Automotive 3/6/9A driver - feeler thread

[DIWdiver]

Hi gang,
I'm considering developing a new driver for the high-power LED automotive market.

My big question is: would anyone want this instead of DerWichtel's driver?

Characteristics of this driver would be:
Similar in operation and based on the IS1006 driver.
Suitable for use in 12 volt automotive application without external components (except hi/lo switch and LED).
Linear Current regulation - ideal for LEDs, won't cause electrical interference.
Current up to 10A, possibly more.
1 or 2 modes, either or both adjustable.
Modes selectable by toggle switch, not clicky, etc.
Thermal protection for the driver, which can also protect LEDs if designed properly.
Linear regulation, which means only moderate efficiency, and requires 3-4 LEDs in series, and good heatsinking for the driver.
Price $35-40, including thermal mounting pad.
Size around 1.5" x 1.5" x 0.5".
Mounting by single 6-32 or M3 screw. A second mounting hole would be provided for optional use.

I'm looking to see what the interest is in such a driver, and any suggestions for features I haven't thought of.

Suggested uses:

Three SST-50s or SST-90s in series, full output when engine is running, reduced output when engine is off and battery is not fully charged.

3, 6, or 9 XM-L in 3S1P, 3S2P, or 3S3P configuration, full output whether engine is running or not.

4, 8, or 12 XM-L in 4S1P, 4S2P, or 4S3P configuration, full output when engine is running, quite reduced output when engine is off. Can be tricky to reach full 3A/LED.

Could also be used for single or dual (2S1P) XM-L or dual SST-50 applications. The efficiency would suck, but maybe that doesn't matter.

Any comments are welcome.

[Mattaus]

I'm open to any drivers developed specifically for automotive use - there are not enough doing the rounds. I've fiddled with developing my own LED light bar for my 4WD in the past (other projects have taken precedence however) but one thing I struggled with was driving the emitters. Realistically you can't use generic flashlight drivers.

Some things that need to be considered (and I'm no electrical expert) is voltage spikes especially when the car engine is running. How will your driver deal with these spikes?

That price seems excellent if you can keep it at that...

EDIT: Der Wichtel goes through a potential protection method in his 9A buck converter thread involving Zener diodes and capacitors. This is all well and good but I am eluding more to an built-in method. Less work for people to have to do, and closer to the traditional wiring job that we are all probably more used to. Just a suggestion :-)

[drmalenko]

This would be awesome.. now if only someone could make a cnc machined housing for some sst-90's we would be all over this....

[DIWdiver]

This would be awesome.. now if only someone could make a cnc machined housing for some sst-90's we would be all over this....

I could do that too, but you wouldn't like the price!

[DIWdiver]

I'm open to any drivers developed specifically for automotive use - there are not enough doing the rounds. I've fiddled with developing my own LED light bar for my 4WD in the past (other projects have taken precedence however) but one thing I struggled with was driving the emitters. Realistically you can't use generic flashlight drivers.

Some things that need to be considered (and I'm no electrical expert) is voltage spikes especially when the car engine is running. How will your driver deal with these spikes?

That price seems excellent if you can keep it at that...

EDIT: Der Wichtel goes through a potential protection method in his 9A buck converter thread involving Zener diodes and capacitors. This is all well and good but I am eluding more to an built-in method. Less work for people to have to do, and closer to the traditional wiring job that we are all probably more used to. Just a suggestion :-)

DW suggests clamping everything at 20V. However, in the worst cases, there can be a huge amount of energy available at 20V. That's why he suggests using 5 diodes in parallel. I'm not sure even that's enough in the worst cases, but then most vehicles will never see the worst cases.

My board would clamp spikes (high voltage, but low energy) at around 80V, and survive surges (lower voltage but higher energy) up to 80V. The driver would survive a short time at 80V, but not a long duration. This approach requires a high-voltage mosfet, which doesn't work well at really low voltages, but that's not an issue here. It also doesn't require large capacitors or multiple diodes.

[Mattaus]

OK, so protection is built in which makes this driver perfect.

[SemiMan]

Unless I am building them by the thousands, I don't really care about EMI. If I am building them in the thousands, then I am not going to pay 35-40 for what amounts to a simple linear regulator.


FYI, the HV mosfet is really not an issue for this application since you would almost never be fully enhanced so RDSon does not really matter too much. That said, since it's linear, you are going to clamp at a low voltage any way.

Just not seeing the value in this too much.

