Maker Pro
Maker Pro

Common Reference for individual references

J

Jon Slaughter

Jan 1, 1970
0
http://focus.ti.com/lit/ds/symlink/tlc5947.pdf

This devices uses a resistor to set the constant current value. I am using
up to 24 of these things all tied together. I'd like to have one common
reference for all. (to be controlled by a pot to set the maximum current for
a dimming effect)

Ultimately what would be nice is if I could just tie them all to the same
resistor but I do not know if this will work? Maybe i'd have to use 24x the
resistance? I imagine they simply measure the current going through the
resistor and use that as the current(maybe current mirrors?)

I'm not sure though if this would be a valid method though? It definitely
would solve my problem though.

I believe the formula for setting the current is

I = 41*1.20 = 49.2/R_IREF as it shows on page 4.

so I'd just have to increase R_IREF by 24x(48k should get 24.5mA per IC).

Anyone see any problems doing it this way? (ultimately I probably will use
an active resistor so I can control it digitally)
 
http://focus.ti.com/lit/ds/symlink/tlc5947.pdf

This devices uses a resistor to set the constant current value. I am using
up to 24 of these things all tied together. I'd like to have one common
reference for all. (to be controlled by a pot to set the maximum current for
a dimming effect)

Ultimately what would be nice is if I could just tie them all to the same
resistor but I do not know if this will work? Maybe i'd have to use 24x the
resistance? I imagine they simply measure the current going through the
resistor and use that as the current(maybe current mirrors?)

It is unlikely that a common resistor would do what you want; the data
sheet quotes the reference voltage as around 1.2V, and it seems likely
that one of your up-to-24 packages would have a higher Vref than the
others, and would supply the bulk of the current going through the
resistor, so that one block of LEDs would end up brighter than the
rest.
I'm not sure though if this would be a valid method though? It definitely
would solve my problem though.

Not in a way that you would like.

What you could use would be 24 parallel current sinks; 24 NPN
transistors, each with its collector tied to an Iref pin, each with
its emitter returned to 0V through a 620R resistor, and all the bases
connected to a common, nominally 1.2V control voltage that you could
turn down to dim all your LEDs at once.

If all the transistors were BC107 or BC109 for which Vbe is specified
to be in the range 0.55V to 0.7V at 2mA and 25C, each transistor would
sink between 0.8 and 1.05mA for a 1.2V control voltage, which you
could then turn down by reducing the control voltage,

The tolerance is based on worst case specifications; actual
transistors from the same manufacting batch tend to cluster more
tightly, and you could do even better by sorting your transistors into
batches by Vbe-on.
 
J

Jon Slaughter

Jan 1, 1970
0
John Larkin said:
I'd suggest a separate 47K resistor per chip, with all the free ends
connected together, then to a single pot to ground.
huh? that would limit the current way too much! I assume you mean something
like 2k. I also don't see the point of doing it this way if you have a
common connection except maybe that the resistors can be close to the IC.

I'd rather have one pot and one inline resistor(to limit the max current).

Or separate 47K resistors per chip, with the free ends connected
together, driven by a voltage source that goes from 1.2 volts (leds
off) to ground (max bright). Brightness would be linear on the applied
voltage (or actually on 1.2-V).

I'm not too concerned about linearity as I will have a lookup table for the
RGB to PWM colors. (it will be done emperically since I don't need an
infinite number of colors... or maybe I'll map it to a function and use
that)

I'm just curious as to why you say 47k resistors per chip? What is the
purpose of seperating them? Surely if I have a common point for all of them
the effect of just using one will be the same? Or are you saying maybe there
will be "back flow" into each chip? If so I guess it's better 1 resistor per
chip but I was hoping to skimp on the resistors if possible ;/ (not that
it's a big deal but problem is that if the resistors are not accurate then
some led's might be brighter than others and I would have to deal with that
somehow. I could fix it in the software but I'd rather take care of it up
front.
 
Well!

Voltage on the IREF pin is +1.2V.

Output compliance goes down to +0.5V.

Take one TLC5947 and use it to reference 24 other TLC5947's... feed
outputs of first into IREF pins of the 24 others.

R_IREF will have to scale up an additional 41X. HOPEFULLY that
regulator can cope with the low current.


nice idea, but I'm not sure it would work, the output of the TLC5947
is PWM.

It might give some strange effects with one PWM referencing other PWMs
all running on slightly different 4MHz internal oscillators.

I'd think the most straight forward solution would be a 1k6 resistor
for each TLC5947
driven by 0-1.2V. (reference, pot,buffer) but it depends on the
parts behind resonably
matched. Don't know if the max 4% channel-2-channel and 7% device-2-
device is too much

-Lasse




-Lasse
 
J

Jasen Betts

Jan 1, 1970
0
http://focus.ti.com/lit/ds/symlink/tlc5947.pdf

This devices uses a resistor to set the constant current value. I am using
up to 24 of these things all tied together. I'd like to have one common
reference for all. (to be controlled by a pot to set the maximum current for
a dimming effect)

if you tie all the "Iref" inputs together they'll all get the same voltage,
but not neccessarily the same current, and it's the current that's
important.
Ultimately what would be nice is if I could just tie them all to the same
resistor but I do not know if this will work? Maybe i'd have to use 24x the
resistance? I imagine they simply measure the current going through the
resistor and use that as the current(maybe current mirrors?)

Figure out what the minimum size of resistor that you would ever want to
use is (eg: 1K6) , and use 24 of that size (one from each chip to the wiper of
your pot adjustment pot) that should share the current well enough.

(if it doesn't you'll need to build a 24 output current mirror...
but from skimming the datasheet I expect the resistors will work for
you.)

your combined adjuster should be about 1/24th the resistance of the
individual adjuster, so, if you were going to use a 20K variable
resistor on one chip you can use a 1k one on the group and get about
the same adjustment range.

Bye.
Jasen
 
Top