Drop 5v from source using LM317 or other?

[XOIIO]

Hey all, quick question, I'm sure you can do this though not sure if it will work with an LM317, I am wondering if there is a way to drop 5v all the time, from the source. Basically modifying a power supply to be variable, has a range of 5.75-14v dc right now, I'd like to knock 5v off that at all times so I can drive LED's directly. Max amperage on the supply is 1.1 amps so I won't need anything too high power.

[Skimask]

String of 10 1N5817 diodes.

[Ian.M]

Nearly all LEDs are current driven not voltage driven.  The exceptions are ones that already have a constant current chip integrated in the package and those can typically be run directly from a 12V supply.

You therefore shouldn't be trying to drop a constant voltage but should be trying to get a constant output current.  There's a circuit for that in the LM317 datasheet.  It needs one resistor to set the current.  e.g. nuse a 120R resistor between OUT and ADJ and connect the LED anode to ADG and you will get a steady 10mA through the LED as long as the supply voltage is greater than LED Vf + 3V.

[XOIIO]

Nearly all LEDs are current driven not voltage driven.  The exceptions are ones that already have a constant current chip integrated in the package and those can typically be run directly from a 12V supply.

You therefore shouldn't be trying to drop a constant voltage but should be trying to get a constant output current.  There's a circuit for that in the LM317 datasheet.  It needs one resistor to set the current.  e.g. nuse a 120R resistor between OUT and ADJ and connect the LED anode to ADG and you will get a steady 10mA through the LED as long as the supply voltage is greater than LED Vf + 3V.

It's not just LED's that I am trying to power though, it's basically going to be a mini bench supply, so I'd like to be able to lower the voltage range for other uses as well but being able to get down to .75 volts (if it was an exact 5v drop) would be good for testing them.

Didn't think of diodes but yeah, that might work well. Don't think I have that many kicking around though, I think I'll have to go scavenging for parts.

[Kdog44]

If you are strictly looking to drop a set voltage, you could just link diodes in series;  1n4001 diodes. Analyzing the non-ideal diode case: each diode would drop about 0.7 volts with 1.1 A flowing through them. See the attached diagram for a visualization of the relationship between the current and the voltage drop across this diode. For the current flow you will be operating at, you would need a different amount of diodes in series. For example, with a current flow of 1.1 A (your max) you would need  7 1n4001 diodes.

[XOIIO]

If you are strictly looking to drop a set voltage, you could just link diodes in series;  1n4001 diodes. Analyzing the non-ideal diode case: each diode would drop about 0.7 volts with 1.1 A flowing through them. See the attached diagram for a visualization of the relationship between the current and the voltage drop across this diode. For the current flow you will be operating at, you would need a different amount of diodes in series. For example, with a current flow of 1.1 A (your max) you would need  7 1n4001 diodes.

Alright, seems that's how I'll do it. A good thing about this is I can add a switch to bypass them as well giving me a lower range or higher range. I'll just need to set them up in a compact/fairly nice little bank to fit in well. It's somewhat limited for space but not too much, and that's what is making this a great little supply to modify, I just need to make a good cutout for the voltage display, I have a nibbler, but I need to find my file and go really, really slowly. I usually end up with gaps that look terrible.

Ah if only I had a CNC.

edit: what the heck image is that lol

[Ian.M]

A diode dropper is about the daggiest way of doing this that has any chance of working.  It will seriously compromise regulation. What you actually need is an amplified Zener.   Take a beefy NPN power transistor, rated for at least 1.5A Ic and 10W dissipation and connect a 5.1V Zener cathode to its collector, anode t its base.  Add a 1K resistor base to emitter, feed your supply to its collector and get 5.7V or so less from its emitter.   If it drops too much try a 4.7V Zener.  The transistor will be dissipating nearly 6W at full load so it will need a heatsink (and an insulating heatsink mounting kit).  I see the PSU is alrready fan cooled, so as long as you can position the heatsink in a good airflow where it isn't going to heat up anything critical downwind, that shouldn't be a problem.

If you need a more precise, adjustable amplified 'Zener' circuit, there is one with a PNP pass transistor in the TL431 datasheet.

[george graves]

Seems like a long way round to getting what you need.  When ever I find my self asking a questions like "Why can't I just do A and get B" and the answer is, "well, you could kinda do it, but it's not very good"....I know it's time to re-examine why I need "B" in the first place. Cause chances are, I'm not doing something right. Just my 2 cents.

[XOIIO]

A diode dropper is about the daggiest way of doing this that has any chance of working.  It will seriously compromise regulation. What you actually need is an amplified Zener.   Take a beefy NPN power transistor, rated for at least 1.5A Ic and 10W dissipation and connect a 5.1V Zener cathode to its collector, anode t its base.  Add a 1K resistor base to emitter, feed your supply to its collector and get 5.7V or so less from its emitter.   If it drops too much try a 4.7V Zener.  The transistor will be dissipating nearly 6W at full load so it will need a heatsink (and an insulating heatsink mounting kit).  I see the PSU is alrready fan cooled, so as long as you can position the heatsink in a good airflow where it isn't going to heat up anything critical downwind, that shouldn't be a problem.

