Coping with voltage spikes

[Ian PParticipant @ianp]

Five pages so far and we seem to be no nearer a solution, I dont think we even know what colour the car is.

Tim said four LEDs had failed but one LED survived after seeing 6 volts, presumably not one of the four.

If I understand what Tim said, one of the none working LEDs had no faults in its wires and resistor but then gave light output when the LED was connected directly to a supply, but it failed after a while?

The plot thickens.

Ian P

[Robert Atkinson 2Participant @robertatkinson2]

Posted by Maurice Taylor on 18/08/2020 16:22:58:

Posted by Robert Atkinson 2 on 17/08/2020 15:23:09:

While this is an interesting demonstration of the magnitude of spikes I note your LED had a 820 ohm series resistor. This is a lot higher than the 470 ohm in the OP's. This gives a much more reasonable 11mA current at 12V (assuming 3 V Vf for LED).

I can't see what wiring fault would cause the LED to fail.
As I stated earlier the OP's LEDS seem to be running close t absolute maximum curret rating at 12V.

Robert G8RPI

P.S whats the box of electronics next to the coil?

Edited By Robert Atkinson 2 on 17/08/2020 15:23:50

Hi Robert

I have since tried the same clear Led with a 270 ohm resistor in series,works OK.

The Led I'm using is CPC Part no SC08786.

I think the experiment has proved spikes are not the problem,as Led is wired across points ,cant be any more spikey.

Regarding the box of electronics,it is a homebuilt ECU ,using an Arduino mega ,it will run the ignition using a homemade crank trigger wheel,a BMW crank sensor and a VW wasted spark coil.I haven't fitted fuel injection yet.

I can change from ECU ignition back to distributer ,by changing plug lead an power connection.

The project is known as Speeduino, plenty of info on internet.

Maurice

Edited By Maurice Taylor on 18/08/2020 16:23:35

Unfortunatly it only proves the spikes from your ignition don't kill your LEDs We have no high speed measurements of the power quality on Tims 1929 car. While the spikes form your igniton are impressive, they are narrow so lower total energy. Also you are using a quality 5 mm / 30 mA LED which will almost certainly have a larger chip than Tim's unknown 3mm ebay LEDs.
I do note that the LED you used has a maximum reverse voltage rating of 5A and a ESD HBM rating of 250 V. This gives an ESD energy limit of 3.2 uJ (1/2CV2). A 10V negative spike from inductive swiching (maybe that solid state dynamo regulator?) 60us wide would equal this with a 470R resistor with (crude) assumption of 5V across LED and 5V across resistor.

We DO NOT KNOW the quality of the power supply on Tims car.

For reference, if this LED was to meet the power surge and spike standard used on a 12 V light aircraft it would have to be tested to withstand +16 V for 5 minutes, +24 V for 1 s, +40 V for 0.1 s, 50 off 10 us pulses of 600 V positive and 50 off negative (RTCA DO10G sections 16 and 17).
A typical automotive "load-dump" specification (for alternator equipped cars) resqures resistance to 10 pulses of + 90 V about 0.2s wide (ISO 16750).

Robert G8RPI.

[Maurice TaylorParticipant @mauricetaylor82093]

Hi,

Robert,

Tried green Led from junk box ,it suvived the spikes. If the car is still positive earth ,those spikes on my scope would be inverted ie 130V -ve and 20V +ve ,was it said previously in the thread that leds didn't like -ve pulses.This could be causing the led to fail.

I've given up with it now ,think I've tried my best,would welcome any comments on the above.

Tim,

You could probably try the idea/circuit suggested below.

The led is CPC Part No SC08786, Relay has 12V coil.

I have tried the circuit and the led draws 2.7mA from AA cells , they should last well.

Circuit is completely isolated from car electrics.

Comments welcome good or bad.

Maurice

[Dave HalfordParticipant @davehalford22513]

Does anyone remember these link 1960's tech to run the gauges

[Robert Atkinson 2Participant @robertatkinson2]

Posted by Maurice Taylor on 19/08/2020 13:46:15:

Hi,

Robert,

Tried green Led from junk box ,it suvived the spikes. If the car is still positive earth ,those spikes on my scope would be inverted ie 130V -ve and 20V +ve ,was it said previously in the thread that leds didn't like -ve pulses.This could be causing the led to fail.

I've given up with it now ,think I've tried my best,would welcome any comments on the above.

<SNIP>

Comments welcome good or bad.

