Hey there!

So I have been working on a project that I think would interest a lot of us laser people (and other electronics hobbyists): a power supply that is constant-current and constant-voltage that is powered solely off of the 12V of an abandoned computer power supply (who can say that they don't have one of those?).

Anyway, I am almost done with the circuit design, but before I finish, I need some input from you guys: would you prefer a circuit where you have two separate settings, i.e. one is constant current, the other is constant voltage? Or would you prefer a single setting where it just balances it out according to the two knobs (like our current Mastech power supplies do)? The former is easier and I think it would be better, but it would make the circuit more complex and adds more to the user-end of the device (instead of just two knobs, it would be two knobs and a switch determining the state it's in).

Anyway, I'll keep you guys updated on this! Should be an interesting endeavor!

EDIT 1:

So I finished the board of the boost portion of the power supply. Anyone wanna check out the board arrangement to give me some feedback, if I arranged it properly, and so forth?

Here is the Eagle .brd file: http://dl.dropbox.com/u/50345936/12V...able%20psu.brd

And here is the Eagle .sch file: http://dl.dropbox.com/u/50345936/12V...able%20psu.sch

If you wanna check out the IC, it's the LTC3786 by Linear. Someone more experienced with board design would be more than welcome to help me out with this project!

EDIT 2:

So I finished up the schematic for the design - I went with a switch for CC or CV - I feel like it will provide more use that way. It *should* be adjustable from 0-24V and 0-5A, but that is still up for grabs, considering differences between real-life and simulations. My simulations *say* that it should be ~80-90% efficient (depending on the output), but I can't be certain what it will be like in real life.


EDIT 3:

Well, I finished the board of the CV circuit. Tomorrow I will do the CC circuit :P

Here are the files:
.brd: http://dl.dropbox.com/u/50345936/12V...able%20psu.brd
.sch: http://dl.dropbox.com/u/50345936/12V...able%20psu.sch

I think it's coming out alright... what do you guys think?

EDIT 4:

Alright, the board and schematic are done! Here they are:

.sch file: http://dl.dropbox.com/u/50345936/12V...able%20psu.sch
.brd file: http://dl.dropbox.com/u/50345936/12V...able%20psu.brd

What do you guys think? I don't have the money to order a board right now (and the next run won't be for a while anyway...), but once I get some money I will order the parts and board and test it out! It probably won't work right away (or ever - switching regulators hate me), but who knows?

EDIT 5:

The new, simpler version of the PSU is up!

.sch file: http://dl.dropbox.com/u/50345936/12v...%20psu%201.sch
.brd file: http://dl.dropbox.com/u/50345936/12v...%20psu%201.brd

This one is much simpler because it uses a SEPIC design for higher efficiency and buck-boost mode. What do you guys think about this one?

sounds fun... i also like the former

+1 bud for doing this because i couldnt ...

Hey, thanks. Working out some kinks on that version now in the simulation... so far, my CV part of the circuit is ideally getting ~90% efficiency, so probably ~80% in real life if I can get it working - MUCH better than the linear PSU. Now to finish up the CC!

will it be as stable as a linear psu? sorry if i sound stupid i kinda warned you already

As far as I can tell, it should be! Of course, there may be some ripple, but that's just a matter of larger output capacitors, which I can always adjust in size. However, my next step is designing the board... how should I do this?

ahahh hope your asking someone else... does eagle cost money?

edit: yes it does....idk what else to tell you

Haha. Not the free version of Eagle Anyway, that was a rhetorical question. Probably going to work on board design tonight.

So I finished the board of the boost portion of the power supply. Anyone wanna check out the board arrangement to give me some feedback, if I arranged it properly, and so forth?

Here is the Eagle .brd file: http://dl.dropbox.com/u/50345936/12V...able%20psu.brd

And here is the Eagle .sch file: http://dl.dropbox.com/u/50345936/12V...able%20psu.sch

If you wanna check out the IC, it's the LT3786 by Linear. Someone more experienced with board design would be more than welcome to help me out with this project!

i knew that i was totally just tricking you.....no i wasnt shhh

Looking good. Keep up the cool project. Looking forward to when its done.

+2

So I think the next step will be ordering or crafting the 30V board and seeing if THAT works. If that works all fine and dandy, the next step will be adding in the constant voltage part. Then, finally, the constant current part. Hopefully I will have a working prototype in a couple of months, but for now, I need to figure out a good way to etch such a precision board.... And I also need to figure out a way to get the parts for this cheaply (been spending a lot of money lately -.-).

CV and CC wouldn't be that hard to do. Just some resistors and pots. Only problem is going to be getting a pot capable of handling the wattage. I guess putting a resistor in series with the pot will help quite a bit.

Also, your gonna need some beefy traces on your PCB. Also, In my opinion, having the ICs connected to the PCB via wires will help. Since your gonna need quite a bit of heatsinking.

what i did for mine was i used 2 4W but they are like 5$ each

See, here's the thing - I am not doing a linear regulator. Those are remarkably inefficient and require huge wattage resistors. I have designed two separate switching regulators that use large resistance and low-value pots for cheapness and ease of sourcing for end-users.

But regarding the heatsinking... yeah. I know =[ In my board design, I have some 5mm wide traces. Haha. That's enough to handle 7A easily =p

EDIT: Well, I finished the board of the CV circuit. Tomorrow I will do the CC circuit :P

Here are the files: .brd: http://dl.dropbox.com/u/50345936/12V...able%20psu.brd
.sch: http://dl.dropbox.com/u/50345936/12V...able%20psu.sch

I think it's coming out alright... what do you guys think?

