LED drivers and Accessories you want, but don’t exist

Sense resistors are a hurdle with regards to efficiency and power handling. MOSFET VDS feedback is a far more reasonable way of sensing. It seems that there already are some commercially available noRSENSE controllers. Look, for example, at this little beauty: LTC3785 - 10V, High Efficiency, Synchronous, No RSENSE Buck-Boost Controller

Cheers :-)

P.S.: that sh1t owns!

I did share some 85 firmware for boards some time ago and shared a couple of boards for it because there was interest, but the problem is that my firmware is locked to my boards. I spent quite a bit of time writing instructions so that people could use it, and in the end it turned out that no one was interested enough to actually read any of it. I totally understand that, most firmware here is compatible with boards from multiple designers, but after that I just don’t bother because it takes too much time to do write ups that hardly anyone will read.

Sure, once the boost circuit discussed here is tested, I can attempt to make a two layer version in Eagle with the vias. If I succeed I can share it with some basic firmware, but as I only work with the 1634 and assign pins to my liking, I kind of have a feeling that a Bistro compatible 25 version will be much more of interest to the community.

In terms of new chips, I think there are a few that are waiting on the 1616 and/or 1617. I didn’t want to wait. They aren’t easily available to me, no AVRDUDE support and higher pin density than the 4x4mm 20 pin package. The 1634 is available, package size is as small as I want to go, 16KB programming space, AVRDUDE support, 17 usable I/Os and has 0.1 uA parasitic drain in power down sleep mode. That’s all I’m interested in really. I don’t use the Full Duplex USARTs, QTouch, I2C interfaces and all that other stuff. If anyone wants some help migrating to it, I’m here, but for my own boards and firmware I’ll just carry on as usual.

@ Mike C:
Hi Mike,
would you mind to share the best drill size of the vias for your programming wires?
I’m currently designing my first flashlight board and intend to use this technique if my additional attempt with pogo pins doesn’t work as expected.

I’m using 0.9. I had 0.8 before but sometimes via sizes are not exact in OSH Park fabrication, so on some boards 0.8 was too tight. Note that I might not be using the same pins as you. I got some pins I bought from Fasttech that work but wanted longer so I’m also using cut up copper paper clips.

0.9 can be a bit loose, so I have to take care when flashing. I’ve bricked a few MCUs on driver boards because of glitches when flashing. Now I use a development board for firmware development where I have soldered on the connections, and when flashing actual driver boards I’m very careful to test the connection multiple times before actually flashing. I don’t have room for those 3x2 connectors, so this is how I do it.

Thanks!
I’m designing a Q8/SRK board, so there is space for a 6x1 row of connector pads. Found some info about pogo pin board design and will try this also.

http://blog.spitzenpfeil.org/wordpress/2013/11/12/the-pogo-key-1-27mm-avr-isp-adapter-take-ii/
https://easyeda.com/feather/ispogo-oA5nFXR93

I had a similar idea but a round one that would match up the vias on my 17mm boards, but for each new design they change places so I just skipped it. For a SRK sized board it’ll work fine.

Funny asking about Vds control (to regulate normal FET drivers) was my almost my first post here, got a lot of negative reaction. Some of that was about software though, which I get now (although easy enough on bigger chips). There's also sensefets, which is a little different I guess. FM I don't know what you're designing, but there is a buck design for SRK already, never tested though. It's slightly degraded by the controller needing PFET's, but it should still hopefully put out 12A to 15A at 6V or maybe even 12. That's a bunch of power.

So Flintrock, you have a version of Bistro-HD that will run on the MP3431 driver Schoki made? This is very exciting to me. As cool as the Narsil ramping is, I ended up changing my Narsil flashed TA driver back to standard mode groups in my Convoy L6, I found that I never really used ramping and appreciated having a few set modes more.

Personally I think between 3 and 5 modes is about right, 3 if just L-M-H, 4 or 5 if there will be a ML, Turbo, or both.

