Driver with PWM input

[Benm]

I'll see if I can find one of those locally. Else I'll try again with the MOSFET and take some screenshots of the noise.

I used a CA3140E, but the chip seems slightly "beaten up". Unfortunately most components bought locally look that way, like they have been in the shelf for years.

Thing is I don't live in the US and we have very few stores that sell that kind of stuff. We don't have good online stores either. Ebay is an option, but packages from abroad take months to arrive :/

Just go for the mosfet that you have then - we should be able to get that to work.

I'm not in the US either though (netherlands in fact) which might change components available to me compared to americans. Most components, including the opamp, have more or less infinite shelf lives, though that should not cause any visible wear or damage.

I don't know where you live, but is it possible it is a second hand part, i.e. something pulled out of some circuit where it was placed in an ic socket?



As for PWM speed: this may depend on the driver. The one i designed i built for a scanner application running with 30k galvo's and modulation speed to match. It works perfectly in that, the only timing problem is in the green dpss channel which is intrinsic to dpss systems in general.

Other drivers may not be intended for high speed operation though. I'm not sure what driver you got, but if its a switchmode driver with an enable pin, chances are that is actually is not very fast. The reason is a bit technical, but most of them cannot be driven at speeds in the order of the switching frequency which could be 100 kHz-ish. With a bit of safety margin, 26 kHz would probably be too much to ask. If you have the schematic for this driver i'd be happy to take a look.

 

[Doucesti]

As for PWM speed: this may depend on the driver. The one i designed i built for a scanner application running with 30k galvo's and modulation speed to match. It works perfectly in that, the only timing problem is in the green dpss channel which is intrinsic to dpss systems in general.

Other drivers may not be intended for high speed operation though. I'm not sure what driver you got, but if its a switchmode driver with an enable pin, chances are that is actually is not very fast. The reason is a bit technical, but most of them cannot be driven at speeds in the order of the switching frequency which could be 100 kHz-ish. With a bit of safety margin, 26 kHz would probably be too much to ask. If you have the schematic for this driver i'd be happy to take a look.

Thanks for the answer!
My driver is the 1.8A Super X-Drive here:
https://sites.google.com/site/dtrlpf/home/flexdrives/x-drive

 

[Atomicrox]

Just go for the mosfet that you have then - we should be able to get that to work.

I'm not in the US either though (netherlands in fact) which might change components available to me compared to americans. Most components, including the opamp, have more or less infinite shelf lives, though that should not cause any visible wear or damage.

I don't know where you live, but is it possible it is a second hand part, i.e. something pulled out of some circuit where it was placed in an ic socket?

I managed to buy the BD677 and LM385Z-1.2 today. From what I understood of the datasheet I should keep the current on the reference <1mA for best accuracy. I suppose I should use a 22k resistor instead of the 10k on R1 even without the Arduino. I also have to wire the reference "backwards", correct?

Hadn't noticed your location before I live in Brazil, not a very good place for hobbyists. But I did get lucky with this project, could find everything locally.

I don't think they'd sell a second hand part, but it could have been abused in storage or transportation. Either way I'll try again with the new parts and let you know.

 

[Benm]

No need to change resistor values, they will not push over 1 mA through the reference when using 5 volt logic control. You should wire it in the opposite direction to the two 1n4148 resistors are indicated in the diagram.


As for the x-drive and pwm operation: It's really hard to tell form the pictures, i can't even make out any chip numbers (perhaps they have been removed on purpose). But it definitely is a switched mode supply which isn't likely to work well at high PWM speed operation. Also this pin seems to be a plain enable/disable, not capable of any analog modulation. Such units can work well in portable lasers (and flashlights) using the enable pin for slow pwm (100 Hz or so?) which is probably their intended purpose.

 

[Atomicrox]

The 1.2V reference seems to be working a lot better. Stays very close to 1.237.


Here's my setup:

I used a 10 ohm sense resistor so I could try it out without heatsinking. The load is a regular LED and a 1 ohm 10 W resistor (scoping was done on this resistor). I know I'm killing the LED at this current but they're cheap


Turns out I could find the noise again, but it's not as random as last time. The 10uF and 100nF decoupling caps helped a bit, but not completely.

First I measured the voltage output from the pot, without the op-amp and the rest of the circuit. It's very noisy while I touch it (1st pic), OK while I don't (2nd pic) and great if it's maxed out (3rd). It's very strange while I turn knob up (4th pic):

These were taken before I added the decouplers.
Here's the noise at low current (about 10mA):

A little higher and it went crazy (that middle one kept pulsating):

The noise seemed to affect the voltage at the PSU leads as well (but this went away with the decoupler caps):

These were taken after I added the decouplers.
Here's before the noise appeared:

And here's when I finally found out the noise frequency (it started to appear at about 70mA):

Here's some crazy stuff that popped up (this first one happened with the pot at minimum!):

This is the signal between emitter and base of the transistor, without and with the noise - note the noise frequency is the double:

I was starting to think it's caused by some crazy interaction with my bench PSU. I increased the voltage from 8 to 12V and the noise got under control even at high currents.



