Coping with voltage spikes

[Maurice TaylorParticipant @mauricetaylor82093]

Hi,

Just done a test with green led from junk box in series with 270 ohm resistor ,to test the effect of negative spikes,

Equipment used 12volt power supply,picoscope and 12volt car ht coil. I was able to produce an 80volt negative pulse ,did this a few times.Led and picoscope both connected across the coil.

Disconnected led ,connected it to 12volt supply still works.

All comments welcome good or bad

Maurice

[Tim StevensParticipant @timstevens64731]

Maurice

Do tell us more about your picoscope, please

Tim

[SillyOldDufferModerator @sillyoldduffer]

Posted by Maurice Taylor on 21/08/2020 10:41:11:

Hi,

Just done a test with green led from junk box in series with 270 ohm resistor ,to test the effect of negative spikes,

… I was able to produce an 80volt negative pulse ,did this a few times….

… led ,connected to 12volt supply still works.

All comments welcome good or bad

Maurice

I love experiments, and this one motivates me to have another go! I failed to kill a red LED with -30V, and Maurice failed to kill his green one with -80V. Seems LEDs are as tough as old boots despite what my books say about them being sensitive to negative spikes.

Tim's problem is proving remarkably difficult. At the moment I suspect bad joints due to faulty manufacture of the LED and its built-in resistor rather than the LED being blasted by spikey electrics. If I understood correctly Tim said one of the failed LEDs lights up when connected directly to a power supply: it's not been killed by a spike.

Dave

[Michael GilliganParticipant @michaelgilligan61133]

After my faux pas, yesterday … I don’t feel worthy to pass any comment except:

Thanks, Maurice

An interesting exercise, thanks for documenting it.

MichaelG.

[Robert Atkinson 2Participant @robertatkinson2]

Maurice,
Interesting experiment, but again producing very narrow spikes.

Tim,
Picoscope is a trade name of Pico Technology (picotech, http://www.picotech.com ) for a range of oscilloscopes that use a personal computer for a display and processing. As well as "ordinary" 'scopes thay do an automotive range, http://www.picoauto.com that include very comprehensive vehicle diagnostics. For example you can check compression just by clipping a current sensor on the battery lead and crankng with igntion off.
They are great units, I have several, including some their early ones, and a full 4 channel automotive kit. There are cheaper similar units but Pico are probably the best.

Dave,
Did you see my earlier explanation of how partial junction damage could produce the effect seen by Tim in the failed LED directly connected to a 6V supply. I've blown up LEDs in the past when pushing the limits under pulse operation and seen this kind of effect. Under a microscope you could see the damaged area and a curve tracer showed a distinct change in the forward voltage characteristic.

Robert G8RPI.

[Maurice TaylorParticipant @mauricetaylor82093]

Hi ,Robert

If you’ve got all this equipment ,why haven’t you done any experiments to show us all what spikes do to an led?

Maurice

[Tim StevensParticipant @timstevens64731]

REPORT:

I have now converted my holder to 5mm, and tested the 5mm LEDs kindly sent by Robert and Duncan (see above). They both worked exactly as they should when tested to 15v and neither showed any trace of flicker etc when starting the engine or switching other stuff on and off. It is not fit for a drive here (persisting down) so I will report further on extended use when the sun next shines. But only at first for the slightly better focussed* of the two, from Duncan.

*Note: For readers who are less familiar with LEDs: the focussing is part of the manufacture and depends on the quality of the plastic dome and the size and depth of the emitting surface.

Having now looked very carefully at one of the remaining supply from which the failed LEDs were taken, it seems that the failures may have been due to the feeble assembly in manufacture of the included resistors. The wiring had to be curved to fit in the holder and this may have started a failure which allowed the circuit to break as the door was closed or the first road bump was crossed. Both the LEDs I have been sent have the resistor further from the diode and not affected by any bending.

Regards, Tim

[Maurice TaylorParticipant @mauricetaylor82093]

Hi Tim ,did you do any of the tests I suggested about 3 days ago regarding connecting your leds across battery?

Robert has already answered question about Picoscope. Go to ‘picotech.com’ for info .Software is free and runs in demo mode .Prices at around £100.

Maurice

[Tim StevensParticipant @timstevens64731]

Maurice – I'm sorry, but if you can point me to the message I will try.

I have been without my PC over the last few days and I may not have a full set of messages.

