| demogorgon |
Hey guys.
I'v caught myself sitting on my *** for to long, doing nothing.
for some time now i'v been loosing faith in the power waisting, overkill side of audio engineering, like gigant power burning class A amps, massive oversized transformer powersuplies and such. My interest has shifted to the more efficient, smart, but more complicated side of audio, like class D amps, dacs and smtp's.
as you can undoubtly guess i'm needing the final kick-in-*** to get moving into this field of DIY. i'v been googeling around, and looking in the forum for some simple ways to get started, but havent found the final extra to get me going, to re.
i was hoping some of you may have some links or such for me so i can start to explore this side of the pond.
my first project on this i was hoping to make a small 2-3 amp smps fron 230 to 12v, hoping to gain the final bit of overall understanding that i lack. i get the basics of trafo efficency rising proportional with frequency, up to the point of non-practicality, i altso have some understanding of general electronics, having buildt a gainclone or two, playing with tubes a bit, and designed and buildt a power\pre system microcontrolled as such. (including my "level of understanding so you hopefully will se better what i'm looking for.)
i would altso like to make use of scrap components as far as i can, having access to tons of such through school and work, and very limited monetary recources. i have access to lots of computer psu's, and smtp's from a lot of other appliences in general.
thanks
-Marius |
|
|
| richie00boy |
| Start with a 12V powered SMPS. A mains driven one is too dangerous for a beginner - it has over 300V DC present for a start. Look at the ESP website, theres a suitable project on there. |
|
|
| demogorgon |
| richie00boy, i'm constantly fixing mains powers at work and school, so i dont see that as a valid argument, and besides, i would like to make something that i can use :) i dont belive there is any mains powered ones at ESP? |
|
|
| richie00boy |
Fixing mains powered stuff and designing stuff for mains power use is a completely different ball game. The ESP site has a 12V design, as I said I wouldn't recommend a mains powered one. Cut your teeth on something safe first :)
If you insist on offline converter, I contributed to a thread on here sometime in the past 2-6 weeks. You can find out how to take the philosophy of adapting the 12V powered ones to mains in there. |
|
|
| demogorgon |
Totally different ballgame, yes, but still alike in the respect that the ball remains. a 12v smps just isn't much to cut my teath on, and besides, i dont have the parts required for it.
i'm not an idiot, and i take my precautions, so dont worry about that. :)
thanks for caring though.
i looked for the post you talked about, but i dont think i found it, if i did then it wasn't what i needed.
I'm not looking for the philosofy, i'm looking for the rock hard and real to put phiposofy into practical use.
still be glad for help :)
-Marius |
|
|
| Eva |
Well... I will give you some ideas.
How about a half bridge? MOSFET or IGBT?
Voltage control (easy), peak current control (medium) or average current control (complex)?
Single output (easy), dual symmetric outputs for audio (medium) or several outputs (complex)?
Slow switching (30 to 40Khz) and low inherent EMI at the expense of bigger output capacitors and magnetics? or fast switching above 100Khz with small components at the expense of high EMI and high switch dissipation?
Do you feel like messing with an active power factor correction stage?
Also, do you have a mains isolation transformer, a variac and an oscilloscope?
|
|
|
| richie00boy |
| quote: | Originally posted by demogorgon
a 12v smps just isn't much to cut my teath on, and besides, i dont have the parts required for it. |
So wrong. It's posts like that that make me say to people trying what you are, to take a step back. You make a 12V and mains SMPS in pretty much the exact same way.
| quote: | Originally posted by demogorgon
i looked for the post you talked about, but i dont think i found it, if i did then it wasn't what i needed.
I'm not looking for the philosofy, i'm looking for the rock hard and real to put phiposofy into practical use. |
I did not put the philosophy in the thread, I put exactly what an engineer would need to do it. I suggest you keep looking, or start like I said with something that won't be dangerous. I'd like to build a fast car engine because that's what I want and need, but that doesn't mean I won't build some smaller ones to hone my skills first - it's not wasting time or parts, it's part of the necessary learning process to make sure that the big one doesn't go up in smoke and blow me up in the process. |
|
|
| Eva |
| Off-line SMPS are not always operated from mains voltages during development. For example, sometimes (mostly when I'm trying new things) I use only 24V AC or 60V AC from a standard transformer. |
|
|
| ingrast |
Some years back I built a simple +/- 25V unregulated SMPS still happily humming along.
