flexmod P3.3 catastrophic failure

[VDX]

Hi Steve,

can you post an image of your 317?

I'm using them as voltage and as current regulators since some +25 years without issues, regardles, if from ST (SGS-Thompson), Texas Instruments, Fairchild or other vendors.

You need a pretty big cooler to distribute the heat when cutting excessive voltage down - tested today 3x LM317 set to 1Amp each on a block-cooler with a 24V PSU and driving 405nm-LED's with around 4Volts voltage-drop - got really fast really hot! ... then found a 12V PSU, reduced the voltage to 10.5V (so only 5 Volts to cut, instead of 19!) and it gets good warm but run without problems for hours ...

Viktor

[kecked]

I do not beleive the flexmod design is at fault. It sounds like a bad batch of boards. One of the issues with this design is the board capacitance has to be spot on. I do wish he would update the design and make the board a touch larger. Then he could elimate this problem by spreading the parts out a little. Never the less when I hear Steve toss Andrew in the bin I am a bit taken back. There is obviously something wrong with these boards but don't toss a successful product to the curb over a bad batch. How many are out there running fine for ten years or more.

Do not bother flaming me. I will not respond. Yes he should take these back and replace them. Maybe if he gets enough failed boards he can find out why. He does not do this for money I can tell you that. In fact he would rather not bother. He keeps it going for you.

[Draco]

I am not going to flame you for that post, i really wish he would at-lest respond to the fault, i don't care if he does this for money or not, it's the total lack of customer service that i have issues with and that goes well past the flex mod, he screwed us on a past show as well by not giving the license code for software the company i work for bought, so we where unable to have what would have been a nice add on to a show we did early in the month.
I would be more then happy to ship him my failed board even at my cost to find out what had happened but i have gotten no replay to the e-mails i have sent, as i have always stated it is his lack of customer service that i have grievances with and this whole thread has reinforced that. That is not yet set in stone, if he would replay and provide feed back i would be open to change my views and continue buying from him, i would much rather support a local company then go overseas but at this point that is not an option i have.

[kecked]

Fair enough. .

[catalanjo]

I do not beleive the flexmod design is at fault. It sounds like a bad batch of boards. One of the issues with this design is the board capacitance has to be spot on. I do wish he would update the design and make the board a touch larger. ........ How many are out there running fine for ten years or more.....................He keeps it going for you.

Are you certain about this .....none of my boards have failed as yet, and most are from many years ago!

@ "when I hear Steve toss Andrew in the bin I am a bit taken back." ... to be accurate what he tossed was a "flexmod" not Andrew, ... but I have to admit, that my original decision to swap out all old flexmod boards, was mostly due to this same reaction.

However I am very concerned about leaving clients with gear known to be faulty, especially when "replacement" will be a very costly exercise for me, not to mention the inconvenience for the clients (but less so than failures during a show).

Is there any way to test "board capacitance" that you know of ? ...and do you know exactly how this is managing to disrupt the circuit in such a disastrous way?

Must get to the bottom of this somehow !

Cheers

[Draco]

I have in the past come across a high gain op-amp being sensitive to stray or parasitic capacitance causing the device to oscillate"forgive me of that spell check is wrong on that" and cause some really odd failures. I have the habit of some times working on small devices on my recliner rather then my bench and have noticed that the "simple drive" from Russia will not reliably operate when it's on my lap, yeah i know that's a self induced problem on my bad habit but that does make board capacitance a plausible fault to investigate, i lack the equipment needed to test that and not having a schematic makes it nearly impossible, my board did not fail under that set of conditions though, it was mounted into my projector and on a non conductive surface but in my career as a tech i have learned to suspect anything, i have seen boards from my former employer fail because of a signal reflection on a trace less then a cm long, tearing a pin off the test fixture corrected the fault so in this case i am open to anything

[VDX]

... I've built my first drivers around current controlling with the LM317/LM338 IC's because all other available 'modern' drivers with DC-DC-converters had overcurrent spikes which would have killed the laserdiodes over time (or pretty fast).

Attached is a DSO-image of a driver with DC-DC-converter, set to 100mA output current and measuring the voltage drop across a 0.22 Ohm resistor as load when switching the output off - a peak with 6 Volts shows a current peak of 27 Amps !!! (I = U / R)

With the laserdiode it gives nearly the same pulses, but 'only' 5-10Amps peak when set to 1Amp output



Viktor

[mixedgas]

Kecked, I'm sorry, I love Andrew like a brother, and the last thing I wanted to do is post that. But I've had too many problems with recent Flexmod revisions having the beam suppression .. Both new batches came from Illumination Supply... 7 devices is not a big sample in the scheme of things, but 5/7 failing out of box or within 24 hours during a year hurt.
!
Steve

[Draco]

on one model of green 532 that uses a factory driver that has a switch mode operation has a large spike when starting up, even with a lasorb installed it does eventually cause diode failure or mode hopping then diode failure, there are good switch mode drivers but they are not generally hobby level devices like the cheap stuff from china

[VDX]

... here for comparison the signal of my 'stable' driver around an double-OpAmp and two parallelized MOSFET's as current regulators, measuring and regulating the current through shunt resistors of 100 Milliohms in the diode path.

The first image shows the voltage drop across the shunt resistor while pulsing with roughly 70µs long pulses with 2.2Amps of 'regulated' current.




The second image shows the voltage drop across the diode itself in the extreme EMI 'contaminated' environment while laser-engraving some parts.




The received EMI signal comes from the servos and is between 0 and 3 Volts and is received by the DSO across the "free floating" diode ... the real pulsing signal is pretty stable between 0 and 4.5 Volts, when the diode emits at 2.2 Amps with around 3.2Watts optical power.

You can see, that my drivers didn't create any spikes or overshoots - and I can switch currents of some hundres Amps with the same quality without any problems, by 'parallelizing' the current controllers for the needed summed current ... a single controller block consists of a double-OpAmp and two MOSFET's, each capable of 20Amps continuous or up to 100 Amps when pulsing with 30:70 PWM ratio

Viktor