Developing a customizable driver

67 posts / 0 new
Last post
sidehack
Offline
Last seen: 5 months 6 days ago
Joined: 03/03/2020 - 21:30
Posts: 53

When first powering up something new, I usually take bets between Work, Smoke or Fail. I never put money on Work.

But good news! The boost driver worked straightaway. I used a 0.1ohm sense resistor instead of 0.01 so the max current should be 500mA through the LED. At 100% output I was seeing 3.2V 1.1A on the input, which would be just about right for a 6V LED at 0.5A output considering losses and big resistor. I’ll up the current sometime and give it a full-range test when I can put it to a better PSU and better wiring but I’m very encouraged. The only other circuit I’m as excited to test is the big 4-switch buck/boost and I don’t have all the parts for it yet.

sidehack
Offline
Last seen: 5 months 6 days ago
Joined: 03/03/2020 - 21:30
Posts: 53

Been testing more scrounge cells. The other day while digging for something else, I randomly turned up an orange-label LG cell that had been scavenged from some old laptop battery sometime in probably 2016. It turned out to be a beast, measuring 2414mAh and clocking 71 minutes at a 7W draw. I only found one other cell over 2000mAh and it’s about 2140.

I also found a bunch from an old Dell battery that I had some trouble charging. After a while at 4.2V the current started going back up and making a lot of heat, so I charged them all at 4.1V and they leveled off. So far they’re running 1000-1100mAh, so pretty lousy but good for quick tests and stuff. I just thought it was weird they were going nuts with a 4.2V charge voltage. Anyone else seen that behavior?

Also. I know y’all are going to hate me but I like the 504b host. It’s a P60 host, so thermally limited without substantial modification. I really like the simple design, good sized head, no sharp edges, but I’m worried about putting any kind of high-spec driver in one. I wonder how possible it would be to build a batch where, instead of the brass pill tapering to fit into the tube, the thread was continued all the way down and receiving threads were cut into the tube to match. Instead of a drop-in, it’d be a screw-in, which should give much more contact for heat transfer. Might require extending the pill somewhat to get rid of the big spring, but that would have a side effect of deepening the pocket which could allow room for better parts on a driver. I assume someone’s already worked on something like that.

June’s going to be a big manufacturing month so I don’t know what R&D time I’ll have, but I need to get a full range test of that boost driver someday and especially see how it performs at low voltage, high output. Might have to mount my LED on a heatsink and fan since “high output” for it will be around 30 watts. I need to get a final design of the 3A buck driver PCB, integrating a temperature sensor on the driver and laying out ISP pins compatible with HQ’s pogo key. Should be able to do a layout for the boost driver at the same time and get some of both PCBs made.

I think if I can put together a buck-boost that will output from 2.7 to 6.5 volts, and input from 2.8 to 8.4 volts, will be really fun. It’d be pretty universal, able to run 3V and 6V LEDs from 1S or 2S configuration, without modification. Slap a good current rating on it and it’d do almost everything.

I’ve enjoyed working on the controls systems for flashlight drivers. I’ve got several other design projects going on at the same time, and it’s fun to learn a new concept for one thing and see how it can be added to another project as an improvement. I’ll be using the temp sensors I found for these lights on another thing, and the improved voltage adjustment methods; a lot of the PWM, ADC and RAM/EEPROM access routines in the flashlight’s micro were borrowed from that project. I should juggle complex jobs more often.

sidehack
Offline
Last seen: 5 months 6 days ago
Joined: 03/03/2020 - 21:30
Posts: 53

Thinking I might drop the 3A buck in favor of a 3A buck-boost for the entry-level driver if I can keep the cost reasonable. I would really like to be able to give someone constant brightness independent of battery voltage.

Still haven’t had time to code temperature throttling, or do full-scale load testing on the boost driver prototype. Might get to it this week, but probably not. Did get some time to work on a driver PCB layout for the boost driver, so progress was made. Course that progress may be moot if the circuit doesn’t hold up to full power testing.