Being unregulated, it gets as simple as can be to debug and build, and being off line I used regular computer PSU rectifier, electrolytics and transformer. To keep things simple, I used a small inexpensive transformer to power the drive electronics (a quad gate, flip-flop and some transistors) instead of going for self-powered auto starting circuitry. Yet it even has a simple, capable overload protection.
If interested, I can post details, schematic and artwork.
Though I did not have isolation transformer or variac for developement, I should recommend them as safety precautions if possible. Be ready to burn some MOSFETS in the process.
Rodolfo |
|
|
| demogorgon |
thanks for the replys! i was hoping you'd stop by Eva. what i'v seen from your post's here you seem to be vastly knowledgable on this and other subjects.
Richie, i take it you have some experience in this field as well?
I'l check out tomorrow what the IC in elliots project cost.
if it aint to much, i'l grab one, and start from scratch, without any execsive voltages. :)
salvages a humongous smps from a canon copymachine today, and there's 6 more of them in storage, waiting to be thrown on the heap.
it looks way good to be thrown away :)
on a slightly other note, what considerations need to be taken when designing a switchmode from scratch?
I have variac, scope and isotrans at school, but if i ask really nice... they have equipment for 15 students in the sound and picture department where i am, but since everyone is choosing car electronics theese days, we'r only 3 in my class. lots of stuff extra.
Gonna read up a bit, by the looks of it, it's just what i'm looking for :)
http://www.smpstech.com/ |
|
|
| demogorgon |
| quote: | Originally posted by ingrast
Some years back I built a simple +/- 25V unregulated SMPS still happily humming along.
Being unregulated, it gets as simple as can be to debug and build, and being off line I used regular computer PSU rectifier, electrolytics and transformer. To keep things simple, I used a small inexpensive transformer to power the drive electronics (a quad gate, flip-flop and some transistors) instead of going for self-powered auto starting circuitry. Yet it even has a simple, capable overload protection.
If interested, I can post details, schematic and artwork.
Though I did not have isolation transformer or variac for developement, I should recommend them as safety precautions if possible. Be ready to burn some MOSFETS in the process.
Rodolfo |
Hi Rodolfo, you caught me typing :)
I'l gladly take the schematics, since you offer them :)
reccon i need all the input i can get, setting of to do the job of an educated engineer :cannotbe:
-Marius |
|
|
| richie00boy |
I have dabbled a bit in very simple SMPS.
The ESP project uses SG3525 which is a nice chip and cheap, but a lot of people use TL494 which seems even cheaper and easier to find. I suggest you start reading datasheets. |
|
|
| ingrast |
| quote: | Originally posted by demogorgon
...I'l gladly take the schematics, since you offer them :)... |
Here is the schematics. It is basically a square wave generator with dead time provision and a half bridge switch. Current sense on the supply rails provide device overload protection.
Rodolfo |
|
|
| ingrast |
And here the PCB. If you are still interested I will gladly provide further details.
As said, it works nicely though unregulated, and has been easy to debug and put to work. For audio work at least, it is just fine.
Rodolfo |
|
|
| demogorgon |
| quote: | Originally posted by richie00boy
I have dabbled a bit in very simple SMPS.
The ESP project uses SG3525 which is a nice chip and cheap, but a lot of people use TL494 which seems even cheaper and easier to find. I suggest you start reading datasheets. |
Tl494 rings a bell, think i have one left over from repairing a car amp.. on to the datasheets.
| quote: | Originally posted by mzzj
[snip..]
...that should be enuff links for a while. :) |
thanks :cool:
| quote: | Originally posted by ingrast
Here is the schematics. It is basically a square wave generator with dead time provision and a half bridge switch. Current sense on the supply rails provide device overload protection.
Rodolfo |
Thanks.
I'l be studying it. |
|
|
| Eva |
ingrast:
I like the way in which you have overcome the use of a control IC with some CMOS gates (even a gate oscillator :D )
There are a couple of improvements that I would do to that circuit. One is to place a 1uF non-polar capacitor in series with the primary of the gate drive transformer, as it's a better way to solve any potential saturation issue due to the drive signals being never exactly symmetrical. Currently, the transformer does not saturate only because severe flux imbalance causes excess current to flow in one direction and asymmetric voltage drop across R25, thus preventing deeper saturation. The capacitor will ensure almost perfect flux balance.
The other improvement is to add active gate turn-off to the secondary sides of the drive transformer instead of relying on primary controlled turn-off. The following figure shows a couple of ways to do it:
The simplest approach includes just R5, Q1, D2 and Q_IC_1 should be replaced by a diode.
