0
\$\begingroup\$

I'm designing some LED strips to use to grow plants along a long wall. There will be hundreds of LEDs run by a constant-current high quality LED driver (Meanwell brand or similar). All the commercial strips I've seen (Bridgelux, Samsung, etc) divide the LEDs in sections, wire the LEDs inside each section in series, and then all sections are wired in parallel, like the first example in the image below.

enter image description here

Could anyone tell me why that's better than wiring the LEDs in each section in parallel, and have the sections wired in series, like the second example in the image?

One problem I see with the first approach is that if there's a section where airflow is limited it will create a hotspot where all the LEDs in that section will drop their forward voltage and start hogging more current leading to thermal runaway.

The second approach seems like it would solve that problem since the LEDs heating up are all competing for the same current so it should stay more evenly divided.

Is my thinking correct? Am I missing some other major pitfall in the second example?

\$\endgroup\$
1
\$\begingroup\$

Thermal Runaway is an analog race condition where the +ve thermal resistance and heat controlled negative threshold voltage coefficient from heat dissipated by large differences in series resistance, Rs when shunted with others cause power hogging. Runaway may be modelled as DC thermal-electric negative feedback when the loop gain is >1 and the sensitivity to heat rise accelerates instead of reaching a steady state.

This results in current hogging by an accelerated drop in voltage and rise in temperature.

It is prevented by reducing;

  1. thermal resistance, Rja ['C/W] {positive} and thermal coupling to shared shunt LEDs.
  2. overall temperature rise and reducing the thermal rise time.
  3. the number of LEDs in parallel
  4. the series resistance mismatch tolerance by smaller Vf bins

    1. and increasing the number of LEDs in series.
    2. and increasing the thermal coupling of all LEDs to the same temperature.

The LED with the lowest Vf at the same rated current always has the lowest Rs. This is the source of Vf variations in similar junction LEDs. The power rating and thus bulk size reduces the Rs while rising temperature mainly reduces the threshold voltage, Vt and not Rs.

The electrically parallel shared grid LEDs will always perform worse.

The LED with the lowest Vf and thus lowest Rs shunts more current from the adjacent strings. It will draw much more current depending on the number other LED's in parallel than if there were just N LEDs in series.

enter image description here

  • Margin drops rapidly as component temperature rise increases, accelerating changes.
  • effects of current change are reduced /N LEDs in series

This is why all COB LEDs are done as shown with the left pattern and never in the S/P grid connected pattern.

e.g. 12S8P COB from Seoulsemicon enter image description here

\$\endgroup\$
1
\$\begingroup\$

I'm designing some LED strips to use to grow plants along a long wall. There will be hundreds of LEDs run by a constant-current high quality LED driver (Meanwell brand or similar).

It would help if you added what type of LED they are, rough wattage, whether they have heatsinks/mcpcb.

All the commercial strips I've seen (Bridgelux, Samsung, etc) divide the LEDs in sections, wire the LEDs inside each section in series, and then all sections are wired in parallel, like the first example in the image below.

Yes this is considered best practice. When multiple LEDs are in series they are forced to share current, so the current through the full set is determined by the sum of the "resistances", which is the same as using the average volt drop or average "resistance" of the LEDs.

Could anyone tell me why that's better than wiring the LEDs in each section in parallel, and have the sections wired in series, like the second example in the image?

Could elaborate, but I think this is covered, to look at it the other way, the image on the right shows 3 parallel sets of 3 LEDs, and the 3 sets are wired in series. When power LEDs are in parallel, the first one to heat up or to start with below average "resistance" will hog current, heat up more, hog more current and burn out. Even 2 LEDs in series is a huge aid in preventing this, and the more in the series set, the more the properties of your LEDs can vary and not matter.

One problem I see with the first approach is that if there's a section where airflow is limited it will create a hotspot where all the LEDs in that section will drop their forward voltage and start hogging more current leading to thermal runaway. If your cooling is remotely adequate, having a set of 3 or more LEDs all heat up is very unlikely. In the event it did happen, it's still much less likely than having a single LED heat up, which is enough to start the cascade failure in the drawing on the right.

The second approach seems like it would solve that problem since the LEDs heating up are all competing for the same current so it should stay more evenly divided.

You're making a bit of an arbitrary assumption that you know which LEDs are heating up, but as soon as they're in a situation where thermal runaway is causing them to "compete" for current, failure mode has begun. At least one of them is going to win that competition for current, heat up and burn out, at which point the remaining two will continue the competition until both are destroyed.

Is my thinking correct? Am I missing some other major pitfall in the second example?

You're thinking about the right stuff, you just got it a bit backwards. Current control will go a long way to preventing runaway, so it's the best way to go, but you want to avoid single-parallel LED circuits with current control because when LEDs start to die, the current controller will maintain current, overstressing the remaining LEDs. Having strings of 3 or more helps, and if you want as an extra precaution you can put a fuse in series with each string so that if you somehow manage to actually thermal runaway and burn out an entire string, the current controller will immediately push too much current through the remaining LEDs, blowing their fuses, at which point it will open circuit/overvoltage and shut down.

\$\endgroup\$
1
\$\begingroup\$

I believe the opposite would happen. In the series/parallel configuration, if 1 LED gets hot, it can’t pull more current than the LEDs before and after it. If one entire entire chain of LEDs got hot, then maybe, but what are the chances for one chain of LEDs heating up but not another chain. So overall the series/parallel configuration is more stable and provides even illumination.

In the parallel/series configuration, if 1 LED heats up and takes more current (has a lower Vf) then the voltage across that LED will decrease slightly and the LEDs parallel to it will use less current and become dim. This will lead to uneven illumination and thermal runaway in the 1 LED because the current from the 3 series LEDs before and after the 1 LED in the circuit are trying to pull 3 LEDs worth of current.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.