Side note, I’m using my experience refurbishing LCDs and newfound LED knowledge to build an HD LCD movie projector out of an old overhead projector from university surplus, so family in the neighborhood can do drive-in movie and BBQ nights. Testing with an XHP50.2 today but it’s getting too hot (6V; starts turning blue above 5A) so I’ve got some DTP boards on order. Got a lot of light spill and overall low transmission efficiency so I may have to cook up a reflector; may just hijack a spare off a P60 dropin or something, at least to test with and see if it helps. The lensing on my zoomie housing also makes a really even spot so that’s possible too. I need to cover about a 10-inch diameter area to light up the LCD.
Regulating 12VDC down to ~6V will be a nice change of pace and more what I’m used to; before this flashlight project I’d never worked on boost and buck-boost circuits.

sidehack
Offline
Last seen: 5 months 6 days ago
Joined: 03/03/2020 - 21:30
Posts: 53

Finally got around to full-load-testing the hefty 6V boost driver. With a 50mV sample range and 0.011ohm sense resistor it should peak around 4.5A into the LED.

With current limiting set where it is, I was able to get 100% output into an XHP50.2 LED from a 4V input. Thresholds for 90% at 3.6V, 80% at 3.2V and 70% at 2.9V

By the datasheet calculations the input current limiting shouldn’t have kicked in that early so I’ll have to adjust it experimentally (while watching driver IC temperatures). Be pretty cool to be able to get the full 100% output from about a 2.8V input. Can’t go much lower before the driver IC UVLO kicks in.

Once that’s ironed out, the next thing to test is the buck-boost. I found a different four-switch driver that isn’t as featureful but can be had for 1/3 the price so I need to draw up a new test board based around it, but I already have all the parts for the big one so I might as well play with it in the meantime.

Also will be testing with a host that somewhat resembles the 504b but has a screw-in pill instead of a drop-in and it’s built to handle heat much better. Still haven’t coded temperature throttling into the firmware yet. Work’s really picked up, such that June has seen more orders than March, April and May combined, and I’m still down a guy.

sidehack
Offline
Last seen: 5 months 6 days ago
Joined: 03/03/2020 - 21:30
Posts: 53

Hey guys. Haven’t been around for a while; manufacturing picked up pretty hard but we’re on track to catch up in the short term now so I have more time to play around with R&D.

I’ve cooked up an improved board for the 3A buck driver to test out temp sensor thermal limiting. I also, I know it won’t affect most people but I really like not having the light taper down in brightness at the bottom end of charge so I’m working on a 3A buck-boost that should be fairly economical and if it works as expected it’ll replace the 3A buck driver as a baseline model.

I also have a 17mm driver board drawn up for the 4.5A 6V booster so once PCBs arrive I can test it in a heavier-duty host with better heat transfer than the P60 host and zoomie I’ve been playing with so far. The host will require some customization, some cosmetic and some not (higher current tailswitch, etc), if the thermal properties are up to par.

I’m using ISP pads which are intended to be compatible with HQ’s programming key, which since it’s just got a pin header for tool connection should be just fine as a go-between for PICs. If it works reliably I may use a similar setup for ISP on another compact microcontroller-based thing I’m working on.

Haven’t looked at the big four-switch buck-boost yet. Haven’t done any programming in a while. I should have some good results by the end of the month though, since if I don’t get too distracted by other projects I should have the whole weekend for design and send off for PCBs next week.

sidehack
Offline
Last seen: 5 months 6 days ago
Joined: 03/03/2020 - 21:30
Posts: 53

So I went to bed like three hours ago but it’s one of those nights where the brain won’t shut off.

Today I laid out what should be a pretty good multi-mode buck/boost driver with a practical cap around 400mA, with low-batt cutoff and ISP pads, on a 12mm driver for use in a 2xAAA penlight. There’s still room for a temp sensor, especially if I switch to an SC70 LDO instead of using a SOT23 part I already stock. Gonna be proud of it if it works.

Haven’t calculated any component values yet but I’ve got some pretty good notes from the last hour or so of sleeplessness covering a basic USB-powered (adjustable up to 3A) CC/CV 18650 charger. Since it’s running USB I may drop in a RS232 chip and have the micro push charge data (including integrated charge capacity) to a terminal, since it’s already going to be monitoring both voltage and current. Be kinda cool to be able to chart that from a simple charger, and the extra parts to be able to do that are like a dollar. Reinventing the wheel a bit, sure, but I’m a hardware manufacturer not a rebadger.

Pages