FET driver questions

11 posts / 0 new
Last post
devryd
Offline
Last seen: 1 month 1 week ago
Joined: 02/04/2020 - 07:11
Posts: 134
Location: Germany
FET driver questions

I am currently looking into different drivers and I was thinking. Are there FET drivers, which use a micro controller to controll the current through the LED?
All I could find, were drivers with different modes, which result in different duty cycles for the PWM signal, but that would mean, that the current is variable, depending on the voltage of the cell.
I was thinking of measuring the current that goes through the LED and adjusting the duty cycle of the PWM signal, so that you have a constant current for the whole battery voltage, but wihtout the loss that a 7135 driver has.

I hope you understand what I wanted to say, if not I am sorry, english is not my native language.
Greetings devryd

thefreeman
thefreeman's picture
Offline
Last seen: 6 hours 10 min ago
Joined: 01/06/2020 - 09:56
Posts: 1575
Location: France

Well usually what we call ”FET” drivers are direct drive, as you said they just have modes = fixed duty cycles.

There are also linear constant drivers that adjust the gate-source voltage of a Nchannel MOSFET to adjust its resistance and thus the current, this is similar to 7135 chips, except a 7135 can only be dimmed with PWM.

Quote:
I was thinking of measuring the current that goes through the LED and adjusting the duty cycle of the PWM signal, so that you have a constant current for the whole battery voltage, but wihtout the loss that a 7135 driver has.

This is mistaken though, driving a LED with PWM doesn’t mean there is no loss, it’s just that the heat is generated (mainly) in the LED instead of in the driver, there are also RMS current conduction losses (significant in multi LED lights due to the on current being very high), in the end a linear driver is more efficient than a PWM driver, although a good buck driver is obviously better.

I designed a few Anduril (e-switch) linear driver here with building instructions (I’m going to receive the new boards in a few days and test them).
Led4Power sell CC linear drivers for clicky switches.
Convoy also sells different sizes in 5 and 6A for clicky switches, they are called ”SST-40” driver.

Skylight
Skylight's picture
Offline
Last seen: 10 hours 12 min ago
Joined: 10/01/2018 - 04:49
Posts: 1197
Location: Europe

I think you mean a so-called linear driver. The drivers from Convoy for example also use FETs but have a limited current. You recognize it because of the current limiting or sense resistor (R010, R020).

The most efficient ones are buck drivers or buck-boost drivers which convert the voltage and require less current if the voltage of the battery is high. The 7135 drivers just convert the eccessive voltage into heat. If a driver has a big grey thing on it like this one it is a buck or boost driver. The Lume1 driver from Loneoceans is a constant current buck-boost driver: https://budgetlightforum.com/node/71616

What would you need the driver for?

I am not a driver expert, I just know what I have read on BLF. Let’s see if some of the real experts can tell you more.

devryd
Offline
Last seen: 1 month 1 week ago
Joined: 02/04/2020 - 07:11
Posts: 134
Location: Germany

I was just thinking, that it would make a pretty easy driver. You would just need a sense resistor to measure the current at the led, a mosfet and a micro controller, possibly easy to program, so that you could set different max current levels. I understand that it would be less efficient than a buck driver, but it should be about as good as a 7135 driver, but without the need to stack serveral 7135 for high current applications. I was just curious about building such a driver and programing it myself. I dont really understand enough about buck drivers, to make one of these from scratch, though.

thefreeman
thefreeman's picture
Offline
Last seen: 6 hours 10 min ago
Joined: 01/06/2020 - 09:56
Posts: 1575
Location: France

The problem with measuring the current directly with the MCU is that the resolution of the ADC isn’t enough if you want any kind of low modes. A 50mV shunt (e.g. 10mR for 5A) gives you 200uV at 1/250 (8bit dimming) or 50uV at 1/1000 (10bit dimming), whereas a 10 bit ADC with 0.55V VREF (in the attiny 1 series for example) only goes down to 550uV. Increasing Rsense would lead to efficiency losses and wouldn’t be sufficient on its own, but if you’re fine with no low/low mid modes then it could works.

We talked about this in the 1 series thread , about the Unicorn and Gekko, they use special techniques to increase the ADC resolution (delta sigma modulation) to be able to read the Vsense at low modes.
After that the FET can be PWM or linear, or it can be used for a buck or boost converter too.

devryd
Offline
Last seen: 1 month 1 week ago
Joined: 02/04/2020 - 07:11
Posts: 134
Location: Germany

Ah I didnt really think about how to do low modes, thanks for the reminder. I’ll think about it, maybe I can find some kind of a solution…

devryd
Offline
Last seen: 1 month 1 week ago
Joined: 02/04/2020 - 07:11
Posts: 134
Location: Germany

The chip used in the arduino nano has a 12bit adc. would that be enough?
Or does the atmega 328 need to much power for a flashlight driver?

EasyB
Offline
Last seen: 4 months 1 week ago
Joined: 03/09/2016 - 15:24
Posts: 2250
Location: Ohio
devryd wrote:
I was just thinking, that it would make a pretty easy driver. You would just need a sense resistor to measure the current at the led, a mosfet and a micro controller, possibly easy to program, so that you could set different max current levels. I understand that it would be less efficient than a buck driver, but it should be about as good as a 7135 driver, but without the need to stack serveral 7135 for high current applications. I was just curious about building such a driver and programing it myself. I dont really understand enough about buck drivers, to make one of these from scratch, though.

I agree this would be a good option to have. For very high current lights (>20A), linear drivers can’t really be used easily unless the FET is specially thermally mounted. A current controlled PWM FET driver would remove that difficulty. This would allow high power lights (like single cell multi emitter or SBT90.2) to have high regulated current capability.

Quadrupel
Quadrupel's picture
Offline
Last seen: 11 hours 4 min ago
Joined: 12/03/2017 - 10:40
Posts: 1058
Location: Lithuania

For low modes maybe its possible., but you will need 1000+uF cap on output ;))

thefreeman
thefreeman's picture
Offline
Last seen: 6 hours 10 min ago
Joined: 01/06/2020 - 09:56
Posts: 1575
Location: France

Re thinking about it, since you want to drive the LED with PWM you don’t actually need a high resolution ADC, the ON current will always be relatively high, e.g. a single led might do between 5 and 1A between 4.2 and 2.8V which gives a high enough Vsense. You would need to synchronise the ADC with the PWM signal to read Vsense only when it’s high (if that’s possible, I don’t know much about this) and since you know the duty cycle you know the average current.

devryd
Offline
Last seen: 1 month 1 week ago
Joined: 02/04/2020 - 07:11
Posts: 134
Location: Germany

I might have learned the basics of an Buck converter today and build one from spare parts I had lying around.
Dont know how efficient it is, but I think thats what I want to do for my next build, but with a decent driver