What about reversing driver regulation polarity?

Flashlight Modding and DIY Parts
1

Most, if not all, of the drivers in our flashlights are switched/regulated on the negative side, with the positive side taking a mostly unobstructed path from Batt+ to LED+ on those drivers. I’m wondering if there’s a simple way to make the driver do its switching/regulation on the positive side instead. From my (very limited) understanding, the ATtiny sends out a positive signal to either the 7135(s) or FET. Both of those components take the PWM on the positive input and use that to regulate the negative output. That regulated negative is what goes to the LED.

Obviously, the LED doesn’t care which side is switched/regulated. It will be “on” only when both sides are connected, and “off” otherwise. But, the switching/regulation components themselves DO care. So, for instance, if you want to add a slave board to your driver in order to add more 7135’s without stacking, you normally must run three wires from the driver board to the slave board. You can’t just connect it to your LED+ and LED- leads. You have to run another lead for PWM+ to the slave board so that it can be regulated along with the driver. This is normally not a big deal. But, there’s at least one case where it might be: remote slave units. The further away the slave board is from the driver board, the more the third wire might matter. Although in flashlights this isn’t going to come up, in other types of LED lighting it can. But, if the driver is switching/regulating on the positive side, then a third wire isn’t needed, because the remote regulators will have everything they need in just two leads. The positive lead from the driver will work for both the PWM input to the regulators and the positive power to the emitter(s).

Also, there are some LEDs still being made with no thermal pad. It is generally recognized that the LED- pad is carrying most of the heat in these two-pad emitters. A while back, BLF member Clemence had some DTP boards made for these emitters, in which the LED- pad is integrated into the MCPCB in DTP fashion, just like the thermal pads on Noctigon and Sinkpad MCPCB’s are made. But, if the driver switches and regulates output on the LED- lead, then the emitter can’t be directly (electrically) connected to the host. There must be some form of insulation layer which will ultimately hinder thermal transfer. While there are ways to work around this issue, if the driver were made to switch/regulate on the LED+ lead instead, it would not be an issue. Then the DTP boards designed by Clemence for two-pad LEDs could be connected in the host in the most thermally efficient way possible with no extra effort, just like a Noctigon or Sinkpad.

Third is the fact that some flashlights actually do power switching on the positive side. I’ve seen several flashlights (I own some) with side clicky switches in the head that interrupt the batt+ to the LED. Some of those flashlights don’t have drivers at all, while others do have drivers which are switched/regulated on the positive lead. Obviously, in either case, in order to upgrade any of those lights, a driver that switches/regulates on the positive side is needed.

So, what will it take to make drivers regulate on the positive side? I know there’s one obvious answer I can think of myself. Add a transistor after the PWM output of the MCU to switch the polarity of the signal to negative. Then, use a P-channel FET for the output of the driver instead of N-channel FET. But, is that the only way? What about linear drivers? This method won’t work, because there aren’t two different polarities of AMC7135, right? AFAIK, there’s just the one, and it’s switched by a positive PWM signal, and output is a negative PWM regulated signal. I guess for those, a P-channel FET could be added after the 7135’s in order to switch the polarity on the outgoing side. But, that would make a much more expensive driver, and would be difficult to incorporate in a lot of linear driver designs because of space constraints.

This isn’t just a theoretical problem for me. I actually have a design I’m working on which already uses two-pad LEDs. Since they’re remote mounted, grounding isn’t an issue right now. But, I’m thinking of adding remote regulators as well, in a branched way that would divide the LEDs into smaller groups for better management. I want the driver to be as simple as possible, but I’d need to either run a third “PWM signal” wire, or change the driver polarity. For what I’m doing, changing the driver polarity is probably going to be the easier fix to implement. Anybody have any ideas?

2

Do you have a diagram? I’m trying to see how this would eliminate a wire. Are you thinking the ground is connected through the host body and therefore a wire is not needed?