If you need a more precise, adjustable amplified 'Zener' circuit, there is one with a PNP pass transistor in the TL431 datasheet.

The PSU actually isn't fan cooled, that's for a 10w LED that I have but I have other small fans that I could add in, I just need to find a 12v source somewhere on the board aside from the output. I guess I might have to wait longer but using a zener and transistor sounds much more elegant, and I think I have some that would exceed 10w quite a bit but I might just look locally for a new one, that or wait for one off ebay. A 4.7v zener would probably be the better option, since I probably won't need to go below 1v.

The case for the PSU is decently chunky so that might work as a heatsink, failing that I have some smaller sized ones from power transistors that I could rig up to sit in the airflow.

Seems like a long way round to getting what you need.  When ever I find my self asking a questions like "Why can't I just do A and get B" and the answer is, "well, you could kinda do it, but it's not very good"....I know it's time to re-examine why I need "B" in the first place. Cause chances are, I'm not doing something right. Just my 2 cents.

Well yes I could probably buy some tiny "bench" power supply off ebay, but it's way more fun to modify something to your needs. I've already tried adding a resistor in series with the potentiometer but the supply will not go below 5.5 volts or so, so I need to regulate it right before the output.

[Ian.M]

If its not fan cooled, you *MAY* be able to get rid of 6W simply with an aluminium heat spreader bolted to bare metal of the case and the transistor to the heat spreader, but its quite likely that you will need to fit a finned heatsink to the rear panel, with a cutout to mount the transistor to that.

I'd use the 5.1V Zener, as, if you are using range switching to bypass the dropper, it doesn't matter if the voltage drop is a little on the high side, and its useful to be able to go right down to zero, especially if you are trying to find a power rail short on a board.

[XOIIO]

Well, here we are, as far as bench power supplies go it's the smallest I've seen. It's a great size to have beside a laptop or something when working on arduino or other microcontroller stuff, and pretty fine adjustment considering the old trim pot was 2k and this is 10k.

Unfortunately I make the hole for the display in the opposite corner  so I had to drill four new holes and pop rivet it together, but rivets are easy to drill out. Problem is the drill slid off my desk and I broke a drill bit :/ Oh well. A new paint job and it might look pretty slick even though there's a bit of a gap for the display, there's room to add a current version of that if they make them in that size, I do like the old labels on it though, so that you can tell it was something specific before. I just need to get that sticky crap off the top now, and add in that other stuff once I get parts, I'm going to spend some time working on a good place to mount a fan/heatsink first tho





For those interested it;s a JVC ac-c712, and holy crap I just looked them up on ebay, maybe they aren't a source for cheap moddable supplies lol

[TerminalJack505]

Here's another idea.  It will need to be compensated, I'm sure.  You'd also probably want to use a unity gain difference amplifier in place of U2.  As shown, the voltage drop can be adjusted by adjusting the voltage at U1's non-inverting input but you can make the voltage drop a fixed 5V by setting it to 5V--use your favorite voltage reference IC for that (TL431, for example.)

[XOIIO]

Here's another idea.  It will need to be compensated, I'm sure.  You'd also probably want to use a unity gain difference amplifier in place of U2.  As shown, the voltage drop can be adjusted by adjusting the voltage at U1's non-inverting input but you can make the voltage drop a fixed 5V by setting it to 5V--use your favorite voltage reference IC for that (TL431, for example.)

Probably a bit complicated for what I will be using this thing for, I don't think I'd need it to be quite so precise. The local electronics place has transistors, and possibly heat sinks for them, and fans if I want to overpay massively for a fan lol, I should be able to get everything there. I need to stock up on heatshrink anyways.

[XOIIO]

Unfortunately there is no listing for appropriate wattage dissipation on their heatsinks, maybe someone with some experience can point me towards one that could handle 6w freestanding (or I guess maybe 4w since I will be setting up a fan) www.be-electronics.com/SearchResults.asp?Search=heatsink

This one looks promising. http://www.be-electronics.com/product_p/647-10abp.htm

[XOIIO]

Crud, the only other supply in this thing is 1.9v for the LED, nothing else, so I can't power a fan unless I tap off the bridge rectifier and massively reduce that somehow, damnit. I'm going to have to see if I have any more ebay USB iphone chargers left, problem is those only put out 5 volts... Hmm. I'll have to find one and poke around that and maybe find 12v or modify it, then stick it in there too. Not gonna be pretty 

[XOIIO]

Sweet! God 3 of them, 2 of which I already sealed up with hot glue, and I hae fans from an old xbox 360 that run pretty good on 5v. Decent amount of airflow, I wonder if it will be enougn though. hmm

[Ian.M]

Wakefield's datasheet says 42 deg C temperature rise when a 647-10ABP is dissipating 6W.  http://www.mouser.com/ds/2/433/board_level_heat_sink-228.pdf

[XOIIO]

Of course, their website shows stock in another province, that only makes sense.