Maurice

Good point Maurice,

Positive earth per se should not make any difference, all else being equal. However if the modern Dynamo regulator works on positive or negative earth (or is based on a negative earth basic design) that could affect the level of spikes generated by it's operation on negative earth.
Additionally a conversion from 6 to 12 V may make any spikes generated by the dynamo work.

Tim,
is the car positive or neagative earth? Was it converted from 6 to 12 V and or positive to negative earth.

Robert G8RPI.

[not done it yetParticipant @notdoneityet]

Dave,

Those are bi-metallic strip voltage regulators. Generally around 10.5V for automotive use to drive gauges. Not a lot of good for this as the voltage only averaged the target.

Edited By not done it yet on 19/08/2020 15:54:01

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Tim Stevens on 18/08/2020 12:20:58:

I have just made a further test: A failed LED complete with its leads – apply rising voltage, nothing. Reverse leads + to -, nothing. Wiggle the wires to check soldering, still nothing. Apply digital ohm meter to outer end of black lead and direct on the back of the LED – zero ohms. Apply meter to red lead and its connection via the resistor at the back of the LED, 470 ohms. So, both leads seem sound, including the added resistor. Chop off both LED leads, and try bare LED on variable voltage. It lights at 2.6V and is bright at 3.5.V continued OK to 6v – brighter still but not much. So, a good LED was connected by good leads and a good resistor, and it worked new but failed soon after.

So – any ideas what is going on? Four LEDs have now failed, in identical circumstances.

 

.

I have just re-read the above, more carefully … and am now as bewildered as you, Tim

You tested a failed ‘12V LED’ and confirmed, after dismantling it, that the individual components were all good.

I have only one question before I creep back under my stone:

Is the resistor marked as 470 Ohms, or is that just what you measured ?

MichaelG.

.

In case you were wondering why I ask ^^^
This is a grab from the ebay listing that I quoted:

Edited By Michael Gilligan on 19/08/2020 17:21:07

[Dave HalfordParticipant @davehalford22513]

Rather looks like the resistor fails to cope with the voltage and goes open circuit, but still able to some how function at 1.5V from the test meter ?? Wonder what the meter reads on the highest range.

[Robert Atkinson 2Participant @robertatkinson2]

The ebay sellers generally offer a choice of voltage ratings for these., The pictured resistor may be for the 5V version. Tim said he measured 470 ohms. A 470 ohm resistor is consistent with a 20mA, 12V white LED set-up when you allow for a 3V drop across the LED.

Robert G8RPI.

[Dave HalfordParticipant @davehalford22513]

Must be a lack of magic then.

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Robert Atkinson 2 on 19/08/2020 18:56:09:

The ebay sellers generally offer a choice of voltage ratings for these., The pictured resistor may be for the 5V version. Tim said he measured 470 ohms. A 470 ohm resistor is consistent with a 20mA, 12V white LED set-up when you allow for a 3V drop across the LED.

Robert G8RPI.

.

If you say so, Robert

But [at the risk of breaking the new forum rules] here is the ebay link: **LINK**

http://www.ebay.co.uk/itm/12V-Pre-Wired-LEDs-3mm-5mm-10mm-Various-Colours-Lamp-Wired-LED-9V-12-Volts-UK/262762466125

This is the listing which I found by using Tim’s search term [but which,  of course, has NOT been confirmed or refuted as being what he purchased]

The fact that they are specified for 9 to 12 Volts with a maximum of 14V is what has worried me all along.

 

MichaelG.

Edited By Michael Gilligan on 19/08/2020 21:03:51

[Maurice TaylorParticipant @mauricetaylor82093]

Posted by Michael Gilligan on 19/08/2020 20:56:25:

But [at the risk of breaking the new forum rules] here is the ebay link: **LINK**

http://www.ebay.co.uk/itm/12V-Pre-Wired-LEDs-3mm-5mm-10mm-Various-Colours-Lamp-Wired-LED-9V-12-Volts-UK/262762466125

This is the listing which I found by using Tim’s search term [but which, of course, has NOT been confirmed or refuted as being what he purchased]

The fact that they are specified for 9 to 12 Volts with a maximum of 14V is what has worried me all along.

 

MichaelG.

Edited By Michael Gilligan on 19/08/2020 21:03:51

Hi,

I can’t understand what is wrong with the specification.It simply says they work on a 9 to 12Volt supply with maximum

of 14Volt.            Please explain.

These will probably be bought by railway modellers,dolls house makers etc instead of bulbs surely all the info is there.

