Looking for an appropriate pull-up with an IRF510

Forum 2005-2010 (read only) Hardware Interfacing
1

I am using a Octobrite DEFILIPPI from macetec (a constant current 12-bit driver). Using this to drive 8 RGB Luxeons off of a 30A 5V supply.

I'm supplying the LEDs with 3.3V. Their cathode is connected to the drain of an IRF510. The supply is connected to common ground, with the pin from the Octobrite connected to the gate. I have a pull-up connected to the gate.

When PWMing the MOSFET, I get a significant amount of flickering when using pull-ups with higher values (greater than 560 ohms) I have been experimenting with various lower values, but it seems the brightness curve is not exactly linear (sending what I would expect to be a brightness of 0 gives me a dim light).

So! I'm wondering how to maximize the range of the apparent brightness from the LED. Is there some sort of formula I can use?

Many thanks.

2

(sending what I would expect to be a brightness of 0 gives me a dim light).

This is the problem when using these types of chips with an external FET. The reason is the inversion in the control. The LEDs are not on 100% of the time when they are on, there is a short period when internal shift register loading goes on. This is so brief you can't notice it dim. BUT when you invert the control you do notice the brief on pulse. Te only solution is to invert the signal before it drives the FET either with a transistor or some logic inverter like a 74LS04.

but it seems the brightness curve is not exactly linear

That's right the eye is not a liner receptor but the LED is a liner generator.

3

Thanks for the explanation.

Would a TIP120 do for a switching transistor? I'm already a bit cramped for space in this project, I'm afraid adding 24 hex inverters would make this a bit more unwieldy than I'd like.

4

IRF510 is not a logic level mosfet. It requires +10vdc gate voltage for full saturation current ratings. The 5vdc or 3.3vdc high digital output voltage will not allow much drain/source current to flow.

Lefty

5

I'm afraid adding 24 hex inverters would make this a bit more unwieldy than I'd like.

You get 6 in a package so that is only 4 chips.

6

Ah, thanks agan, I hadn't looked at the datasheet. Looks straightforward enough. Will still be a crowded board, but if it works, excellent!

7

I've been telling people to just use a pullup and invert the PWM commands, but I have realized that approach hasn't actually seen real use. On the other hand, back when I had TLC5947 chips to sell, a customer built a circuit on my suggestion using pullups to hex inverters and then to MOSFETs (with pulldowns), which worked perfectly. They used resistor networks to reduce the pad and part count even further.

8

Just a followup, I had some TI SN7404N laying around, and they work great. Thanks again.