Coping with voltage spikes

[Anonymous]

Posted by SillyOldDuffer on 16/08/2020 12:45:50:

A half watt Zener and resistor would safely drop a continuous 20V overload to 12V, so should deal OK with spikes. I agree an oscilloscope is really needed to confirm what's going on, but the Zener circuit has a good chance of fixing it.

Was an OCP71 really an OC71 with the paint scraped off?

Agreed, although you won't get an accurate or stable 12V limit. Since we don't know what the problem is there's no way we can say the circuit has a good chance of fixing it. It's the multiple monkey approach to problem solving.

Yes, the OCP71 was an OC71 with the paint scraped off. Until the manufacturer got wise and started sealing the OC71s with an opaque resin. The original clear resin was reserved for the OCP71s, at a premium price of course.

Andrew

[Maurice TaylorParticipant @mauricetaylor82093]

Posted by Andrew Johnston on 16/08/2020 12:09:24:

Posted by Rod Renshaw on 16/08/2020 11:18:58:

The LED that Tim is using has a series resistor or other circutry built in, as supplied, to render it suitable to run off a 12 v supply.

If that's the case then I agree that the Zener might do something, albeit not necessarily that useful. The problem with Zeners is that they're not overly precise, the knee is quite soft at low breakdown voltages and they have a temperature coefficient.

An interesting test which might help solve the problem is to put an example of the LED on a bench power supply and see what happens as the voltage is increased. That should tell us if there is a maximum input voltage before the LED goes phut

Hi, this test has been done previously in this thread. Why haven't other posters done this simple test. I've just done it myself using a clear led in series with a 820 ohm resistor, lights at 2.7v 0.2mA at 12v 11.4mA at 20v 21mA at 32v 38mA still did not blow (power supply doesn't go any higher ). I don't think zeners etc will make any difference.

Maurice

[Tim StevensParticipant @timstevens64731]

speedy Builder suggests a fibre optic solution. Interesting idea, but no go in this case because the LED in question is mounted on the drivers door (Where there was historically a switch – I think perhaps a horn button) and the connection moves as the door opens. Not only that, getting the feed from either the tail lamp or the side lamp is tortuous in the extreme. And the rear lamps are red LEDs, while the front ones are white/yellow flip-flop also acting as flashers. Perhaps it would be nice to have a flashing coloured lamp shining on the oil gauge, though …

The problem LED is connected to the sidelight switch (avoiding yet another hole in the dashboard), and the same feed goes to the light inside the glove box which also permits map reading. All the other LEDs fed by this branch seem OK (so far).

The suggested LM317T is described as variable – so how is it set to the desired 12v?

Regards, Tim

[duncan webster 1Participant @duncanwebster1]

You wire it as a current source, see **LINK**

but I think it's overkill.

[Maurice TaylorParticipant @mauricetaylor82093]

Hi ,Tim

Where is the earth wire from the led connected ,is it connected to the main car body or to the door ?

Maurice

[Anonymous]

Posted by Maurice Taylor on 16/08/2020 14:25:02:

Hi, this test has been done previously in this thread.

Can't immediately see where? The test needs to be done with the actual device being used, not something out of the spares box. A generic LED and resistor should behave exactly as Mauruce details – it's a crude current source. But we need to know if there is anything odd about the actual device.

Andrew

[duncan webster 1Participant @duncanwebster1]

Andrew, see my post of 14/08 (the last one). Tim's LED is advertised as 20mA at 12v, so if the car is giving more than 12V bye bye LED

[Tim StevensParticipant @timstevens64731]

Maurice: The 'earth' connection is by wire to the main earth feed in the harness. Almost everything is wired-earth – except stuff that is not possible such as sparking plugs. This is because old rusty chassis rivets do not provide wonderful continuity.

And while the 'how to set the LM317T' undoubtedly makes sense to an electronics PhD, it doesn't to me. That is why I asked …

Cheers, Tim

[Anonymous]

Posted by duncan webster on 16/08/2020 14:56:56:

Andrew, see my post of 14/08 (the last one). Tim's LED is advertised as 20mA at 12v, so if the car is giving more than 12V bye bye LED

Ok, but it's the same test as Maurice did and it tells us the same, ie, nothing about the problem in hand. Just because the LED being used is advertised as 12V it's not necessarily going to go phut at 12.01V. The whole point of doing the test with the actual unit is that it tells us what it will withstand. If it goes phut at 13.2V (+10%), say, then there's almost certainly a problem. If it survives to 20V then probably not. Either way we have a fact that helps solve the problem.

Andrew

[Maurice TaylorParticipant @mauricetaylor82093]

Hi Tim,

I thought it could be a faulty earth between door and body but obviously not.

I always go for the simple things first with car electrics as earth faults give as many faults on new cars as they did on old ones

This is the reason for suggesting simple tests instead of complicating everything.

