Converting AC PSU to DC-DC for 12V lighting

[amberwolf]

Normally I'd stick this in my CrazyBike2 thread as I figure it out, but this one is probably gonna take some outside help to figure out; I'm at the head scratching stage (I'm not much of an SMPS guy, really).

I will upload the pics as soon as I can get the bandwidth on the wifi I have access to, but I'm having a lot of trouble with that at the moment. For now, some descriptions will have to do (and I know it's not enough). Also, the pics I do have are pretty bad, as I can't recharge the android phone with the better camera on it yet; it's USB port needs repair and I've not had time for that either).

For now, it looks like this:

So...I found a 12V 16.5A power supply, apparently originally from an old XBOX, for a few dollars at Goodwill, and verified it works normally (green LED) under my lighting load on AC power, and it will power on to Standby mode (yellow LED) when hooked up to my ebike traction pack (at 57.4V at the time), with only one polarity on the AC input--the other polarity does nothing. More on that in a bit.

The standby mode is when it isn't plugged into anything, because it has a 5V pin and a Power Good pin on the output plug, whcih must be shorted together to turn it on (no 12V output until that happens). I did the shorting using a small-value resistor, in case it doesn't handle full 5V on that line very well, but it appears it would probably work fine even with a direct connection.


On the polarity thing: It appears there are a couple of places on the PCB where the AC side does cross over to the DC side, though I haven't traced the circuitry fully to see where they go. If I connect the DC input from the bike battery to the bridge rectifier's + and - "dc" outputs, then unexpectedly it does nothing at all. The reason for this is that those AC input crossconnections into the DC section seem to require the opposite voltage that they get if I hook up the DC input that way (but it would blow things up if I hooked up the DC the other way, of course).

When I hook the DC up to AC input in the polarity that works (for standby, at least), the voltages that appear at those crossovers are opposite of what they would be if I hook up the DC to the DC side of the bridge. If I hook DC up to AC input in the polarity that doesnt' work, then those voltages are the same as if I hook the DC to the DC side.

In either case, I can get it into standby either from DC applied to DC side, or one polartiy on the AC input side. But if I enable it, it shows Red LED for "fault" condition.

So...what I have to do is figure out what I need to change in the circuitry to put the correct voltages at those points, without shorting out anything else (probably will have to cut traces?), just so I can see if it will actually fire up and provide any 12V power on the output when running from the bike's traciton pack, as I want to use this as my lighting and accessory 12V power source, instead of the tiny little DC-DC I'm using now (whihc doesn't even put out a full 12V on the headlight cuz it needs more current than the thing can provide...much less the nearly 14V the headlight really ought to have).

Note that normally it runs on 115VAC, whcih measures out as 170VDC at the bridge's DC outputs (and at the main cap hooked to them). I'm sure I couldn't get the full 16.5A (probably not evne half that) out of it on the voltage I will use, but even 6 or 7A would be enough to run the things I really need it for, and perhaps for short periods I could even use my headlight on "high beam", with both filaments running--I don't need that all teh time, but I do need it sometimes, and right now I can only do it if I carry a 12V battery with me for the purpose. The DC-DC I have is too small.


I've done some googling with what bandwidth I can get, and found a lot of places to tell me how to just do what I already did--enable it with the power-good pin to turn it into a "lab PSU", but nothing yet on how to power it from HV DC instead of AC.

 

[amberwolf]

Pics I took before, not really good enough, but hopefully I'll fix the other cameraphone and get better ones.

 

[friendly1uk]

I suspect the control electronics are happy until you call for the switchmode which won't initialise properly. Perhaps because the output voltage swings up to slow and this is a high precision supply. Mine is bang on 12.00V having twisted the 5v to s/w

Cheaper supply might be needed.
.
Edit: What control chip is in it? I Imagine a small family of them, but what I have worked with has not been that complex. Just a single chip that won't give up unless told to by a few issues that just involved some opp amps. Nothing that would notice charge speed.

 

[amberwolf]

It works perfectly fine on AC input, with or without a load.

On DC input, it only operates on "standby" mode, with the 5V output. Doesn't appear to matter what voltage--I forgot to note that I had tried putting my packs in series, whcih is almost 120VDC and it doesn't work that way either.

So I'm pretty sure what's wrong has to do with the "crossover" circuits from the AC to the DC side, that go around the bridge. Hopefully they just need the right voltage to trick them and not an actual AC input. But I have to do some tracing out of the circuit to find out.


As for "cheaper" PSUs to do this with...I can't imagine anything that's much more cheaply designed than this thing is that I'd still trust to run my lights and stuff (whcih are almost as critical to me as the motor is!), that would also be able to output as many amps as I need. Or at least, not one that would be any smaller and lighter than a big 12V battery. :lol:


All that said: Do you mean that you have one of these same PSUs and are successfully using it as a DC-DC supply? And that all you did was connect the 5V and "power good" wires together? If so, what exactly are the model numbers/etc off the bottom label of the case? Maybe I can find the differences between models and modify mine to be like yours....