Tim

[Maurice TaylorParticipant @mauricetaylor82093]

Posted by Maurice Taylor on 15/08/2020 20:24:53:

Hi Tim,

I would do the following,

1 Make a test lead ,with a 12v led ,length of black wire ,black crocodile clip,length of red wire,red crocodile clip.

2 Measure battery voltage with engine running.

3 Turn engine off

4 Connect test lead , led should light.

5 Leave test lead connected,start engine ,hopefully led should stay lit.If it stays lit there must be a problem with wiring to pressure guage.

6 If it blows ,fit new led to test leads,then disable generator either by disconnecting it or removing fan belt.

7 Connect test lead and start engine,if led stays lit ,this indicates a problem with the charging system,if it blows it suggests a wiring problem.

8 Make another test lead 12volt 5 watt bulb ,put this in place of led at pressure guage to see if it works or not.

Hope this helps,I know it seems a bit long winded but it takes longer to read it than do it.

Regarding voltage spikes ,put a supresser capacitor on th e HT coil , like when fitting a car radio.

Maurice

Hi Tim,

The above was what I posted ,if you still have one of your Leds connect it across the battery and see if it works as you won’t need to bend it.

Maurice

[SillyOldDufferModerator @sillyoldduffer]

Posted by Robert Atkinson 2 on 21/08/2020 17:36:12:

…

Dave,
Did you see my earlier explanation of how partial junction damage could produce the effect seen by Tim in the failed LED directly connected to a 6V supply. I've blown up LEDs in the past when pushing the limits under pulse operation and seen this kind of effect. Under a microscope you could see the damaged area and a curve tracer showed a distinct change in the forward voltage characteristic.

Robert G8RPI.

Hi Robert,

No I didn't see it! Very strange, I thought I was paying attention to this thread, but missed it.

Very interesting. Another divesrion from amusements and duty galore, but it would be fun to spend time investigating LEDs. Not just blowing them up! For example, I notice they behave like a noisy zener and wonder if that's useful as a noise source. I don't need a noise source, but…

Look's like Tim has cracked the problem. Hurrah!

Dave

.

[Tim StevensParticipant @timstevens64731]

Aha – that message. I did try to explain that there were no leads to the oil gauge as it was completely non-electrical.

Secondly, there is no fan belt to remove, as the dynamo is gear driven, and I commented that fan belts came in about 1930.

And yes, if I connect one of the original batch of LEDs to the battery, it does light.

But having said all that, I'm sorry, but I'm not sure how much further it gets us.

Did I misunderstand you somewhere?

Tim

[Robert Atkinson 2Participant @robertatkinson2]

Posted by Maurice Taylor on 21/08/2020 17:47:51:

Hi ,Robert

If you’ve got all this equipment ,why haven’t you done any experiments to show us all what spikes do to an led?

Maurice

Hi Maurice,

Unfortunatly I'm pretty busy with the day job (mostly from home) and the workshop is full of stuff due to some re-rranging and decorating for SWMBO. so no time. Additionally I can't replacae a 1929 car.

However Tim's recent posts indicate it might have been mechanical failure of connections. I've not seen the car or LEDs and I guess we may never know for sure. I'm not convinced by the broken connection theory though.

Tim,
The LED I sent is a wide angle type. It's not a quality issue, they make them with different viewing angles. The one I sent was left over from a backlight modification on a digital meter.

Robert G8RPI.

[Robert Atkinson 2Participant @robertatkinson2]

Dave,

LEDs do make fairly good low frequency noise sources. .I've seen them used for that in the past.

Robert G8RPI.

[Maurice TaylorParticipant @mauricetaylor82093]

Hi Robert ,thanks for your reply ,faults like this usually are bad connections ,that’s what I tried to say early in the thread. I’ve had cars for 50 years and most electric faults have been bad connections etc. I appreciate your a busy man ,and don’t have time.

Maurice

 

Edited By Maurice Taylor on 21/08/2020 21:02:40

[Maurice TaylorParticipant @mauricetaylor82093]

Duplicate ,deleted

 

Edited By Maurice Taylor on 21/08/2020 21:01:51

[Maurice TaylorParticipant @mauricetaylor82093]

Hi Tim ,

I asked you to connect led to battery and start the engine to see if it stayed lit,this would have proved your led was ok.

I suggest connecting the bulb to where the led was connected to check that wiring was ok.

Did the led across the battery stay alight with engine running ?

Sorry if I wasn’t very clear

Maurice

[Tim StevensParticipant @timstevens64731]

Before I contacted this forum, I did all the 'standard' tests including checking that a new LED would work at the battery and at the switched connection. I didn't mention doing this specifically because (as I think was clear) the LEDs worked in situ before the engine was started, and then ceased to work. Checking that another LED would do the same thing – work to start with – didn't seem to add to the fund of knowledge.

