adjusting the output on a smps used as a DC-DC

[auraslip]

I couldn't easily find (without ordering from china) a dc-dc to buck my new 20s pack down to 12v, so I went out and found an older 12v 5a smps on ebay for $10.

It works, but of course not as expected. It only outputs ~8v. The adjustment pot does nothing, but I assume it won't be too hard to figure this thing out. Something else I'd like to mod (remove really) is the rectifier so that it won't fail per-maturely.




The output dual op-amp ties into the feedback of the uc3843 through an optocoupler. Since this output has a voltage divider that is the pot, (and since the pot does nothing) I'm going to guess that nothing I do on the output stage will help. So this leaves the uc3843. I attached the data sheets to these devices.

 

[dnmun]

i am not sure how much you know about switch mode power supplies, but i suspect there is some key information that was left out of this. is this 120V AC input or some kinda DC input?

 

[ZOMGVTEK]

20s might just be a little low for the supply.

You could bypass the full wave, but beyond that I'm not sure what else there is. Does it put out any current? Did you put a load on it? It might just fall on its face.

I just put a 7.5v switcher on my 24s pack, and it won't turn on with the headlight connected, it just flashs on and off. I ended up having to run the switcher on the ignition wire, and switch the output to the load. Not ideal, but it only takes 1-2W unloaded. It worked fine off 120VAC, but just barely off a dead 24S pack.

 

[auraslip]

Yeah, it runs my halogen.... at 8v it's pretty dim though. Based on the early threads where methods and fetched tested some of the smps down all the way to 60v I was surprised mine didn't work.

I'm running it off a 20s pack 76v nominal or so.

I was looking at the board and the feedback pin on the uc3843 goes to ground before a resistor. It's not set up at all like the sample application schematics provided by various datasheets. Arhg. By the time I get everything working on this bike I'm going to know more about smps than I ever thought I would.

 

[dnmun]

there are very few textbooks on SMPS devices using the latest ICs, but there is a lot of literature in the specs.

i used to have a white paper for the everbright 816, but i think that is on a dead disk. this is for the pwm on the voltphreak. maybe what you think is a resistor is a capacitor. notice how the cathode of the voltage reference and the cathode of the opto transistor are tied together through the resistor divider bridge. not sure why you see the output of the op amps as going to the opto.

 

[nechaus]

I say put this away for a rainy day and just put some bulbs in series ??

And halogen, why dont you order some LEDS...

i have about 30 leds on my bike, man its bright..... i only use it when im going into a super dark area,
@ night i perfer to have no lights unless i need them

 

[ZOMGVTEK]

Some of these work perfectly down to 60VDC in? That's surprising. I have a few larger switching supplies that don't like 80-90VDC, but run off 100 and up. 60 is well off what most of this stuff was designed for.

 

[auraslip]

yep http://endless-sphere.com/forums/viewtopic.php?f=16&t=2264

 

[docnjoj]

As usual there is an exception to every rule. I have an Averatec power supply that put out 19volts @ several amps from my 54 volt Fatpack set up. It powered the Magicshine and rear view screen till I went back to 36 volts.
otherDoc

 

[auraslip]

well, I'm in contact with my supplier of dc-dc convertors from last year. looks like I might be ordering some higher voltage models as well as the lower voltage models I got last time.

 

[auraslip]

So i've found what SHOULD be a great solution : http://www.ebay.com/itm/Mean-Well-SD-100D-12-DC-DC-Converter-100W-12VDC-8-5A-/150676294775?pt=LH_DefaultDomain_0&hash=item231501cc77

meanwell 72v-144v 100w dc-dc converter. For $40 it's not a bad deal.... BUT the thing is BIG like 8"x6"x1.5" .... OTOH... it's meanwell quality....

wish they made a 60w smaller version with high input voltages.

I've also noticed that some of their ac/dc power supplies also have dc voltages listed in their input range.... like 90vac-230vac and 135vdc-150vdc....

seems confusing.

 

[Ricky_nz]

If your original converter is only outputting 8V with the pot having no effect you could try adding a few turns to the secondary of the transformer. looks like there might be enough space to get a few turns in.

The fact you see 8V suggests that the mains input side is switching but you don't have enough volts on the secondary for the regulation circuit to have an effect.

