[WIP] 17mm DD+single-7135 driver / single sided / Dual-PWM

That’s the same thing I discovered, and I wasn’t able to do much about it in software. It seems led4power ran into the same issue, and used a separate power channel for the lowest modes. The FET is incredibly unstable at the low end.

Up front: Sorry if I miss anything folks - I don’t have a lot of time so I’ve skimmed the last 10+ posts.

I used the ATtiny13A for a couple of reasons: (a) because I’m rusty and I wanted something I knew I could program properly. (b) I also knew that it would be stable regardless of the hardware, so I’d be able to move through the modes properly and view them on my oscilloscope.

Having done those things, it looks to me as if this driver is ready to use with the more sensitive ATtiny25/45/85 MCUs. The extra components are present for a variety of reasons - one resistor is for the tailcap LEDs, one is for OTC pulldown, one is an FET gate pulldown resistor, and then I think the rest are there for regulating the voltage to the ATtiny25/45/85 (spike reduction). For more details look at #’s 301, 314, 326. I’m sure there are more details in there somewhere. (I’m under the weather today and my mind is elsewhere to boot. Sorry!!)

R5 hasn’t been established, but Sharpie made a suggestion a few posts ago. I left the OTC pad at 0805 with the intention of testing moving to 10uF w/ a pulldown. Sharpie suggested 100K. His suggestion was also to change R1/R2 out for 10x the resistance, which I find to be a very reasonable suggestion. (so approximately 191k / 47k) Rather than bypassing R3, 10ohm might be a good start. Sharpie seemed to think so too. I still haven’t ordered 0603 sized components, but once I do I’ll test that myself.

Looks like you and comfy may be onto something with that SiR404DP. Good work.

[quote=fixed it]

Unfortunately I think you’re exactly right. In order to avoid a large increase in component count it may be possible to use a current sense resistor and measure it with ATtiny25/45/85 using the 20x gain setting. The ATtiny13A doesn’t have this 20x gain setting and would be unsuitable. I assume that an RC circuit would be used to maintain the voltage on the gate pin.

lol

[quote=wight]

Yeah, I can build a UI I’ll be happy with within those restrictions but it won’t be something most people will like.

Would people bypassing springs and trying to find FETs with slightly lower resistance accept a current sense resistor? I really don’t know. Haven’t worked out the values needed either as it’s not something I can easily do in any case.

What I might try, as it’s a simple change, is to connect the LED negative to an ADC input and try keeping a stable LED Vf. I’m thinking I could self-calibrate it with the 7135.

As for voltage on the gate, that would be easiest indeed as you could use built-in PWM. Or you could do it with just a large enough resistor to ground, using the gate’s capacitance and the internal pull-up resistor of the MCU. Somewhat harder to program though. It can even be done with none of that by fully draining and recharging the gate every time a change is needed (what I did in my testing). It takes about 50 µs so it should not be noticeable.

Read this piece of ancient history. Feel free to laugh.

It does work, I've built the circuit with a 10K pot (higher would be better, since it's wired direct between B+ and B-) as texaspyro describes there. The FET does get uncomfortably hot in the upper half of the range, like between 70-98%. At WOT it stays cool.

I did come across that thread when searching for previous similar attempts. And no, I did not laugh. I actually tried something similar with good old transistors about 10 years ago, trying to put a touch sensitive switch on a project. But I’m getting off topic now so I’ll spare you the details of that failure.

My first build failed. Nada, nothing, zip. I have a short across a couple of components due to not having proper sized smaller parts. Will punt later. Today has been a grandpa day, 1st time for everything I suppose. They don’t know it yet, but I sprinkled some small resistors and XP-E2 emitters in the diaper bad, see if anything grows there… :wink:

Though I admit that I am week at circuit level electronics, this isn’t an FET from a typical budget flashlight that overheats if you take its series resistor out. It is bigger than the LED and, properly cooled, it should be able to absorb a comparable amount of power. I didn’t get this out of the blue. I don’t remember which thread but the EE who suggested a gate resistor said it could absorb power. If the gate resistance of the FET is high enough, then probably a gate resistor is better, using the FET’s gate capacitance.

I have tested it with the big beefy TO-252 Vishay 70N02 and it gets hot enough to burn you. FET hasn't died yet, but what kind of a reduction in lifespan would be acceptable to you, to get something that can be had by other means that doesn't cause components to fail?