No zener diode, and no heatsink, I got a better one off of an old motherboard though, just need to hook it all together, I still need thermal compound and the damn diode though, I doubt anywhere else will have one so that would mean waiting 3 days at least.

[Ian.M]

If you've got a TL431, use it in its datasheet Figure 21. High-Current Shunt Regulator configuration.  They used to be fairly common in the voltage regulation feedback circuit on the secondary side of the SMPSU in old tube TVs etc. so you may have one on a scrap board.

Otherwise, kludge in a Vbe multiplier till you get the Zener, with a J-fet wired as a constant current diode  as a pulldown to ground from the pass transistor's base to provide the Vbe multiplier with enough bias current to work reliably.

[XOIIO]

The transistor I got is an NTE152, there doesn't seem to be a spec for un heatsunk (IE freestanding) current capacity, maybe hooking it to the metal chassis would be enough, I'm having trouble finding a heatsink that will fit, my collection has vanished during a move of course :/ http://www.nteinc.com/specs/100to199/pdf/nte152.pdf

[XOIIO]

The back of this thing's enclosure is pretty thick actually and I also found this relatively wide finned heatsink, a couple holes for mounting too, I could rivet the heatsink to the case, and then the transistor to the heatsink through the case, pulling it flush with the case but also having the metal from the rivet to transfer heat directly out. No way I'll find specs for these heatsinks, they were for this size transistor though, I might even be able to find the board.

I'm not sure it seems like that might work out alright.

[Ian.M]

Orient it vertically for better convection.  Do a soak test and, worst case, if it cooks, look for another transistor and a bigger heatsink.  You can also off-load some of the dissipation into a resistor - 3V Vce should be enough for the transistor to behave fairly well, and a 1.8R 3W resistor will drop just under 2.2V at max current.  Insert in the collector lead, between the transistor and the Zener cathode.  The same can be done with a stack of 3 beefy silicon diodes.  The worst case dissipation in the transistor will be reduced be between 40% and 50% with negligible effect on the regulation.

[XOIIO]

It was quick enough to set up I just figured why the heck not, easily reversable too. I think it might work fairly well. We'll find out once I get the rest of the parts and I whack a thermocouple on it.




Well, read that reply too late lol, I guess vertically would have been butter but it would also stick up a bunch too.

What is a soak test though? Never heard of that before.

[XOIIO]

Oh and hey there's a part number for the heatsink after all. THM 7022

http://www.newark.com/aavid-thermalloy/7022b-mtg/heat-sink/dp/99R8258?ost=99R8258&CMP=AFC-NEDA

Looks like it would be a 39 degree temperature rise?

Edit: I think I'll mount the fan to that heatsink to pull air through, I bet it will work pretty good, if the rivet gun doesn't crack the fan.

[Ian.M]

Soak test A.K.A burn-in test
Apply max rated load for maximum probable period of continuous operation, or at least several hours after a steady state has been reached. 

You are aware that you are going to have to rip that off and start over as the rivet will short the transistor collector to the chassis, and you haven't used any thermal grease?   Its *ESSENTIAL* to use a proper top hat insulating heatsink washer on the transistor's mounting bolt.  You also need thermal grase both sides of the mica washer and between the heatsink and chassis or you might as well not bother.

[XOIIO]

Soak test A.K.A burn-in test
Apply max rated load for maximum probable period of continuous operation, or at least several hours after a steady state has been reached. 

You are aware that you are going to have to rip that off and start over as the rivet will short the transistor collector to the chassis, and you haven't used any thermal grease?   Its *ESSENTIAL* to use a proper top hat insulating heatsink washer on the transistor's mounting bolt.  You also need thermal grase both sides of the mica washer and between the heatsink and chassis or you might as well not bother.

The chassis isn't grounded, and yeah I'll have to put some thermal grease in at some point, it's not difficult to get back to it, though I didn't think about that when putting the fan on, that will be a bit more... interesting. The smaller washer wouldn't fit over the rivet, and I don't think I can find any small enough screws around here without driving haflway across town. Nice to finally have a use for the rivet gun anyways. I just want to make some sort of progress before I sleep in10 minutes or so since I waited all night for parts, then they didn't have all of them.

Fan is definitely too fragile for full force but partially and the rivets hold it in well, just need to file them down. I'll do that on the final ones though.



I think this should work quite well overall though.

Edit: ah since those stems are there and they aren't fully... riveted I guess, they just popped down and it's good to go.

[XOIIO]

*palm* alright never mind, that's what I get for going on the scuffed lettering, I checked the datasheet and the mistake was on my end