Ebike 36-43V Headlight / Tail light: Configuration note about DC-DC converter

[Kiriakos GR]

This topic is the recording of my very own last research regarding of how to properly select the necessary electronics so to use on your ebike headlights or tail lights coming from small motorcycles (scooters) or similar.

The original plan was to use one ready circuitry with LM2596S back DC-DC converter so to step down the almost 42 volts of my Lifepo4 36V battery (when is fully charged), as needed according the default volts for its one incandescent light bulb.

The problem with this LM2596S seems to be it specifications which do not say the complete truth.
While the input voltage range is up to 40V the problem seems to be of how it handles the instant current in Amperes.
In my attempt to power up one 6V 15W incandescent light bulb by setting the input voltage at 37V, in just a second of time this LM2596S fried instantly prior any light to come out of the light bulb.
By having a spare and last LM2596S PCB, I did try again and the same thing happened again.

My first conclusion is that from 37V by stepping down to 6V and expecting 2A this is not going to happen.
Could be design limitation? Either way it does not work.

While the same circuit handling beautifully the 12V 3W incandescent light bulb at my tail light, for the headlight I had to change my strategy.

The new plan is to use 24V incandescent light bulb (commonly available) and also one better and stronger DC-DC converter that would be indestructible in my application.
Regarding wattage the plan is to use 24V 15W or 21W bulb, unfortunately at eBay the choices for higher in watts converters are very few.
And I did end up getting the one pictured at the last picture of this topic.

While now I feel safe that as soon the parts arrive, my success would be at 100% regarding my headlight, I did purchased too some additional PCB with  LM2596HV that is the new improved version for higher volts.
This called as LM2596HVS and has an input Voltage of 4.5V ~ 53V which makes it a top solution for ebike applications up to 10W incandescent light bulb.

Regarding the LM2596 what is still unknown to me is if it can really handle 15W at 12V?
15W at 6V is not possible.
15 - 21W at 24V is something that I am going to test even as experiment in few weeks.

Thankfully my headlight is huge and street legal, originally used at Piaggio SI FL2 motor scooter, and there is plenty of space inside for this little beast at 90W of power which unfortunately for the fellows with 48V battery is not compatible.

You can see the complete story of my ebike build at the link below. 
http://www.ittsb.eu/forum/index.php?topic=963.0

Additionally if you have performed any experiments with these LM2596HVS regarding Watts and Volts handling, feel free to post your experiences in this topic.

[GeorgeVita]

Hi Kyriakos!
I think that the problem is the resistance of the cold filament into the lamps. A 6V/15W lamp draws 2.5A of current which means 6V/2.5A=2.4Ω (hot filament) and 2.4Ω/15=0.16Ω (cold filament).

One solution is to use an NTC thermistor to limit inrush current. The NTC thermistor must have enough cold resistance acceptable for the regulator, but a minimum resistance at working temperature to minimize wasted power. Possibly the output voltage will be set to 7-8V (ex. for 6V lamp) with excessive voltage drawn on the NTC.

Alternatively the design will be approached including a current regulator after switching voltage regulator (ex. 16-18V regulated and then LM317T for 1.25A regulation. The lamp will be a 12V/15W type. Some calculations/tests must be done due to dropout on LM317. If you change the switching regulator, some other types work in constant current mode with short circuit protection.

Although I like designing electronic circuits, your application seems to be perfect for a custom LED light lamp solution!
Use some highpower LED chips, calculate appropriate resistors and fix all assembly into the old headlight. Some kind of heatsink will be needed.

P.S. If you insist to use incandescent lamps, use 3x12V/5W in series!

Regards,
G

[Kiriakos GR]

Hi George !

Thanks for your analysis which show the root of the problem in detail.
My own ( theoretical at the moment ) work around, is to use 24V 15W incandescent lamp and the powerful version of DC-DC converter which is capable for nominal 10A and is protected by an 20A Fuse.

Basically just my selection of the 24V incandescent lamp, it does remove the high inrush current issue.
15W divided with 24V equals to 0.6A , but even so the new issue is the limit of LM2596 regarding total handling power in long term.

In my research for another product solution it seems that there no scalability, and therefore I was forced to get something three times more powerful in Watts than needed regarding converter.
Hopefully in three weeks the most I will have the parts at hand, so to confirm my theory about the LM2596 and the 24V incandescent lamp.
 
Regarding LM317 I did see some one to use it as DC-DC converter, but my quick thought was that I can not trust it.
In the end I did prefer to combine bulletproof solutions which will simply work.
It is a fact that electricians they do think differently than product designers. 

