What happens when "Flashing front headlamps on a car"

[treez]

Hello,
In UK, when you are driving, and you wish to be courteous to another driver, and allow him right of way, then you flash your headlamps at him ,on/off/on/off......... in quick succession.
How is this done?.....is it a relay in the power line to the headlamp that gets repeatedly switched on/off/on...?

Also, with LED headlamps, constantly turning the power supply to the lamp on/off/on like this will cause violent oscillations in the LC EMI filter of the headlamp.......and since its a boost converter used to power the leds, there will be violent oscillations between the boost inductor and the boost output capacitor..............this will mean huge oscillatory currents, and of course, these will end up flowing in the switching FET.

Therefore, do you agree, that with LED headlights, the following type of current sense/comparator circuit is needed to stop these sort of oscillations from happening?

 

[Audioguru]

In addition to flashing my high beams to allow another driver the right of way, I also flash at stupid imported and/or old people who drive at night without their headlamps, tail lights and side lights turned on. They see their daytime running lights blinding on-coming drivers (but they cannot see the road) then think wrongly that their car has nighttime lights. Instead, other stupid and/or old people drive at night with only their parking lights turned on. They also cannot see the road. No headlights.

My car (Chevrolet) is made specially for stupid people because it automatically turns on the required lights when it is getting dark then automatically turns them off when it is getting light or when the car is turned off. I do not know how to turn the lights off at night when I am driving.

 

[permute]

I would have thought that would be much worse. You are alternating the source impedance to a power converter. I would think its internal feedback loop would attempt to correct for the lack of current at the input by trying to draw more current, until the inductor enters CCM mode. Are you sure you don't want to add in something that adjusts the reference level (at least conceptually) at a predictable, slow rate. Perhaps with weak feedback from the output of the EMI filter, but probably not.

 

[FvM]

I'm not aware of LED headlamps or even full beam lamps. Modern cars have LED daytime running lights but still incandescent or xenon HID headlights.

 

[Dan Mills]

You are grossly over thinking this thing, all that is reuired is to design the LC filter with a sufficiently low Q such that the thing rings up to no more then 40V or so, at least if you have a well behaved 12V supply.

However the automotive 12V bus is a notoriously nasty environment (Key phase to google "Automotive load dump transient"), and in that context your circuit is woefully inadaquate as well as tending to protect against the wrong things (And drawing way too much quinciesecent current).

Modern practise usually favours mosfets for conventional lamp control with CAN messages used for signalling (and fault reporting), with very low standby current boost converters used for LED again controlled and monitored by CAN message, relays went out of fashion years ago for most things. Given a boost converer I can actually ramp the current demand over a few ms if I need to, but usually simply picking the correct bulk cap will keep the Q low enough to avoid problems.

There are some very good 'smart' high side switches that are automotive qualified and include such things as current sensing and overcurrent shutdown that are quite capable of handling the 10A or so required for conventional headlamps in a D2PAK, no reason to mess around with relays for medium current switching these days.

Regards, Dan.

 

[treez]

**broken link removed**


...actually FvM I see your point, for example, the Bentley, which i thought was total led heads, as shown linked here, has a main xenon beam...I thought xenons needed high voltage , like 250V to light them?

Dan, I see your point too, but even with the low q filter can you guarantee low current in the oscillation that takes the voltage up to your said 40V?....remember, when the boost fet turns on, that current flows in that fet.

 

[Dan Mills]

Short arc Xenon actually has a very low operating voltage, with even big projection lamps in the KW region requiring less then 25V once lit, ignition is another matter however, particularly is the lamp is still hot (Well over 25kV at hundreds of KHz imposed on the DC, and a RC arrangment to provide a large current surge to get the electrode surfaces up to temperature)!

The current required of the boost fet is drawn from the LV side capacitors, which are actually surprisingly highly stressed components, running at significant ripple current.
Seriously, ringing in LC RFI filters is just not that big a deal, particularly if you pick your core material correctly and don't get too wild with very low ESR caps, worst case, just slug the inductor with a few ohms of parallel reistance.

Using the soft start on the DC/DC also helps, as does setting a reasonable undervoltage lockout to provent the fet and inductor being damaged if the bus voltage is too low.

Regards, Dan.

 

[treez]

I see what you mean about uvlo, but I heard that during cold cranking, where the battery goes down to 6V transiently, the headlights and rearlamps still have to work, and so the regulatory bodys actually check operation with a constant 6V battery........I think so anyway?

 

[Dan Mills]

So set the lockout to 4.5V or so, and size the switching components to operate at the resulting current levels, no big thing.

 

[treez]

the switcher will draw more current with 6v = vin......would you bother rating the switchers components for that current continuously,?...since it is only at 6v transiently.

 

[Dan Mills]

Depends on how you wish to trade the marginal cost Vs reliability thing.
Personally, I would, but I don't do highly cost constrained design these days, and us industrial control types tend to view reliability as being the thing to prioritize.

The current levels are not huge in either case, and a fet and inductor good to a few amps seem unlikely to break the bank (Those LT parts you seem so fond of however might well blow the budget).

Regards, Dan.

 

[betwixt]

I'm not sure I'm right but Xenon lamps and 'Xenon arc lamps' are I believe two entirely different technologies. Xenon is an inert gas and can be used in the same way as halogen gases around a conventional filament. I believe it is also used as a trade name for conventional halide lamps.

