Narrow Version of DIY 317 Voltage Regulator Board, with Build Files

Rod, looks good.  How long is the board in this configuration?  BTW  Thx for all your time and trouble on these pcb applications.  😉

TedW posted:

Rod, looks good.  How long is the board in this configuration?  BTW  Thx for all your time and trouble on these pcb applications.  😉

Hi Ted. The board is 45 mm long, about 1-3/4". Sorry I did not include that in the description. To be clear, its exactly the same board as the earlier post circuit-wise, just narrower, which may help in some folk's applications. 

Rod

Rod, after linking back to earlier LM317 threads there is the dangling participle of isolated AC vs. DC ground.  Does the 3-D rendering function allow you to "install" a 2-cent 1N400x diode into 2 of the 4 bridge pads (AC to +) and, say, a jumper wire between the other 2 bridge pads (other AC to -)?  This would allow the same board to be used for 1/2 wave rectification where the AC and DC grounds would be the same which can be handy in some rolling stock applications.  This can also be handy in track-side applications making it directly compatible with insulated-rail triggering.

Separately, I may have missed it but I didn't see discussion of using the LM317P insulated package which would allow you to bolt the LM317 directly to the chassis/frame from heat management. 

That is, finding those insulating washers and related hardware pieces can be a nuisance.  The insulated package is harder to find but can be expedient albeit at a price.  You can't get a whopping deal on eBay as you can with the LM317T (metal tab) for less than a dime.  While there are several eBay sellers of the insulated package, this may be the rare case where buying from DigiKey is cheaper! 

FWIW, no extra components required, just a set of jumpers.  I just use one of the diodes in the bridge for the half-wave option.  Here's how I did a switchable board, in this case this is mated with an eBay switcher to provide power at greater efficiency than linear regulators.

This is the board and the regulator board sandwiched below it.

stan2004 posted:

That is, finding those insulating washers and related hardware pieces can be a nuisance.  The insulated package is harder to find but can be expedient albeit at a price.  You can't get a whopping deal on eBay as you can with the LM317T (metal tab) for less than a dime.  While there are several eBay sellers of the insulated package, this may be the rare case where buying from DigiKey is cheaper! 

That's true Stan, Digikey and Mouser have them for less than a dollar in quantity one, I didn't find one cheaper on eBay or places like Aliexpress.com

gunrunnerjohn posted:

FWIW, no extra components required, just a set of jumpers.  

IIRC, we had this discussion a few years back.  I believe if all 4 terminals of the bridge are soldered-in, you have to cut trace(s) on the board?  OTOH, if you know ahead of time that you want half-wave rectification for common-ground operation I think you can "simply" bend up one or more terminals of the bridge and solder less than 4 terminals.  Then add a jumper or whatever.  The point being cutting a trace with an x-acto knife or whatever is a bother.

In any case, the idea is to document these options, alternatives, etc. in the same spirit that the guys are consolidating the various PCB projects.

I was curious if these 3-D renderings allow you to "install" a 2-terminal diode into 2 of the 4 terminals of the bridge and show what it would look like.  Or, for example, to cut off or "bend up" one or more terminals of the bridge rectifier and show what it would look like installed. 

Stan, I added the jumper field to solve any cutting issue.  That allows me to build the board with both capabilities and just jumper it for the desired operation.

There's a little fooling around to try to graft components in that aren't in the design.  I suppose it could be done, but I'm not sure how much effort it would be.  I think the way to do the resistor substitution would likely be just change the 3D file to a resistor and then try to position it.

Just as a FYI, the version Rod posted here has been added to the main Projects list in the Electrical Reference thread. (Link below in signature line)

Stan; thanks for these two very good ideas. You are the master of timely helpful suggestions! 

I actually have in my user notes that the grounds must be kept isolated, and that if the supply is DC you can just jumper the bridge rectifier pads. But I never considered the idea of adding a simple diode to the + side and jumpering the ground side to achieve common-grounded half wave rectification. Think I'll go back and add that to the user notes.

The insulated 317P is also a nice option. I checked Digikey and they can be had for less than $1, compared to anywhere from $3 to $5 each, and more, on ebay. So definitely DK is the way to go. I'll also add that to the user notes.

Thanks again for your valuable input! 