[GeorgeVita]

In case you need an NTC for the 24V/15W lamp (filament Rcold=2.67Ω, Rhot=40Ω), take a look at:
EPCOS P11 NTC thermistors for inrush current limiters
Part# B57211P0470M3 has 47Ω initial resistance and aprox. 1Ω at your working current.
A small overhead in your regulated voltage will fix luminusity to normal levels.
G

[Kiriakos GR]

Your messages are always food for thought !

I = V / R   ->   24V / 2.67 Ohm = 8.98A  Peak inrush current.

Therefore even the lower value NTC at 2.5 Ohm is going to assist about to drop down the Peak inrush current in to a safer 5A.
But the new problem seems to be the  power specification of 3W at the specific NTC family.
I am a bit confused if this value is about the constant Watt which it can handle ?   
 

[GeorgeVita]

As I know, power rating of NTC thermistors apply to the power consumed on the component itself @25°C.
From their characteristics table, if you calculate (Imax@25°C)^2 x Rmin@Imax@25°C the result is 20-25% below the power rating value which is within limits of resistance tolerance (+/-20%).
Concluding, just choose the lowest ohmic value with Imax 2x or 3x your normal working current.

You cannot use an NTC alone for short circuit protection. As the current flows, the temperature on the NTC will force it to minimize its resistance increasing more the current flow.

Using NTC thermistor for your application, the only problem could be while flashing the light (fast turning on-off-on...). If you choose a component with a small Imax, may not cool enough compared to the filament between flashings of the lamp. You have to do some tests.

[Kiriakos GR]

Just received my new LM2596HV Step-down Module and thought to sacrifice just one of the five, by testing it so to discover if it is capable to handle the stress which one 24V 21W incandescent light bulb creates.

The good news after the first 25 minutes of stress tests which has not end yet, is that yes it can handle the load of 21W with out getting in to flames.

The only downside by working at 21W instead of 15W (PCB design with out cooling) is that now it gets warm up to 62C at the highest heat spot over this PCB.

My room temperature is at 15C (winter time) and even if the PCB does not seem to work at it destruction limits, definably it is helped to operate for 30 minutes up to now with out needing any heat sink.
The temperature has stabilized to about 65C and does not climb up any further.
The voltage stability is good but I had to readjust it after the first 10 minutes of operation because from the target voltage of 24V it drops down 150mV when the PCB started to become warmer.

My news report includes definably good news, but personally I going to use the larger version of Step-down Module in Watts which has not arrive yet.
In my project, the headlight is plastic and I do not wish to have inside another heat source other than the incandescent light bulb.

Until to finish writing this text and taking pictures, one complete hour of flawless operation has been completed, and the plan is to let it run for just another one hour,  mostly to check individual components temperature.   

[Kiriakos GR]

Just received this 160W DC-DC module which is specified as adjustable step-up converter by it description.

Unfortunately it output can not be adjusted below the input voltage, yes it is called as step–up for a reason, but due the not that clear description regarding specifications I had the hope that it can serve after all.
Even so now this ZS-PD9 module it does seems useful for other use than the original reason which I got it for.

Technically ZS-PD9 module is based at quad Darlington switches and actually the UDN-2843B IC.
The actual IC over this PCB is 2843B PEZR which seems as unknown to all search engines.

The first use that I can think for the specific module is to boost my regular professional bench top power supply 0-37V 0-10A so to act as high performance charger for my e-bike battery at 43V.

Now that there is no other option, my alternative is to use LM2596HVS with a true heat sink at the back of it.
At my first experiment I tested it for three hours with out heat sink, and by turning it off the PCB temperature recovered to almost room temperature in few seconds.

Below are few pictures of ZS-PD9 in case that some one is interested to see more details about it.

[Kiriakos GR]

And now I will present my very own solution of getting safely most or all performance from LM2596S (LM2596 HVS).

By using a regular heat sink from a dead SONY PC monitor. 
One piece of aluminum was created so to serve as PCB raiser so all PCB to elevate just a little so input / output pads and pot contacts to float on air, and at the same time the back plate to contact with the heat sink.
Few brushes with what ever type thermal past will help too.
The most crucial in this design is the unseen... you do not have to tight the screws, just a bit of gentle pressure is all that it takes.
The spacer with the thermal paste will do it job perfectly, the screws needs just a softly tightening as to were a spring holding down the PCB.

Neither the thermal paste placement is important, this is power electronics and not a CPU.

I am up to soldering the cables so to enjoy the 100% successful results.   


.