LED driver (Cree Q5) in SOT23-3 format and without external diode

[0xdeadbeef]

I currently stumbled over one of these Chinese "Cree Q5" LED flashlights (1xAA) that doesn't use the typical fake 2803 boost driver with current regulation but obviously just a SOT23-3 device named AS657G without an (external) diode. I'm aware that drivers like this exist for lower currents, e.g. the Zetex ZXSC380 or Prema PR4402, but I wasn't aware of something like this for the 300mA range. Besides, the 1.5Ohm resistor between cathode and the 3-mode chip obviously isn't a shunt/sense resistor, so it's most probably there for current limitation.

My main interest was in converting this from 3-mode to 1-mode for which I just needed to bypass the 2nd SOT23 chip (HDI4G) which works as expected, so I didn't investigate any further.  I'm still curious though about this design and that AS657G converter. Like I initially thought it might have an integrated diode but then again it could be actually a design without a diode that doesn't have a real DC output but something pulse shaped.
That 1.5Ohm resistor should be a 0.125W type, so the average current can't be higher than 280mA. So since the forward voltage of the LED (~3.1V @300mA) and the voltage drop over the resistor increase with higher current, I guess a boost regulator with ~3.5V output would create a somewhat regulated current around 250-300mA. Just guessing and rule of the thumb, was too lazy to really measure or simulate anything.

In a nutshell: is anybody familiar with this design or this AS657G IC?
And also regarding these 3-mode chips like HDI4G or CX2859A: is there any chance to find datasheet for these obscure Chinese parts? Google doesn't seem to find anything.

[ocset]

You could contact Mass Power in China...

http://mass-power.com/contactus.html

Say that you have a potential volume order for LED drivers, and that you think using their AS657G looks good but please could they first send you ther datasheet so you can check its OK

Otherwise, i reckon theres little chance to get a datasheet of a Chinese driver.
I requested a smps controller datasheet when i worked at a company that got an smps deisgned by Mass Power, and they kindly sent me the datasheet.

The company was Braun (cosmetics).....say you have a friend there who had some good work done by them, etc etc

[floobydust]

Take a look at Shinginic, they have marking codes like HDI13, HDI14 but I forgot where I got those from.
SOT23-3, SOT23-5 linear, boost LED flashlight controllers with 3-modes dimmer/blinky.
White LED flashlight controller IC's

[0xdeadbeef]

I got my lazy ass up this morning and did some quick measurements. They more or less look what I expected assuming that the AS657G provides just some fixed PWM step-up conversion.
I guess the step-up works at 400kHz with a 20% DC or so. At 1.5V there are ~750mA flowing into the circuit and the average voltage at the output is 3.12V. At 1.2V, the current is reduced to ~500mA and the average output voltage is reduced to 3.04V. In like 80%, the voltage is above 3V though and slowly rises from 3.3V to 3.4V (1.5V supply) or from 3.2V to 3.3V (1.2V supply).

I also tried to measure the average voltage drop at the 1.5Ohm resistor and the diode with a DMM (Fluke 87V) but I'm unsure if I can trust these measurements due to the high frequent period and shape of the voltage.
Anyway, at 1.2V supply, I measured an average voltage drop of 2.684V at the diode and 0.1036V at the resistor. At 1.5V supply, I measured a drop of 2.7V at the diode and 0.14V at the resistor. Obviously neither do these values add up to what I measured with the scope (where the average values where like 0.3V higher) nor do the current values look plausible. Like 140mV/1.5Ohm would mean an average current of only 93mA which seems much too low. Again, I think the bandwidth of the 87V might not be sufficient to accurately measure the average voltage drops here, so I think the only thing I can see from this is that the input voltage is somewhat proportional to the average current. So higher battery voltage means higher output current and brighter LED.

Anyway, I guess the maximum current surely can't exceed 300mA since this would mean a 0.45V drop at the resistor leaving a maximum (!) voltage drop of ~2.95V at the LED. Actually, even assuming 80% efficiency of the step up, the average current can be >200mA for the 1.2V case as (1.2V*520mA*0.8 )/3.04V = 164mA.

While I was at it, I tried how much I could lower the voltage without the LED stopping to work. Surprisingly it worked down to 0.2V but of course the brightness dropped with the voltage, so it was very dim at this point. The good thing is that this flashlight will work down to 0.9V for sure, bad thing is that it will exhaust a rechargeable battery which might damage it.

Well, yeah, not the best circuit obviously. Which is the shame as the flashlight itself is rather nice.

[0xdeadbeef]

I recalled I recently acquired a differential probe, so I decided to measure the voltage drop at the resistor with a scope. Surprisingly, this time the currents flowing into the circuit where much higher, so I also measured the step up voltage at the diode again. Guess I had contact issues in my first measurement.

At 1.5V, there are 900mA flowing into the circuit and the mean voltage drop at the resistor is around 204mV which means a current of 136mA through the diode. The mean voltage measured at the output of the step-up (and this the total voltage drop) was ~3V this time.

At 1.2V and 660mA input current, the mean voltage drop is 142mV which means 94.7mA through the diode. The average total voltage was 2.89V.

So yeah, letting aside that all of these measurements are not very accurate, all of this shows that it's a super cheap circuit that drives rather low currents and has a somewhat linear relationship between battery voltage and LED current.