite well except for one small thing. If you look closely at the picture, you will notice that the mosfet (measure on the Source pin) is switching between 3v and 15v...not 0v and 15v. I wonder why it is doing that?? I can't be having my mosfets doing that, other wise there will always be 3vdc going through the transformer windings :":: Any way to fix this?

Datasheet IRF9640:
http://www.datasheetcatalog.com/datasheets_pdf/I/R/F/9/IRF9640.shtml

F1010E: This one is an N-channel Mosfet. The Source is connected to ground and the Drain is connected in series with the 15k resistor to the 15v supply rail. Waveform is measured on the Drain pin. Now what I don't get is why it is switching it only between 0v and 5v. Is there not enough current to turn on the mosfet hard so it is partially turning on, then turning off? This mosfet is pretty beefy unlike the IRF9640...it can handle somewhere around 80A and 64v, perfect for this switching supply i'm building! I just need to know how to get that switching waveform correct:/

Datasheet F101E:
http://www.datasheetcatalog.com/datasheets_pdf/I/R/F/1/IRF1010E.shtml

07N03: This is also an N-channel mosfet and is set up exactly the same as the F1010E. You'll notice that this waveform is switching from 0v to 10v, which is better, but nowhere near good enough as the thing looks like a crap triangle wave! Again this mosfet would work well for the switching supply as it is also rated for about 80A.

Datasheet 07N03:
http://www.datasheetcatalog.com/datasheets_pdf/I/P/B/0/IPB07N03L.shtml

So really if at all possible, I would love to be able to use one of the latter two mosfets because they have such a high rated current, but I don't know how to make them switch well yet. Would anyone like to educate me? -)

 

[e1000]

Hi codex, with the f1010 you need to drive the gate with a 100 ohms resistor and try lower values until you get the waveform right, you'll have the source right from the ground and thats all. Look at the schematics you posted first, then you have to measure the waveform from the drain pin to the +15 pin, hope this help.

Saludos.

 

[codex653]

Hi codex, with the f1010 you need to drive the gate with a 100 ohms resistor and try lower values until you get the waveform right, you'll have the source right from the ground and thats all. Look at the schematics you posted first, then you have to measure the waveform from the drain pin to the +15 pin, hope this help.

Saludos.

Ok i'll try out some various gate resistors then. Yeah I already am measuring from the drain pin. Thanks!

 

[McMax]

i need to increase the caps on what?? I'm not sure I understand you

You're right sorry, I've been too fast in replying....
Maybe I'm wrong but looking at the waveforms you have posted, it looks like you have the too low (or far from the IC) capacitor for the driver section. If you look at the schematic you posted (f9) there is the capacitor C17 (100nF ceramic) which is needed to be mounted very close to pin13 (VC) and is used to supply the gate charge. I suggest you to use a multilayer X7R ceramic cap 100-220nF mounted as close as possible to pin 13. It looks like you didn't used this cap on your breadboard....

 

[codex653]

You're right sorry, I've been too fast in replying....
Maybe I'm wrong but looking at the waveforms you have posted, it looks like you have the too low (or far from the IC) capacitor for the driver section. If you look at the schematic you posted (f9) there is the capacitor C17 (100nF ceramic) which is needed to be mounted very close to pin13 (VC) and is used to supply the gate charge. I suggest you to use a multilayer X7R ceramic cap 100-220nF mounted as close as possible to pin 13. It looks like you didn't used this cap on your breadboard....

Ok i'll go ahead and add that real quick! I'm not so sure i have that type of cap you mentioned, but i'll see what i can find. i'm at my breadboard right now so you should be getting some updates pretty frequently on my tests for today.

 

[codex653]

ok, I've just put in a 470nF polyester cap (don't have any ceramics that were just laying around) and as far as I can tell, it made no difference to the waveform. Although the only mosfet i have tried this on has been the IRF9640...gonna try another mosfet in just a second.

 

[codex653]

Same problem with the 07N03 mosfet...I forgot to mention that I also used a 10 ohm Gate resistor for both of them, but that did nothing to improve the waveforms. Time to try and build a higher current capable totem pole