LD driver configuring questions

[VDX]

... when next time setting up my "micro-station" for aligning fibers, I'll bend the wires, so they will touch, then connect them with a small droplet of solder ... did this type of "fusing" with some nano-nailheads, to clamp them to a probe needle ... the "heads" were 0.03 mm big, so similar in diameter as the bond wires.

Gold is conducting better, but the solder droplet is bigger than the wire, so won't be much difference in volumetric conductivity

... and yes, overdriven until damage -- IIRC this 2W-diode (1,7A max!) did survive around 3 Amperes from an overtuned driver for maybe 4 seconds, before the wires blow!

Viktor

[catalanjo]

Many thanks Viktor ... this kind of confirms my suspicions concerning Maltes LD.

So dodgy gain pot (primary suspect), or very spikey PSU (next up) = cause of trauma.
Question is how to figure out which, with no scope

Perhaps a sacrificial LED and series/paralell resistors using different PSU would be best bet to check gain pot.

Cheers

[VDX]

... my drivers are OpAmp based constant current regulators, so testing them is pretty simply with a low Ohms resistor in place of the LD.

I'm using 0,1 Ohm for easy spotting the current by their voltage-drop value with the 1/10-factor ... this could be 1 Ohm too -- but for my high speed controllers (with up to 2 MHz pixel-clock) lower resistance is better to get the timing more exact.

So I'm clamping the resistor into the LD output pins, switch the driver to "constant ON", adjust the output current for the whished value (stay below 90% of max rated current for longer LD-lifetimes!) -- and last replace the resistor with the LD.

Other drivers could be different, but constant-current types will behave most identically.

One critical point is eventual over-/under-shoot of the driving voltage and current - some of the modern drivers with integrated DC-DC converters aren't fast enough to regulate high speed switching in time! - I've found some with either hefty overshooting or too slow to follow a fast signal.

Here is a DSO-image of massive over-/under-shooting -- the current was set to 10 Amps (1Volt voltage-drop for 10 Amps current flow), but the regulator "overreacted" a bit and added 10 Amperes on his own, so the diode with a max rated current of 11 Amps got spikes of up to +20 Amps or (even worser!) -10 Amps, and (surprisingly enough) survived only some ten seconds, but managed to melt a plastic lense, which was stated for safe with powers of up to 15 Watts IR!





And here an image of a driver, which was not fast enough to get the current synchrone with the controller signal:




Viktor

[jors]

I dit it!!
https://www.photonlexicon.com/forums...es-experiences

@ Jors "RIP diodes" thread ...great idea ... why not go ahead and do it?

[catalanjo]

Good stuff Jors !
Lets hope people start sending dead (open circuit) diodes to Viktor for microscopic post mortems!

I still haven't figured out how to test for fast spikes for people who have no (or slow) scopes!


Cheers

[VDX]

... you can try with this "Arduinoscope" - http://www.homebrew-tech.com/arduino...e-arduinoscope

But don't think, it's much faster than a sound-card hack with up to 44kHz normally ...

Viktor

[catalanjo]

Yeah I agree ..... but I was hoping for something far cheaper than that if at all possible,...all that is really needed is to capture a "one off" fast event and translate it into a much slower one, so that it becomes visible. A "trigger/gate" setup springs to mind, even if it kills part of the mechanism in question (as long as it's cheap)
Cheers

[VDX]

... for devices with "slow" triggered inputs, which can't detect the only 0,5µs long step pulses of my controllers, I'm using a 74HCT123 (retriggerable monoflop) to define/elongate the 500ns pulse to a longer duration - actually with a RC couple of a 10 Kiloohm resisitor and 1nF cap to around 8 microseconds pulse-width.

With bigger resisitors or capacitors this can be set to much longer pulses too ...

Viktor

[catalanjo]

I thought we were looking for much shorter pulses than 500ns, more in the 50+ns range
But that is definitely the kind of direction I was referring to, since it could enable a LED to advise of the presence of spikes

Cheers

[VDX]

... the 74HCT123 triggering sensitivity is much faster! - they're testing the chips with trigger pulses of 6ns duration.

I was only refering to the 500ns long pulses of my ArduinoDue controllers, what's not long enough to activate the trigger input of the paste-dispensers I'm developing actually ...

Viktor