FETs and gate resistors - scope images

Cap moved from C1 to B+/GND:

1 cap at C1, 1 cap at B+/GND:

Adding a cap back at C1 also brings back a little bit of boosting, Vin was below 4v for those pics and Vcc is up above that.

Voltage at pin 7, for voltage monitoring, is nice and flat with either setup (and even in the first config, with the voltage boosting and spikes and the rest). If it's stable and doesn't misbehave with just a single cap relocated, does the oscillation on Vcc hurt anything?

And I found the thing the diode across LED+ & LED- is supposed to deal with...

Yikes. :O (using different dividers to avoid the 'out of range' warnings, 0.8 is actually 8.0 ZERO YOU STUPID AUTO-SMILEY B.S.)

But again, as scary as that might be, I've used lights with these drivers without the diode extensively and haven't had anything croak yet. Is that another something that really should be addressed with a board redesign?

Thanks for testing - the results seem reasonable considering the other test results you’ve already shown. I don’t think the oscillation is going to hurt anything. Hopefully others will chime in one way or another. I would still like to move the diode & cap towards VCC, but no sense fooling with that until we’re reasonably sure of things.

I’m a little lost with the most recent trace. I see the spike, but I’m confused about the scales despite your explanation. It seems that the resting voltage of LED- shows up as 2v? And when the FET turns on LED- dips to 0v, and then spikes to >8v when the FET turns off?

I read some about margins recently. Apparently a 20% margin is normal for engineering. As in, if all the parts subjected to the spike can handle an 10v potential from LED- towards B+ (or whatever), then 8v would be OK.

That's with the probe on LED- & ground clip to battery negative. I only changed the dividers to get rid of the out of range message. The waveform didn't change, just the voltage scale on the side.

By the way, what IS the proper connection to look at the signal to the LEDs?

Sounds legit to me. Remember I’ve got about the same experience with a scope as you :-). Knowing where that clip goes is just part of keeping us on the same page.

Please try the same test with one probe on LED- and one probe on the FET’s tab (eg the other end of the wire that goes to LED-). I don’t understand well enough to know whether the spike should be worse there or not.

Well I dunno if this is normal for all oscilloscopes, but on this one grounding the clip on either probe provides ground for both of them. Nothing on screen changes if you have both clipped to ground and then remove either one.

Ground clip to LED-, probe to LED+?

Now that I think about how I had it connected, I think the trace is inverted... it looks almost like a mirror image of the gate signal.

OK…the BLFDD redesigns are in the works…Mattaus is currently reviewing, finalizing my tweakage of his builds and if they check out he will be re-releasing with the design suggestions you guys recommended

I hope I did him proud!

Thank you all you guys. I could never have figured any of this out by myself and yet I can follow most of the discussion guing on here. You guys are brilliant! TY :-)

If I want to do this with a zener mod would I replace the diode with the R200 resistor (like normal) and add the zener diode to the now empty C1 pads or does the zener diode have to also have the cap in the factory position?

We haven’t tested it (AFAIK), but the zener mod drivers simply do not have this problem (we assume). Build them as normal but do not bother with a gate resistor (use a jumper instead).

Specifically:

  • Without the reverse polarity protection diode there is no boost circuit, so the overvoltage condition does not exist
  • Without the reverse polarity diode in place the decoupling cap (C1) should act to smooth out any spikes generated on B+ by the FET switching.

even with the AOD510 I can get away with building them regular and no gate resistor?

thx

edit Sorry I wasnt asking a repetitive question here about the 510’s working in general, I got that, I have one together now on the bench with a 510- I was asking specifically about zener modded old style 17dd’s with the 510…

That’s definitely what I’m seeing at this point.

YES AOD510 WORKS WITH NO GATE RESISTOR* AFTER MOVING C1 TO B+/GND! All of them work with no gate resistor after moving the cap. Even the 15DD with the IRLM2502 works with no gate resistor after moving C1.

*replaced with a jumper on the versions with gate/pulldown pads, pulldown location left open, of course

I think I have a copy of the BLF15DD, I can re-design it like the BLF17DD is that is what you guys want

In fact, I might be able to make it single sided (nothing on the battery + side of the board)

Working on it now

I’ve found some strange behavior…

Since getting my 858D station yesterday I’ve built 6 17dd’s (5 this new way and one zener modded), 3 of them with the AOD510’s and 2 with the old Vishay’s are working fine however the one with the one FET I have out of an M6 is giving me some issues with the exact same firmware. It’s heving problems when cycling threw the modes, after turbo sometimes it’ll miss moonlight completely and just go back to low (5-mode off-time with no memory and moon defaulted on without the star needing to be soldered), other times it will go to moon but with a bad preflash.

I’m not really complaining, I have exactly one of these FET’s and dont really mind discarding it to the random parts drawer, just reporting my findings.

What's the p/n on the problem FET? Curious to find a datasheet on it and see what it has in the section for gate characteristics. Both the 70N02 & AOD510 have a total gate charge number known to be something the attiny is capable of driving. Above something like Qg=40 they can get troublesome, or at least that's the general number I've heard to stay under.

DTU
30N02
DD22U

then a little circle symbol with an Lv

Clear PCB images are here that show the FET’s markings and M6 layout: Supfire M6 - resistor mod (Update testing XP-G2s on copper in post 59-60)

Wow that thing is awful. Throw it away. Rds(on) is around 0.030 ohms, 70N02 under the same conditions is 0.004-0.006. lol

This is really great stuff you’ve found out CC and precisely the kind of thing that only comes out after new ideas are tried out and someone takes the time to troubleshoot odd behavior. Good job.