Maurice

Edited By Maurice Taylor on 19/08/2020 21:39:51

[Robert Atkinson 2Participant @robertatkinson2]

Hi Michael,

While this is a nominally UK seller the item and listing are typical of the chinese sellers. Just because this seller is only listing 12V LEDs does not mean the photo is of the actual item they are selling. These sellers often "share" photos. Even professonal online catalogs like Farnell do not use correct images. I'm an electronics designer but a couple of weeks ago I asked why a mechanical designer used a flanged riv-nut (not something he had used before), causing a panel gap, instead of a countersunk one. He said Farnell (the supplier I suggested) didn't do countersunk riv-nuts. I gave him the stock number and he pointed out the picture on the web site was of a flanged one. You can't trust photos in adverts.

The fact that these LEDs are rated at 12 V and seem to be running at the absolute maximum current, is probably part of the issue. The described failures still point to negative spikes being the cause.

Robert G8RPI.

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Maurice Taylor on 19/08/2020 21:38:22:

Posted by Michael Gilligan on 19/08/2020 20:56:25:

[…]

Hi,

I can’t understand what is wrong with the specification.It simply says they work on a 9 to 12Volt supply with maximum

of 14Volt. Please explain.

These will probably be bought by railway modellers,dolls house makers etc instead of bulbs surely all the info is there.

Maurice

 

.

The ‘problem‘ is the value of the resistor, in the context of Robert’s dismissal of my recent question to Tim.

… That and the fact that these may possibly be the devices that Tim is using.

MichaelG.

Edited By Michael Gilligan on 19/08/2020 21:55:58

[Michael GilliganParticipant @michaelgilligan61133]

I simply asked Tim a straight question.

Is the resistor marked as 470 Ohms, or is that just what you measured ?

Edited By Michael Gilligan on 19/08/2020 21:55:14

Edited By Michael Gilligan on 19/08/2020 22:09:37

[Robert Atkinson 2Participant @robertatkinson2]

Maurice,

None of the LED specifications or suppliers linked to in this thread are the ones being used by Tim in his car. A typical 3mm white LED has a absolute maximum current rating of 20 mA We know that Tims LEDs hve a 470 Ohm series resistor. with 9 V across the resistor and 3 V across the LED (typical for white LED) you get 19 mA (9/470) for 12V supply. At 13.8 V this rises to 23mA (10.8/470 voltge across the LED is fairly constant), so out of specification. I would not expect this 15% overload to kill the LED. It is almost certainly NEGATIVE SPIKES generated when the engine is running by dynamo, ignition or the starter motor turning off that are killing the LED.

I'm stopping replying on this thread until either Tim reports back on the parts I sent or SOD sent or somone nearer to tim goes and puts an oscilloscope on the car.

Robert G8RPI.

[Tim StevensParticipant @timstevens64731]

I have just found to my considerable chagrin that if you start a message and look back to check a previous message beyond the current page the message box is wiped clean. So, I will start again:

Sorry about the delay in responding, my PC was being 'upgraded', and my printer sill does not work on Win 10.

The car is 12V from new, negative earth, believed original dynamo (but no way to tell – every old car has had 100 POs). The resistance was measured with a modern digital meter, and also looks identical the the one M Gilligan shows – 470 Ohms.

The two alternative solutions have arrived – many thanks. I will try Duncan Webster's first and report, then Robert Atkinson's.

The connection from the harness to the LED in question is via a twin-core red/black cable as used on many after-market LED offerings thus: LED leads – resistor fitted by supplier – new soldered joint to cable, shrouded in heat-shrink, and held to the door by a clip (so no flexing). From the door to the dashboard is the same cable with a min radius of about 25mm as the door closes.

Finally I have abandoned the idea of 3mm for the LED, and will try 5mm for future experiments, in view of the reports of the marginal nature of what I have got compared with what I need – for which also thanks.

And if I still don't succeed I will never go out in the dark.

Regards, Tim

[Stuart BridgerParticipant @stuartbridger82290]

At the risk of thread drift, I couldn't resist commenting on "never go out in the dark ". Many years ago not long after moving into our village, we tried to book the local village taxi to take us to a pub in the next village for a meal.. The answer was "I can take you, will you be wanting to come back", "errr, yes" was my reply. "Sorry I don't go out after dark" was the answer…

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Tim Stevens on 20/08/2020 15:00:59:

[…]

The resistance was measured with a modern digital meter, and also looks identical the the one M Gilligan shows – 470 Ohms.

[…]

.

