Problem driving MC-E with Recom driver from 29VDC

Hi all, first post, but I've been researching this forum for a few weeks.

I have a problem with my first high-power LED build that I hope someone here can help me with.

I am building a headlight for my electric bicycle and want to drive it directly from the main battery which is a nominal 26VDC, 11.5Ah LiFePo4 pack. The battery voltage ranges from 29.1VDC straight off the charger to about 21VDC at the BMS cutoff.

Based on research here and on the Mountain Bike forum, I decided to go with a single MC-E LED, series-wired on a round base and a Recom RCD-24-0.70 (wired) driver (both bought from Cutter). I chose the Recom driver as it has an absolute max input voltage rating of 40V and a recommended max input voltage of 36VDC which is well over my max battery voltage of 29.1V.
I intend to mount the LED into a DX E27 housing, but that is the next step.

As a prototype, I mounted the LED on a decent sized heatsink designed for a TO-3 power transistor, with a couple of screws clamping down opposite sides of the pill. I used heatsink compound to ensure good thermal conduction.
I wired up the LED to the Recom driver and also wired up the analog dimming circuit shown in the Recom datasheet using a 4k7 resistor and 1k pot to drive the analog Dim connection. The circuit is shown operating from 24VDC, but I calculated the voltage from the resistive divider at 29V and the max voltage to the Analog Dim input was 5V which is well within the 15V max rating for that pin.

I connected the driver to a 15VDC bench power supply and everything worked fine. I was able to slightly dim the LED with the Pot, but as the 15V input through the resistive divider only gave about 2V max at the Dim wire, this would only dim to about 50%. None of the components got hot and the LED output was nice and bright.

I then connected the Recom driver to the 29VDC battery and switched on. There was a brief low-level flash from the LED and then nothing. I reconnected the driver to the 15VDC power supply, but nothing happened. After a few checks, I discovered that BOTH the Recom Driver and the MC-E LED were dead! :mecry:

Can anyone explain what I did wrong as I don't understand why the circuit didn't work at 29V as I don't believe I exceeded any of the absolute maximum ratings for the Recon driver. I understood that the constant current driver would automatically adjust itself to deliver a max of 700mA through the 4 LEDs in series. Also, in the Recom datasheet, there is reference to a circuit where 36V can drive "1-9 LEDs", so my 4-LED connection is within this range.

I have another driver and another LED, but I don't want to connect them up until I know what went wrong last time. I've just wasted $50 and don't want to do that again.

Any advice would be greatly appreciated. Thanks, JB.

Hi JB- Welcome to CPF, and the led world.

I have blown pleny of LEDs - kind of exciting isn't it. Welcome to the club.

I looked at the data sheet for that driver (found it on allied electronics). It is not obvious at all to me what went wrong. The driver seems to be similar in concept to the leddynamics driver in form and function.

I will throw out some wild ideas just for you to ponder - not knowing if any of them are useful.

a) Is there any chance that the polarity was somehow reversed?

b) Was the motor on when you turned on the driver? Sometimes, these magnetic drivers are really sensitive to electrical surges and noises. There is a diagram on the data sheet that shows an optional class b filter (basically, rf suppression)

One of the few drivers I have seen with real isolation from for automotive type use are the ones from taskled.

c) Technically the driver is indicated for 24 volts, but it seems like 29V is within the spec.

d) I am not sure about your resistor divider approach to the control voltage. It should work, but one way to test things down is to temporarily run it from a 5 volt source just to test it.

e) Is there any chance that you connected all of the grounds together? The LED "ground" is very different than electrical ground. I am not sure if the MC-E star you used is electrically isolated from the LEDs or not. Perhaps that is something to check.

It might be useful to try running the driver through a power resistor - maybe 5 - 10 ohms, to see if things work that way.

I will ask a moderator to move your post to the electronics / battery section. Perhaps you will get more answers there.

Thanks for the response HarryN.
In this case it wasn't exciting or explosive and I didn't even see a blinding flash of light as $50 bucks of bits died.

I'll go through you thoughts:

a. I'm pretty sure the polarity was correct. As I said, I had it working on 15V from a power supply and just moved the input wires from the supply to the battery connections. Maybe I had a brain freeze, but I did check things after the catastrophe and everything looked ok... (I wouldn't be posting here if I just connected things ***-about - I'd be too embarrassed.)

b. The motor was not running, I had the battery on the bench and just connected the LED driver to it and used the switch to turn on the 50A relay I use to isolate the battery. But thanks for the comment. I will look at suppression filters as I know the Brushless DC motor controller is a bit noisy.

c. Why do you say the driver is "indicated at 24V"? in the datasheet there are several circuit examples that show 36V inputs and the absolute max input is quoted at 40VDC! I downloaded a datasheet from a link on Cutter Electronics which took me to the Recom website. (I don't know if I am allowed to include the link here?)

d. I took the resistive divider circuit directly from the datasheet. I also measured the voltages at the pot wiper at both 15V and 29V inputs and they were very close to the calculated values. (The current flowing into the Analog Dim input is about 1/10 the current flowing through the resistor + pot series circuit which is the usual rule of thumb for divider cicuits.)

e. The LED ground was not connected to the input ground. (I also just re-checked that the LED connections are isolated from the star.)