 

[friendly1uk]

No, I just use my brick on 230v~
Mine has the 5v and s/w soldered, and with an inductive load I can still flick it on at the wall and it all powers up.
I expected it to be complex (I edited my post). What chip is running things? I can't picture the input end or fathom much from what you describe, but many of these chips have a single high/low pin for work or stop. Very easy to make them run regardless

Edit: Did I say easy? I meant something else I'm sure

 

[dnmun]

can you attach your DC input to the capacitors on the input section?

 

[Farfle]

I have used many of these guys with great success at the full 8A draw from 44v up to 116v

http://www.amazon.com/gp/aw/d/B0057KIHAI/ref=mp_s_a_1_2?qid=1380611323&sr=8-2&pi=AC_SX110_SY165

 

[amberwolf]

I guess if I can't get this one running then that would be a possible option--but the one I have here was like $6 or $7 at Goodwill, and that one at Amazon is $25. So definitely don't want to spend that money unless I have to.

can you attach your DC input to the capacitors on the input section?

Already done (well, actually it's "attached" to the bridge's +/- DC output pins, which are themselves connected via PCB to the main DC side input filter cap, but it's the same thing electrically AFAICT). I guess the pics are not good enough to see that, since I don't have any clear pics of the component side (not sure I can get pics of all of it cuz of the heatsinks in the way).

Problem appears to be that AFAICT there is some enabling circuitry that requires input from the AC side, separate from the bridge input or output. I haven't yet had time to go back to it and get better pics, but I finally got the better cameraphone's charge port fixed, so now it's charged and ready to take pics as soon as I have the time to do that (hopefully tonight or tomorrow, which will be my first day off this week).


I am hopeful that there will be a simple way to feed an "enable" voltage via diodes or something to those "ac" side inputs, from the battery voltage---as long as it doens't require an actual AC voltage it should be easy once I figure out what it acutally needs.

I hope.

(I just wish I'd been lucky enough to find someone here that had already done this on this unit...I guess I'm lucky enough that anyone at all had even used one of these on AC input. :lol: )

 

[dnmun]

if it doesn't switch at the voltage you applied to the input caps then it may be possible to change the way the oscillator is started.

on the switch mode power supplies, there is one leg off the transformer in the back end that specifically is used to start the IC current controller to make it start oscillating. it has a 1 meg or 10 meg resistor from the transformer over to one of the pins in the corner and when the power supply is first energized, the pulse of current in the transformer is carried on that specific input to the IC current controller and that is how the oscillator is initiated.

after the oscillator is running then the transformer does not need to send a signal any more and that is why it has the high value resistor, to keep the current to that pin low. i think.

can you read the part number? is it a 3842 IC current controller? you should be able to find that pin, then see if you can pop it with a voltage, or maybe reduce the value of that resistor by soldering another surface mount on top of it so it can get enuff juice to the IC to make it start oscillating.

hope this helps.

 

[amberwolf]

It isn't that it doesn't switch, because it does startup in Standby mode and give 5V output, which is used to power it's logic and whatnot, and that plus it's Power Good pin are used to turn it's 12V output on.

It's just that it "faults" when I do that, if it's powered off of DC, instead of AC, because of the previously described problem.

AFAICT it isnt' the *amount* of voltage that is hte problem, but rather that it isn't getting the other voltage input from those AC crossover traces to the DC side.

I'll have more info on it's control chips/etc once I can get back into it, and will post that info and any pics I can get at that time. I guess we'll see about it hten.

 

[friendly1uk]

Do you have a variac or something to try more dc?

I can't think why it would watch the AC side. If it was truly basic if would just watch the output. These are very accurate though, they run a pc. They could have brown out protection, where the output is killed before it has time to drop.

Mine is not here. Sat around 2 years then went out on loan just last week. Typical.

 

[amberwolf]

I can always put more cells in series for higher DC, but a variac won't work with DC, only with AC.

However, if it won't work with the DC voltage my pack puts out presently, then it isn't going to work on the bike--I don't have a practical way to configure a higher voltage pack and still use the controllers I have on there in the way that I want to. (I definitely don't need the higher speed this would make possible, and it would make smooth throttle control especially at low speeds even harder than it is already, even with the Throttle Tamer).


I did not get back to this yesterday; both then and today I have had shaky hands and tired brains, and didn't think it safe to be working with the voltages involved. I also couldn't hold the camera still enough to get unblurry-enough images to be useful.