Finally, that circuit relies on C4 and C18 for proper flux balancing of the power transformer. This is fine for an unregulated converter that always employs the same duty cycle, however, shall voltage mode regulation be employed, the dynamic duty cycle fluctuations with its related asymmetries would require a non-polar capacitor with a smaller time constant (1 to 10uF) in series with the primary. |
|
|
| ingrast |
Eva, Thanks for your interest.
| quote: | Originally posted by Eva
....One is to place a 1uF non-polar capacitor in series with the primary of the gate drive transformer, as it's a better way to solve any potential saturation issue due to the drive signals being never exactly symmetrical. Currently, the transformer does not saturate only because severe flux imbalance causes excess current to flow in one direction and asymmetric voltage drop across R25, thus preventing deeper saturation. The capacitor will ensure almost perfect flux balance.... |
True, in fact R25 does not show asymetrical drop, but I put it there as a malfunction safeguard. Though it is true the driving waveform cannot be perfectly symetrical, I divided the clock frequency by two to ensure that in a cycle by cycle basis it is almost nearly so. The little dead-time window provided by R60-C28 and R2-C1 help in allowing any residual flux to decay.
| quote: |
...The other improvement is to add active gate turn-off to the secondary sides of the drive transformer instead of relying on primary controlled turn-off. .....
Finally, that circuit relies on C4 and C18 for proper flux balancing of the power transformer. .... |
Agree also. Interestingly the converter runs (I don't recall exactly now) at about 100 KHz or more, yet swithching losses are extremelly moderate. When I built it I left it on 24 hs. with a 300W load and though it went hot, it was bearable to touch (no forced air cooling). In fact, the output rectifier was far hotter than the switching stage.
Rodolfo |
|
|
| demogorgon |
| Any eventual progress to this thread is held on ice, as i'm going to portugal for a couple of weeks on saturday. i'l be getting back to both norway and smps building after that however. thanks for the help so far :) appreciate it. |
|
|
| demogorgon |
I'm not nesecarily very sorry to be leaving, but it'l be good to get some time off from school. no ice down south last year at this time for shure :)
I'l take at look at the site. thanks.
at the time i'm busy goint through the smps tutorial here:
http://www.smpstech.com/tutorial/t00con.htm |
|
|
| Curmudgeon |
| I'm surprised not to see this:Linear Technology App Notes or this LT Design Aids; Some of the LT app notes are classics. And although it's been a while since I used the tools, they worked very well. |
|
|
| demogorgon |
hey guys.
just letting you who helped me getting started know that i'm on track and i'm starting to play around with practical designs today. pretty basic stuff, but i'm aiming low for now, and then i'l see what comes later. i'l even bring you a pic or two ;)
-Marius |
|
|
| N-Channel |
| eagerly waiting........... |
|
|
| demogorgon |
hehe, dont hold your breath ;)
i'm testing out a smps in it's simplest form, using an external signal generator and a low voltage source. just to get the feel for it.
i'd like to ask someone of the knowledgeble here the following:
what is the deffinition of a halfbridge and fullbridge smps?
and when using an IC for switching controll, how about voltage suply?
is a dropdown resistor used? are there high voltage IC's available?
so many questions... :)
thanks
-Marius |
|
|
| richie00boy |
| Half bridge is like a single ended audio amp, full bridge is like a bridged audio amp. |
|
|
| demogorgon |
| quote: | Originally posted by richie00boy
Half bridge is like a single ended audio amp, full bridge is like a bridged audio amp. |
ok, i suppose i get it.
halfbridged it is then. |
|
|
| demogorgon |
and wheres the protoboard when you need it?
nowhere to be found.
well, at least i got the oscilloscope and signal generator in house.
and i'l se what can be done on protoboards tomorrow.
-Marius |
|
|
| demogorgon |
got the demoboard, the oscilloscope set up, the sig-gen wired up and the time on my hands to play around a litle.
thinking i'm gonna go for the simplest possible first, anything you see thats terribly wrong with this simplest of circuts?
thanks
marius |
|
|
| richie00boy |
| You would usually use an n-channel device and connect the load to the drain and 0V to source. |
|
|
| demogorgon |
hehe, thanks richie, i'm gonna do that. :)
i'm using 2sk1217 for switching, but i'm uncertain as to if it's a depletion or enhancement piece. gonna try it out tomorrow, se if it runs. had to go to an extra meeting at work, just home to eat and shower, then i'm back again.
