Using BJT's with a 595 shift register not outputting expected current.

Hi all,

Currently I am in the process of muxing a TLC5940 and a shift register for controlling a matrix of RGB LED's. What this means is each LED will need approx 20mA of current, and when I mux across 8 inputs from an SN74hc595, I only get 25mA total per pin sourced. So clearly this is an issue that can be fixed with a FET or a BJT. Currently I am testing with a 2n3904, a BC517 (darlington pair), another high power BJT that i don't have the product # for right now and a few different high-power mosfets.

The issue is that none of these are actually supplying a high current to my LED's!

What i'm thinking is that my VBE simply isn't high enough, as VB = 3.3V coming from a teensy 3.1, and my VE needs to be from 2 to 3.2 volts depending on the color! So when 3.2V is required, my VBE wont be able to match that, as it needs to be 0.7V to be saturated and on, or else the linkage current will be really low.

Any suggestions on what I can fix to see a high current pass through this NPN transistor?

Thanks!

Swap LED and R1

I'd put a 300 ohm resistor on the base (while you swap them around)

Move L1 to between R1 and the transistor. How else is the 595 wired? You don't show power, Gnd, OE/, MR, and the data and clock lines.

CrossRoads: Move L1 to between R1 and the transistor. How else is the 595 wired? You don't show power, Gnd, OE/, MR, and the data and clock lines.

I'm not sure that your suggested approach would work. The LED's are passed to a TLC5940 then to ground, so the shift register controls whether or not to pass VDD and the TLC decides whether or not to ground the circuit.

I didn't include OE/MR/GND as I didn't believe them relevant, OE is set low, GND is at 0V (relative). Power is 3.3V

Here's a better diagram (still missing a lot of pins, but has the info that you need):

What about this for a solution?

Then NPN is the wrong transistor choice - you should be using PNP to source current to the anode, while TLC5940 sinks current from the cathode. Pull the Base low to turn on the PNP.

CrossRoads:
Then NPN is the wrong transistor choice - you should be using PNP to source current to the anode, while TLC5940 sinks current from the cathode.
Pull the Base low to turn on the PNP.

Ok, not sure if that will work with the shift register i’m using though, as it only can turn on one pin at a time, not the opposite. (sn74hc595n)

2nd diagram would work - LEDs would be powered when the NPN was off (base low) and the NPN being on would sink the 12V away. Don't put in R2 - that will just elevate the LED anode voltage when you want it off. Not the most efficient design power-wise.

as it only can turn on one pin at a time, not the opposite. (sn74hc595n)

Huh? Has 8 outputs, any combination of highs/lows can be shifted in. It's not a demux chip.

CrossRoads:

as it only can turn on one pin at a time, not the opposite. (sn74hc595n)

Huh? Has 8 outputs, any combination of highs/lows can be shifted in. It's not a demux chip.

Sorry, little sleep was had. Makes sense. I was confused as the 595 is a SIPO device, but all I would do is pass in 11011111 if i wanted to light up row 3.

Thanks so much, i'll pick up some PNP's.

Does this mean that when I hook up to the PNP transistor, i connect L1+ to the Emitter and GND to collector with only the emitter resistance?

Reply #5. top diagram, replace npn with pnp, no resistor frmo +12.

PNP is not enough to make a high side switch, there 'd be +12 V at emmiter, and base at max +5, it's not going to off. Why there is 595, do you need more pins? If you replace 595 with ULN2803 and 8 PNP , than you 'd be o'k

Like this:

CrossRoads: Pull the Base low to turn on the PNP.

Isn't the problem going to be how to switch the PNP off?

Even when the 595's output is high, its still only at 3.3 or 5V. The PNP's emitter will be at 12V. Won't it still be switched on? If so, I think you may need to get the PNP's base close to 12V to switch it fully off. If I'm right, an NPN could be used to achieve this. So for each 595 output, a ~330R connects the output to an NPN's base. The NPN's collector is grounded. The NPN's emitter is connected via a resistor to the base of the PNP. The PNP's emitter is at 12V and its collector connects via R1 to the LED anodes.

Apologies if I'm talking rubbish!

Paul

Or just single LEDs, that string was left from prior drawing I did.

You guys are right about the PNP needing 12V to turn off, I had overlooked in my earlier replies, too caught up in the mis-used NPN.

to OP: find UDN2981 /2982 - high side switch,

PaulRB:

CrossRoads: Pull the Base low to turn on the PNP.

Isn't the problem going to be how to switch the PNP off?

Even when the 595's output is high, its still only at 3.3 or 5V. The PNP's emitter will be at 12V. Won't it still be switched on? If so, I think you may need to get the PNP's base close to 12V to switch it fully off. If I'm right, an NPN could be used to achieve this. So for each 595 output, a ~330R connects the output to an NPN's base. The NPN's collector is grounded. The NPN's emitter is connected via a resistor to the base of the PNP. The PNP's emitter is at 12V and its collector connects via R1 to the LED anodes.

Apologies if I'm talking rubbish!

Paul

Hey Paul, I'll give it a shot! What is the point of the 330Ohm R? Unless i'm trying to set my base close to 0V when it's off, which would make sense..

Magician: PNP is not enough to make a high side switch, there 'd be +12 V at emmiter, and base at max +5, it's not going to off. Why there is 595, do you need more pins? If you replace 595 with ULN2803 and 8 PNP , than you 'd be o'k

At this point i'd rather not completely redesign my circuit (it's more complex than just what's pictured). That being said, if it's practically the same to switch to a ULN, then it may work. Can you whip up a small schematic for me to see? The issue was that I was indeed running out of pins.

The issue was that I was indeed running out of pins.

Then go with TPIC6B595 controlling PNP transistor, P-channel MOSFET. Or as input to UDN2981 or UDN2982 if you can find anyone that still sells them: http://www.digikey.com/product-search/en?lang=en&site=us&KeyWords=udn2982