When you suggest "running the driver through a resistor" do you mean a resistor in series with the Vin to the driver, or a resistor in series with the LED (or both)? This is a good idea as it would possibly reduce the maximum currents and voltages the components would see.

Moderators- If you feel this question should be moved, please do so as I really want some answers before I connect up the remaining components as I don't want to do this again..

Thanks, JB

Moving it to Batteries/Electronics...

yes do provide the LINK to the driver, more info is always better,
http://www.alliedelec.com/Images/Products/Datasheets/BM/RECOM/394-0171.PDF

supposed to be a DC-DC , some of the specs dont make sence, and not enough USE information is provided.
then they call it a "step Down" which means it should have worked in the configuration you had it in.
there is a large voltage drop through the driver at high currents, but claim of less than 1W heat in it .

its a DC-DC with huge input output variables, but no physical size or component capacity to fully convert everything like an old school style DC-DC , unless it uses magic (they do), to deal with vast offset voltages .

no use info provided in specs to show that huge offset voltages are recommended, just the usual I/O possibilities, which tells me nothing. see any graphs which show a 2x input and what the output would be?

Voltage control on gate input basically designed for 0-5v probably intended originally for things with duel voltages, one error on voltage dividing , like one side of the resistance lifts and the voltage to that control could go bonkers. which still shouldnt ever send the whole thing over in current. still shouldnt have fried the leds themselves even if the driver croaked.

i would have to put it on a power supply, keep the voltage at the control gate below 5v (thats 2 power supplies), slowly raise the voltage , with current controlled , then watch what happens. dont trust that stuff

did you make an error, heck you fully tested it on the bench, shoulda expected to be able to transfer it from there and had it work. which just leaves being able to Slowly bring it up in voltage on the bench, and see if it maintains the current , when the offset in voltages is very high.

cool project, take pictures

Hi VidPro,

Here is the link to the datasheet:

http://www.recom-international.com/pdf/Innoline-2008/RCD-24.pdf

I bought the RCD-24-0.70/W/X3 (wired version with both analog & PWM dimming controls (six wires)). This has a maximum output current of 700mA which can be reduced by applying a control voltage to the analog dimming wire.

I believe this is a Buck type Constant Current driver.

The battery (input) voltage was 29.1 volts.

The MC-E multi-chip LED was series wired so from the MC-E datasheet, the total Vf at 700mA should be 4 x 3.4V = 13.6V. (When the driver was connected to the 15VDC supply, I measured 13.5V across the chip, which tallies with the datasheet.)

I thought the whole idea of a Buck type Constant Current driver was to allow LEDs to be safely driven from a voltage source that was much higher than the Vf of the LEDs in series.

I thought the driver would automatically increase the voltage across the load (the 4 LEDs) until the current reached 700mA and then hold this voltage/current. I thought you could have one LED or multiple LEDs in series as long as the total Vf of all LEDS was about 1 V less than the input voltage.

If this is not the case, then it is obvious that around 28 volts was applied across the 4 LEDs which would cause their destruction. Is this where I went wrong? This is my first experience with driving high power LEDs from these CC drivers, so please help me understand the limitations here.

Thanks, JB

VidPro, looks like our replies crossed in the ether.

I had a quick look at your link and it says 2009 at the bottom of the page. The link from Recom that I used is Rev: 1/2010 and has a few more example circuits.

I understand your comments about the supply to the Analog Dim, but the datasheet says you can leave the input open if not required. It looks like it is tied to Ground internally as 0v corresponds to 100% output current. so even if the pot had an insulating bit somewhere on the resistor, it should have just gone to 700mA which the MC-E can take.

Maybe you are right. I just should connect the other driver and LED I have and try to power it via a variable voltage supply. The only problem is I don't have a variable supply that goes up to 30V.
I just don't want to waste anothe $50 as I'm not sure what caused the first problem. Also, why did the driver also die? Maybe, i was just unlucky and the driver failed due to infant mortality and took the LED with it??

Regards, JB

i dont know, dont have one, for that price i would buy a taskled driver that does the same thing and has all the specs needed to figure out what to do.

but this is one cool project , and one cool chip thing.
i cant believe you/it toasted both items :-( that sooo gets in the way of testing.

dont stop and be sure to post your end results.
if you had bench tested the 24v , which should work, then you would KNOW. then i would know, then i would shut-up

in the meantime i am just hoping the best for you.