 

[AussieJester]

Lyen sells dc-dc converter good from 12-60v input, outputs 12v 5amp IIRC iits 1/4 the size and only 25 bucks...why bother
with these 120vac units when you can buy the correct dc-dc converter for near same price
as what your spending?!

KiM

 

[auraslip]

a) the lyen unit couldn't handle the 50w halogen bulb I run let alone the 5v phone charger

b) cause I run a girly frock motor I need to run +70v batteries to get some fun power


Is rewinding a transformer easy?

 

[Ricky_nz]

Is rewinding a transformer easy?

Not too hard provided it isn't potted in resin. If you can figure out which pins are the output winding.
You will probably only need a few more turns of wire. you can usually remove the clips off the transformer provided its not potted and remove the ferrite cores to make things easier. The turns per volt is quite low in high frequency transforms like this.

You might have to use a solder sucker and wick to get the transformer out the circuit board to make it easier but I think you could take a bit of thin wire and thread it on top of the exsisting windings and then connect it in series with the exsisting output winding. if you connect it the wrong way the output voltage will drop. you can use several; strands of thin wire if the space is a bit tight rather than one thick one.

The alternative which may be possible if the input is on the outside layer of the transformer is to remove some turns from the input side but messing with the input side may cause problems if you remove too many turns.

 

[dnmun]

i posted up a schematic for you to educate yourself with about the switch mode. you should be able to find all the resistors in the divider between the output voltage and ground so you would then be able to make the changes that will get you in range of your needed output voltage.

did you test it to see if it worked on 120AC? you may wanna do that, and measure the voltages on the output of the bridge, across the input caps. and there should be voltage on the high side of the schottky diode from the transformer. then compare that to the voltages if you connect up the battery pack, which i assume is what you were doing, but was never stated yet.

not sure where this idea you needed to add more windings on the transformer comes in.

what is the voltage on the gate of the switching mosfet in the front end? you will have to adjust that power resistor on the source leg of the power mosfet too in order to keep the total power through the transformer down within spec for the transformer.

 

[Ricky_nz]

not sure where this idea you needed to add more windings on the transformer comes in.

It really depends on the converter topology but if its designed to operate with higher input voltages it would make sense to add a few turns to the ourput to move the turns ratio back to nominal as it could be too low. possibly less of an issue if it is a flyback converter though.

It is clear that his converter is switching but the output voltage is too low. I guess the aux on the input could be a bit low causing poor switching but at 8V outt it sounds like it just ran out of headroom.

 

[dnmun]

i'm sorry, but we do not even know if it is switching. the oscillator requires a kick to start and i thought that had to come out of the transformer, so just to have some numbers even to guess would help in giving him advice.

like i said, i only found one textbook on them and it was old from the 80s, even though it was considered the 'bible' on SMPS. and i had to go find a used one in powell's technical department to read, but just a few hours made a lot of difference in my understanding of them.

this is where google helps because there are several good white papers prepared by the pwm manufacturers that explain how the newer IC current controllers work. that schematic though is simple enuff that he should be able to find the resistors, and see that there is a cap on the feedback, not a resistor to ground.

 

[Ricky_nz]

Yeah, it is a bit hard playing guess the circuit :lol: . I do remember that chip from back when i worked in telecoms.

My assumption of switching could be wrong although he did state that he measured 8V out. But I guess that could be pulsing or unstable if the chip supply isn't stable.
That chip does have a low voltage lockout with hysteresis so thats why i assumed if he had some output it had enough to run the control chip but yeah checking the chip supply should be an easy first step. If thats there then it is likely that there isn't sufficient output voltage for regulation caused by either FET not turned on hard enough or winding ratio. Nothing on the output side circuitry should need to change.

The chip may be powered off a separate winding on the transformer to avoid losses once running but be kick started by a resistive dropper.

Edit: just looked at the data sheet in more detail to refresh my memory.
You should be able to measure between pin 7 and pin 5 of the UC3843 and see a voltage that is stable and above 8.3V. if that is the case the converter chip will run/be running and should be able to drive the MOSFET. You need to make sure this voltage is stable and never drops below 8.3V though.
It is likely this is supplied through a dropper resistor and then fed from a winding on the transformer through a diode once running.