-Marius |
|
|
| richie00boy |
If it's a TO220 case power device (which is at least what you should be using in an SMPS) it will be depletion. Usually only JFETs are enhancement.
Are you using a 12 volt supply for everything? |
|
|
| Onra |
Hi,
@richie00boy:
It´s the other way around:
2SK1217 is an enhancement type MOSFET and most JFET are depleteion types |
|
|
| Eva |
| Marius, I'm sorry to say that but your circuit is not likely to work. Rethink it. What are you expecting to get in the output? What waveforms are you expecting to see in each node? |
|
|
| demogorgon |
small updated schematic, this time, with a rectifier bridge.
i reccon thats what you reacted on eva?
if not, then i'm at loss as to what's missing or wrong:xeye:
Eva, Richie00boy and Onra, thanks for helping me out :)
appreciate it.
-Marius |
|
|
| richie00boy |
That is better, I don't know how I missed you had no recitifer before!
That kind of circuit can work (I used something very similar for my first design) but it is very crude. |
|
|
| Eva |
Your secondary side looks right now, however the transformer will saturate because the secondary side inductor will short it through the diodes during the off-time, thus preventing it from resetting.
Try to find a way to drive the primary in a symmetrical way, for example in push-pull. |
|
|
| richie00boy |
| When I built that kind of thing I didn't use an inductor on the secondary, I guess that would help against the saturation problem. It is a crude way though and a centre tapped primary would be better. |
|
|
| demogorgon |
ok, the circut is up and powered, though only wierd resoults have arisen so far, and no squarewave is yet connected. gonna troubleshoot a bit.
i altso have taken a few pics for your viewing pleasure :)
the switching frequency is 25kHz, Vp2p is 2v, the load resistor is 1k.
anything anyone would like to add to this?
i'l keep you posted.
-Marius |
|
|
| demogorgon |
ok, connected the source and drain wrong, nothing more.
now, i altso connected the squarewave, and more wierd resoults emerged. never the less, here another picture ;) |
|
|
| demogorgon |
after a bit of investigation i would draw the conclution that my fet isn't switching. i sort of fixed it now, by refering my squarewave to gnd through a 20kohm resistor, but the readouts still puzzle me, the ones on my secondary side that is. i read squarewaves on both sides of the diodebridge, over the load rsistor. i figure i need to refere the secondary zero volt to gnd altso? gonna try.
altso you spoke of saturation and a-symetrical loading Eva, i'm afraid you lost me there, could you be so kind as to elaborate a bit?
thanks
marius |
|
|
| Eva |
You are going to blow the that way FET :D
Search on "push-pull primares". |
|
|
| ifrythings |
Here's some information help you out. (Please do what Eva suggested and do some googling on push-pull primaries or flyback design)
See attached image.
The top circuit (flyback desgin) is what your tring to build, if you build it that way in the schematic you should get farther than before.
In the second one is a push-pull desgin, as you can see there are two mosfets drive half of the primary hence push-pull, but this one requires that both mosfets NEVER turn on at the same time or they will cancel fields can cause a short circuit.
Check out this page for more on push-pull
BTW try driving the MosFet with 5-10V (the maximum voltage on the gate can be +/- 30V) of gate voltage, it may help your circuit better and if your using a signal generator you must put the ground with the ground and the positive on the gate of the MosFet or you'll get very crappy result or as Eva mentioned you'll blow it up. |
|
|
| demogorgon |
gnd from the siggen is connected to gnd from the psu, and gnd in the circut and the osciloscope. the "positive" (the squarewave is both positiv and negative referenced to gnd). is connected through a .22uf and has a 20k resistor to reference it to gnd.
ifrythings and eva, thanks for your tips, and thanks ifrythings for your schematics! i'm not completely ready to give up on the circut yet, but i'm gonna give your circuts a go. :)
-Marius |
|
|
| demogorgon |
I'm gonna have to set up a gainstage in order for my to reach those signal voltage values. perhaps that will take some load off the siggen, i didnt like the waveforms i were getting out of it, think perhaps i was drawing to much current from it? the signal wasn't squarewaves anymore to put it like that.
bah, first vecation in years, the sun is shining, and my ski's are ready to se some action, and all i'm wanting is to get back to my smps:clown: |
|
|
| ifrythings |
This should give you a nice gain factor (depending on transistor used) and the one with the mosfet has some protection for the fet incase it get too much gate drive.