I’m real happy with using PWM level of 2 on the single 7135 of a FET+1 driver. I can now set the moon mode PWM value via the UI from 1 to 7 in my latest Narsil version, and 2 has been working well so far, even on a low cell. It’s soo nice now to set the PWM level of moon via clicks to see the differences easily, in the the light.

Nice. I always end up having to add a bunch of moon modes then test them in a light box to figure out which one works best, then remove all but that one. I can get it reasonably close just by shining it next to a reference light, but the light box is still necessary to make the final choice.

A value of 3 seems the best for the lowest, stable value for a 350 mA 7135 and that’s what I’ve been using lately, and I’ve found you have to raise to maybe 5 for a 380 mA 7135, so I’ve been using 350’s now exclusively. What value to use for moon mode has been bugging me for a long time, so best to make it configurable, just wish it was easier in the UI to adjust. Right now it’s in the configuration settings, but would love to be able to tweak it while in moon mode. Sometimes I like it super low (2), other times 3, or higher at 6 or 7 is nice to have. Should treat moon mode as a separate state with it’s own button actions.

My configuration settings are now stored in 3 bytes to cover the 10 different setting values.

It also depends on the exact 7135 chip and the emitter, and even how many. Some 350mA chips work better at PWM=2, others work better at PWM=3, and the 32x7135 SRK driver lights up even at fast PWM=0. Older emitters require a higher PWM level than newer ones, usually. The highest I’ve seen was an old XM-L which needed PWM=9.

If I have an e-switch I don’t bother with specific modes. I just use a smooth ramp from moon to maximum. On a single-channel driver (or tiny25/45/85 FET+1 with a slightly different pin setup), the low end can have extra-fine adjustment by using PWM+PFM, so you can get lots of precision in the moon range instead of just a couple levels. I haven’t actually seen a driver set up with two channels on independent PWM counters though, and now I kind of want one.

Some happiness arrived in the mail yesterday. Those 0805 SMD pads look soooo tiny.
I’m ok with learning to reflow the one sided boards but how do you reflow a two sided board and not lose all the parts off of the other side?
Is it time to add a hot pencil to my soldering station ?

Surface tension of the solder is greater than gravity. They stay floating in place on their pads just fine. Unless you bump it while the solder is still liquid. A big heavy part with small pads can sometimes fall off, but that's why you do the side with heavy parts last (or hand-solder them afterwards).

Cool! 8~~) I had been wondering this for a long time as well. But, I’m not doing any re-flowing yet, so I never asked the question. O:~~)

So assuming I use a toaster oven,

  1. I would populate one side of the boards
  2. Ramp up the heat.
  3. Hold max temp till reflow is complete.
  4. Ramp down till cool.
  5. Flip over, repeat.

The other less sophisticated method that I am considering is to place the populated boards on a solid bar of copper and heat from below with a torch. Then again, maybe the copper bar would only work good to reflow LEDs.

Spend the ~$50 for a hot air station. Search Amazon for '858D'.

Hot air is so much more versatile than an oven. You can heat just one area and pluck off one component without disturbing anything else or re-melting all of the solder on the whole board. It's also easier to control the total heat input - time can be just as critical as temperature. With the board right out in the open in front of you, you can watch the paste flash over and then remove the heat right away. Hard to do that when it's inside an oven.

About fifteen years ago I bought a Weller WSR1002 Rework Station. I never purchased the hot air pencil for it. It’s dis-continued model now so I don’t know if I can find hot air pencil that will work with it. The 858D looks like a lot of tool for the money so that’s my backup plan.

I’m not personally familiar with the industry, but I believe that mass produced boards are sometimes done using two different temp solders. Neat-o.

Personally I do my reflows using a large 1500W heat gun - mine is the MHT3300. Those run about $40 and are good for other types of project. They are definitely not as flexible for soldering work as the little stations. I do own a hot air station, but never learned to get the proper use out of it. Since I’m space constrained it’s in storage, unlike most of my other soldering gear.

At the end of the day hobby reflow will probably still require skilled soldering touchup - especially as you learn the ins and outs of reflow. It’s quite possible to solder the entire boards by hand, take a look at my videos here: Nanjg stripping and A17DD-SO8 building videos

I may have to get one of those…

Any idea if it would run on a 15A 120V circuit which is already at ~30% capacity? Some reviews suggest it may need its own dedicated circuit or even a 20A circuit. Pretty big change from my weak little 40W soldering iron.