Rod

Whoa, I see that by the time I got around to responding to Stan's suggestions, there had been all kinds of posts added. Lots of good comments here. 

Rod

Of course a lot of this is just for sport.  That you got the cost down to a 85 cents all-in is amazing enough!  

So, for those applications that need common AC-DC ground, why not save a nickel using a 1-cent diode instead of a wallet-busting 6-cent bridge!

OK I removed the original user notes from my opening post and added a new set R2.2a instead, that incorporates both Stan's suggestions. Tom, you may need to re-link the user notes because of the different name; not sure?

Rod

stan2004 posted:

gunrunnerjohn posted:

FWIW, no extra components required, just a set of jumpers.  

IIRC, we had this discussion a few years back.  I believe if all 4 terminals of the bridge are soldered-in, you have to cut trace(s) on the board?  OTOH, if you know ahead of time that you want half-wave rectification for common-ground operation I think you can "simply" bend up one or more terminals of the bridge and solder less than 4 terminals.  Then add a jumper or whatever.  The point being cutting a trace with an x-acto knife or whatever is a bother.

In any case, the idea is to document these options, alternatives, etc. in the same spirit that the guys are consolidating the various PCB projects.

I was curious if these 3-D renderings allow you to "install" a 2-terminal diode into 2 of the 4 terminals of the bridge and show what it would look like.  Or, for example, to cut off or "bend up" one or more terminals of the bridge rectifier and show what it would look like installed. 

 

I think if you know ahead of time that you need a common ground, the easiest thing to do is leave the bridge out, put a diode across from the near ac connection to the #1 pad (+), and a jumper from the other ac connection to board ground. Presto, done. No trace cutting, no bending, no muss, no fuss! 

Rod

All is well, the Projects post links to this entire thread. All the projects are linked to their original threads for getting further info and/or files.

I realize the train has left the station, but two other comments for the record.

1) Will a 3296W trimpot physically fit in the space allotted for the 3362P - without bumping into the cap or bridge?  It's a little longer/wider and it would require a new pad to be dropped on the board since the terminals are in-line albeit on a 2.54mm grid.  The idea is to allow either style of trimpot to be inserted.  The 3296W is multi-turn as used in the ubiquitous LM2596 stepdown regulator modules that we have come to know and love.  Same cost on eBay when I just looked. This might be a nice project for someone getting started with DipTrace and could take Rod's existing design and make a minor modification with a high likelihood of success while contributing to the cause (in my opinion of course).

2) For some applications with low-current, it might make sense to install the TO-92 version of the LM317.  Might be even cheaper depending on where you buy.  The idea here is very few designs come anywhere near requiring the >1 Amp capability of the LM317T.  In fact I seem to recall real-user data where the current required in a passenger car LED application was 5 mA (or less)!  In such a case there's no need for the TO-220 version which of course is a rather large package.  Commensurately, you would not need a 1000uF capacitor to achieve a suitable flicker reduction...and could use, say, 220uF or less which would allow even a smaller overall space requirement.  Again, the idea here is to create what amounts to an Application Note for Rod's board design.  I suppose if someone learning DipTrace wanted a simple modification project to get one's feet wet, they could add 3 pads on a closer grid to install either the TO-92 package or the TO-220 package (same PCB board)...and contribute to the cause (in my opinion of course).

Would strongly suggest that a heat sink be added to the LM317.

You could also add a pad for a diode under the bridge so that you didn't have to guess where the diode goes. 

FWIW, the board could also be smaller, no reason to have all that space between the choke and the pads.

stan2004 posted:

2) For some applications with low-current, it might make sense to install the TO-92 version of the LM317.  Might be even cheaper depending on where you buy.  The idea here is very few designs come anywhere near requiring the >1 Amp capability of the LM317T.

I did a fixed voltage version using the TO-92 chip. 

It's less than an inch long and .3" wide.  You can use any voltage three-terminal regulator in the board to pick a voltage.  You can also stick the CL-2 constant current regulator in the board with no changes for a constant current application like LED lighting.

Bob, the heat sink is up to the end user, and will vary by application.

John, the space between L1 and the pads is to permit the use of connectors like the small green 2.5mm screw terminals. For the final run I did in fact shorten the board a little, it is about 1-3/4" long.

I do like the idea of the additional diode pad under the D1 pattern; will likely add that to the design. 

