Discussion about FET vs 7135

Ok everyone I wanted to start a discussion about FET drivers vs 7135 drivers. I was curious about a few things and Think there could be some good posts on this topic. First off let me list some assumptions I have, hopefully I’m not too far off the mark.

1. If your reading this and spend any time on a flashlight forum then you are serious about your lights and are not the average consumer, i.e. Just buying the first flashlight you see on the shelf, or buying the cheapest thing you can find at walmart (and not modding it) is not your style.

2. Again if your reading this you like your lights to be bright!! Willing to sacrifice runtime and comfortable operating temperature to get as many lumens as possible

Now those aren’t absolutes but I think you get what I’m trying to say. I get the impression from most of the posts I read is that the majority of people seem to prefer FET drivers when buying or especially modding a light. My preference is for a current regulated driver and I feel like I’m in the minority on that one. I feel that FET drivers get too hot, not that 7135 drivers aren’t hot, but in my limited experience not as hot. Of course it depends on the situation but going direct drive with imrs gets intense. FETs also seem to really drain the cells fast, again each case is different but letting the cells push the light as hard as they can really hurts runtime. I also don’t like pushing the emitters so hard, LEDs are cheap but I don’t want to shorten their service life too much. Basically I feel the output increase is not enough to justify the extra heat, lower runtime, and shorter lifespan. Maybe I’m just being a puss or I’m totally not understanding something.

If you disagree, and I think most will, why do you prefer FET drivers? Or if you prefer current regulation, why?

No wrong or right here just personal preference. 7135’s allow you to LEGO up to a certain top end somewhat below maximum cell capacity and fets take you straight to the top. Also, because the voltage drop across 7135’s is usually a bit more than across a good FET the FET will have a higher max output(I think). An FET sure beats stacking a bunch of chips to get a 6A driver.

You are getting that impression because FET's are currently the hot trend here as we always like to push the envelope in any way possible As RBD said above, FET's allow a bit more voltage to get to the emitters. Most of the FET lights being built are reserved for the builder him or herself or for someone with enough knowledge to safely use them. Until more reliable controls are readily available, I certainly wouldn't give a FET light to a novice or even use one myself in a situation where I am not actively monitoring the light.

If you’re stacking more 7135 than can all saturate with a fully charged cell, it is better to use an FET. If you want constant output, for a while at least, without changing modes to manually regulate the current, then you need 7135s.
Cheap flashlights usually have direct drive or only resistors to limit current, while expensive ones usually have regulation. So FET drivers are, in a practical sense, a step backward, but they make it brighter.

One of my simpler mods is to solder the ground tab of one or two 7135’s to the bottom of a 10mm SinkPad and mod a solitaire for it. It could also be done with an FET but the solitaire won’t take the heat produced. You have to at least give passing thought to the what you’re ending up with.

I don’t understand. What is a solitaire? What is the 7135 doing so far from the rest of the driver? Why does an FET that acts like a wire produce more heat than a 7135 that acts like a resistor?

Maglite Solitaire. There is no driver, just the 7135 chip(s). The FET causes more heat to be produced in the led because it allows way more current (~2.5A from a 10440 vs .35A through a single 7135). Heat is a function of voltage drop and current. In both case most of the heat is generated by the led but the FET allows more current hence more heat. The 7135 acts like a resistor with a variable voltage drop but the limited current it passes keeps the led much cooler.

Since we have high-current-tolerant leds-on-copper-boards, there's the option to direct drive them, making current regulation not a neccessity anymore, but a choice.

I don't think the lifespan of the led is so much affected by direct drive, and overheating has not killed any of my direct driven copper-board lights yet.

So using current regulation has become a preference, for limited but more constant output and increasing efficiency and runtime.

I personally prefer 7135s for all of my lights because they are often used in the lower modes. To me, getting lower modes using a FET with PWM is a bad idea. My own drivers use a variable amount of 7135s with constant current, and only use PWM on a single 7135 for those lowest modes. My brightest light (triple MT-G2) uses 16 x 7135 per LED, which is theoretically about 6A per LED. Using a FET for the highest mode is probably more energy efficient on the highest mode, but for the lower modes it wouldn’t be. When I’m using it for photography I don’t always use it on the brightest setting, and I can use cells with lower discharge capacity for those not so bright shoots because I can regulate exactly how much constant current is being pulled, instead of always pulsing what ever the cells can give on all modes.

I’ve considered using a FET on the highest mode and 7 or 8 x 7135s per LED to choose from for the lower modes, but I just like the idea that I can regulate constant current with 0.38A steps all the way up to 6A per LED (theoretically at least).

I’ve been looking at components for variable constant current like led4power’s drivers, but so far it’s a bit over my head so I’m sticking with the 7135s for now because they are so easy to use. For me the biggest joy of this hobby is using my own drivers and firmware, so I’ll continue to use the 7135s until I find a better solution I can use in my own drivers.

FET drives are not just FET drivers, a driver like the LD-2 works like a bunch of 7135, except it is better.

The advantage of a current limited driver (i.e. 7135 or LD-2) is that even a new super high current battery will not blow the led. The disadvantage is often that you loose some voltage across it (Not really the case with LD-2) and this will reduce the brightness slightly.

Any current limiting driver will also have to get rid of some heat (If it does not get warm, it is not really doing anything).

One advantage of using DD drivers if that you can pair a triple of low VF leds with a high drain cell and not worry about the driver overheating.