[SemiMan]

Let's also keep in mind:

- 10Amps
- 14+ volts + when the car is running
- 3 leds in series, say 10.5V forward
- That's a minimum of 35watts of dissipation and could be higher

Takes a heck of a good sized heat sink to dissipate 35 watts ... much more than can be mounted with a #6 screw.

[Moddoo]

I've been running one of Don's automotive drivers to 4 XML in series @ 5A on the front of the truck for maybe 6 months now.
Thermal interface is a screw to the light case with some Arctic silver.

No problems at all so far.

I am getting ready to expand the setup to 20 leds.

These work great, and the high/low function is nice.

[Mattaus]

Out of curiosity, and not to sound too silly, but which Don are we talking about and which driver?

[DIWdiver]

Out of curiosity, and not to sound too silly, but which Don are we talking about and which driver?

The Don he's talking about is me. The driver is the IS1006 discussed extensively here:
http://www.candlepowerforums.com/vb/...w-and-Improved!

While not originally designed for automotive applications, with a few component changes and two external components, it adapted pretty well for that.

What I'm considering now is a driver designed specifically for automotive applications, so it wouldn't need external components. Because of the large power dissipation pointed out by SemiMan (and discussed elsewhere as well), the FET will be in a larger package to help with heat transfer. I've also thought that temperature protection would also be a good idea, given the high power levels that are readily available in a vehicle, which can easily overdrive even a very large heatsink. People are discussing builds up to 500W (LEDs + driver) with multiple IS1006s.

[DIWdiver]

Unless I am building them by the thousands, I don't really care about EMI. If I am building them in the thousands, then I am not going to pay 35-40 for what amounts to a simple linear regulator.


FYI, the HV mosfet is really not an issue for this application since you would almost never be fully enhanced so RDSon does not really matter too much. That said, since it's linear, you are going to clamp at a low voltage any way.

Just not seeing the value in this too much.

You might care about EMI if you were trying to listen to the radio in the vehicle with the lights on. Others have commented on severe interference from cheap imported drivers. I would expect DW's driver to be better, but any switcher will generate broadband EMI. It takes careful design and mitigation to minimize (never eliminate) this, and that don't come cheap.

If you ordered 1000 pcs, I'm sure I could dramatically reduce the price. The estimated price is for 1 pc. with total production in the dozens.

You're right that the high-voltage FET isn't a big deal. But it isn't trivial either. You want one with good heat-transfer characteristics, which means a big die. And if you want to run 4 XM-Ls in series when the vehicle is off, you don't have any overhead on the regulator. Four times 3.35V is 13.6V, so in this case your overhead is less than zero and ANY resistance in the FET costs you output. In this situation the FET would be fully enhanced. The question isn't whether the output is reduced, it's how much is the output reduced, and that depends partly on the FET.

As far as not seeing the value, I welcome that input. In fact, I question the value myself. The reason I started this thread is to see if enough people see the value to make it worth my time to build the driver. Also, I wanted to see if anyone had ideas that could add to the value.

[Mattaus]

Thanks for clearing that up - there just seem to be many 'Don' characters here on CPF it can make keeping tabs on who's doing what a little difficult. It also doesnt help that Der Wichtel makes drivers, and his drivers are commonly referred to as DW drivers which, as you can probably see where I am going here, is close to DIW.

Or maybe I'm the only one getting confused


I think demand for these drivers would increase if most people realized that while you can use normal off the shelf drivers in automotive applications, it's not the best or most robust long term solution. At the moment I don't think that sort of knowledge is out there, and it might take a few actual builds to get the word spread. If you ever get these going I am definitely in for a handful (lets say 5 or 6) but that would only be after I get my new Amarok, and that's still a number of months off. The girlfriend won't let me build anything in advance...apparently 'enough is enough'

So for me personally I sincerely hope others jump on this just so that there is something for me to get down the track!!!

Cheers,

- Matt

[SemiMan]

You might care about EMI if you were trying to listen to the radio in the vehicle with the lights on. Others have commented on severe interference from cheap imported drivers. I would expect DW's driver to be better, but any switcher will generate broadband EMI. It takes careful design and mitigation to minimize (never eliminate) this, and that don't come cheap.

If you ordered 1000 pcs, I'm sure I could dramatically reduce the price. The estimated price is for 1 pc. with total production in the dozens.

