Oshpark Projects

comfy, if you have a look on the scope of the waveform of an ordinary 105C, would that also have the spike?

Yes it does. Battery voltage while running measured 3.78v. Those peaks are easily above 5 volts.

That's completely stock 105C hardware, 8x7135s, flashed with the same STAR clicky FW as in the other scope pics of the BLF drivers.

The size of the spike depends on the load being driven by the PWM pin. 7135s are a very different load than the FETs. If the spike doesn't hit the MCU's overvoltage limit, it runs reliably.

I have seen some BLF-DD drivers that are only just a little bit flaky, right on the margin of working right when testing out in the open on the bench, but do OK once inside a light. So these waveforms are likely exaggerated somewhat due to the test setup - battery box, clicky switch, longish wires from the box to the driver, and so on. But what you can know for sure, is that if it works reliably in this bench setup (and the scope shows the spike issue is down to acceptable levels) it will be solid when installed in a light.

There is no over-voltage limit circuitry built into the MCUs (there is an under-voltage brown-out detector that can be enabled). Over-volt them and they can die. That voltage spike does not appear to be enough to harm the chip, but it may be causing it to reset itself.

On the ones with issues, one thing they do pretty often, other than just shut off completely after a mode change, is they blink and end up 2 modes lower instead of one. It does act the same as if you tap the reset pin to ground.

This is a different driver, no gate resistor, and the original D1/C1 setup, without the extra cap in parallel w/the diode. CH1 (flat blue line, at 6.000V) is after the diode, CH2 (yellow) is B+ before the diode. These voltages are correct, I moved the probes to the contacts in the battery holder and verified with a separate DVOM. Battery no-load voltage was 4.08 - I'm really not cheating here and feeding the driver 6 volts! It's making that extra 2V all on its own.

The spike is most likely coming from the inductive kickback voltage from the emitter wires. When the FET switches the magnetic field in the wires collapses and generates a voltage spike. A cap across the protection diode is not a good way to stop it. Try a diode across the emitter wires at the driver (bar side to LED+, triangle side to LED-). The diode should have a decent current rating.

Comfy has a dedicated thread over here where I was just suggesting the same thing.

1A schottky diode (IN5817) between LED+ and LED- had no effect at all. On a driver with the gate resistor replaced with a jumper, the scope shows no change and the driver doesn't change modes reliably.

I just built a few 17mmblf drivers with rmm kit, does this voltage spike have an effect on those? I haven’t used them in any lights yet.

If it's one of the versions using the gate resistor and it changes modes correctly you can use them as-is.

-----

17DD with the extra cap & resistors eliminated:

Oh, and the AOD510 that didn't work before works properly now in the extra cap/no resistor setup. That's why I added it to the Digikey shared carts.

I'm sure there's history on this, but does the AOD510 MOSFET perform as well as the Vishay 70N02?

We might need to get some current numbers to find out for sure but it may do even better. On state resistance is hard to compare between fets since they are often spec’d at different gate voltages and aren’t exactly the same from one FET to the next.

I don't have a current measurement, but I have checked voltage drop across both in a like-for-like setup. 70N02 = 32.6mV, AOD510 = 41.5mV. Higher voltage drop means higher resistance. So they're not equal, but the AOD510 is readily available and relatively affordable and they're only 'not as good' when compared to a really excellent one.

And you can buy it :wink: (that’s always a plus!)

Comfy what are you using for the + button on that driver there? Very clean install (as usual). Also very good news about the 510’s! I was really getting worried as I’d been testing different gate resistor values (I have every resistor pack coming from fasttech so I wasn’t gonna give up yet but I tested every value resistor I had in the spare resistor tube and nothing had worked yet so that’s great to see). Thank you for your time/money.

Side note my 858D will be here Tuesday!

Just a piece of scrap sheet copper, I think it's around .040" thick. Not much room in that little Z1...

Revision revised again (probably won't be the last time, either)!

dave_ suggested an alternate fix in the 'scope images' thread (post #50, in case the pagenumber thing breaks the link for you). Instead of adding a second capacitor in parallel with the diode (which does work acceptably well), just relocate the one original cap from its original spot between the diode output and ground, to just plain between B+ and ground.

This is the extra-cap version:

Gives waveforms that look like this:

Acceptable, and it makes the driver reliable with the gate resistor deleted. But requires an extra capacitor.

dave_'s suggestion:

Gives you this:

Waveforms look better, needs fewer parts, and works just as well.

The capacitor can go anywhere on the board where it can connect between B+ and ground. There are many options on top or bottom of the board that will give the same connections. On the topside, there's a single ground via right beside the four B+ vias at the LED+ wire pad, if you'd prefer that it not be on the spring side of the board.

A larger value capacitor will flatten the oscillation more than shown above, but this works well enough and the part is the same as what comes on the Nanjg boards, for those who are using that as a source instead of buying bare parts in bulk. For those wanting to get more involved, you can change to a larger value cap if you wish. Experimentation required, and probably needs to be verified with your own scope images if you want to try something different.

ANOTHER SET of shared carts - same components as before, just with 1 cap per driver instead of 2:

shared cart- BLF-SRKDD | Digi-Key

shared cart- BLF-17DD & BLF-20DD | Digi-Key

shared cart- BLF-15DD | Digi-Key

I hope you dont mind me posting this here but I got my first Osh Park shipment today. I just need to know now how it all goes together.

Uh-oh! Those spacer boards seem to be missing the center hole?!?