LED driver has bogus compoents on it to protect against LED overcurrents?

[treez]

Hello,
Please could you help us because we believe an electronics design contractor company is trying to embezzle money out of us. We believe they deliberately designed a 140W LED power lamp for us with “problems” in it so that it would fail on thermal grounds. We believe that the electronics design contractor has pre-disposed the LED driver to fail on thermal grounds, knowing that since only they hold the PCB Layout files, we will have to go back and pay them to re-do the PCB to make it thermally acceptable.

There is also a microcontroller on the PCB and some signal circuitry, and he has layed this out so incredibly tightly that even though we have the schematic, we cannot replicate the PCB ourselves. We cannot make the PCB any bigger because all the enclosures have now been made.

The “problems” in the circuit relate to a lack of thermal spreading copper on the small LED driver PCB –this will likely lead to thermal failure…..the reason there is little thermal spreading copper on the PCB is because the designer added a large number of bogus “protection” components to the LED driver circuit, taking up lots of PCB space. These bogus components provide what the designer tells us is a “LED protection circuit”, which shuts off the LED load with a transistor switch whenever an overcurrent surge flows into the LEDs…..the designer tells us that if the LED connector ever becomes intermittent due to corrosion…..then the LED load will keep going open and then reconnecting…the designer tells us that when the LEDs go open, the output capacitors will over-charge (as the LED driver is a current source) and then when the connector re-makes, the overcharged output capacitors will deliver a surge of damaging current into the LEDs.

Thus the designer has added all of the components that you can see in the below schematic which he says, prevents damage to the LEDs and allows the microcontroller to detect that the connectors have become intermittent.

I believe this circuit is bogus….certainly running the simulation with and without the “led protection” components makes no difference….(the intermittent connector is simulated with a voltage controlled switch added to the simulation)
I admit that the comparator would trip on overcurrent, but does anyone seriously add such components to LED drivers?…I mean…how common is connector corrosion?…to the point where the contacts become intermittent?…I suspect bogus circuitry here.
I don't believe that the opamp that he uses as a "comparator" would trip quickly enough to stop the surge of current that he speaks about...and the simulation bears this out.

The designer also did not add much thermal spreading copper even where he could have done…and used no thermal vias to bottom spreading copper, even though he could have done this as the bottom layer is unpopulated. He also uses a 1206 high side sense resistor that dissipates 310mW.

We believe that he predisposed this circuit to fail, in the hope that since he holds the (Easy PC) PCB files (we do not have them), we will go back and offer to pay him to improve the PCB thermally.

What do you think? Do you think the “LED protection components” are bogus?
Why would the LED connectors be likely to corrode anyway?……even if they could, wouldn’t it be better to stop the corrosion by squirting silicone over them at assembly time.?

Schematic of LED driver with suspected “bogus” led protection components attached, as well as an LTspice file, showing the workings of the LED driver with the suspected “bogus” led protection circuitry.

 

[dick_freebird]

It looks over elaborate to me, and not like the sort of simple buck
I'd expect.

I suppose my questions would be:

- If the connector fails, do you care about LED damage (and, too,
if it fails open how can the LED be damaged anyway)? It's not like
somebody's going to repair the assembly, most likely. If it's really
disposable (as most consumer electronics) then your concerns
devolve to fire safety, pretty much, and that might be had at
less complexity. Maybe you want to spend some time at FMEA,
and figure out what's needed vs "good idea" vs "not even a good
idea".

- What do you hope to get, from these people? Some sort of
do-over, by proving some malfeasance? And is that (do-over,
by an unwilling and perhaps vengeance-minded contractor),
going to put you (with any confidence) closer to your goal?
Or should you just reverse engineer until you understand, and
then find a more capable and willing Plan B contractor?

 

[D.A.(Tony)Stewart]

When there is no spec. or contract to define the deliverables and test acceptance criteria, the fault is not the designer but the person who wrote the spec. Normally all design documents for boards and software are not included in a contract unless specified.

The protection would be necessary if hot inserted with a large energy storage Cap and could fuse the I^2t limit of the LEDs, but otherwise not. There is no need for a large storage cap.

But there certainly is a need to operate at the specified ambient to meet Spec.

I wish you contacted me sooner.

 

[treez]

The output capacitance is just 9.4uF.
I don't believe that the disconnect fet would switch off in time to stop the overcurrent surge which would occur each time that the led connector re-makes....the opamp that gets used as a comparator would surely not be fast enough, and even if it were, there will be a delay before the PNP/NPN disconnect fet driver would act...thus any overcurrent surge would not be staunched?

As for the circuitry making it easier for the microcontroller to detect the incidence of intermittent connection, I don't believe this circuitry does make it any easier for the micro to detect this?

 

[mtwieg]

So you hired someone to do a design, but didn't include the full design files and "works properly" in your list of deliverables? You invited this sort of situation. Perhaps you could get it working by using heavier copper on the PCB?

Also a couple 4.7uF caps aren't going to hold enough energy to damage a string of high power LEDs. A simple overvoltage detection function is sufficient to deal with hot plugging the LEDs.

 

[treez]

we cant do ov detect as the highest vf led string voltage is too close to the actual supply rail......the supply rail is only 48v.....in our newer circuit we do led open detect by repeatedly reading the led current via a high side current monitor into micro adc

 

[D.A.(Tony)Stewart]

How much does the hot spot temp rise exceed 25°C ie 50C at 25°C ambient?