You're right that the high-voltage FET isn't a big deal. But it isn't trivial either. You want one with good heat-transfer characteristics, which means a big die. And if you want to run 4 XM-Ls in series when the vehicle is off, you don't have any overhead on the regulator. Four times 3.35V is 13.6V, so in this case your overhead is less than zero and ANY resistance in the FET costs you output. In this situation the FET would be fully enhanced. The question isn't whether the output is reduced, it's how much is the output reduced, and that depends partly on the FET.

As far as not seeing the value, I welcome that input. In fact, I question the value myself. The reason I started this thread is to see if enough people see the value to make it worth my time to build the driver. Also, I wanted to see if anyone had ideas that could add to the value.

80+V N-channel FETs under 10mohm are pretty inexpensive these days (rated in the 100+W range too). Even P-Channel are pretty much there and with obviously simpler drive circuitry. 10mohm (over temp) at 10 amps is 100mV. Odds are wiring is going to be a bigger issue than the drivers. Softening the edges on the switcher so that you get 80's instead of 90's efficiency can eliminate much of that noise issue as can a bit of filtering. I find that once you hit high powers, it is cheaper to put the money into filtering versus aluminum, etc. for the heat sink. It is great if you can use the vehicle, but steel makes for a poor heat sink. Can't say I have ever heard any noise on my car stereo when I have had LED drivers hooked up to the battery, some of which have been 50+ watts. Many car stereos have switchers for higher voltage for their amplifiers these days ... home stereos too. If you are going to build 10, then yes likely not worth the time to work out the noise issues, but lots of high voltage, high powered switch mode controllers that will take load dump without any issues.

Semiman

[DIWdiver]

80+V N-channel FETs under 10mohm are pretty inexpensive these days (rated in the 100+W range too). Even P-Channel are pretty much there and with obviously simpler drive circuitry. 10mohm (over temp) at 10 amps is 100mV. Odds are wiring is going to be a bigger issue than the drivers. Softening the edges on the switcher so that you get 80's instead of 90's efficiency can eliminate much of that noise issue as can a bit of filtering. I find that once you hit high powers, it is cheaper to put the money into filtering versus aluminum, etc. for the heat sink. It is great if you can use the vehicle, but steel makes for a poor heat sink. Can't say I have ever heard any noise on my car stereo when I have had LED drivers hooked up to the battery, some of which have been 50+ watts. Many car stereos have switchers for higher voltage for their amplifiers these days ... home stereos too. If you are going to build 10, then yes likely not worth the time to work out the noise issues, but lots of high voltage, high powered switch mode controllers that will take load dump without any issues.

Semiman

Clearly a properly designed switcher is a much better choice than a linear at high power levels. The fact that today's cars are loaded with them (including, I imagine, high power LED drivers for headlights) is proof of that. Unfortunately, there doesn't seem to be a DIY one readily available for high-power LEDs in an automotive environment. DW's is the closest, and it needs 10 external parts to provide reasonable (not bulletproof) protection.

Since I don't have the ambition to go to the effort necessary to develop a proper switcher that will survive, won't wipe out your radio, will be efficient, and sell at a reasonable price, at small quantities, I wonder if anyone else will either.

If I wanted to build a light bar like people here are using the IS1006 for, I'd probably build exactly the board I'm talking about. It just isn't worth the effort to do anything else. Or maybe I'd just use a few IS1006s. Oh, who am I kidding; I'd build the board. It's what I do.

When I designed the IS1006 a few years ago, I expected it would be obsolete within a few months, given that two CPF members were working on switchers. I decided to buy 30 PCBs, planning to break even on the cost if I sold 20 pieces, make a few bucks if I sold more. I lost track somewhere around 70 pcs, and just shipped two yesterday. So it's not a huge commercial success, but it filled a niche need, which was the original intent.

The big question today is 'Would a similar driver tailored to the automotive market fare as well?' Given that the response to my feeler is very excited people who are very few in number, I don't think the answer is yet clear.

D

[register_990]

Hi,

Glad to see I'm not the only person in the world struggling to find a driver that will work in the automotive world.

Wouldn't you only have to design for 40 volts if you used AUML's in the protection circuit?

http://www.littelfuse.com/products/Varistors/Multilater+Varistors+(MLV)/AUML.html

Good luck

Mark

[DIWdiver]

Hi,

Glad to see I'm not the only person in the world struggling to find a driver that will work in the automotive world.