 

[amberwolf]

A couple days or more ago (don't remember) I was trying to post and upload pics, but local wifi bandwidth was so low nothing would upload, and then battery died...had no time till now to retry. So:

Got a few higher-resolution pics. Can't get much on the component side cuz the heatsinks are in the way and the phone's camera will only focus on the right spots at certain angles. Have not been able to do any more circuit tracing or investigation than the below:

NTC inrush limiter runs from DC + out of bridge to + of main DC cap (220V 860uF).

Only control chip I see in there is "5H2 6 1001DL" which comes up as:
http://www.ic2ic.com/search.jsp?sSearchWord=1001DL

 

[boo]

I believe you have two separate Converters. The 5V standby and your 12V main converter. Standby will power up just fine and works at a lower input voltage than the main converter.

The main converter has AC line monitor (R7 R8 and R9 + diode on the component side?), this may be the reason it cannot power up the 12V when supplying DC. This pin requires a pulsating DC input (not so smooth as the rectified waveform from the bridge's DC out).

 

[amberwolf]

Ah, well, I'll have to see about simulating the pulsing DC there then.

Thing that I don't quite understand about that, though, is: why does it partly "work" when I don't bypass the AC section, at least with one polarity, but doesnt' when I do bypass the AC section?


Anyway, I still havent' had time to get back to this. I'm hoping tomorrow, maybe even tonight when I get back from my round trip of Goodwill, grocery shopping, wifi, etc. But probably something else will need my attention/time, like usual.

 

[boo]

Thing that I don't quite understand about that, though, is: why does it partly "work" when I don't bypass the AC section, at least with one polarity, but doesnt' when I do bypass the AC section?

Oh, that's good news . Main converter can work, just need enough signal to fully enable it. This also confirms that you have a diode connected to one AC line to R7, R8 and R9. Only one direction will forward bias this diode so only works when your DC input+ is connected to AC line where this diode is tied.

 

[dnmun]

can you explain what you are talking about? you are saying that the two high value surface mounts R6 and R7 are connected to the AC in front of the rectifier diode and that it somehow causes the switch mode circuit to function for one voltage but not the other? i could not follow the traces on that pcb and cannot even see how they are connected to the AC. also i do not follow how you meant that his statement that connecting in one polarity versus another somehow is proof of this idea you are alluding to but never explained.

those three resistors are there as some sort of resistor divider bridge providing feedback to some part on the other side of the board i had assumed but i cannot see what part they are connected to by the via in the pcb. i don't see it.

 

[boo]

you are saying that the two high value surface mounts R6 and R7 are connected to the AC in front of the rectifier diode and that it somehow causes the switch mode circuit to function for one voltage but not the other?

Yes in front of the Bridge diode via small diode.

If you were to read carefully I posted this:

The main converter has AC line monitor (R7 R8 and R9 + diode on the component side?)

with a question mark "+ diode on the component side?" I initially suspect that a diode links R9 to the bridge's ac pin.

those three resistors are there as some sort of resistor divider bridge providing feedback to some part on the other side of the board i had assumed but i cannot see what part they are connected to by the via in the pcb. i don't see it.

You don't see it because its on the other side of the PCB (component side) .

In summary, One of the bridge's AC pin connected to anode of a small rectifier diode (possibly a DO-41 package) then cathode in series with R9, R8, R7.

PS: My apology for not explaining clearly. I'm an engineer not a teacher...

 

[Arlo1]

I would like to get one of mine working.

 

[Arlo1]

I have used many of these guys with great success at the full 8A draw from 44v up to 116v

http://www.amazon.com/gp/aw/d/B0057KIHAI/ref=mp_s_a_1_2?qid=1380611323&sr=8-2&pi=AC_SX110_SY165

Farfle did you have to mod these or just pop them in and go?
Is it isolated?

 

[dnmun]

i just could not see any of that. i cannot see those on the front side. but i don't follow how it has two separate switch mode oscillators since there is only the one transformer.

 

[boo]

There are two transformers, big and small. One on each side of the daughter board.

 

[dnmun]

i thought the small one was the one that is driven by the small npn transistors from the TL494. that is how we usually see them.

the output from the small transformer usually drives the base current on the big npn switching transistors that drive the oscillator through the big transformer.

that is what i though was happening with the 16 pin IC on the daughter board being the TL494. i could not see the little npn transistors but they are usually on the daughter board.

congrats for seeing all that. i will keep looking at the pictures to work through it. you really did good to see all that in those pictures.

 

[amberwolf]

I might get back to this soon; I picked up two more of these for $1.50 at goodwill today. (the original I started the thread with has vanished (hopefully just into a box from when I moved from the apartment back to Bill's and then back to the house, but I don't remember where it might be).)


They are also both the 16A+ versions, and both at least light up the yellow standby light. Haven't tested otherwise yet as I couldn't remmeber what pins to short to turn them on, so I came back to this thread to find that info, to discover that I did not actually post a pic or diagram of what pins were what on the output connector. :/ If I had any other diagrams I made on paper, they're gone.