New Driver: PD68 "DoubleDown" - 17/20/22mm FET+4

For FET drivers it’s important to stick with reasonably low Vf on the protection diode. This keeps gate voltage as high as possible.

wight, can you explain the need for the higher gate voltage? Does it matter if you aren’t PWMing the FET?

Sure. Gate voltage corresponds to RDS (ON), right? The MCU drops a lot of voltage (the ATtiny13A is spec’ed at 0.7v drop I think) on it’s own. That’s after the protection diode has dropped whatever number, maybe 0.2v. Subtract that from your battery voltage - which at high drive currents is sagging.

For an example, let’s imagine a light where the battery voltage drops to 3.7v under full load. Subtract 0.2v, then subtract whatever the MCU drops (0.7v?)… so gate voltage can be no higher than 2.8v at this point. Look that up on your RDS (ON) graph and see what’s happening… we are probably flirting with poor FET performance at that point.

I think you may be overestimating my understanding of the subject.

I understand and can follow the voltage losses going through the diode and mcu before getting to the FET gate, but I have no idea why that needs to be above a certain level or what “RDS” is.

From reading discussions previously I somehow came to the assumption that that only mattered if we are PWMing the FET. Are you saying it always matters?

Yes, it always matters. Take a look at Figure 8 on Page 7 of the PSMN3R0-30YLD datasheet. Another “telling” place to look is Figure 10 on the next page.

A minute or two looking back and forth between those graphs and the back-of-the-napkin math will likely get your mental cogs turning! :GRADE:

EDIT: RDS is the “Drain-Source” resistance, it’s very important for our applications. It must be kept as low as possible in a DD driver.

So let’s see if I understand this (in extremely simpleton terms). I had assumed that applying voltage to the gate pin on the FET just served as a notification to the FET like, “Hey, FET! Open the door!”.

But really, the voltage on the gate is what is actually “opening the door”? So if the voltage is too low on the gate, it will take longer to “open the door” and it might not be able to “hold the door” all the way open, causing extra resistance?

edit: BTW the graph was helpful. I never realized how close we were operating to the cliff on these voltages

Is Rfet needed here? If so, what is the suggested value?

Thanks Matt

I thought it would be. We are that close to the cliff!!

You are correct, the voltage is “holding the door open”. We never get the door 100% open (zero resistance), but we can achieve very low resistances. As has been pointed out to me in the past: at low battery voltages the resistance becomes a moot point because we don’t have enough voltage to drive the LED hard anyway. Ideally I’d like to see decently high drive currents through 50% to 75% of a battery if possible. I’ll certainly take whatever I can get, but a lot of that depends on the LED Vf in most cases. At a higher drive current the problem is exacerbated. I’m not sure how that fits in to the “door” analogy, but it doesn’t really need to - you get the idea.

Eyeballing that “knee” in the graph is one of the things which motivated me when moving the battery voltage divider. Measuring “before” the protection diode means that we can put in whatever diode makes sense without having it drastically affect voltage monitoring.

Your assumption is a common one. I have made the same assumption in the past. An FET seems a lot like a solid state “relay” but it’s really not very similar at all. What a letdown, right? Carefully looking over those graphs is what makes DD FET selection take forever.

Home work: OK, I kept this one in reserve. It’s an easy assignment, don’t worry. Now that you’ve achieved a handle on this concept… go lookup the same graph in the SiR800DP datasheet.

Well that explains why some people love it. Still not sure if I’ll jump to using it though, it’s a whole $1 more! :smiley:

Thanks for the education. :GRADE:

a $1 more? Where? Not from Richard: $0.80 —> $1.35, or am I missing something?

Received my boards today looking good PD

Looks like there’s about an 80-cent difference per piece when purchasing QTY=10 on Mouser or Digikey.

Will the parts from a Banggood A6 driver work on this board?
If not, how can I get a controller that is already programmed?
(After using computers heavily for almost 50 years I don’t feel like learning how to load the programs in another system.)

Just finished the basic build on the 22mm version of this board, it caught me by surprise with a new set of pads called Rfet. I did not populate it because I am not using the Vishay FET, I am using the standard LFPAK56 FET on this build. From reading many threads, it appears that the Vishay FET has a better state of high conductivity/lower resistance during a nearly fully driven state and it was causing the issue with the full to moon flicker. I also did not include it because I could not find a reasonably solid value for that resistor, it looks to me like 12k ohms might be the desired value with the Vishay… Do I need something on the pads for the driver to work? I could use a little guidance here, looking at the board layout, it looks like it is optional with the traces in the area of the FET.

Anyway, on to the board! I think I am finally getting the hang of building these well the first time. I have simply been using to much solder paste. I think this is my best build yet! :slight_smile:

!

!

Matt

Looks good! R-FET is definitely optional. What host is getting that big thing?

Going in this old Ultrafire. Do not need the biggest, baddest FET. I assume Rfet is needed if the Vishay is used? BTW, did ya catch the MCU? tiny25 with Bistro and a bleeder!

!

!

I reflowed my first 17 mm PD68 DD v2.1 board last night, using just one 7135. Made one mistake - thought I’d apply solder paste to the LED- pad to pre-tin it for later, but with all that solder paste on the single pad of the FET output, when the solder melted, the FET slid up to the round LED- pad and just did not want to cooperate in positioning… It was a nightmare for a while, til I could clamp down the FET and apply more hot air to re-melt the solder paste.
This was with a SIR800DP FET. Pretty sure it was my own dang fault for using too much paste on the FET output padding. The spacing for this FET though is really, really tight to the LED+ hole. Anyone else have problems or issue fitting the FET on there?

I buzzed it out and the way I positioned the FET turned out good, but kind of slanted it around the LED+ hole. Think’n bout it now, I could have sanded/grinded off the small tab on the side of the FET that is very close to the LED+ hole… Sorry, no pics available right now, though I took pics.

Haven’t tested the driver yet, but it has a ATTiny85 on it, bent pins, and will go in a ThorFire JM07 running the latest Narsil firmware (full programmable UI). The JM07 stock driver board is all prep’ed for it already (stripped, JB Weld added to support the driver mounted e-switch, switch wired to pads).

Tom, got my JM07 yesterday in the blizzard. I like it and may leave it stock, no it is not as hot as my other 26650 tube lights but I do not think it needs to be with the OP reflector. I think it will replace my hopped up C8 in the truck, in favor of runtime. It is the only light I have that will fit the RadioShack protected 26650’s I picked up, dang they are long. I just got in my first 2 SIR800DP FETs and my first 2 tiny25’s this week. I may now have to take about the D01 pill and install that FET on the board!

Tom, I know the 17mm Rev2.1 is really tight around the hole. I actually don’t like to use that board myself because of how far it pushes stuff out to the side, I mainly made it at the request of another user.

That and I’ve never had a SIR800DP , I always forget to plan for their different case shape.

Cool JM07, literally! I got my 1st one on Monday I believe, still waiting for the 2nd - tracking #'s are totally useless - both tracking #'s say left China, USPS never got them. I'm not a fan of power ON/OFF at the tail, then mode changing on the side. Narsil supports both switch's, so if I want to, I can configure mode switching w/memory on the tail and have full control on the e-switch from 1-7 modes, lock-out, battery status blinked out in voltage, etc. - all the features, plus full FET power .

Stock throw I measured was 26 kcd and 3.0 amps on a top IMR 26650 cell (EFEST). Throw is great for an OP reflector light of this size.