Edit: oh ya, the a2 regulates the positive side.

3

I only scanned your post but here is the basics of why we switch ground instead of positive.

You can switch the positive side of the LED no problem, a lot of (most?) buck/boost drivers do this already.

The issue are the components to actually do the switching. 7135’s will not work with a positive switched setup at all.

Right now we use N-channel mosfets which only work with negative switching. For positive switching you need a P-channel mosfet.

This is the largest reason for using ground switching. N-channel mosfets are cheaper, lower resistance, faster reaction and just generally better then P-channel in every way. Now we can use p-channel just fine but it is easier to use n-channel.

If you use a p-channel and adapt the circuit to use it then it would work just fine.

4

no matter which polarity the body has it will always wont work if one of the LEDs has an electrical contact on the DTP board

That is also the reason why Mosfets with metal case have to be isolated to the cooler as others have the opposite polarity on their case

Having the negative on the body
With an N type Mosfet or 7135 you get a forced unregulated DD, if your switching channel gets shorted to the body
Even worse with P type Mosfet you get a direct short circuit of the battery, which will end in a destoyed light and battery

No matter if N or P type regulation you have always to wire 2 thick and one signal to a remote switching unit

5

In order to make a remote regulator work, the driver output needs to do switching on the positive side, while the regulator itself turns that into switching on the negative side. If you’re tagging a remote regulator to a normal driver, there is no switching on the positive side of the driver’s output, because it connects directly to Batt+, not to PWM. The reason for that is current handling. If the MCU PWM output pin could handle large current, we wouldn’t need regulators at all. The whole reason they’re on the board is to do that heavy lifting. So you need to run another wire from the PWM+ output of the MCU to the remote regulator if you want it to be able to switch anything. If you look at slave boards that are readily available, you’ll see that at least three wires are specified in order to make them function properly.

6

Wait, what? Why does using a P-channel MOSFET on a driver automatically make it a direct short circuit of the battery? Are you saying that there is no way to successfully design a driver which uses a P-channel MOSFET?

7

My bad I thought the Mosfets metal side would be as well contacted to the body for cooling, but if that would happen the battery would go pop

8

Lexel I don't what you're saying. You can definitely have a DTP electrical contact to the body and do P-channel switching.

I also don't see any reason you couldn't turn the battery upside down, and make a driver that still does N channel switching! Who says the case has to be negative? Would it be tough to make clicky contacts work well that way?

Is there a greater risk of shorts then? I don't know, maybe. Doesn't seem so different. You can always get shorts. There's no inherent short though. There's nothing truly stopping it except as ACE said, worse transistors. Even in the buck we designed, I ended up wishing the driver had been setup for N-channel, and even looked at P to N FET drivers to do it.

Anyway, DavidEF I think you were the one who told me this has been done before when I asked in djozz's 219 copper thread, and it has been, but after working on the texas buck design, I get why it's not usually. TA has the answer.

I'm also going to paste this from another time I wrote it:

https://budgetlightforum.com/t/-/25978

The clear conclusion is overall DTP matters more than aluminum vs copper **BUT** up to 3 Amps, copper without DTP still does just as good as DTP (ok maybe there's a 1% difference or something). The only thing that doesn't work so great at 3A is Aluminum without DTP, at least going by those tests.

Of course this may depend on the particular copper board.

... but of course who at BLF is happy with 3A, right? (Any of us who are probably wouldn't admit it openly)

9

If you want runtime, no overheating and efficiency, 3A for an XM-L(2) or XP-L(2) is great.
Keeps the Vf low too, so you can enjoy regulation for the full capacity of the battery (without loss of output under (say) 3.7 Volts).