Thanks for that, Tim … but I regret to say that the two statements appear to be inconsistent

Admittedly the ebay photo is poor, but the bands look like Brown, Black, Black [100 Ohms] to me.

The reason I asked about the resistor was: I had a nagging suspicion that your measurement may have included a dry joint … simply because this was the only reason I could think of to explain the mysterious result of your test on the dissected assembly that had previously ‘failed‘. … I await enlightenment.

MichaelG.

.

P.S. ___ Although he lacks Robert’s credentials … ‘The Evil Mad Scientist’ suggests 100 Ohms as a suitable resistor for a White LED  on a 9 Volt DC supply.

https://www.evilmadscientist.com/2012/resistors-for-leds/

 

Edited By Michael Gilligan on 20/08/2020 16:37:49

[SillyOldDufferModerator @sillyoldduffer]

Posted by Tim Stevens on 20/08/2020 15:00:59:

…

The connection from the harness to the LED in question is via a twin-core red/black cable as used on many after-market LED offerings thus: LED leads – resistor fitted by supplier – new soldered joint to cable, shrouded in heat-shrink, and held to the door by a clip (so no flexing). From the door to the dashboard is the same cable with a min radius of about 25mm as the door closes.

…

Regards, Tim

All sounds tickety-boo and suitably professional. If it's a mechanical problem, suspicion is now concentrated on the LED, resistor and supplied wires.

Try testing all the wires with a magnet: I've noticed a number of cheap components recently have tinned copper-plated steel-cored leads, that are more liable to snap than copper. The connecting cable itself might be steel, more likely just the resistor and/or LED.

Michael mentions dry-joints as a possibility, as are bad crimps. Crimping makes better and more reliable joints than soldering but only if the crimp is made at the correct pressure. As with dry joints due to dirt or temperature issues whilst soldering automatically, an incorrectly set machine could turn out several batches of marginal joints before anyone notices.

In the bad connection scenario the LED fails because vibration and heat cause one or more joints in the supplied item to lose contact. I can't think of another reason why a LED would start working again – they don't rise from the grave after being zapped electrically!

Good chance Duncan and Robert's offerings will fix the problem simply because they're better made, not because they're super spike proof.

Dave

[Robert Atkinson 2Participant @robertatkinson2]

Posted by Michael Gilligan on 20/08/2020 16:31:44:

Posted by Tim Stevens on 20/08/2020 15:00:59:

<SNIP>

.

P.S. ___ Although he lacks Robert’s credentials … ‘The Evil Mad Scientist’ suggests 100 Ohms as a suitable resistor for a White LED on a 9 Volt DC supply.

**LINK**

Edited By Michael Gilligan on 20/08/2020 16:37:49

There is nothing on that page I can see that suggests 100 ohms is correct for a single white LED and 12 volt supply.

The closest is 100 ohms for TWO white LEDs on 9 volts and that is for 30mA not the 20mA of Tims LEDs.
I would never design for any component to be at ait's Absoute Maximum rating(s) under nominal conditions.
I broke my first LED in 1974 at the age of 12, the lead fractured close to the body. That LED cost me about 2 weeks pocket money.

[Michael GilliganParticipant @michaelgilligan61133]

Robert,

Sincere apologies for my stupid mis-reading of the Mad Scientist’s table

I confess that I was looking for 100 Ohms [to match the ebay photo] and saw what I wanted to see

mea culpa

MichaelG.

[EmgeeParticipant @emgee]

Posted by Michael Gilligan on 20/08/2020 19:20:05:

Robert,

Sincere apologies for my stupid mis-reading of the Mad Scientist’s table

I confess that I was looking for 100 Ohms [to match the ebay photo] and saw what I wanted to see

mea culpa

MichaelG.

Never mind the table Michael, very informative page of reading, well worth a read, thanks for the link.

Emgee

[Keith PetleyParticipant @keithpetley53472]

Hi Michael,

Try reading the resistor in the ebay photo from the other end – looks like green,blue,black,black,brown which would be 560 Ohm at 1% tolerance.

Keith

[Michael GilliganParticipant @michaelgilligan61133]

 

Good point Keith

Oops !

MichaelG.

.

.

Reminder to self: 

1.Don’t assume that things are ‘written’ left-to-right !!

2. Digikey has a nifty little calculator:

https://www.digikey.co.uk/en/resources/conversion-calculators/conversion-calculator-resistor-color-code-5-band

Edited By Michael Gilligan on 20/08/2020 22:51:23

[Author]

Posts