The popular QX9920 buck controller as we’ve seen on a lot of controllers. Marked LEDA 1328.
Two tiny 8205S N-Channel FETs (with a ridiculous pinout, is that normal for the SOT-26-6 footprint?). This is probably the same FET configuration we see in the “new DRY” driver, the lightmalls driver, and probably others. Note that these FETs are not in parallel, the capacity cannot be added together.
IC marked 00VM - probably the modes chip. Also that’s probably a batch code or whatever, not the name of the IC.
R100 sense resistor
Diode marked 39 12 - I assume that this is a 3A diode.
As usual there is a zener present to support the QX9920, it’s next to the IC marked 00VM.
As pointed out by the IOS listing, this driver has no low voltage protection. Running 3s into an XM-L will let you pull the cells down to a very low voltage with a low current draw from the cells. I suspect that some protection circuits may not trip. (I’m a pessimist.)
This driver seems very similar to the Lightmalls one I linked above which ohaya has put a bunch of work into (component pics on second page).
I don’t remember if we did definitively, but FYI, l have been successfully replacing those with 5amp alternates mentioned in the LM thread since awhile ago.
Whoopdy! My order with the LD-44/LD-40s should be arriving from CNQ this week some time (shipped 9/20)! Looking forward to comparing these to the LM ones, and maybe killing a few :)…
If you’ve got everything setup and easy to get to I’d just hookup the MT-G2 for a quick test of output current. I’m sure it’s probably fairly close to the 2.3A you’ve already observed though, so it’s not a big deal. If you wanted to take tons of measurements (I rarely want to do that, it’s just a matter of needing the data sometimes) you could get current in, current out, volts in, and volts out. All measured under load. That’s our standard data for doing efficiency calculations. I’m not asking for that myself, because I don’t think it will be anything impressive. Probably 80% stock and <70% with a resistor mod. EDIT: in fact, I see little point in any of that testing. I’m pretty confident that we already know all of these answers. Don’t bother unless you get super bored!
You increase current the same way it was done on your LM driver. Decrease resistance through the sense resistor. I assume I posted math over there in your thread somewhere, but it should be this:
Iout = 0.25/resistance
so for example the stock resistor gives us:
0.250 / 0.100 = 2.5 Amps (close enough)
I didn’t see any evidence that this driver was significantly different from the LM driver. Comfy’s pic is down right now, but the BOM looked very similar to the LM driver you photographed for us.
Efficiency aside, this driver seems to be more robust than the LM one.
The 8205s seems to be spec’ed for 4.8-6 amps , and so far I’ve tried stacking 1 R200 and also 2 R200, on top of the original R100, and got about 3.2 and 4.48 amps respectively.
I tend to be a pessimist. It’s rated at 4.8A @ 70°C under these conditions: “Surface mounted on 1in2 copper pad of FR4 board; t≦5 sec. 180°C/W when mounted on min. copper pad.”
I’d call that a “safe” 3A for our uses. I think that 5A will probably kill it inside a flashlight, but there’s only one way to find out for sure!
Hi wight, is this always the case or is this calculation only applicable on LD-44?
I wonder why LD-44 has only 80% efficiency (I know it’s your “pessimistic” estimation lol)… is 80% efficiency considered good and normal for a buck driver? I thought buck driver should be quite efficient at about 88% especially it is a good quality driver apparently.
bibihang, it is not always the case. It is based on which controller is in use. The QX9920 is used in quite a few of the driver boards around here and uses a 250mV sense voltage. That calculation is valid for any driver which uses the QX9920 controller. Maybe I typed up a rough list somewhere on BLF? I’ve been thinking of starting a thread to keep track of drivers with some of the most common controllers.
Non-synchronous buck circuits only get so efficient. I’m not sure how good they can get, probably up to 90% in really ideal conditions. I think those conditions would be something like this: Vin and Vout would be close together and very high. Iout would be low (eg amperage would be low). Since we usually have high current requirements and want a large step-down, I think 80-85% is a good goal.
Something of note that I've thought about but haven't done anything with yet: has anyone measured the duty cycle of the QX9920 at different outputs?
There is a different current limiting strategy going on with these drivers than what you see with the HX-1175B1. Here we have a 0.1 ohm resistor operating both as a current limiting resistor but I can't tell what is between the input and the CS pin of the driver. On the HX1175B1 we have the equivalent of 0.034 ohms of resistance on the input to the FET (2x0.068 ohm resistors), but with an additional 100 ohm between that and the CS pin. In both cases, the CS pin is looking to maintain the same voltage on the CS pin, but it seems like the duty cycle used would be different which may affect other aspects of the output, such as current ripple (less big inputs vs. more smaller inputs). It seems like it would also affect the amount of losses through the resistor(s).
Using the ILEDS=0.25V/RCS formula in the datasheet, this appears to be pretty close to what we actually get.
This driver: ? I can't see what the small white component is by the larger 0.1 Ohm resistor or exactly where the traces go there because of the white silk screen.
If it were only the 0.1 ohm: 0.25V/0.1=2.5A (I think that this is wrong, please read the rest of the post).
The HX1175B1: 0.25V/0.034(0.068/2)=7.35A, which is about what we see, but how does that explain the "101" 100 ohm resistor also between the 0.034 ohm and CS pin? It seems that our calculation should be off by a lot, but it isn't, unless Iled out is supposed to be mA instead of A! I can't read Chinese but this must be it!
Looks like there is more fun to be had.
Whoever has the driver, could you tell us exactly what the component is by the R100?
I’m really not sure what you are getting at. The duty cycle of the QX9920? Tthe duty cycle is whatever the input and output together dictate, right? Duty cycle is another way of saying “on-time percentage.”
Previously (read on ;-)) with the HX1175b I declined to think hard about what that resistor is doing. Now, since you brought it up specifically, I have thought about it. It is doing nothing. We are simply measuring voltage so we have almost zero current draw on the CS pin. Since we have almost zero current draw when we measure the voltage on the shunt, we get almost zero voltage drop across the 101 Ohm resistor. Shorting it probably won’t change anything.
That's what I'm getting at though: there is a difference between having the MOSFET pulse 10 big pulses vs. 100 small pulses in the same period of time to average the same current, right? That would be a big difference in the amount of time the FET is on to average the same amount of current and it would also change what the inductor sees on the other side.
I get what you're saying about the amount of current across the resistor, and they may be using it only to filter what CS pin sees, but I still want to know what that small component is on this driver even though I'm guessing it's a small input capacitor. (edit: looks like it is a cap. Thanks for the picture!)