We need to remember that one of the objectives here was to stick with easy thru-hole design so that most folks (like me) can build it without too many issues.

Rod

Come on Rod, live a little!   Gosh, I'm 76, and if I can stick SMT parts on a PCB, I think there should be lots of folks that can manage it as well.

The major reason I use SMT stuff is I'm usually trying to make something I'm going to stick inside a locomotive or rolling stock.  It always seems I don't have sufficient space, so I try to make them as small as possible.

stan2004 posted:

I realize the train has left the station, but two other comments for the record.

1) Will a 3296W trimpot physically fit in the space allotted for the 3362P - without bumping into the cap or bridge?  It's a little longer/wider and it would require a new pad to be dropped on the board since the terminals are in-line albeit on a 2.54mm grid.  The idea is to allow either style of trimpot to be inserted.  The 3296W is multi-turn as used in the ubiquitous LM2596 stepdown regulator modules that we have come to know and love.  Same cost on eBay when I just looked. This might be a nice project for someone getting started with DipTrace and could take Rod's existing design and make a minor modification with a high likelihood of success while contributing to the cause (in my opinion of course).

2) For some applications with low-current, it might make sense to install the TO-92 version of the LM317.  Might be even cheaper depending on where you buy.  The idea here is very few designs come anywhere near requiring the >1 Amp capability of the LM317T.  In fact I seem to recall real-user data where the current required in a passenger car LED application was 5 mA (or less)!  In such a case there's no need for the TO-220 version which of course is a rather large package.  Commensurately, you would not need a 1000uF capacitor to achieve a suitable flicker reduction...and could use, say, 220uF or less which would allow even a smaller overall space requirement.  Again, the idea here is to create what amounts to an Application Note for Rod's board design.  I suppose if someone learning DipTrace wanted a simple modification project to get one's feet wet, they could add 3 pads on a closer grid to install either the TO-92 package or the TO-220 package (same PCB board)...and contribute to the cause (in my opinion of course).

Good suggestions Stan; my take for what it's worth:

1) I haven't tried to implant the 3296W trimpot, but may give it a whirl. IMO the readily available 3362P pots are nice and compact, and offer only a 270 deg turn from max to min, and will allow 0.1 volt output resolution in this circuit, so that's what I went with, for now! That's only me though.

2) This board was not really intended for low power apps like led lighting. Grj's constant current board already does a great job of that. This board was intended for GP vreg use up to about 500 ma, AC or DC input, engine, car, or trackside mount. Having said that I tried adding the TO-92 pattern, as well as grj's suggestion of a single diode pattern over the bridge diode pattern. Here is what I came up with:

This is certainly workable, though I had to scrunch the components a bit to kept the new length down to 48mm, just under 2". But it fits and it would work. Width is still 1/2". I used a DO-35 pattern for D2 and angled it as you see to take up less space. 

But then I got thinking, is this really necessary? The TO-92-100 pattern has the same pad spacing as the TO-220, and the pinouts are the same. So why not just stick the TO-92 leads into the same holes and call it done? Sure, the TO-220 drill holes are bigger than needed for the small version, but no big deal. Then I tried doing this fit and a 1N4002 diode across the ~ and + pads of D1. Here is the mock up:

As you can see, it's a no-brainer fit. This of course is not the same board as posted above, but you get the idea. For a low power 317L version, you could just use a 1N4148 diode for D2, which would be even easier to fit.

I see 4 power options available in using just the bridge diode pattern as it stands:

1) Full wave AC to DC, with isolated grounds, using the bridge diode.

2) Half wave AC to DC, with common ground, using a diode as pictured, and a wire jumper from ~ to -.

3) Straight DC input using jumper wires for both sides of the bridge pattern, as per 2).

4) Straight DC input with the bridge in place, accepting that there will be a 1.4 volt loss in feed voltage to the 317. 

This is pretty good versatility. All in all, I really don't see a burning need to put yet another set of Gerbers out there, but that's only my opinion of course!! If we think it should be done, by all means I will put them here for rtr12 to add to the sticky post of build projects. 

Rod

gunrunnerjohn posted:

Come on Rod, live a little!   Gosh, I'm 76, and if I can stick SMT parts on a PCB, I think there should be lots of folks that can manage it as well.

