LED lights

[Gibbo]

So, Chris, are there any worthwhile simple things that can be done to help guard against the effects of these spikes ? - i.e fitting Transient Voltage Suppressors, capacitors, in line inductances etc ? Even if they won't guarantee protection, might it help / be worth doing for the cost of the bits ?

 

i.e. What can us average boating punters fit, or do, to help protect our equipment from at least some of these spikes, which we all have ?

 

Thanks,

Nick

 

There are no guarantees but there are a few things that will certainly be a big improvement.

 

First thing is a series reverse protection diode. This must be a fast recovery diode. A 1N400X won't cut it. This will protect against reverse voltage spikes. An FR105 run at a maximum of about 750mA would be a reasonanble choice for an "easy to get hold of " diode. But remember that you'll be losing some voltage over the diode.

 

For protection against forward spikes a series inductor followed by a transient suppressor of some sort is one of the more practical solutions Don't bother with MOVs, they're crap. A true transient diode is the way to go. I haven't got any part numbers handy as I only use surface mount stuff which aren't really practical.

 

The problem is selecting the right value. If the LED is going to be damaged at (say) 20 volts, then it seems simple... "Oh I'll get one at 18 volts". Hmmm, when you carefully check the datasheet and realise that the 18 volt one starts to conduct at 18 volts but is actually only fully clamping at 24 volts or so. If you get one that fully clamps at 18 volts you realise it's starting to conduct (and thus getting warm and wasting power) at about 13 volts. Also, the high energy ones have quite a wide tolerance. Also fit a "cap in series with a resistor" in parallel with the diode. Resistor about 50 Ohms. Cap either a low value one of around 470nF (they work better at high frequencies than high value caps) or an electrolytic of around 50uF to 100uF but make sure it is a low ESR cap designed specifically for high frequencies, ie one made for switch mode power supplies. If you don't fit the resistor in series with the cap you can end up making the problem worse. But the resistor, in turn, reduces the effectiveness of the snubbing. You starting to see some of the problems?

 

All in all that should cost well under 50 pence per lamp.

 

Less than ideal but it's really the only practical and simple solution. There are more advanced techiques but it starts to get rather complex.

 

If someone says "Put a 7812 regulator in there" just laugh at them and walk away

 

The absolute best solution is really to run a completely separate feed for all the lights and regulate the entire feed.

[nb Innisfree]

Blimey, LEDs sound a right pain, easier to fit a bigger alternator and battery bank and use 'proper' lighting.

[Nickhlx]

Thanks for that....

 

I have a spare couple of 110 AH batteries, that I could arrange to be charged by day by shoreline / solar panel / engine, and run the led lights only from them at night, so no other inductive loads would see the sensitive LEDs...... It would be easy to move the supply to the lighting circuit breakers to this other source..

 

Actually, I did get some transient diodes, rated at voltages around 15 to 18 volts, and they seem to start conducting at about 1.5 volts "below rated voltage and be "hard on" at around their rated voltage... I assume there is some "standard" for the voltage between the start of conduction and when fully conducting. If these were put in front of a regulator ( say a LM 338 - 5 amps adjustable reg with up to 35 volt input) would not the transient suppressor diode cut off above say 20 volts and the regulator control up to where the Transient Diode "takes over" ? My 'scope is not "storage" so I can't be sure of seeing (fast) spikes...

or do spikes go through the regulator chips as if "transparent" ?

 

(I have been sprinkling the bi-directional 1.5KE18CA's around on electronic circuits I have put on the boat, in the hope they may provide some protection...) Too small a sample period to know if they have done anything yet, although nothing "blown"

 

Thanks,

 

Nick

Edited May 15, 2010 by Nickhlx

[Gibbo]

Thanks for that....

 

I have a spare couple of 110 AH batteries, that I could arrange to be charged by day by shoreline / solar panel / engine, and run the led lights only from them at night, so no other inductive loads would see the sensitive LEDs...... It would be easy to move the supply to the lighting circuit breakers to this other source..

 

That's actually the best way to preserve the LEDs. There's no doubt about it.

 

Actually, I did get some transient diodes, rated at voltages around 15 to 18 volts, and they seem to start conducting at about 1.5 volts "below rated voltage and be "hard on" at around their rated voltage...