And connecting the LED across the battery under the driving seat makes it difficult to start the engine as I need to work the pedals and so need to sit.

It is not that your message was not clear, but it arrived after or crossed with messages explaining that the tests you mentioned had already been done. Or so I thought.

Regards, Tim

[Tim StevensParticipant @timstevens64731]

Latest info: I have now obtained some plain, no wiring, no resistor, 5mm white LEDs. The pack is marked 3.2V – 3.8v.

Applying an increasing voltage gives light output at 1.8v, dim, at which point the milliammeter registers. Increasing to 3.2v and the brightness becomes 'standard – not measured but what I have come to expect visually – and the current rises to 17mA. Increase again and the brightness does improve but not a lot, and the current at 3,8v is 27mA.

Increase the voltage further and at around 5v the light output starts to dim, losing the whiteness and becoming bluish. Then after about 20 seconds at about 6.5v the light stops and the current rises to over 100mA.

Turn off for 5 minutes. Turn on again – current at 3.2V is now 69mA, and at 3.8v is 84mA. These currents seem to stay steady – and I conclude that failure involves the destruction of the phosphor coating, and causes something else inside to change reducing the 'resistance'. But a failed LED does not go open-circuit.

I hope this is useful to those among us who are not yet up to speed on 'modern' wizardry.

Cheers, Tim

[Anonymous]

Posted by Tim Stevens on 24/08/2020 16:07:58:

These currents seem to stay steady – and I conclude that failure involves the destruction of the phosphor coating, and causes something else inside to change reducing the 'resistance'.

Based on what evidence?

What has actually happened is that semiconductor junction has been overheated and the doping levels have gone pear-shaped. So the junction no longer operates as a diode. You'd see exactly the same effect (albeit with different voltage/current values) if you repeated the experiment with an ordinary silicon diode.

Andrew

[Robert Atkinson 2Participant @robertatkinson2]

You should never connect a LED to a voltage source without a series resistor or other means to limit the current.

The voltagge on the datasheet is so you can calculate the correct resistor value and work out how many you can use in series with a given supply voltage. Resistor value is (supply voltage – LED voltage) / desire operaing current.

Fo your 3.2V leds, to operate from 13.8 V at 20 mA you get 10.6 / 0.02 = 530 ohms. The next highest standard value is 560 ohms. 680 ohms would b a safer choice for a car applicaion.

Your experiment confirms that you can change the charateristics of LEDs without destroying them completly.

The original failures were almost cetainly due to eceeding the LED ratings, not broken connections.

Robert G8RPI.

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Robert Atkinson 2 on 30/08/2020 18:40:17:

[…]

Fo your 3.2V leds, to operate from 13.8 V at 20 mA you get 10.6 / 0.02 = 530 ohms. The next highest standard value is 560 ohms. 680 ohms would b a safer choice for a car applicaion.

[…]

.

Sorry to labour this, Robert … but I need to better understand

Why are you working around a voltage of 13.8 when the manufacturer of Tim’s regulator states that it charges at 14.2 ?

MichaelG.

[Robert Atkinson 2Participant @robertatkinson2]

Hi,

I had missed that the regulator was 14.2 V. 13.8V is the traditional nominal voltage for a 12V lead acid battery automotive system. 14.2V is actually a bit high for a vintage car with low electrical load and conventional (as opposed to newer long life or stop / start) battery. This is offset by less frequent use if the batery is not charged when the car is not in use.

That said, 14.2 V and 20 mA gives 550 ohms for a series resistor so the next higher standard value – 560 ohms (that I suggested as a minimum for 13.8 V) is fine for 14.2V as is my suggested 680 ohms.

We are not talking precision electronics here. The LED forward voltage varies with temperature and forward current never mind form device to device. The difference between 13.8 V and 14.2 is only 3%. I doubt the regulator is that good and standard low cost resistors are only 5%.
(I can do precision when required, voltage measurement to 7.5 digits)

Robert G8RPI.

[Michael GilliganParticipant @michaelgilligan61133]

Thanks, Robert

Just for info though …

The resistor on Tim’s original LEDs [confirmed, I think, in the blur of this discussion]

is 560 Ohms1% … see P7

Which seems to leave not much ‘wriggle room’.

Please don’t bother replying, I’m comfortable with it now.

MichaelG.

 

Edited By Michael Gilligan on 31/08/2020 14:08:54

[Robert Atkinson 2Participant @robertatkinson2]

No,

Tim's original 3mm LEDs had 470 ohm reistors giving a 15% overload at 14.2V
The 560 ohms on page 7 came from a random ebay image.

Robert G8RPI.

[Author]

Posts