The major reason I use SMT stuff is I'm usually trying to make something I'm going to stick inside a locomotive or rolling stock.  It always seems I don't have sufficient space, so I try to make them as small as possible.

Well you've got a couple of years on me John; but you must have a way higher level of patience than me! 

Not to mention better eyesight, haha!

Rod

I'm a masochist.

But wait!  There's more!  

Given the compact size of this board, it reminds me of a $1 module you can get on eBay that is essentially just a bridge rectifier and capacitor.  OK, there's an LED but ignore that for a second.  The idea is Rod's board can be used to perform the useful task of AC-to-DC conversion with large capacitance on the DC side.  And of course it has the DCS choke which I can't imagine any eBay module would ever have!

So if compiling a book called "101 things you can do with Rod's board":

1) there are thumbnail sized mini DC-to-DC modules though some are limited in their allowed input voltage range.  If, say, driving a smoke unit in a diner car or caboose, you need at least 5 Watts of power at maybe 5V or 6V  DC.  Because the LM317 is a linear regulator, it might be overwhelmed by the heat dissipated going from 18V AC track voltage down to 5V DC.  So you can set the LM317 to, say, 20V DC and then pair it with a mini DC-to-DC module to efficiently reduce the voltage to 5V DC.

2) or if a car is used in both conventional and command environments, the AC-to-DC converter could be paired with a buck-boost DC-to-DC module for constant voltage lighting.  Most of the buck-boost modules do not have enough input capacitance to support flicker free lighting so that's where the capacitor on Rod's board comes to the rescue!

3) though probably not for rolling stock, but one could choose not to install the LM317 in Rod's board and simply use the board for the AC-to-DC bridge and capacitor and pair it with a DC-to-DC LM2596 voltage regulator with built-in voltmeter.  It has come up several times that there are no low-cost AC-to-DC stepdown voltage regulator modules on eBay WITH integral voltmeter.

And so on!

Rod is helping me. I needed a small board.

My question Stan brought up. Is ground. So dc and AC can use same ground?.

On the half wave. Which I'm using.  Once I get the boards so I can see where things go my plan is to jumper AC in to + of bridge. With a diode.

Then jumper - of bridge to dc neg. A.c. ground will share common ground.

When using fixed resistors. Do you leave one hole open? No jumper...

riki posted:

Rod is helping me. I needed a small board.

My question Stan brought up. Is ground. So dc and AC can use same ground?.

On the half wave. Which I'm using.  Once I get the boards so I can see where things go my plan is to jumper AC in to + of bridge. With a diode.

Then jumper - of bridge to dc neg. A.c. ground will share common ground.

When using fixed resistors. Do you leave one hole open? No jumper...

Riki, yes jumper the diode from ~ to + with the band to the +, and wire jumper the other side from ~ to -.  That makes the ground common. For fixed R2, jumper the pot pad from pin 1 to 2 with the resistor. No connection to pad 3. 

Rod

Check out this interesting use for an LM317 regulator.

Stan, interesting suggestions. I would never have thought of any of those potential uses! 

Rod

Thanks rod.

No problem, hope you recieve your boards soon! 

Rod

This is my wiring plan for half wave. 

Riki, the half-wave method requires a diode or a modified (bent-up terminal) bridge to perform the ac to dc conversion.

Stan..yap forgot to add that.

Here is where I'm at.  The neg side of diode I can put in other hole at edge of board.

Either one will let me add AC line. Think it's easier to solder  AC wire on the edge.

That looks like it will work Riki. Don't forget the wire from the nearest ~ pad on the DB107 pattern, to the - pad. That leaves both pad holes of the AC header open for attaching wire. Hope that makes sense.

Rod

Ok.. I see..

Thanks rod. .

1 down

That's the idea Riki. Looks like you have it fitted up very nicely. Good idea mounting R2 vertically like you did. Fits better. I would maybe use a screw driver to bend that diode lead a little, where it is very close to the jumper wire; just to give a bit more clearance. You don't want them to touch.

It looks like you are nearly ready for a test run! 

Rod

Rod. I did bend over diode ..but it was away from jumper.  But more clearance now.

Have one that works out of three.

Think the voltage reg are bad on other's. No regulation. 

Made one with bridge . but . if negs have common ground. Dead short.

If I put a diode on the AC ground will that work. With shared ground.

I think I need a heat sink on 317 while soldering .