I think that there must be some fundamental thing that is different with this CNQ driver, vs., say the LM driver.
I just went back to a light (an RC-G2) that has the LM driver, resistor-modded with an R100 stacked, and a new/upgraded Schottky diode and FET (only the one nearer the diode). The light has a quad XP-G2. I had to jury-rig to get 2 x 14500, and here’s a pic:
So that was 3.042 amps at the tailcap with 2x14500, so “theoretical” (not counting efficiency losses):
Power = 3.042 x ~8 = ~24.336 watts
So, the Iemitter (total) would be approximately:
24.336 watts / ~3 volts = ~8.112 amps
and Iemitter (per emitter) would be:
~8.112 amps / 4 = 2.028 amps / emitter
The thing here is that the LM driver seems to be behaving “correctly” with respect to output, whereas the CNQ driver seems to be constraining output?
Page 4 of the datasheet discusses the effect of the offtime cap (they call the cap Coff, not Toff. oops.)
I’m having a trouble figuring out how I’m supposed to handle units, and since Roff isn’t present I’m assuming that the Roff fraction just cancels out… yeah I don’t know where I’m going with this.
Where I’m really going with this, is would you be willing to strip the Coff/Toff cap off of your board ohaya? I think that output may get much, much dimmer without the cap present. Or something could blow up. Like I said, I’m having trouble figuring this section out. From what I can figure the offtime without that cap should go way up. So a higher value cap could reduce offtime.
Or all that could be wrong and Toff actually (sort of) sets the period. I don’t know. I’d like to know what happens when you take it off, but if it increases the period a lot (which it might do by a few orders of magnitude… like 3 or 4 orders of magnitude) it could make the output so rough that it kills LEDs.
Purchased 3 of these to play around with. I think I will stack it with an R150 instead of an R100 to get a theoretical 4.16A vs 5A. I’ll be pairing this up with an MT-G2 in a Dipper D19 (L2 clone).
I prefer stacking resistor chips as I am better at it than stacking 7135s, so no zener modded driver for me.
I now have an LD40 and one of the LM drivers (extensive thread here) in hand for testing and comparison thanks to another BLF’er.
So far I haven’t learned a ton, but I’m working on it.
Here are the connections for the pads on the bottom of the LD40:
V+ = The zener-regulated voltage supplied to the modes chip (00VM) and the QX9920 (LEDA)
GND = having a little trouble here. It beeps out to Pin 1 & 3 on each MOSFET package… which the datasheet says is Gate? Really? GND for the logic section.
3 = Pin 6 on the modes chip (00VM)
1 = Pin 3 on the modes chip (00VM)
0 = Pin 1 on the modes chip (00VM)
When the driver is powered Pin 6 is pulled up to V+ and Pin 3 and Pin 1 are pulled low.
In the stock configuration the mode order is L/M/H (at least for the 3-mode one I’ve got). The modes chip does not give a 100% duty cycle. The duty cycle of each mode is: ~0.5% (L), ~17% (M), ~84% (H ).
When Pads 3 & 1 (Pin 6 & 3) are connected the driver appears to be locked in High. Duty cycle appears to go up a smidgen to 85%.
When Pads 3 & 0 (Pin 6 & 1) are connected the driver appears to be locked in Medium. Duty cycle is still ~17%.
EDIT: be sure to see updated info 2 posts below this one.
I was about to post that these 8205s chips are certainly marked upside down as compared to the datasheet I used. More likely is that they are just a different 8205S as you’ve shown. That datasheet turns the pin configuration inside out!
The connections are as I stated. Pins 1 & 3 connect to the “GND” pad as well as VSS/GND on the QX9920 and to VSS/GND on the modes chip (00VM). Pins 5 & 6 of the 8205S next to the QX9920 (LEDA) chip connect to DRV from the QX9920. The other 8205S is no doubt hooked up in a convoluted fashion, as usual (it’s also the one which connects the GND net hooked up to QX9920 and “00VM” to battery GND).
The pinout for “00VM” is:
Pad 0 (Connect to Pad V+ or Pad 3 for Fixed MEDIUM mode (~17.65% duty cycle))
Vss (GND) (Pad GND)
Pad 1 (Connect only to Pad 3 for Fixed HIGH mode (~85% duty cycle))
PWM (PWM output)
VDD (regulated voltage input from Zener diode) (Pad V+)
Pad 3 (Connect to GND pad for Fixed LOW mode (~0.5% duty cycle))
I measured 2.0 Amps output to my XM-L2 on a Noctigon. (2.00 - 2.05A) I used my modded HF meter. This corresponds pretty well with the R100 current sense resistor, which sets 100% duty cycle at approximately 2.5A. Since we only operate at 85% duty cycle we’re getting about the right current reading.
The IOS version may be using a modes chip with a higher duty cycle.
Unfortunately my measly 3A CC buck PSU gets upset with the driver when I do that. The PSU claims that it’s outputting ~2.4A and something like 4-5v (when set to 7.1v). The CC light flickers. I tried turning the voltage up, but it still does it. It seems that the driver’s inrush or whatever must spike above 3A with the R200 stacked on top of the stock R100. I got almost exactly 3A output from doing this. I should get a bigger, better PSU.
So to get more than 5a from this i would have to use a bigger diode? Im going to build a tripple xpl and think the person im building it for might like the thought of being able to use more batteries in series for longer runtimes.
Do you have some of those small magnets? If so, you can use them to put two Li-ion batteries in series. Some of the testing that I’ve been doing most lately were with this, using 2 x 14500 in series.
I have magnets somewhere, but I prefer other methods of securing the battery such as a clamp.
The trick with using the batteries is that output voltage isn’t constant and neither voltage or current gets automatically reported like when you are using a PSU. That’s more measurements than I had time for at that point yesterday. I’ll knock out an efficiency calculation like that soon.
With the battery type testing, I just clamp a clamp meter around the tailcap lead or the emitter lead to get tailcap or emitter current (I only have 1 clamp meter, so with a battery test, I can only get one of the current readings).