 

If that's true (and to be honest I suspect you have missed something) then you have some transient diodes that are meeting a spec, the likes of which I have never seen.

 

Here's a typical (quite high spec actually) transient diode datasheet....

 

http://www.datasheetcatalog.org/datasheet/...tor/mXvusty.pdf

 

I'll pick one at random, the SMBG14A. See page 2, breakdown voltage is 15.6 to 17.2 volts. That's where it starts to conduct. That's a pretty wide range when trying to pin down a protection scheme within a narrow voltage range. Now look at the last column, the maximum clamping voltage is 23.2 volts. So unless absolutely everything is known about the entire circuit and all other devices likely to cause spikes (which is basically impossible) all we can say is that diode will clamp somewhere between 15.6 and 23.2 volts.

 

That's too wide a range to be of much real use. But, unfortunately, that's the reality of trying to clamp voltage spikes and the naive idea of "oh a zener or transient diode will do that" just gets completely blown out of the window when the datasheet is checked properly.

 

I assume there is some "standard" for the voltage between the start of conduction and when fully conducting.

 

It would nbe nice to think so, but no there isn't.

 

If these were put in front of a regulator ( say a LM 338 - 5 amps adjustable reg with up to 35 volt input) would not the transient suppressor diode cut off above say 20 volts and the regulator control up to where the Transient Diode "takes over" ? My 'scope is not "storage" so I can't be sure of seeing (fast) spikes...

or do spikes go through the regulator chips as if "transparent" ?

 

That's no better than using a 7812. Again. check the datasheet properly. The LM338, at full current, under normal temperature conditions, needs the input voltage to be 2.4 volt higher than the output voltage. So when not charging, and the batteries at (say) 12.2 volts, that means the output will be at 9.8 volts absolute maximum.

 

The general rule of thumb for that type of reg is that you make sure the input is at least 5 volts higher than the output. Anything less and the whole idea falls apart.

 

The voltage drop on the reg quoted above, and the wide variation in clamping voltage on the transient diode can be reduced by running them at much lower currents than the spec allows, but in the case of the diode you have no real control over that. In the case of the reg, even if you run it at 1/10th its rated current it will still drop almost 2 volts so with the batteries at 50% and no load on them the output can never be higher than 10.2 volts.

 

Never look at the marketing stuff for semiconductors. Read the entire datasheet and none of them are even remotely as good as they at first appear.

 

If it really was so simple then LT wouldn't have spent a few million dollars developing this...................

 

http://www.linear.com/pc/productDetail.jsp...06,C1065,P38400

 

That chip exists purely and simply to address this problem. It's all it does. It doesn't do anything else. If it wasn't such a huge problem there wouldn't be quite complex ICs dedicated purely to this task.

 

You can't replace that chip with a 10 pence diode or linear regulator!

 

Edit: I just checked a proper datasheet for the 1.5KE18CA and it's clamping voltage will be somewhere between 16.2 and 26.5 volts depending on current and manufacturing tolerances. That's the same sort of range as the diode I quoted. It really is about as good as it gets.

Edited May 15, 2010 by Gibbo

[WotEver]

Thanks for all the clarifications Gibbo

 

So... I was wrong then

 

Tony

[nicknorman]

Totally agree with Gibbo (who would dare to argue!). RS sells various wired transient-absorbing diodes, £0.30 each in quantities of 25. They seem to clamp +ve and -ve voltages equally so perhaps it would be best to add a fast diode reverse biased to clamp the negative swing to near zero, with a low value resistor in series as Gibbo suggests. However I suspect its the 100s of volts that kill the LEDs rather than the momentary slight over-voltage that results from the non-ideal behaviour of these transient absorbers.

You could fabricate the whole thing for under a £ per circuit. This is what I intend to do when our new boat nears completion.

[Nickhlx]

Totally agree with Gibbo (who would dare to argue!). RS sells various wired transient-absorbing diodes, £0.30 each in quantities of 25. They seem to clamp +ve and -ve voltages equally so perhaps it would be best to add a fast diode reverse biased to clamp the negative swing to near zero, with a low value resistor in series as Gibbo suggests. However I suspect its the 100s of volts that kill the LEDs rather than the momentary slight over-voltage that results from the non-ideal behaviour of these transient absorbers.