Maurice

[Rod RenshawParticipant @rodrenshaw28584]

Didn't the old loco drivers have an oil lamp to light the water gauge?

[Michael GilliganParticipant @michaelgilligan61133]

Posted by duncan webster on 16/08/2020 14:56:56:

Andrew, see my post of 14/08 (the last one). Tim's LED is advertised as 20mA at 12v, so if the car is giving more than 12V bye bye LED

.

… or perhaps more strictly 14V

See my post of 14th

[quote]

Tim,

I tried your ebay search phrase and found a seller whose ‘description’ reads thus:

”Lens Type : Water Clear Lead Length : 20cm Optimum Supply Voltage : 9v – 12v for highest brightness Minimum Supply Voltage : 5v with much reduced brightness Maximum Supply Voltage : 14v Current : 20mA @ 12v Viewing Angle : 20~25 Deg”

IF those are the same LEDs then 14v looks a risky Maximum

MichaelG.

Edited By Michael Gilligan on 14/08/2020 00:01:52

[/quote]
.

MichaelG.

.

Edit: 

What voltage will I measure at the battery using the DVR3/DVR4

Nominally 7.2 V in a six volt set-up and 14.2 V for twelve volts. More is definitely not better here. For example 14.5 V as commonly stated as 'correct' will mean higher trickle charge leading to more topping up and shorter life for your battery.

Ref. http://edsltd.ddns.net/tech_talk_regs.html#what_voltage

Edited By Michael Gilligan on 16/08/2020 16:14:48

[Tim StevensParticipant @timstevens64731]

Rod – if you really think that I am going to drive a 92 year old car while holding an oil lamp by my right ear, then you do not understand the complications of a crash gearbox, or the subtleties of double declutching.

Tim

[Rod RenshawParticipant @rodrenshaw28584]

Hi Tim, it was just an idea!

The experts seem to be strugging to assist you to find a workable solution to the expiring LED problem. I can't remember a thread as long as this about one little lamp to read a gauge by.

I hope you do get it sorted and can then enjoy driving your mother in law around.

Regards

Rod

[Jeff DaymanParticipant @jeffdayman43397]

Just some out of the box thinking –

Maybe decouple the LED's power source from the car entirely. A pound shop book reading LED lamp with a switch and small battery could be deconstructed and the components arranged near the gauge to be read. The wires and bits could be camouflaged behind panels etc. to keep original appearances. This way you would have a trouble free light for use at night to read your gauge, with its' own power supply – no fancy circuit or phD in electronics required.

Forgive me if this approach is too basic or crude, but I could probably adapt the pound shop book lamp in less time than it took to write this. A cheap LED penlight type flashlight would likely work too.

Offered purely as food for thought.

[Robert Atkinson 2Participant @robertatkinson2]

I've not been keeping up to date with this thread and there have been some misunderstandings. Andrew, the 180R nd 12 V zener was my suggestion. SOD's diagram was a little simplistic in not including the resistor (unknown value) that is part of the "LED" and had a polarity error. My suggestion was based on a presumed "LED" specification of 12 V and 20 mA. An additional 180 ohms will drop 3.6 V at 20 mA keeping the "LED" within it's 12 V rating even at 14.4 V (typical maximum charging volte for a 6 cell lead acid battery.The zener will keep the voltage at the "LED" terminals at 12 V (give or take a bit) even if the LED does not draw it's full current. It will also provide suppression of positive spikes. I did suggest in a later post that a 10 or 11 V zener would be even better. The resstor / zener combination provides very good protection against negative spikes (clamping to less than 1 V most LEDs will withstand 3-4 V). I think the most likely cause of the failures is negative spikes but unless an oscilloscope is put on the car we are just making assumptions, some based on experience, some guessing.

I'm happy to put my money on my suggestion. I'll send the OP a 180 ohm resistor, 12 V zenner to try with one of his "LEDs" on the car. If the LED fails I'll make a donation to a charity of his choice. @ Tim Stevens send me a personal message with you address and I'll post them ready wired.

To use a LM317T (or LM317L) regulator its input voltage must be at least 2 V higher than the required output (datasheet says 2.5 V). So would have to be set to 10V output to work with the battery off charge. Even worse, the LM317 is also susceptible to spkes, especially negative, on the supply, To use it safely in a car it needs additional components. As a minium it is a LM317, 0.1uF capacitior. diode and two resistors. A lot more complication for little, if any, advantage of one resistor and one zener.

This stuff is part of my day job – on aircraft.

Robert G8RPI

[MacolmParticipant @macolm]

The point about the LM317 solution is that it would supply a simple LED (not a 12V component of unknown content) with a constant current (constant light output) irrespective of applied voltage. Though primarily intended as a voltage regulator, it is equally suitable for this application.

