I was thinking last night about my current project, and the heat generated by 7135 chips. How “real” of a condition is this? I mean, running a double stacked board at ~4-4.5 Amps off a bench power supply (theoretically perfect power), I can see that getting them hotter. The power supply is going to maintain ~4.2V and the 7135 is going to convert that to heat as it bucks down. As compared to a single cell battery (real-world) thats going to voltage sag under load right? So in the real world scenario the 7135 is really only bucking ~3.6V down to Vf… a difference of only .2-.3 Volts?
Run a bench test and hit the chips with an IR thermometer?
I’ve looked at lots of graphs showing voltage vs runtime at different discharge rates and they all sag some. A common figure was ~20% runtime to drop to 3.7V for 1C draw or less but it’s a curve so I’m curious about their performance in the first few minutes. How hard can you push them with a good heat sink(separate from LEDs). I know you can mount these chips off the board; maybe to a dedicated ground/heat sink?
You could look at your batteries discharge curves to see how much they sag under real world load and set your bench supply to that voltage. . For example, a test i did with a sanyo 18650 fully charged to 4.18v, the cell volts instantly dropped to 3.97v under a load of 2.8amps. In my case the 7135 chips were dissipating 3.97v - 3.3v x 2.8a = 1.9w total. If i fed them with 4.18v from a fixed supply 4.18v - 3.3v x 2.8a = 2.5w total. Not much difference compared to the heat from the led, so at the end of the day i would feed them with 4.2v as your leads will be dropping some voltage anyway.
These will get hot for two reasons:
- they are a linear regulator, so the extra voltage above the vf times the current is wasted heat
- the construction of these boards (specially with chips on both sides of the board) is the worst possible scenario for heat dissipation
If you can find a way to heatsink the boards, they will perform better and last longer (heat is not good for long term reliability).
I just finished a custom 3x 7135 board for a customer of mine, and I not only put the 7135’s on the top side, but I also added thermal vias to efficiently conduct heat to the bottom of the board and then to the heatsink:
Custom Linear LED Driver
Will
If you do a “7135” search on ebay, you can see a 20mm Soeul board preloaded with them. I agree that the 8x boards could be improved, but at least they’re out at the ground ring where the improvement can be made.
Yes, you can find the generic Asia boards with various quantities of 7135’s, but my customer needs custom software and access to pins for I/O, as well as a slightly larger hole for screw mounting and two holes for wiring through the board.
Here is a mock-up of the screw mounting and the two holes for the LED wires:
Will
I was thinking about soldering some copper stock, cut into strips or ~22 AWG solid core copper wire to the large solder tab sticking out of the side of the stacked 7135s. Its for a 4-4.5A stanley fatmax mod with a lot of extra room, so I could make the copper up to.5” long to serve as a heatsink.
Would that be worth the effort? Or does the 7135 need to be SMT soldered to really get rid of the heat?
Curious… what kind of heat do these things generate anyways? I mean… ballpark are they in the 40C (comfortably hot shower) vicinity?
No Criticism intended, implied or otherwise. That’s a nice board. I only mention the stars as; 1, I had not seen them mentioned elsewhere and 2, they were mounted to a heatsink board and not pcb material.
I may have misunderstood the data sheet( easy to do for a dabbler like me), but I think the sot-89 package is rated to 700mW with good heatsinking. This equates to dropping 2V at 350mA and I’m not sure if this is a max or if better performance can be achieved with better heatsinking. The only relavent max I found was Temp at either 125C or 150C depending on model. To me, better heatsinking would be more mass under the ground tab AND a separate path to ambient from the LEDs.
What kind of heat? The chips will generate all of the extra voltage * current as heat, until they hit their internal temperature setpoint and shutdown (per the data sheet). But in my own experiments with just 3x of them on air, without a heatsink, the output current decreases as the chips heat up (which makes sense). So any/all that you can do in terms of heatsinking “is” worth the effort.