High should be set as the maximum continuous output the light can sustain thermally, with low and medium appropriately spaced based on that. Moonlight should be about as low as the driver can handle with turbo as high as the emitter or driver can handle.

Does Bistro-HD have temperature compensation built in? On my drivers I have a thermistor mounted on the driver PCB and one remote mounted on the MCPCB and would like to use both of those to automatically ramp down the output at set temperature limits.

I’d also like to potentially ramp down the output based on battery level at the bottom of discharge rather than just a hard cutoff.

Not sure what an M3431 is honestly. Is that the boost IC? He asked about a firmware for this board shown on the last page or so, using an attiny25. Yeah, I have a firmware config that should work fine on that assuming the hardware is up to the task of OTSM at those low voltages.

I have simplified temp compensation in HD (although it can be reconfigured easily for the old way) so that it just steps down to a sustainable level if it gets to hot. No attempt to regulate (previously not well usually). You can instantly tap and get a 10 second extension, but the idea is that you set the temp threshold really at a safety level, so you wouldn't want to, but it's only 10 sec. You can also wait for it to cool a little and tap and get more than 10 seconds, until it gets too hot again. Or of course can tap twice and click out to another mode entirely.

Voltage control is the same as original bistro. It's pretty dynamic. Yes, if high current pulls down the voltage, that will cause it to ramp down to a lower current. I'm not sure how well this stabilizes in practice, haven't done much testing with it. Feedback is always a very tricky thing to really nail down.

I tend to agree, a few modes is fine for me, and having some idea of power output is also useful.

I'm not that keen on doing ramping, but it could happen possibly, eventually. Who knows. TK already has that I think. HD is just really easy to configure for many different board configurations.

I’m wanting to build a phatlight so a high amp driver would be nice 9a+

Those already exist:

https://pcb-components.de/led-abwaertswandler-buck-step-down/ampere-5000ma-oder-9000ma-7v-25v-detail.html

Well that first one is a complicated example. That's >9A at 12V even. not quite sure I'd say it "exists" either. Nobody made it yet as far as I know. But just the basic stock Q8 driver does 20A at 4.2V.

Not sure what's with the guy grumping about driver design being uneccessary on a forum where people design drivers.

NightHike before you get a 6G toaster driver, maybe you can try to fit code to regulate output of low voltage emitters that works on people's existing attiny25 boards so they can use the latest LED's without frying them. Is that too easy for you or too Oracle?

@Jensen, so yeah, the MP3431 is the boost IC. You don't run firmware on it. It has an MCU controller, the qfn attiny25 in this case, and yeah, bistro-HD runs on that, with a trial configuration for that board ready.

But here it does. Think about it.

Can somebody tell how many vias from one board side to the other I need for a given current? Which diameter?

Just wondering if you order the 25V then on your board is the diode after the LDO missing for OTSM

It doesn’t need a diode if you get the right LDO. I have the TPS706, which has a lower reverse current than some diodes (10nA out of IN pin, 100nA flowing into OUT pin). Here’s the datasheet: Datasheet TPS706

And for the current per via, there are some websites, just google it. I used roughly 1A per 10mil via on the plus side.

Yeah, you have to check diode current when hot, not the quick-spec sheet, as it can be wildly different (exponential with temperature). The plots look like about 2uA. That should be good.

I just posted 1.7.1 of Bistro HD. https://budgetlightforum.com/t/-/44344

The OTSM-LDO-fetonly hex should work for your driver, although with no real moon mode solution yet (I'd prefer the hardware two-PWM channel solution to some PWM hack, but we'll see). I guess I should compile an OTSM-debug hex for you too, but we can see how it goes.

Thanks!

Found some information meanwhile:
http://circuitcalculator.com/wordpress/2006/01/31/pcb-trace-width-calculator/
http://circuitcalculator.com/wordpress/2006/03/12/pcb-via-calculator/
http://www.blackstick.co.uk/pcb-design-calculators/