Wouldn't you only have to design for 40 volts if you used AUML's in the protection circuit?

http://www.littelfuse.com/products/V...MLV)/AUML.html

Good luck

Mark

Yes, those look like pretty good parts. They are expensive though. I ran across them the other day while researching load dump protection, for this project.

However, in this particular case, the difference between designing for 40V and designing for 80V is pretty minimal (often it isn't). I haven't been through a careful design process yet, but I suspect that when that is said and done, it would turn out to be better to design for 80V and save the cost of these parts. Still, it's a good tool to have in the toolbox.

D

[SemiMan]

Actually for real automotive design, it can be far harder to design for double battery (boosted wrong) and failied regulator (19 volts typ for test) as now you can be dissipating some serious power with a linear solution. Usually a fold back is the best option and simplest. With a switch mode, this can be normal operation.

Semiman

[DIWdiver]

That could be a problem if the LED lights are on when you try to jump the vehicle or if the regulator fails. I suppose it would be a good exercise to investigate whether the thermal limit would protect the driver before it gets destroyed. FET die temperatures would rise pretty fast if the input were 24V and the output was 9A @ 10V! 126W isn't unreasonable for the FETs I'd be using, subject to proper heatsinking. If the thermal limit sensor (thermistor) is well coupled to the die, it would be okay. If not, there could be a failure.

Perhaps an 'overvoltage cutout' circuit would be a good addition. I'll have to look into the thermo-mechanics of it to know.

Thanks for the input.

[SemiMan]

You can do wonderful things with a 12V zener and resistor connected at the right spot. It makes for a simple voltage driven current fold back. There is no expectation of proper operation past about 14.8V or so in automotive. Things just need to survive past that voltage, not operate normally.

[Foxx510]

I swear I remember reading that alternators in some new cars were going above 15v to charge calcium batteries properly.

[mvyrmnd]

I've always had my eye out for a proper, dedicated automotive LED driver. This sounds promising, and I hope it comes to fruition.

[mulderfox01]

Would have to see actual specs before ordering, but i'd probably take a few

[Imapilot]

I am looking to build something for a small motorcycle's electrical system as well. I obviously don't know the complexities of it yet. But ready to go drivers are a good start! I think after seeing the rallys this year with all the LED lighting, Its a market that can only grow from here.

[Trev]

Hell yes i would be interested in a few of these!

Any word on if your going to go through with production?

[DIWdiver]

Yes, it looks like I will be doing it. I'm starting a PCB layout now.

Don't get too excited yet though, it may take a while.

I'll let you know when I know for sure what it will look like.

[DIWdiver]

Okay, the board layout is done and I'll have prototypes in a few weeks. Here are some specs on the protos:

Board size: 1.25" x 1.375" (32mm x 35mm)
Max Height: about 0.36" (fixed voltage version), about 0.42", adjustable version
Output Current: up to 13.5A
Input Voltage (operating) 10-15V
Input Voltage (withstand) 0-85V
Dropout voltage: 0.2V at 10A
Modes: 2 (high, low)
Mode selection: jumper, switch
Control:
-fixed version can be controlled by PWM or by external potentiometer
-adjustable version has on-board potentiometer for 0-fullscale adjustment
-compatible with HALLSW and HALLTG from Taskled (by design, not tested)
-High and Low modes are proportional to control value.
Thermal protection:
- onboard thermistor to protect driver
- input for external thermistor to protect LED (compatible with thermistor on CSM modules from Luminus)
Transient suppression: 85V unidirectional TVS diode.
Cable recommendation: For external control with cables exceeding 12" in length, it is recommended to use shielded cable, with shield grounded only at the driver end. Control can be on wire as small as 30 awg, power cable should be determined by max current, cable length, and allowable voltage loss.
Behavior below 10V is not guaranteed. I expect output current will decrease monotonically with falling input voltage, but this is not tested or guaranteed.

It is important to connect the LED(s) to the driver only when the driver is not powered. This includes momentary interruptions, as with poor connections. It is recommended to solder all connections between the driver and the LED before the driver is powered.
Please note: due to time lag between junction temp and thermistor temp, thermal protection will NOT protect against large overloads like shorted LED, missing heatsink, etc.

Thanks to Semiman, the driver will shut down if the input voltage exceeds 15V. It will restart automatically when the voltage drops below 15V. This will protect the driver and the LEDs from things like 24V jumpstart, load dump, etc. It will have no effect on a vehicle in normal operation.

Cost: TBD (in the next few days)
Adjustable version will be about $2-3 more than fixed current.

