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.
I don't know yet, my transformers and MOSFETs aren't here yet. :)

I have a board layout done, but I haven't etched it yet. When I have results, I'll post 'em up.

Note:

I have a lot of respect for Rod and his designs. Chances are that I will submit my artwork for the boards to him, which he may choose to modify or use as-is to possibly produce boards for this design. If he does so, then I will not post artwork.

Personally, though I don't think that he will choose to make a board available for this SMPS. There is far too much that can go wrong in construction, and if it does, I don't think that he wants to have any responsibility for the failure of the SMPS. I could be wrong, though. :)

I've got a bunch of Rod's boards, sitting on my desk right now, and the design and layout are top notch. I'm going to try to do my own PCB for a couple of these designs, too.
 
soundNERD said:
I want to build this project:

http://sound.westhost.com/project89.htm

but PCBs aren't avaliable.

Has anybody made this project and has a good, working PCB layout?

Thanks!

National Semi, On Semi, Infineon (and probably TI and Maxim) have board outlines on their websites -- for their chips -- these have been "vetted" so you have a modicum of insurance that they are going to work --

take an outline off the chip manufacturer's website, download the ExpressPCB freeware and have boards made for some tens of dollars.

I believe that Infineon has a board outline for one of their highpower switcher development kits. Check their site.
 
Disabled Account
Joined 2004
I have this that may be interseting for you.
The upc494 is a very common chip used for pwm power supply in car amplifiers.
This power supply drive a very powerfull power amp (500w)

Pat Allen
 

Attachments

  • power supply.jpg
    power supply.jpg
    71.1 KB · Views: 1,556
soundNERD:

There's some problem with the schematic: push-pull topology in SMPS is vulnerable of transformer core saturation .:magnet:

The MOSFET and the switching transformer has very low resistance, so a slight unsymmetry on duty cycle or on resistance could cause large DC current passing through the transformer and saturate it. Then high current will flow and.....:hot:

It's a classical problem in SMPS papers, and somebody even used DSP (!) to control the bridge to overcome it.

Perhaps using other topology like half bridge (with a big capacitor series with the transformer primary ) or use some current-mode PWM controller to control the bridge and ensure the current symmetry will be safer.

Another simple choice is : use a charge pump made by capacitors, diods and a half bridge switch(something that's used in class D) to boost the voltage up. We used it on our robot to get 24V from 12V and it's reliable. :)

Use more capacitors and diods can get a higher voltage,but the output voltage is fixed to several times of the input voltage and can't be adjusted continuously.
 
Disabled Account
Joined 2004
that schematic is a commercial power amp, its been used since many years in many car amps and i never saw one of those with problems you related. This almost same exact schematic is used also in the higher series of Alpine car audio amp, they were still using the upc494 int the mid 90's. Its an OLD schematic, shown just as a guide.
 
Hello all.
Glad to see that there is still some interest in the Project 89 at ESP. I am the writer of that article, with the unvaluable help of Rod.

A long time ago I started with that design, that has proven to be very reliable. I have developed an improved design for a car SMPS, and plan to sell the finished boards in a short time, as long as some other stuff that I think can be interesting for audio entusiasts, like Class-D amplifiers and also off-line SMPS for them.
You are invited to have a look at:

www.coldamp.com

The webpage has only minimal information by the moment, as it is on construction, but please be patient, it will be filled with lots of info soon.

Best regards,
Sergio
 
Hey i brought an alpine amp that has a pwm powersuply in it similar to this one

i brought it pretty cheap and found it was broken and never took it back cause it was that cheap i opened it up and the powerfets had been major cooked

they are irfz44 about a 50amp fet that switches ground just like in the diagram

with the transformer what kind of load should it put on the fets
cause mine is putting mega load on the fets and practically a short on the fets

sorry to rabble on a bit

what kind of resistance should be on the 12v side of the transformer

sorry for my stupid question.. im only 16
thanks
 
Not a stupid question, congratulations if you are only 16 and are involved in these things, then the only thing I can say is: well done, keep working on it!

You wont be able to measure the load that the transformer puts on the fets, as with a multimeter you only see DC, and the primary of the transformer is almost a shortcircuit from the fets to +12V.
The load the transformer puts in the fets effectively at the working frequency depends on the load of the power supply, that is, the current the amp is demanding. To put it simple, if the amp is demanding 1A rms, the primary current will be that multiplied by the turns ratio of the transformer (for example, about 3 if the supply is +/-36V).

If your mosfets have exploded there are some simple possibilities:

a) There is a shortcircuit in the secondary: try to disconnec the amplifier from the power supply (if there are coils you can simply desolder both of them), and see if there is shortcircuit. Sometimes the capacitors fail to shortcircuit, what cause huge primary currents and even fet destruction.
b) There is a problem in the driving of the mosfets, that cause its spurious destruction, although not likely in a commercial product.
c) A problem in the input supply (like someone has connected polarity backwards or 220V or whatever).