I think the OVP protection was to protect the contacts from burning the plating from the surge current, not protect the LEDs.. The connector current needs to be derated by the number of contacts used, I believe as the local air temp will accumulate between pins.

At this point I will give up unless some you are willing to contribute required info.

 

[treez]

if the contacts heave become intermittent due to corrosion, then they are already damaged so it doesn't matter too much about the contacts being damaged then

 

[dick_freebird]

That was kind of my point, about the thing being "disposable"
vs repairable. What failure mode can you conceive, that this
"protection" style would benefit?

 

[treez]

I think the OVP protection was to protect the contacts from burning the plating from the surge current, not protect the LEDs..

..the "protection" circuit is bogus, and doesn't actually manage to stop the overcurrent due to break/re-make of the connector. It is absolutely bogus, it doesn't work. Theory and simulation say so.

We all know simulations can NEVER be used to prove something works, but generally, if a circuit comprising simple components does not work on the simulator, then its pretty much certainty that it won't work in real life...the tripping of the "comparator" is simply not fast enough to prevent the surging discharge current.

We are talking bogus.
We are talking predisposition to failure of the PCB....and further payment to the contractor to correct the mistakes that he inserted into the circuit...

This is like a plumber leaving your pipes leaking so that you need to call him back to stop your ceiling caving in from the weight of water.
This is like a garage shovelling iron turnings into your gearbox, and coca cola into your oil, so your car breaks down and you come back for service quicker etc etc

Sorry SunnySky, we will get the temperature info soon.
As you suggest, doing a thermal test to see temp rises above ambient at ambient temperature is a good idea.

 

[FvM]

I wonder if implementation of protection circuits has been adressed in the original design specification and if so, what has been exactly specified in this regard. Similarly, what has been specfified about efficiency, heat disipation, board to ambient thermal resistance, acceptable overtemperature etc.

Which part of the design has been prescribed to the contract developer and what has been added by him?

 

[treez]

the contractor was not asked for such protection against open led load in this manner...........we need to detect open load, but that can simply be done by a micro reading the sense resistor voltage via a small LTC6101 current monitor.........the contractor uses this current monitor, but uses a whole load of other nonsense components....we did not put in the spec, "do not include bogus circuitry that does nothing, and takes up room that could otherwise be used for thermal spreading copper, or correctly sized components", because we expected him to be professional enough not to do that...obviously we were wrong.

..its indisputable, that "comparator" does not trip the series fet quick enough to prevent the overcurrent surge from flowing following a break/re-make incidence.

Bogus full stop.

 

[mtwieg]

we cant do ov detect as the highest vf led string voltage is too close to the actual supply rail......the supply rail is only 48v.....in our newer circuit we do led open detect by repeatedly reading the led current via a high side current monitor into micro adc

Oh yeah, it's a buck topology, not a boost. So based on that alone the supposed problem of the output cap overcharging without a load is just bogus (except maybe in the case of really bouncy contacts, where the output could ring up to higher voltages for short periods).

Anyways, IMO the whole issue with surge current due to bad contacts is bogus, at least when your capacitors are so small and you're using a buck topology. And even if it were an issue, putting output OVP on it is the easiest solution (the fact that the output voltage is near the input voltage doesn't matter, so long as your OV detection isn't horribly inaccurate).

 

[treez]

thanks, though the OV threshold on the LT3756-1 has a tolerance on it, and we could end up triggering ovp when it was just a high vf led string...in any case, the led drivers are three in number , in order to spread the current out, and if one goes open, the others pulling current drags the rail voltage down slightly, and so you wonldnt have the voltage go up high enough to reliably trigger the ovp.

..so anyway, we detect open load by just every now and then reading a current monitor ic which is also connected to the existent high side sense resistor....that does the job fine.

Another point about this buck led driver , is that the designer uses a 20 milliohm low side sense resistor, and then bizarrely divides its voltage by two with a 2-to-1 divider....thus the 1206 dissipates 230mW max and is mounted on minimal pads....poor thermal situation for a 1206 resistor........we wonder why he didn't just use a 10 milliohm low side sense resistor, but then again, why would a car mechanic fill your car with fresh new, beautifully viscous oil when he can fill it with a mixture of water, mud and sour milk?...and have your car fail sooner ready for the nxt lucrative service?

 

[D.A.(Tony)Stewart]

It appears the contractor did not receive clear specifications for environmental conditions , protection limits, test acceptance criteria nor was a design review mandated.

As a consequence of insufficient design experience to make the difficult tradeoffs , some ineffective circuits took scarce board space that would be useful for other solutions.

Verdict: inexperience on both sides.
Recommendation: define the acceptance criteria in a spec and return to contractor for a Revision, request for quote and avoid derogatory comments, then negotiate for design files .

Ensure you schedule a Design review next time and a spec. Before starting.

 

[treez]

Thanks , will do.....mind you, in UK, this is par for the coarse, we rip each other off, period. In UK, we even have a company called commsaudit.com, who exist to build RF electronics for the military....they don't design anything, (or at least they didn't when they started out) they just build what other contractors have done for the military, in order to check that these contractors are not ripping off the ministry of defence, since they were typically ripping the MOD off big time.

Commsaudit have started expanding now , in order to keep their RF guys 'tuned up'.