10

I just want to clarify what I’m saying, in case I’ve miscommunicated somehow. With the body being solidly connected to ground, it is always negative. Right now, our drivers are switching and/or regulating power on the negative wire to the LED. So, the positive wire going to the LED is always solidly positive. In order for this to work the way we expect it to, the LED- has to be electrically isolated from the Batt-/GND through the driver. This isolation is what allows switching/regulation to the LED. But unless I’m missing something, it isn’t the only way switching could be done. There is still the Batt+ side, which is already always isolated from the Batt-/GND anyway, and could also be switched/regulated if the components were changed to operate that way.

The switching/regulating components we’re using now (AMC7135 and N-channel FET) only take input from the positive side and output to the negative side. So, they’d have to be replaced with components that take input from the negative side and output to the positive side. I don’t understand how this could ever be a problem in itself. But, drivers like this have never been made here at BLF. I think the reason is simply that it’s easier to do it the way we’ve been doing it, because the AMC7135’s we’ve always used don’t have the other option available. Now, since we’ve already been using FETs a lot in our driver designs, it seems to me that it shouldn’t be as big of a deal to flop the output to being switched/regulated on the positive side. I’m not saying to make all of our drivers that way, just make some for people like me who could use them.

11

Yes, side switch can be done, but you will still need 3 wires with regulation on either side. So I don’t think there is any gain in switch the regulation. Unless I’m not correctly visualizing how you plan to wire the remote regulator.

12

You can use N-channel MOSFETs for High-Side switching. Look up “bootstrapping”. Also there are ICs that drive N-Channel Power MOSFETs on the High-Side with internal charge pumps.

13

Reversing the battery orientation to have positive on tube with DTP on negative LED connector is a direct short if the insulation fail, so no good idea
Normal batt orientation just gets a DD when shorted

What I would like to see on high power lights designed for unprotected cells would be a proper sized SMD fuse in the tailcap to protect the battery, while allowing normal use, even if it adds a little resistance

14

LegolasGT presented his version at TLF a while back. link

15

If there's a will there's a way. You could use a plastic sleeve too, but yes, that's getting pretty messy. It's a good point.

16

Yes, you can, but you need a charge pumped fet driver or something like it as nothing in the circuit has a enough voltage to drive a high side n-channel FET. It's not just about polarity. When you're already on the high side, there's nothing higher. And that's one more thing to squeeze on already packed tiny boards. It's possible though, just have to choose your trade-offs. I looked into for the buck driver and basically we ended up deciding it wasn't really worth it, but maybe part of "worth it" was just the complication. Some was maybe cost. Some was space. It wasn't that we found it to be impossible though.

17

Well, maybe we should consider stacked driver configurations, like PD has already done. Would be nice to have some vias (to plug male jumper wires into from the spring side) for real in-system-programming.

18

I wasn’t considering reversing the battery polarity, just switching some components on the driver so that the PWM signal ends up on the positive lead instead of the negative lead. As it is now, the drivers we use send two leads to the LED (s). One is a positive lead which is directly connected to Batt+ and the other is a negative lead which is connected to the PWM- output of the regulator(s). What I want to do is reverse it so that the negative side is connected directly to GND, (just like the host is, so no problem for two-pad LED’s on DTP boards) and the positive side is connected to a PWM+ output from the regulator(s).

19

As it is now, the LED+ lead is solidly connected to Batt+ and the LED- lead carries a PWM- signal to the LED (s). I want to reverse it so that the LED- lead is solidly connected to GND and the LED+ lead carries a PWM+ signal. Now visualize the LED- and LED+ leads going to the LED (s) from this “reversed” driver. Strip some insulation from the LED+ lead and solder it to the PWM+ pin of a 7135 chip. It still feeds the LED+ directly, but now also feeds the PWM+ of the 7135 also. Cut apart the LED- lead and solder the incoming (from driver) end to the GND pin of the 7135 and the outgoing (to LED) end to the output pin of the 7135. There are still two wires from the driver to the LED. One of them is “tapped” and the other is “interrupted” by the remote regulator!

20

Bump for the ‘night’ crowd and/or all those in another time zone. :innocent: