Go Down

Topic: How to measure current draw in a multiplexed display? (Read 1 time) previous topic - next topic


Oct 16, 2011, 05:56 am Last Edit: Oct 16, 2011, 08:12 am by Chris Magagna Reason: 1
I have a 24x16 LED matrix - the 16 rows are connected by transistors driven by a pair of 74ac138s and the 24 rows by 3 TLC5916s. Refresh rate is 10K Hz (0.1us).

I want the lights to be as bright as possible so I'm overdriving them with 100 mA pulses...at least that's what things should be if I did the calculations right.

When I measure a row's draw I know I won't see 2400 mA (24 LEDs @ 100 mA ea) since it's only being powered 1/16th of the time, but my meter says ~38mA instead of the 150 I'd expect (2400 / 16).

On each column the meter says ~63mA. I know it won't say 100mA because the display is off for 3-4ms each update, but 63mA seems low...

Maybe my meter is so off because it doesn't average pulses very well?

My meter only goes up to 500mA so I can't do the easy thing and just measure total circuit draw.

My question is should I just get a new meter with a higher current capacity, or is there something like an oscilloscope etc. to more accurately measure pulsed current? If so are there any models you recommend?

Thanks in advance.



I don't think any normal meter can read this, maybe if it has a peak hold facility.

I'd use a scope to measure the voltage across the current-limiting resistor then consult Mr Ohm about his law.

If so are there any models you recommend?

Anything will be able to do it, I can't recommend a modern one though.

EDIT: Just noticed you're using a constant-current driver chip, can't you just trust the 5916 to do as advertised?

Rob Gray aka the GRAYnomad www.robgray.com


Oct 16, 2011, 09:36 am Last Edit: Oct 16, 2011, 09:39 am by Chris Magagna Reason: 1
Thanks for the scope info. I'll start researching.

BTW I do trust the TLC5916s, they're working great.

I'm trying to figure out if the row transistors are supplying enough current. 74ac138 outputs go to DPLS350Y-13 transistors via 43 ohm resistors. If I did the math right then 24mA from the outputs and a gain of 100 from the transistors will give me 2.4A on each row (for 24 LEDs @ 100mA ea).

The problem is with the setup I have right now is the lights aren't as bright as when I drive them directly with 20mA. I realize I won't be able to get exactly as bright, but I was hoping for more than what I have.

Here's the transistor data sheet:


So worst case is that a transistor has to drive 2.4A, they have a max pulse rate of 5A so I guess that's OK (Transistors aren't my strong point).

They have an Hfe of 80 at high currents, so the base current has to be 2400/80 or 30mA, can the ac138 provide that much on a single pin?

One way to measure the current would be to break out the TLC5916 GND pin and measure the current flowing out that. I assume that with all LEDs on there will always be 16 LEDs driven at a given time. Different LED but still 16 so you could use that measurement /16 minus the chips quiescent current to see what each LED gets.

Rob Gray aka the GRAYnomad www.robgray.com


43 ohm resistors. If I did the math right then 24mA

43 x 24 = 1.032 V.   Where it came from?


Oct 16, 2011, 08:37 pm Last Edit: Oct 16, 2011, 09:27 pm by Chris Magagna Reason: 1
@Graynomad - Hfe is 130 @ 2A and 80 @ 3A, so I'm estimating I should be able to get at least 100 @ 2.4A. The data sheet on the 74ac138 (http://www.fairchildsemi.com/ds/74/74AC138.pdf) says it can provide up to 50mA; in another thread (http://arduino.cc/forum/index.php/topic,72866.msg547333.html) Crossroads said it should be able to do 24mA.

I think you're right, if I measure current from a TLC5916's ground pin I'd hope to see about 800mA (8 outputs @100mA ea) but my current meter only goes up to 500mA. I know I need a new meter but am trying to figure out if I should get a new basic multimeter, or if there's something fancier that would let me measure these pulses.

@Magician - sorry if I was unclear. The output pins from the 74ac138s are connected to 43 ohm resistors, which are connected to the bases of the DPLS350Y transistors. I'm using the formula from "Choosing a suitable transistor" at http://www.kpsec.freeuk.com/trancirc.htm ; Rb = (Vc x hFE) / (5 x Ic), so (5 x 100) / (5 x 2.4) = 500 / 12 = 41.67.

Thanks again.


Your formula isn't correct for pnp transistor.
Rb = Vr / Ib; Ohm law
Vr = 5V - Vbe - Voic; 5V - power line, Vbe = 0.7 base-emitter drop of transistor, Voic - Low voltage IC = 0.8.
Vr = 5 - 0.7 - 0.8 = 3.5 V
Ib = Ic / hfe;
Rb = 3.5 / Ic / hfe = 3.5 x hfe / Ic = 3.5 x 100 / 2.4 = 145.8 Ohm.
For R = 43 Ohm , output current of IC is : Io = 3.5 / 43 = 81.4 mA


Thank you for that information. Where do Vbe and Voic come from? Are they standard values or from the specs of the transistors or the 74ac138?

Thanks again!


Vbe from Fig. 5 datasheet of the transistor you provided above, Voic is standard, Fairchild gives 0.44V / 24 mA  (74AC138SC)

Go Up