Alright, the board and schematic are done! Here they are:

.sch file: http://dl.dropbox.com/u/50345936/12V...able%20psu.sch
.brd file: http://dl.dropbox.com/u/50345936/12V...able%20psu.brd

What do you guys think? I don't have the money to order a board right now (and the next run won't be for a while anyway...), but once I get some money I will order the parts and board and test it out! It probably won't work right away (or ever - switching regulators hate me), but who knows?

Cool project! Let me know when the group buy begins.

Or even a board run would be great!

Well, FIRST I need to make sure it works. So that may be a while due to funds :\

i may be onboard for a group buy, depending on prices. i dont have much money for it . anyways, +1.

Well, one of the boards (from DorkBot) would probably run... I think $19? In bulk I could probably get it down. Parts... I haven't even sourced most of them yet (not planning on it until I know for sure that I can afford some of these parts, etc.).

Alright, so I found another great IC which may not involve boosting and then bucking... and it will do it all in one package! This one is a SEPIC converter, so it means buck-boost ^_^ I don't know MUCH about board layout specifics for SEPIC converters, but I do know how they work, and this one has been simulated to work and it should I will update this post later tomorrow or something when I stop feeling like a lemon wrapped around a brick is beating my brains out with the .sch and the .brd file.

Why don't you do what we do when designing a new project...

1) design the circuit on paper (or simulator)
2) build a prototype on a Breadboard to test the circuit design...
3) build a prototype on a perforated Proto PBC (Vero Board) to
test the device parameters....
4) Then and only then when the design has been tested and
proven to be within design specs... design the PCB...

BTW... I can't seem to find the data sheet for your LT3786...


Jerry

Due to the nature of the high switching frequency and the high currents flowing, a normal protoboard would be no good for such a situation. Also, they only come in SMD parts, unfortunately. I realize I could use a breakout board, but then I would have to custom make THAT board in order to make sure that it fits for the application. I'm not sure how much knowledge you have about designing high-power switching PSUs, but they are a nasty bugger get right, from what I can tell.

Also - the IC is actually LTC3786 - not LT3876. That was my bad. Typo.

To fix any additional output ripple problems, you could add a small filter in the form of an inductor and additional capacitor on the output....

I could probably give you a hand at designing this. I've made a buck converter before using an LT11070T IC from linear, and it worked perfectly. It was fully adjustable too! The chip itself is pretty awesome, with the correct heatsinking it can handle up to 5 amps! Plus, circuit design is also greatly simplified. It can do SEPIC, Cuk, boost, buck, buck boost, forward converter, and flyback converter mode.

It's a pretty versatile chip.

Hmm. I never knew that inductors are often used for filters o.o What kind of specs for that inductor are we looking at? I assume it would require full saturation current of 5A, but how much inductance?

And I would love your help on this! And actually - I have a new version to throw up there - it's quite a bit simpler. Lemme update the main post.

EDIT: So the first post is updated. What do you think about the layout of that SEPIC design? I've never done a SEPIC before, so I am not sure if I did it right and all that.

I can't look at the schematic just yet, I'm still at school so I apologize.


So anyway, yes, inductors are used very often for filtering. It doesn't need to be very high inductance; Maybe like 15-20 uH, on just a ferrite rod or a toroid. It gets rid of the higher frequency ripple. Higher capacitance doesn't always fix everything

Also, you might want to try perhaps a forward converter design. It's how most PSU supplies work (except for nicer ones...they use a half bridge inside)

The only problem with forward converters is that they require a transformer. BUT, this allows you to use rectified mains!

SEPIC converters allow for boosting and bucking, depending on which needs to be done. (the voltage is always constant, regardless of the input voltage) The only problem is the amount of components required to get it working. And, since you'd only be stepping down voltage, a SEPIC converter really doesn't make sense. There'd be no need to ever step up the voltage.

To help with the "large resistance" values for the feedback (or, shunt resistor) you just need some sort of error amplifier, just an op amp. It basically multiplies the voltage a certain amount so that 30 mV becomes 3 volts, or whatever you need. That way you can use much much much smaller resistance values. Then to adjust it, you just use a potentiometer to change it.

A buck converter would probably be the best way to go, or forward converter if you can afford to add a transformer to the board. Either way, the feedback methods are still roughly the same. Or if you wanna get really fancy, a resonant converter.

If you want, I can shoot you a PM and give you my email address so we can talk a little quicker than on the forums.


EDIT;

I just looked at the schematic... I can't tell how you've got it hooked up. xD

May I try and see if I can find something else that may work a little better?

Ah. See, the intent behind this initiative is that I want to provide a good PSU that can be powered off of a 12V line, because I know a lot of us have old computer PSUs which can easily supply 15+A on their 12V line, which is what this is based off of.

Anyway, the schematic is a little messy - I had to make all the symbols which didn't make for a nice, efficient schematic, but it's all there. I can throw up the LTSpice simulation file too if you want. But, the schematic I used for this design is nearly the same as the SEPIC schematic found in the LT3756 datasheet - I just calculated values, simulated it, and added some features for switches and what not.

The reason I went SEPIC though is because I couldn't find a nice, high-powered buck-boost that could provide 5A at 24V from 12Vin :\

Anyway, later tonight I will probably throw a coil for reducing ripple ^_^