My vehicle (a Mazda) uses old fashioned relays to operate the flash. Not high tech but it works fine!

Brian.

 

[Dan Mills]

Yea, but I figured a Bentley would be going for the arc sources rather then just using it as an inert fill in a TH variant.
High end cars tend to do things like that.

Nothing inherently wrong with relays, and if you have a design that already works there is usually little reason to change, but if you are doing a new electrical system fo a new vehicle line then for most things it is no longer the way to bet.

Regards, Dan.

 

[treez]

Actually earlier we considered there were no pure led headlights on cars...but this ti.com design appears to be for a 20W LED headlight for a car
https://www.ti.com/tool/pmp5518

Its SEPIC when it could have easily been a boost...they don't say why , but presumably the inherent short cct protection of the sepic is the desired feature.?
They use no RC damper across the sepic capacitor....so I don't believe their design is for real.
The LLC resonant frequency of that sepic is 3580Hz, so, in order to avoid instability, they would have to have the crossover frequency of that sepic >10KHz, (or <1khz). Greater than 10KHz I just don't think is reasonable, even with the 300KHz switching frequency that they use.

 

[FvM]

I was under the assumption, that LED headlights won't yet achieve the technical requirements, but I stay corrected. They are on the way, still fighting some national regulations.

https://www.hella.com/MicroSite/soe/en/buses/highlights/led_headlights.html
https://www.thecarconnection.com/ne...lights-banned-in-the-u-s-okay-everywhere-else

Referring to the technical problem that motivated this thread, I'm sure that you can design voltage converters for effectively any dynamical specification, once it has been defined. I don't see a serious problem. And yes, the problems are almost trivial compared to the design of xenon arc lamp supplies for cars, which are in use since fifteen or twenty years. (But not for the high beam, because they need a few seconds to heat-up and more time to cool-down before re-ignition, you probably know about their behaviour from video beamers).

 

[treez]

From what you say, FvM, I now think xenons are totally unsuitable for car headlamps, because they cannot be quickly switched on and off as in the way motorists signal to each other....Being able to flash your headlamp at say a lorry driver who is wanting to pull out in front of you, in order to signal to him that you are aware of him is an "absolute must have" property of front headlamps......on any motorway journey, I flash my headlamps at lorry drivers over 30 times over the journey's course.....how else do they know its safe for them to pull out into the middle lane?........they then respond to me, after they have pulled out in front of me , (or anyone) by alternately blinking the right and left rear blinkers.

If LED headlamps cannot be flashed , then they are useless...and if the flashing means too much resonant oscillation in the L's and C's then they cannot be done.......but that is why I proposed the snap-off comparator which senses high current, and cuts off the supply by a series pfet.

 

[FvM]

I think disussing the properties of xenon arc lamps is somehow out-of topic here. The power supply has some challenges but is state-of-the-art now.

Related to LED lamps, you doubt if a specific converter topology is suited for "flashing" operation. You may be right or wrong, but the problem doesn't apply to all possible LED power supply designs.

And it's not the way how engineers handle a design problem. First have a specification (e.g. flashing capability required, then choose a suitable topology and design a circuit.

P.S.: I have a short link explaining how flashing operation and slow xenon HID come together, to further clarify the state-of-the-art.

A comment on the "oscillating LC" thing. I think, many LC ciruits, e.g. filters have a potential of being excitated to unwanted oscillations. But there are also various active and passive means to suppress the oscillations respectively reduce it to a specific acceptance level. It's design detail problem, no more, no less.

 

[treez]

I appreciate that, but when you have a front end LC filter in your led car headlamp, then realistically, without an awful lot of circuitry, one is not going to be able to prevent the LC oscillations from traversing up and down the supply wires from the relay which switches the light on/off

The following two schematics show the problem.........with the SMPS, when the relay opens then there is a large voltage across the relay contacts......wheras with the incandescent bulb, the voltage spike across the relay contacts is not nearly so large.
I believe this is a great problem with LED lamps used in cars....the simulation bears it out....

Ive also attached the LTSpice simulations

 

[Dan Mills]

So stick a small cap across the relay contacts (A good idea anyway, it supresses arcing as the contacts open by limiting dV/dt) and a few ohms of resistor across the EMI filter inductor (To kill the Q), whats the problem?

Seriously, you are way over thinking this this thing, you appear to be seeing a big issue where everyone else sees a trivial one.

Does your spice model include for parasitics and things like ESR and stray capacitance? If not then you cannot use it for analysing this sort of thing.

Regards, Dan.

 

[treez]

I agree but i fear that the car manufacturers will not want to go about putting caps across the relay fittings....instead, they will just say, "lets forget led lamps if its this much trouble".
am I right in believeing that we have agreement that the relay opening sparking will be worse with the led bulb than the incandesant?...thus the relay may well wear out quickly with the led lamp.
I can put some stray capacitance in there...but I still expect that resonating LC current to spark like mad when that relay opens...wearing out the relay.

By the way, I once worked in a car led headlight place as a contractor........the issue of LC filter ringing was a big issue to them.....they would not tell me why though....drat!.....they asked me if the LC ringing (with the led lamps EMI LC filter) could be stopped....and I told them that it could not be stopped without very significant extra amounts of circuitry.