You could fabricate the whole thing for under a £ per circuit. This is what I intend to do when our new boat nears completion.

 

 

Some of these LED modules with 12 5050 SMD LEDs are good for up to about 30 volts, by which time a "16 or 17" volt device will be clamped hard on at about 25 volts, and the device should survive, if I read that correctly....

 

Thanks again for the input, Chris.

 

Nick

[Gibbo]

Some of these LED modules with 12 5050 SMD LEDs are good for up to about 30 volts, by which time a "16 or 17" volt device will be clamped hard on at about 25 volts, and the device should survive, if I read that correctly....

 

Thanks again for the input, Chris.

 

Nick

 

Right if they're good for up to 30 volts then you can probably get away with this idea (unless it's a 24 volt system in which case you've got a real problem). You want a series reverse protection diode. It needs to be a fast one as previously stated. FR105 would do it (see previous posts).

 

You also want a series inductor and then the transient supressor diode.

 

DON'T use the bidirectional transient diodes. They're rubbish. They have a much sloppier curve and tend to have an even wider range between starting to conduct and clamping fully.

 

The inductor needs to be very low resistance and somewhere in excess of about 500uH. Anything less won't really help much.

[Ally]

this is a subject that has been doing my head in all day....why didn't I think to check here first???????????? doh!!!!!

[Smelly]

Hopefully I'm going to answer my own question here...

 

I'm sure, Gibbo, you once said that an inverter would clamp voltage spikes... I'd imagine if it's on it's own spur then it'll only clamp the voltage once the potential works all that way back from the pump/fridge that has switched off, hence we still need to think about suppressing them.

 

Just a thought, but I think I'll be picking up some of those diodes come payday...

 

Our leds have been fine so far

[nb Innisfree]

Wouldn't it be easier if LEDs had their own individual (rechargeable) batteries? no bother with wiring in and should be a decent time before replacement/recharge.

 

Just a thought

[Gibbo]

Hopefully I'm going to answer my own question here...

 

I'm sure, Gibbo, you once said that an inverter would clamp voltage spikes... I'd imagine if it's on it's own spur then it'll only clamp the voltage once the potential works all that way back from the pump/fridge that has switched off, hence we still need to think about suppressing them.

 

That's rught, most decent inverters do indeed have some serious spike protection inside them. But for it to be effective the sensitive loads would have to be powered from the back of the inverter otherwise the inductance of the wiring would still allow the spikes to get to them.

[nicknorman]

You want a series reverse protection diode. It needs to be a fast one as previously stated. FR105 would do it (see previous posts).

 

I don't think it should be a series reverse biased diode - surely it should be in parallel to short out any negative voltage. In series, when reverse biased it has significant capacitance that would allow a transient through - that is why reverse biased diodes are used as variable capacitors in rf tuning circuits.

 

(I seem to recall saying something about not arguing, but never mind!)

[Gibbo]

I don't think it should be a series reverse biased diode - surely it should be in parallel to short out any negative voltage. In series, when reverse biased it has significant capacitance that would allow a transient through - that is why reverse biased diodes are used as variable capacitors in rf tuning circuits.

 

(I seem to recall saying something about not arguing, but never mind!)

 

Sort of, but no

 

Firstly it won't be a series reverse biased diode. It will be a series forward biased diode. But I think you just typed the wrong words

 

This is one of the reaosns I stressed that it has to be high voltage and fast (as well as several other reasons for the job requiring a fast diode).

 

High voltage fast diodes are specifically designed to have extremely low capacitance (they wouldn't be much use otherwise).

 

Shorting "crap" out is always the last resort that is only used when there is no other option. It's far better to prevent it getting in there in the first place rather than hoping you chose something big enough to, perhaps, maybe, possibly, clamp it without something going bang. This is what a series forward baised diode does.

 

PS, Varicap diodes used in RF tuning circuits are designed to have high capacitance. Orders of magnitude higher than normal diodes. They're also not much good at handing 50 amps

Edited May 18, 2010 by Gibbo

[nicknorman]

Firstly it won't be a series reverse biased diode. It will be a series forward biased diode. But I think you just typed the wrong words

 

Oops you're right - typo!