But I would like to suggest another alternative to direct soldering to the board itself: a thermal pad or thermal adhesive. On the thermal pad, you would need something to keep the pad under pressure to allow the pad to do its job (you need some mechanical way of keeping the thermal pad pressed onto the 7135’s AND you need this sandwich to them be pressed against a real metal heatsink). This pad is very good:
TGF120K
As to an adhesive, something like this is dual-side sticky, so it does not need a machanical holding “thing” to keep the board attached to the metal heatsilk:
Bergquest Bony ply 100
Will
They are rated to 125C. How hot the actually get depends on how much voltage they must burn off vs the quality of the heat sink.
All of the 1/2” tabs could be the same piece of copper as this is gnd.
I was thinking of a section of copper pipe, as large a diameter as possible and long enough to make notches to solder the 7135 chips into. Probably the equivalent to awg10 solid in mass. This puts the pins pointed up where they can be connected with 2 rings of wire to the controller and led-( the middle pin is a redundant gnd pin). A notched piece of 1/4” cu tubing formed into place might work as well. At some point it gets difficult to solder the chips to the ring so pretinning is advisable( I do it as a matter of course). I’d like to see some anecdotal evidence of success but it may cost me some chips in failure analysis(fancy for poof).
Hmmm…
I have XP-E N3 reds which are regulated using 7135s, they have a forward voltage of <2.3V at 700mA (what they are each being driven at). Using the temperature probe on my DMM I am getting over 60 degrees C when it is placed on the large ground at the back of the 7135 (the only double stacked pair on the board). Temperature climbs about 1 degree every 5 or so seconds beyond 60 degrees C.
Total current being delivered is 2.1A while the battery is at ~3.75V however each chip only deals with 350mA current and those double stacked would likely have just as good heat transfer to the outer ring compared to the single ones. To be honest I do not see heat build up as a problem provided the ground ring is connected /pushed up against something metal. Temperature on the ground ring taken as far as possible from the 7135 chips was <30 degrees while the chips where at 60.
The circuit board the 7135s are attached to is not inside any enclosure (so any heat being dissipated must do so via the air that surrounds the board).
Here are my quick calculations (worst case, with battery voltage at 4.2V):
LED type: XML
forward voltage at 1A = ~3V
f———- v——— — 5A = ~3.5V
4.2-3=1.2V
1.2V x 350mA = 0.42W per chip
for 5A that is 3.6W of heat (with 3.5Vf) that’s about the same as a XML being driven at <2A right?
XP-E Red
2Vf at 200mA, 2.3Vf at 800mA
4.2-2.2=2V
2Vx2.1A = 4.2W total heat (more then the XML at 5A)
No typical battery is going to maintain 4.2V under 1A load let alone under 5A load. More realistic is about 3.9@1A and probably 3.6 at 5A… And that’s from a quality cell.
Hence, the calculation is:
3.9 - 3 = .9V
.9V * 1A = .9 Watts of heat in the regulators.
What the heck is this calculation? You reference 2V@200mA, 2.3V@800mA and then do the calculation based on 2.2V@2.1A
Makes no sense at all. You also reference battery voltage as being 3.75V when delivering 2.1A above, but again, do calculations based upon a 4.2V battery supply. Additionally, An XML at 5A generates about 17 Watts of heat. 4.2W is NOT more than 17W. I do not understand your calculations.
My reply was a little rushed however to summarize my first reply above:
Temperature measurements were taken with a battery reading 3.75V, if I had a battery that was fully charged I would have used that but I do not.
The different forward voltages were given to show the difference in Vf with a lower current from the same LED. Not really relevant I know however it does show that the regulators have less voltage to burn off the more current the LED is using.
The 3.6W of heat comes from 7135s delivering 5A total, not an XML at 5A (heat-sinking an LED is a different matter entirely). Sorry I did not make this very clear.
4.2V was used as a “worst case scenario” (such as a regulated power supply as mention at the start of this thread). The word “battery” was not suppose to be there. 
That last one I did not put too much thought into: The 2.1A is being delivered to 3 XP-E reds all individually regulated (in parallel) to 0.7A (hence 2.1A total) from the same board. 2.2V comes from the LEDs being driven at that 0.7A.
If you are good at reading between the lines all of the above should be there however I could have made it clearer.