JB_LED said:

so even if the pot had an insulating bit somewhere on the resistor, it should have just gone to 700mA which the MC-E can take.

Regards, JB

Click to expand...

yup how could it go beyond its capability, , jam 24v into the gate, and some diode passes it across to the output and bam.
its just like micro electronice to do something like that

cheap tricks, if you dont have a bench supply that will do the 24v, get a 16ohm wire wound adjustable resister (2-5Watts big fat round potentiometer), stuff it in on between the battery and the curcuit, make sure it is on the 16ohms side providing its most resistance.

little safety resistance there for testing, that can be easily adjusted.
then slowly bring it up, and then you can test safely without loosing the leds.
if everything is kosher after you have turned the pot to its lowest setting, then remove it, if things start going crasy as your turning it up, back it off and find out what is wrong.
cost some $3-5, savings >$25+ if needed.

Hi JB_LED, your understanding of how it should be working is the same as mine. The data sheet on that part looks very much like the buck driver from leddynamics. I have the led dynamics version and it works fine, although frankly, I have never tried to run it at that high of voltage.

The resistor comment was just to consider using a resistor as a "test load" instead of the more expensive LED. The driver should just ignore the load type and pump 700ma through it. With a 10 ohm load, you can easily measure the voltage drop across the load resistor under both your power supply and battery pack to see if something is going wrong.

Since there isn't anything obviously wrong, it might be worth it to contact the mfg. for an opinion.

It probably won't make you feel any better, but 5 years ago we used to pay $50+ each for LEDs with 1/2 the capability of those MC-Es. Blow a couple of those a night and you really felt the budget hit.

One last thing to check would be the ground paths for the MC-E star itself. On many LED packages, the thermal pad is not electrically isolated from both the anode and cathode. It sometimes is, and sometimes it is wired with some protection diodes across the leads and thermal pad to provide some static protection.

If the thermal pad is the type that is not electrically isolated, then the typical method is to used "something" that provides electrical isolation but thermal conduction, such as specialized thermal pastes or a insulated metal core board (such as most stars) It sounds like you did all of these things though so that is not a clear path to understanding the failure either.

VidPro & HarryN,

Thanks for all your comments & suggestions. I don't feel so bad now and you have encouraged me to continue.

I have also worked out an easy way to get around the lack of a 30V variable power supply.

I have some old 12V Sealed Lead Acid (SLA) Batteries from my earlier e-bike builds. If I connect a 12V battery in series with the 0 – 15VDC supply that I have, I should be able to get a 12V to 27V variable supply. Since we know the driver / LED can operate safely off 15V, I would have a method of slowly increasing the input voltage up to 27V which is significantly over the total Vf of the LED and close to the target of 29V. This will allow me to do some measurements to observe how the circuit actually operates. I have 3 DMMs, so I can simultaneously measure Input voltage, Voltage across the LED and Current through the LED while I slowly wind up the input voltage.


I will also make the following improvements before connecting to the power source:

• Add a Diode in series with the driver input to protect against wrong polarity. (In this case, the 0.6V drop across the diode is a Good Thing. I might even add a bridge rectifier so polarity doesn't even matter.)
• Temporarily use a 10 ohm resistor in place of the LED.
• Add the filter components suggested in the data sheet as I think this will be needed during operation to prevent the noise from the motor controller affecting the driver and I can check it all works under controlled conditions now.

At least this gives me a way forward that reduces the risk of blowing up the next set of components and allows me to actually take some measurements during the process.

Thanks again for your suggestions and encouragement. Your support for a newbie is really appreciated.
JB.

Hi JB_LED

I've just had the same problem, although mine was slightly more costly.

I had 3 strings of 5 cree leds each ( red, green and blue) and a recom driver for each string.

Out of that I managed to blow all the recoms and all but 4 leds (luckily the first red died open circuit)

I think the recom passes through the full voltage to the load from startup and then a small but significant time later the current limiting sets in and the voltage drops appropriately. The filter in the data sheet should help/negate this.

One thing I am going to do is put a fuse on the led side to at least protect the leds. I might try 1 amp. I recommend you do the same.

The other thing is that it is very easy to blow components while doing development with a battery as your power supply, put an appropriate fuse straight off the positive lead of the battery.

Meanwhile I'll have to put another order in for more leds.

hey there i also got the RCD 24-0.70/pl/A and tried to run 9 cree mc-e in serial only the white led on a 24v 500w powersuppy connected every thing like it should left anlog and pwm dimming open plugged the power supply and the leds flashed one sec and then dead :-O

what could be the problem? hmm

best janosch