 

I think we are proposing two different approaches, either one of which would probably help / has merit (but of course I think mine is better!). Your version is to put up a brick wall to bounce the transient away somewhere else, mine is to absorb the energy of the transient to smother it. I think my approach is better because you don't need the diode to be able to pass the normal operating current of the system, this allows smaller junction therefore intrinsically faster. Ditto in that you don't need the voltage rating of the diode to be greater than the possible max transient voltage. I would use a small series resitor rather than an inductor, since the aim is to absorb the energy. An (ideal) inductor doesn't absorb energy, but it does risk the setting up of a tuned circuit with the capacitance of the diode which, with carelessly chosen values the ringing of which might make things worse

 

Even though you are right to say that the sorts of diodes we are talking about have much less capacitance than a varactor, they still have capacitance and will let the higher frequency components of a transient through initially, before cutting off suddenly as the depleted layer of the junction reaches its limit. Just the cutting off suddenly of a diode is enough to create its own transient. And don't forget that the Transzorb-type device we both agree is a "good thing" works primarily by shunting the transient to ground (whilst absorbing the energy).

 

Maybe we shouldn't persist in this discussion about electronics for too long - even though its interesting to me, I am sure its not to many since this is supposed to be a boating forum, not an electronics one! So maybe we should just agree to differ, bearing in mind we both know that EMI protection etc is a bit of a black art?

[Gibbo]

<cough, ahem>

 

Oops you're right - typo!

 

I think we are proposing two different approaches, either one of which would probably help / has merit (but of course I think mine is better!). Your version is to put up a brick wall to bounce the transient away somewhere else,

 

No it isn't. It is to ignore the transient. I'm not "bouncing" it anwhere.

 

mine is to absorb the energy of the transient to smother it. I think my approach is better because you don't need the diode to be able to pass the normal operating current of the system,

 

No, instead you need your diode to be able to handle the fault spike current which could easily be in the region of 50 amps or more.

 

this allows smaller junction therefore intrinsically faster.

 

Well as your diode has to handle 50 amps and mine has to handle 1 amp I think your diode will have a slightly larger junction

 

Ditto in that you don't need the voltage rating of the diode to be greater than the possible max transient voltage.

 

High voltage diodes have lower junction capacitance not higher.

 

I would use a small series resitor rather than an inductor, since the aim is to absorb the energy.

 

No it isn't. The aim is to stop it damaging the piece of equipment being protected. If you're going to try to turn that round to protecting the entire system from spikes then you are looking at transient "absorbers" (of whatever kind) about the size of a housebrick.

 

An (ideal) inductor doesn't absorb energy, but it does risk the setting up of a tuned circuit with the capacitance of the diode which, with carelessly chosen values the ringing of which might make things worse

 

The junction capacitance is FAR too low to ring with anything in a vehicle wiring system. It simply isn't happening. I also specced a capacitor in series with a resistor (just in case). The effectively swamps the junction capacitance because it is orders of magnitude bigger and the series resistor drops the Q to almost zero. It's called a snubber and is pretty standard practice in these sort of situations.

 

Even though you are right to say that the sorts of diodes we are talking about have much less capacitance than a varactor, they still have capacitance and will let the higher frequency components of a transient through initially, before cutting off suddenly as the depleted layer of the junction reaches its limit.

 

Well it will do, if one is daft enough to put a resistor in there. That's why I specced an inductor. Because it limits the rate of rise of current and effectively removes the spike so there simply isn't one to "get through" the diode. Without the inductor, it will never work.

 

Just the cutting off suddenly of a diode is enough to create its own transient.

 

And it won't cut off suddenly because I was smart enough to put an inductor in there instead of a silly resistor

 

And don't forget that the Transzorb-type device we both agree is a "good thing" works primarily by shunting the transient to ground (whilst absorbing the energy).

 

I don't agree they are a good thing. I agree they are rubbish.

 

Maybe we shouldn't persist in this discussion about electronics for too long - even though its interesting to me, I am sure its not to many since this is supposed to be a boating forum, not an electronics one! So maybe we should just agree to differ, bearing in mind we both know that EMI protection etc is a bit of a black art?

 

Hmmm

 

We shall agree to differ.