 

For a white LED, say the forward voltage at the required light output is 4V, allow 4V minimum drop across the LM317 and the associated resistor, and the supply can be anywhere between 8V and 35V. If there are spikes greater than that, agreed additional clipping components might be necessary.

 

The version of the LM317 needs to be chosen for the dissipation that results from dropping the excess volts. However, provided that at normal supply voltage the power is within the working limit, the device does contain thermal shut down that would allow it to weather occasional spikes. Note also that the 0.1 micro-F input capacitor is recommended to swamp the inductance of the wiring, which might otherwise cause instability.

 

F

Edited By Macolm on 16/08/2020 21:07:27

[Robert Atkinson 2Participant @robertatkinson2]

But a LM317 in constant current mode won't work with the LED + Resistor lights that the OP is using.

[MacolmParticipant @macolm]

Quite so, but he has had limited success with that.

[Robert Atkinson 2Participant @robertatkinson2]

We need to be sensible here. There are dozens of ways to run an LED off a car supply, with different pros and cons. No disrespect to the OP but if his electronics knowledge is such that h is buying "12V LEDs" with resistor already fitted he is not going to be working out a LM317 constant current design.

[Tim StevensParticipant @timstevens64731]

Just to reassure folk – the main reason I am using an LED with already attached resistor is simple. I already had a bunch of them for a different project. Not only that, the lamp is on the door, and needs a neat but flexible visible connection to the wiring behind the dashboard. There is nothing stopping me from using a different arrangement, as long as it is not bulky at the LED end. And it would be almost as easy to mount a different LED alongside the steering column and aim the light at about the same angle but from the left instead of the right. I await developments (and the postman.)

Regards, Tim (OP)

[SillyOldDufferModerator @sillyoldduffer]

Posted by Robert Atkinson 2 on 16/08/2020 17:42:20:

… I think the most likely cause of the failures is negative spikes but unless an oscilloscope is put on the car we are just making assumptions, some based on experience, some guessing.

…

I concur! It can only be a guess, but negative spikes fit the bill.

Summarising the evidence:

• 92 year old car, 12v electrics from a dynamo but the regulator is modern.
• Lighting system is LED throughout.
• LEDs appear to be automotive types apart from the failing 3mm type which is an ordinary panel LED with built in dropper resistor.
• Failing LED is switched on & off with the sidelights and glovebox.
• LED fails when the engine is running, works OK off the battery. Other LEDs on the same circuit are reliable.
• Failing LED is in the door with the oil-pressure gauge? Presumably the gauge and LED wiring run in parallel for some distance, and the gauge wires connect to a sensor on the engine near the ignition. Possibility of inductive or capacitive pick up
• Flasher indicator LED flickers at low rpm despite being off. Suggests inductive or capacitive pick up of ignition pulses.
• Maurice's test shows LEDs aren't overly sensitive to positive spikes; his similar 12V 3mm example survived 32V.

Provided the current doesn't last long enough to overheat it, LEDs can be deliberately pulsed to increase peak power output substantially, perhaps to increase the range of a TV remote. I think LEDs are relatively immune to positive spikes. The same can't be said about negatives! Although failure at -4V or -5V is typical, some LEDs avalanche at -1V, while others can stick -15V. Not sure what happens if the negative current is limited by a resistor; but I suspect LEDs are much more vulnerable to negative spikes than positive. Perhaps an experiment is in order!

Car electronics are beefed up to cope with grubby automotive power supplies. I think Tim's working automotive LEDs are protected against spikes, while his 3mm indicator LED has no built in protection. As Robert's Zener circuit protects against positive and negative spikes, there's a good chance it will do the necessary.

Love to put an oscilloscope on Tim's car. I guess the trace with the engine running would show an average 13.8 vdc regularly modulated by short positive and negative going spikes at ignition frequency, which is what Tim suspected at the outset. But without an oscilloscope or peak voltage detector it's guesswork.

Dave

[Michael GilliganParticipant @michaelgilligan61133]

Posted by SillyOldDuffer on 17/08/2020 10:59:00:

.

I concur! It can only be a guess, but negative spikes fit the bill.

Summarising the evidence:

< etc. >

.

So perhaps try inserting a bridge rectifier between supply and lamp

MichaelG.

[Michael GilliganParticipant @michaelgilligan61133]

Posted by SillyOldDuffer on 17/08/2020 10:59:00:

• Maurice's […] similar 12V 3mm example survived 32V.

.

I suspect that Maurice was fortunate in his choice of LED

The ebay description that I found quoted a maximum of 14V

MichaelG.

[Bryan Cedar 1Participant @bryancedar1]

"Never before in the history of Model Engineer Forum has such a small item attracted such intrest"

[Author]

Posts