This circuit must be driven with a square wave only (though you already know that right?), a sine wave will cause massive problems.
P.S I just notice that the zener value should be higher like 18V. |
|
|
| richie00boy |
| I just hung a MOSFET on the output of a 555 timer running equal mark:space ratio @ 50kHz. Worked fine and with a 12V supply you don't need any zeners or anything fancy. |
|
|
| ifrythings |
| I only thrue in the zeners for gate protection but as you(richie00boy) mentioned a LM555 or equivalent will work just as fine or better than the signal generator. |
|
|
| LuigiDJ |
Hi.
I wanted to build an smps for a car amp, but was thinking on going further building it for a 120v ac mains for quasi's amp, I'll begin with esp projet 89 and advance from there.
Recently got some Vortex rectifiers that were going to the trash, and inside them there's a bunch of mosfets, diodes, capacitors and some nice (and big) ferrite transformers that i'm planning on using. I'll post something when done. |
|
|
| demogorgon |
Figure i'l post a small uppdate.
i'm in the process of building a smps around the ua494pc i had lying spare. dunno what'l be used for, but i'l think of something.
the datasheet calls for a fullbridge, and so does my transformer.
finally got my veroboard, so i'm jumping forward a bit.
probably be asking you guys a lot of questions in the near future :)
-Marius |
|
|
| N-Channel |
Mzruis,
DO keep us posted. There are a number of schematics for the TL494 available, so you should have no problem coming up with a good schematic.
| quote: | | i'm in the process of building a smps around the ua494pc i had lying spare. dunno what'l be used for, but i'l think of something. |
Hey, I don't know, this might sound crazy, but how about making it power........ an amplifier. :D
Steve |
|
|
| Eva |
| How about making it *drive* an amplifier? That is even crazier :D |
|
|
| N-Channel |
| Depends on which side of the road it is on. :D |
|
|
| Eva |
| A class D amplifier is just a four quadrant switching mode power supply :D |
|
|
| demogorgon |
Sometimes in life you get theese sudden realisations.
often theese are the kind that automatically triggers your hand to your face and a "DOH!" typy respons.
i just had one.
the bloody mosfet doesent open for anything less than 4Vgs.
why in the ninth circle of hell havent i been informed of this before?
here i'v gotten used to .65V as the standard of component activation, and pow, face slap.
i'm gonna give my old teacher a wegie for this :clown:
at least it clears up a thing or two.
-Marius |
|
|
| Eva |
:D:D:D:D:D
Be careful with what teachers told you. Some of them are not more skilled than you are now when it comes to make some real thing with real components work ;) |
|
|
| ifrythings |
| quote: | Originally posted by demogorgon
the bloody mosfet doesent open for anything less than 4Vgs.
why in the ninth circle of hell havent i been informed of this before?
here i'v gotten used to .65V as the standard of component activation, and pow, face slap.
-Marius |
I did tell you | quote: | Originally posted by ifrythings
BTW try driving the MosFet with 5-10V (the maximum voltage on the gate can be +/- 30V) of gate voltage, it may help your circuit better |
also the .65V is for transistors and mosfets require several volts before they do anything. |
|
|
| demogorgon |
Yes, you did. so you did.
I didn't make the connection between what i consdered to be a try-out-this type advice, and "mosfets need 4Vgs in order to open." that you said.
on another note, i may have hit jackpot recently:
the low resolution doesent really do the construction justice.
suffice to say, i'v found myself a bunch of sweet parts, and a new ampchassi :) |
|
|
| demogorgon |
That, Eva, is a Eltek 20A, 48V smps.
norwegian buildt and produced, elna 3300\100v caps (huge) are on the bottom of the two pcb's. 5 caps i all.
the xformer is the square one, on the left, and one hell of an inductor is the one on the right.
gorgeus construction.
no controller ic though, based around the tl555 with a discrete circut.
it was a pleasure to dismantle, everything is just great, sturdy and no-nonsense. obviously not made in china to put it that way.
the case will se different use from now on. |
|
|
| Eva |
| I would have believed it to be a linear supply, or a thyristor based mains sinewave chopping supply at best :) |
|
|
| demogorgon |
well, it says 20A48V smps on the front, so thats the story i'm sticking to. :D
nice find i'd say. |
|
|
| demogorgon |
well, i promised you pics.
here's the most current one.
going slow. very slow. understanding takes time. |
|
|
|