 

I shall also agree that I know I'm right

 

If you were right, the problem wouldn't exist, and there wouldn't be an entire industry dedicated to trying to crack a very complex problem that you have seemingly sorted with an idiot diode and a resistor

 

I'm not just "trying to be right" here. But if you really think you can crack this problem in such a simple way then I know several companies who will employ you for ridiculous amounts of money. LT have thrown millions of squids at this and have made good inroads but still haven't got it licked. Funnily enough, none of their solutions involve a series resistor or a transient diode

 

Edit: PS. If you think you have some worthwhile ideas on this problem then PM me and we can argue about them in private I'd be very interested to look at any new ideas but to be hinest I really think I have tried and analysed every possible combination. I, along with others, have been working on this problem for many many years and no real solution yet exists.

Edited May 19, 2010 by Gibbo

[nicknorman]

Well Mr Gibbo, that was an interesting discussion (now over!) - useful for me because, as a new poster to and fairly new reader of this forum I thought you had credibility. However those that respond to courteous posts that nevertheless register disagreement, with discourteous words such as "daft", "silly", "idiot", lots of bold, smilies and sarcasm hint of bluster and are not deserving of respect nor credibility. So I shall not follow up your suggestion of a PM dialogue thanks. I shall however read your future posts with new-found insight.

Edited May 19, 2010 by nicknorman

[WotEver]

Well Mr Gibbo, that was an interesting discussion (now over!)...

A pretty typical response from someone who's been comprehensively proven to be wrong.

 

Tony

[Pretty Funked Up]

I think Gibbo must be having a bad day/week

 

He's normally polite and bordering funny with his responses.

Dont write him off, he really knows his stuff and everyone has the odd bad day or two

 

 

(hope that cheque is in the post gibbo )

 

PS I bought SMT/SMD (not sure T or D?) type at ikea nearly 2 yrs ago I use them unregulated (engine is off where possible as theyre just too bright with it running ) I planned to just change em when they blew but none have yet.

[WotEver]

PS I bought SMT/SMD (not sure T or D?) type at ikea nearly 2 yrs ago I use them unregulated (engine is off where possible as theyre just too bright with it running ) I planned to just change em when they blew but none have yet.

D (Surface Mount Device)

 

Tony

 

He's normally polite and bordering funny with his responses.

His references to a "silly resistor" made me smile

[nicknorman]

I think Gibbo must be having a bad day/week

 

He's normally polite and bordering funny with his responses.

Dont write him off, he really knows his stuff and everyone has the odd bad day or two

 

(hope that cheque is in the post gibbo )

 

Mmm... well nevertheless its difficult to have an adult discussion when the counter-arguments are emotional and lack logic and consistency (eg one moment he is saying he'd use a Transzorb, next moment he is saying they're crap, he talks about snubbers which are normally R-C but then says resistors for absorbing the energy are "silly" etc). I guess he is used to being the "darling of the forum" on electronic subjects and doesn't like his territory invaded! Never mind, I will keep my own council on the matter as clearly the forum isn't big enough for both of us and he got here first! But I will re-think buying one of those rather expensive voltmeters with 408 numbers in the programme for our new boat...

Edited May 20, 2010 by nicknorman

[Pretty Funked Up]

I think I've used one of those resistors....

 

[WotEver]

I think I've used one of those resistors....

That's the very one

 

However, to be strictly accurate it's only silly if it's in series with a diode. If it's all on its own it's just a heater... oh wait... that's pretty silly too

 

Tony

[Gibbo]

A clear case of someone who's read a few books and done nowt. The world is full of them.

[Gibbo]

Things don't sink in do they?

 

eg one moment he is saying he'd use a Transzorb,

 

No I didn't. I said the only "practical and simple solution" involved a transient suppresor diode in conjunction with an inductor. I also had to explain why the inductor was needed and what the possible pitfalls and problems were. I also stated quite clearly that it was far from ideal and most definitely not a perfect solution.

 

next moment he is saying they're crap,

 

Which they are.

 

he talks about snubbers which are normally R-C but then says resistors for absorbing the energy are "silly" etc

 

They are. The resistor/capacitor I suggested is there to lower the Q of the circuit to prevent ringing. Or did you not read/understand that bit either?

 

The rest of your post isn't worth a response. Child.