Almost everything is still up for discussion. If anyone has suggestions, comments, ideas, or requests, please make them known. LED configurations, current settings, control methods, anything you want to discuss, please speak up.

[Mattaus]

This looks very cool. My mind is going nuts as to what I can use them for. Not being too technically inclined please forgive me if the following question is a bit silly but can you explain the dimming/modes etc a bit more? I'm confused because you say it has 2 modes (low/high) but then say a fixed version can be control by a pot, then an adjustable version has an on board pot...I think I understand what's happening (on board versus separated potentiometer, but what's the deal with the high/low modes!?

Also looking at the input voltage and max output current what LEDs and in what quantity do you feel would best be driven? 4 XM-Ls could be driven based on the max output current, but the voltage required is very close to the Vin max. Obviously a car battery wouldn't be able to do this, so it's likely 3 XM-Ls would be the best, though driven at 4.5A each which is a bit much. Is this where the on board pot comes in handy? Set your max output current, and then control the dimming with an external pot?

Or am I way off the reservation here...

[DIWdiver]

Yeah, I suppost that warrants a bit finer detail.

The output current is determined by two things: the sense resistor, and the voltage divider.

The sense resistor determines the maximum output current of the regulator. I would expect settings of 3, 5, 9, and 13.5A would be popular, but that's open for input. The sense resistor is something you probably won't fool with after getting the driver, but you certainly could. For example, if you bought a 3A driver and decided later to use SSR-50's, you could change the resistor and have a 5A driver.

The voltage divider is where all the options are. There are two separate dividers on the board, high range and low range. By connecting the H terminal to the C terminal, you engage the high range driver. By connecting instead the L and C terminals, you engage the low range divider. I usually recommend a 4:1 change in current for High/Low, but I can do whatever the customer wants. I'll offer one or more standard configurations, and do custom ones for a small fee.

The high and low dividers share one resistor. The adjustable version replaces this resistor with a potentiometer (pot). The pot will adjust both ranges proportionately, between zero and full. So for example if you had 10A high/2.5A low adjustable driver, and you adjust the pot down to 80%, then you have 10*80%=8A high, 2.5*80% = 2A low.

An external pot can be used as an additional divider. You connect the output of the pot to either the H or L terminal, allowing you to select modes in conjunction with the external pot. You can even use the adjustable version with the external pot. In the previous example, if you connected your external pot through a switch to either H or L, and set the external pot to 40% and the adjustment to 80%, you'd have 10*80%*40% = 3.2A high, 2.5*80%*40% = 0.8A low.

***WARNING: TECHNOBABBLE FOLLOWS***
You might ask why I don't just offer a 13.5A adjustable driver and be done with it. It would simplify things a lot, and I'm considering it. But there is one small problem with that. The driver has an uncertainty of about +/- 0.1% of full scale. This may seem small, but a 13.5A driver would have an uncertainty of +/- 13.5 mA. When you set it to zero output, there might still be 13.5mA flowing. To counteract this I added a -0.1% offset, so that when you set it to zero output, you get zero output guaranteed.

This has the effect of making the uncertainty 0 to -0.2%. On a 13.5A driver, that's 0 to -27mA. If you then adjust this driver down to 1A, your standard low mode is 0.25A or 250mA, the uncertainty (still 0 to -27mA) could now be more than 10% of the setting. Even in this extreme case, that's not a terrible uncertainty (the human eye is barely capable of detecting changes of 10% in light intensity).

***END OF TECHNOBABBLE***

With all that in consideration, I'm seriously considering offering only the 13.5A adjustable version, with 4:1 high/low. Anything else would be custom. Currently the customizaion fee would be $5.00 per order.

[Mattaus]

OK, so let me get this straight(er) - you have a driver with a maximum output pre-set, and then you can switch between high and low (high being the max output and low being a predefined fraction of that maximum).

Adding a potentiometer allows you to adjust the output of either. So if you're in high mode (say 13A) you can adjust it to 50% (6.5A). I'm assuming this can be done while running giving you a 'dimming' effect? Say your low mode in this example is 5A - if you adjusted your high mode by 50% and then switched to low mode, the low mode would be 2.5A as soon as you switch. The same changes work in reverse as well.

So all if this is controlled by an external switch? I assume it would cycle? So for each press you get on-high-low-off? Then a pot to dim the driver?

Sorry if this is all very basic stuff. I'm one of these 'visual aids' sort of people lol.