Why not just one (vs. 7) resistor(s) with 7-segment LED display

Thread Starter

ericwertz

Joined Aug 26, 2009
14
Can someone tell me why one never(?) sees the use of just one current-limiting resistor on the common side of the LED display, rather than one on each of the segment connections? It's not immediately obvious that it changes the current-hogging situation... does it?

Assume that an adequately rated resistor is selected to handle the worst-case 7x current.

TIA
 
Last edited:

Ghar

Joined Mar 8, 2010
655
Diodes have large changes in current for very small differences in voltage and all diodes will be slightly different.
If you put them in parallel you can't really predict what diode will conduct what current, they won't share equally.
Even worse, as a diode heats up its forward voltage drops, making it conduct even more current.
 

eblc1388

Joined Nov 28, 2008
1,542
Can someone tell me why one never(?) sees the use of just one current-limiting resistor on the common side of the LED display, rather than one on each of the segment connections? It's not immediately obvious that it changes the current-hogging situation... does it?
The main reason is with the display brightness variations between "1" and "8". Only two LED segments for "1" but all seven for the "8". Remember your resistor value is fixed and all segments are in parallel after this resistor, so the "per segment" current will vary widely, depending on what is displayed.

The effect may not be so apparent alone with only one digit but it looks very ugly when several digits are placed side by side, displaying different numeric values.

Assume that an adequately rated resistor is selected to handle the worst-case 7x current.
Nothing to do with the wattage of resistor.
 
Last edited:

Thread Starter

ericwertz

Joined Aug 26, 2009
14
The thing is that I do realize that the FVs are going to be different from LED to LED, and that the resistors are going to vary by as much as 20%. So right off the bat, there's a realization that there's significant variations between branches even if you parallelize LED-resistor pairs. So at that level, we know that the branches are not going to share equally anyways, and it just feels like such variations shouldn't be an issue unless one is operating near the margins.

The problem seems to be that there's going to be a voltage at the common node that produces a larger than "average" forward voltage across one of its siblings, if any varies substantially from the others. Is it that the voltage drop across the shared resistor is too loosely coupled (because it's being driven by N-1 other branch currents) to the current flow through any individual branch to do its current limiting job adequately?

Once the voltage is established at the shared node, the LED with the lowest FV (forward voltage) starts pulling some more current, which in excess is clearly A Bad Thing.

Even though the voltage across the common resistor increases due to the increased current through the lowest FV LED, the resistor's voltage rises too slowly to hold down the forward voltage across the diode (as the two of them are mutually voltage dividing) because the current rises at a monsterously faster rate. And past some point out on the I-V curve of the LED the whole thing runs away.

Is that what's going on?

I still think that I need to sit down with some numbers and I-V charts and go through it. I guess I need to work though the exercise of convincing myself that there either is or isn't a workable value of R(shared) for N parallel LEDs that makes this work. It seems like there has to be such a value for R(shared) that guarantees safety -- perhaps it's just too large to viably drive the LED at any reasonable absolute (or relative) brightness.

I'm not looking to get maximum luminous output from an expensive HB LED, just something that reasonably displays a number in ambient light with a lot fewer resistors. Perhaps it's already well-known that such design window doesn't exist...?

Edit: @eblc1388 -- didn't see your response until after my reply. Digit-to-digit variance would certainly be a concern. Ah, right about number of on branches being insignificant. Nice -- thanks!
 
Last edited:

Ghar

Joined Mar 8, 2010
655
The thing can't run away entirely because the resistor is still there. The absolute maximum current possible is V/R with VF = 0.

I think you've got it when you say that the resistor is loosely coupled to the current in each branch. Its current is equal to the sum of all the diode currents.
That means you could have 20 mA in the resistor and 10 mA in two diodes or maybe 5 mA in one diode and 15 mA in the second.

The diode with the lowest voltage drop will clamp the voltage on the other diodes, reducing their current. For example if you had 1 diode in parallel with 2 diodes in series, the 2 diode string will conduct essentially nothing.
 

Thread Starter

ericwertz

Joined Aug 26, 2009
14
The thing can't run away entirely because the resistor is still there. The absolute maximum current possible is V/R with VF = 0.
Yes I suppose, if it survives the trip... :)

So it's pretty clear then that if one chooses R(shared) for the worst (w.r.t. survivability) case with only one segment on, that the 6/7 or 7/8 current drop with all segments might not cut it.

Thanks for all the help. What's sad is that I'm 100% sure that I knew the answer to my own question 20 years ago, back when.... uh, I forget.

thanks again!
 
Top