I use FET drivers in most of my “show lights”, and the ones which get real usage they have 7135 drivers, but soon to be replaced by LD-2. Flat regulation is extremely useful for outdoors, your night vision will adapt and even the lower modes can be considered bright enough.

I think pushing the boundaries of what you can get out of a flashlight is fun, but not always exactly what I want. So, I look at each flashlight as a separate entity, and I decide what I want from that one flashlight, and that’s what determines whether it gets FET, 7135’s, LD-2 style CC, etc.

For predictable output and run-time, I like the 7135 drivers. For the highest current possible, I think FET is currently king. I haven’t done a lot of personal testing yet (I still can’t even solder correctly). However, I tend to agree with will34 that the FET drivers are good for showing off, but the 7135 drivers are more practical for real-world use. I have a couple lights I’d love to have a FET in, but I won’t do it because I need those lights to be more practical, more dependable.

I’ve been thinking about doing something about that. I have a 17mm driver with 12 x 7135 (6 on each side), I’m thinking about making a version with a FET on one side, and use the 6 on the other for the more usable modes. I could probably squeeze in two 7135s together with the FET on that side, making a total of 8 x 7135s and a FET for boost. Maybe someone has already designed such a driver?

I don’t know of any, but then again, I’m a noob. :stuck_out_tongue:

The only way I would consider it more practical than a “normal” FET driver is if the FET was only used for the highest (Turbo) mode, or maybe the two highest modes. However, truly, FET driver based lights can be reliable and dependable, if used in the right way. I didn’t mean to make it such a huge difference that it really isn’t. But, there is a small difference in that FET drivers are not regulated. Mathematically, that makes them unpredictable. A change in the cell used changes both the output and run-time. Even the same cell, over time, will change properties and that will in turn change the properties of the FET driver(s) that cell is/are used with.

All that being said, I’d like to see your driver come to fruition, because I’m sure it will prove to be useful. I don’t know how this stuff works. Is there a non-PWM output channel that could be used with the two PWM channels of the MCU, to make a three-channel driver? If so, then a modification of the FET +1x7135 driver could be an awesome thing. Connect the FET to the non-PWM channel (if there is one) for Turbo only. Then, connect some amount of 7135 chips to the first PWM channel for all the middle modes. Then, the single 7135 to the second PWM channel for the lowest modes. Would this be possible? It might be the best of both (FET and 7135) worlds if so.

Using 7135s while the FET is on would be absolutely pointless. I would only use the FET for boost/turbo, and only 7135s for all other modes.

Yeah, the 8 pin ATtinys have 6 input/output pins. I use them all.
All 6 can be set to constant on/high output, and simple on/off inputs (like E-switches).
2 are PWM capable.
The remaining 4 are analog to digital conversion capable (which is what voltage monitoring and off time cap measuring uses).
One of them is the reset pin. Setting this to an input/output pin means you will not be able to flash it again unless you use high voltage programming to “reset” it. I have the parts for a “MCU resetter” (I haven’t needed it yet) so I’m not to concerned about it. If you are, don’t use this pin (PB5) and you’ll have 5 input/output pins to play with. Go up to the 14 pin MCU (ATtiny84) and you have 6 more. I use the ATtiny84 in a few projects where I need more than 6 input/output pins to achieve what I want, but it’s not very practical for 17mm drivers because of it’s size.

I don’t think it’s pointless. The BLF A6 GB flashlight driver uses a version of wight’s FET +7135 driver, and the firmware by TK has been tested by many and found to be pretty awesome! In that firmware, she mixes the PWM values of the FET and the single 7135 on some of the modes, in order to cover up, or fade out the color shift. Since the FET is giving full power to the emitter on a PWM cycle, and the 7135 is not, there is a difference in the color of the light coming out the front. This tint shift is well known around here, but AFAIK, it’s never been observed in a single driver/flashlight until somebody put DD and linear together, as wight has done.

So, does this mean you’d be interested in making a three (or more) channel FET + 7135’s driver like this? I’d love to see it done, but I’m not qualified to pull it off. :bigsmile:

I see. I didn’t know this, I’m a noob too you know :slight_smile:
I personally wouldn’t be interested in using PWM on the FET though. I’d use it for full on boost/turbo mode only.

I guess I could do it, I’ve never used a FET before so I’d like to try it out. However, I wouldn’t use a spring. I find springs a waste of component space and would use a design that many might not find very appealing:

I’ve shuffled some components around and have switched to the smaller 0402 sized resistors for the voltage divider, and 0603 for the capacitors, so my 17mm driver now has six 7135s on the MCU side, and eight on my 20mm driver.

With the three-channel driver, there would be no reason to use PWM on the FET, because the multi-7135’s channel would take care of all the middle modes. :wink:

I can see how they might find it unappealing. :open_mouth: I would ask that at least there be an open pad in the center of the board for battery contact, even if it wasn’t big enough for a spring. I wouldn’t trust the SMD components to hold up under pressure from the battery over time.

Then I’m not your huckleberry.

It’s only the MCU that would take any the load and it has eight legs to support it. The copper wire I use is solid and soldered to a via. It rests ever so slightly above the MCU, and would have to give way first before the MCU does. Regarding the pressure, I’ve used a smaller springs on tailcaps to reduce some of it. I’m not at all concerned. I don’t use the ATtiny13a footprint either. I now use only the ATtiny85 for “normal” lights, and with bent legs they take less space than the 13a so I’ve made a custom footprint for 85s with bent legs. Firmware is far from finished, all that is on hold while I’m working on another project.

So I guess what ever I end up making for myself might not be so useful for others.