a laser router

[The Lightning Stalker]

The 4A driver is now fully operational and verified using the "mother load".

I am going to need a larger heatsink. It is getting very hot very fast!

 

[The Lightning Stalker]

Before connecting the power supply to the diode, I want to be absolutely
sure it is producing clean stable output, so I put a scope on the output.
It's a good thing I did, too because there is some really bad oscillation.

This is what the waveform looks like during startup. It does this for about
1/4 of a second, then "flips" into a slightly different mode of oscillation.

I have no idea what is causing this. The input voltage is fairly clean,
aside from the usual ripple. The only thing I can figure is that maybe there
is too much gain in the system. High gain can cause all kinds of stability
issues with opamps. The only other idea is that the 100Ω resistor is
somehow causing it to oscillate. Unless there is noise getting into the
scope from somewhere. The multimeter sees exactly 4A.

All measurements were taken across the 25mΩ sense resistor.

 

[Sigurthr]

What's the V per division? I'm not used to reading modern scopes.

That second photo shows parasitic oscillation in addition to the switching spikes. I'm on my phone so I can't search it now but did you post the schematic? If not PM me it and I'll look for some fixes for you. A hi res photo of the board would help too but not necessary.

The strange startup could be several factors but my bet would be too much (or none) feedback hysteresis, too hot of gain, or too long of 5RC charging the input/power stage causing undefined or "third" state transient state. The last one sounds likely given the long time period it occurs for.

Depending on the schematic there's a few fixes for the oscillation.

 

[The Lightning Stalker]

V/div is 200mV. Here is a schematic.

:thanks:

 

[Sigurthr]

What's the value you used for the hysteresis cap marked question mark?

I'd add hysteresis across the sense resistor first off. Only problem is it's such a tiny resistance that you're gonna need a whopper of a cap. Got any ~6.3v >10,000uF super caps?

I don't see any decoupling caps either, it could be that simple honestly. 1-47uF on supply rails.

You could also try a C-E snubber cap, but it could make it oscillate worse if there isn't enough hysteresis in the system.

 

[The Lightning Stalker]

What's the value you used for the hysteresis cap marked question mark?

I don't know. It's not mine. It was on the board before the modifications were made. It would
have to be removed to find out what it is.

 

[Sigurthr]

Ah. When in doubt add another in parallel.

 

[The Lightning Stalker]

After a few slight modifications, the
circuit now looks like this.

The waveform is now much better and cleaner.

The spikes could be noise, I don't know. It is picking up
noise from the aux supply. Here it is with the power off.

The real problem is that it appears to be
switching completely on and off. That
would be great if this were a buck driver,
but this is supposed to be linear!

 

[Sigurthr]

The real problem is that the op amp appears to be
switching completely on and off. That would be great if this
were a buck driver, but this is supposed to be linear!

That is textbook for lack of hysteresis on the sense feedback. It reads high then turns off to compensate then reads low and turns on.. viola oscillation!

 

[The Lightning Stalker]

You're right. I did some searching and think I got it figured out.

I think the noise is from grounding problems and not actual
current spikes. An isolated probe would help.

What the board looks like now

 

[trussmonkey25]

Wow. I didn't realize i started such a great thread. You guys are awsome on getting this to work right. I enjoy watching your enthusiasm on my project.

 

[The Lightning Stalker]

We are definitely excited about this new diode.

The power supply, not so much. There isn't enough
space inside to fit any of the larger heatsinks I have
on hand. I've tried 2 and it just wants to overheat.
There isn't enough adjustability in the pot to get it
below 11.9V. All that excess has to be dissipated
by the transistor. I don't know what to do.

 

[trussmonkey25]

Would the tec thingy from an electric cooler work? It does a great job cooling my sodas and lunch
Another thought. Water cooled tec cooled closed loop like a car's radiator with fan and housing

 

[Sigurthr]

You can drop voltage with a series string of diodes. It will create additional points in need of heatsinking but that is one way to distribute the thermal load away from a single point.

 

[The Lightning Stalker]

Would the tec thingy from an electric cooler work? It does a great job cooling my sodas and lunch
Another thought. Water cooled tec cooled closed loop like a car's radiator with fan and housing

It would work. The trouble is the heat still has to go
somewhere. Plus we have another component taking
up space and generating more heat.

You can drop voltage with a series string of diodes. It will create additional points in need of heatsinking but that is one way to distribute the thermal load away from a single point.

That might be something to try. I would need 5-7
isolated diodes >4A. Maybe a couple of these.

DigiKey

If I could somehow get the pot to adjust lower, that
would be the way to go.

 

[trussmonkey25]

The heat is so much we may need to make a water jacket for psu AND laser head. The psu can be separated from the head. A water line to and fro the head can follow the power line to the diode laser head
That way the psu cooling can be separate from the diode head. The head will be moving around a 3 x 4 foot work surface