Hope this helps diagnose your problem.

Best regards,

Sergio
 
I'm right there with you man, only 16, got me some good sounds subs, but just blew my amp and am out of money for a new one that can push 500x1 at 1 ohm, so looking to build one myself...

I have been all over, reading everything I can get my hands on. I have really been interested in the smps project at esp and have all of the major parts to build one from my blown amp...

just a few questions, toroid or e-core, i have a very large toriod, at least 2 in. in diamter, but i also have an EDT49 core.
also looking at the mosfets, going with 4 irf540 per rail
got good capacitors out the wazoo cause my blown amp was filled with low esr 105 degree rated capacitor for the supply

ready to build, but i should i make the pwm circuit on a seperate vero board and transfer the signal to the mosfets by wire, or have the mosfets on the board too, keep in mind that this supply will be pulling about 30A at 12V per rail...
Thanks
 
For the core, make sure it is ferrite (not iron powder or similar).

About making the modulator in a separate board, keep in mind that the gate connections should be as short as possible. If that's completely impossible, use a pair of twisted wires to gate and source of the mosfets to minimize inductance of the loop, that could cause ringing and undesired turn-on of any mosfet just when it can do more harm.

Best regards and have luck!

Sergio
 
By the way, IRF540 is a very good choice. Easy to get and very robust. If you can, use IRF540N variation, it is even better and have lower Rds(on), that is important:

IRF540: Rds(on)=77mohm. For 30A, if your use 4 per primary side, you have 77/4=19.25mohm. At 30A, P dissipated due to resistive losses = I^2*R=17W aprox.
IRF540N: Rds(on)=44mohm. P dissipated is now 9.9W in the same conditions.

BTW: Try to use International Rectifier mosfets, they are rated for avalanche, etc. Are much more reliable than other brands (and I don't have relationship with them :))

Best regards,
Sergio
 
If it's real 500W RMS amp then current consumed from battery can be much higher, over 50A. IRF540 is 100V device and not quite suitable to operate from 12V battery, the usual and cheap FET choice for 12V car amp SMPS is IRFZ44N ->55V, Rds(on)=16mohm - about 3 times lower losses. Slightly better and more expensive are IRFZ46/48N. Even better are IRF1010 or IRF3205 (the last you will need only 2 pairs).
Toroid core is preferrable to ETD at input voltages as low as car battery. It's been mentioned a lot about it here.
Your 2-inch toroid core is sufficient for such SMPS. Use 25-30kHz switching and 4 turns each half primary winding for good efficiency.
 
Alme is right, although IRF540(N) has the advantage of being readily available and cheap.
However, if you need the ultimate in efficiency, use for example the ones indicated in my article in ESP: Onsemi's NTP75N06, they feature Rds(on)=9 mohms, so with 1 or 2 of them per phase you can be ok. Besides, they worked very reliably for me.

About the toroid, if it is ferrite (that's very important), the size is ok in principle.
ETD39 is also suitable and easy to wind.

About frequency, the lower you go, the less switching losses you will have in the mosfets (they can be nearly as important as Rds(on) losses). Although the size of the transformer required will be bigger too. But I agree in that 30-60KHz is a good start.

Best regards,
sergio
 
SoundNERD-

Have you tried the Onsemi SG3525 or the MC33025? Both are available as samples from ONsemi.

The SG3525 PWM IC has totem-pole outputs (makes driving the '75N06s much easier), and the MC33025 is the current-mode version of the '3525.

The MC33025 has a very high speed PWM and error amplifier, and with current-mode operation, it all but eliminates core saturation in c.t. push-pull topologies mentioned a few posts ago. Its totem-pole outputs will drive up to +/-2A, good for MOSFETS with really big gate capacitances. :smash:

Best of luck,

Steve
 
SmarmyDog, appears that you use SG3525 at about highest possible frequency of 200kHz (400kHz oscillator). I can imagine that maybe PNP pull-down helps to discharge gates at such frequency, with only 5 ohm in series. In my practice with this chip ut to 100kHz operation, no significant improvement was noticed with changing gate resistors 10 to 68 ohm, with up to 4 pcs. FETs driven from each output (usually IRFZ44N type). But, I noticed improvement at once after decreasing dead-time, certainly, not too critically to keep preventing shoot-thru. If used more than 4 pairs to drive (yes there were such cases), then I had added npn-pnp follower to each output and shorted 'discharge' resistor to have minimum dead-time.
Anyway from my experience, transformer design influences more on total efficiency than those gate things - especially in PWM topology with DC choke between diode bridge and output capacitors.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.