Then I tried it out with 2x14500 batts and the noise was still there. Wasn't the PSU after all. Here are two shots of the noise with batts, the 1st shows the current around 80mA, when I increased it a little bit the noise got real bad:

 

[Benm]

Wow, that seems really odd!

One thing that pops in mind right away though: is that a wire-wound resistor you are using as a test load?

Just to eliminate it's inductance as a source of problems, could you run the test again when using normal film resistors and/or diodes/leds as a load? It doesnt matter if you need to put some in series-parralel configuration to handle the power, but i am curious to see if that would solve the problem entirely or at least have a big impact on oscillation frequency.

One reason for me to suspect this is that the 3.2 MHz frequency you found actually is fairly close to the CA3140's gain-bandwidth product of 4.5 MHz. This might mean that while the opamp is trying to ramp up the current the resistors inductance actually causes it's current draw to respond so slowly the circuit initially overcompensates, and then backs down again after sensing the current is too high once the inductance is saturated and starts acting like a normal resistor again.

 

[Atomicrox]

It looks like a wirewound type to me. Didn't know that could make a difference in this sort of thing.

For what it's worth it has a measurable inductance of 0.273uH, while a low power 1ohm film resistor shows no inductance on my meter.

I'll try again when I get some free time and let you know.

 

[Apocalypse]

Is it possible to use the LM3410 like this to power a LD and also adjust the brightness with PWM using the DIMM pin?

 

[Benm]

It looks like a wirewound type to me. Didn't know that could make a difference in this sort of thing.

For what it's worth it has a measurable inductance of 0.273uH, while a low power 1ohm film resistor shows no inductance on my meter.

I'll try again when I get some free time and let you know.

Sometimes it can make a difference. For slow applications those resistors are fine, but that 0.27 uH works out to be an impedance of 5 ohms at 3 MHz. This could make things unstable. I'm not 100% certain this is the cause, but it's something simple to change and see the results.

You could also have a problem with the way it is wired: both the output and input circuitry has fairly large loops in it, perhaps this causes output current to induce some voltage on the input site making the whole thing oscillate. This is unusual, but while prototyping on breadboard you sometimes get unexpected coupling (loops, but also capacity between adjecent rows of holes etc).

 

[Atomicrox]

Turned out to be the resistor. All low current tests went well.

Also did a 1A test with the power components off-board. Works OK but when it heats up (and that's quite fast with the small heatsink I used on the darlington) it goes crazy. Pretty sure that's because of my wiring or lack of heatsinking.

Tested with the MOSFET as well and it works until a couple hundred mA, then regulation goes bye-bye and it heats up very fast. See the scope below - it only stopped around 600mA because the PSU was limiting current. With a higher limit it went up to 1.5A, even though that shouldn't have been possible with the 1ohm sense resistor.

What does that strobe pin on the op-amp do? Can I use it to control both current (with the smoothed Arduino output) and PWM (from another Arduino pin) at the same time?

 

[Benm]

Good to hear that eliminating that wire wound resistor solved the main part of the problem.

The strobe is something like a latch, useful for sample-and-hold based opamp circuits but not really useful here.

For the higher current tests: can you measure the gate/base voltage and the opamp output voltage under these conditions? It could be that the opamp simply cannot drive enough voltage over the g-s junction to get the mosfet to turn on far enough.

When using the mosfet i'd also suggest putting a resistor between gate and source (something like 1k). This creates basically 'something to drive' for the opamp instead of only the capacitive load the fets' gate presents.

If you are testing analog performance you could also add a capacitor between the opamp pin 3 and ground. This will impede pwm performance obviously, but when testing with long leads and such it could help to tackel any induced voltage picked up by the wiring. Try 100 nF or something in that order.

Cooling the transistor/mosfet is essential at higher currents. This circuit will be fine pushing an amp through a crowbar for a load, as long as your thermal design is good enough.

 

[Atomicrox]

I only tried the mosfet to see if it'd work, the darlington is smaller and cheaper

Next test might take a while as I wait for a proper heatsink for the laser. Might run a test with the whole thing soldered up with a test load first, though - it's hard to test on the breadboard with the high current stuff outside of it. I'll try your tips if it doesn't work properly.


Edit: is there a way to use both PWM and analog at the same time (using two PWM outputs from the Arduino)? I ask this because some diodes shift wavelength with current and it'd be cool to have a "weak" ~470nm laser.

 

[Benm]

You could combine PWM with analog, in the sense that you would for example drive it with 2.5 volts instead of 5 volts input, and a duty cycle of 50%. This would run the laser at 25% of its maximum current. For the 'analog' part you could create a rudimentary DAC using a few of the digital outputs (build an R-2R network). The PWM part would just involve turning the DAC to zero for some perecentage of time.

I have no idea how this would affect the output wavelength of the laser though: I think the shift is thermal (something like 0.2nm/K ?) so it will not matter much if you reduce the output/temperature by pwm or analog.

 

[Atomicrox]

There have been some tests lately indicating the WL shift on blue diodes is due to current, not temperature, while the shift in red diodes is due to temperature.

I'm not absolutely sure it's true, but I'd like to find out
If it's true this would be the coolest handheld blue laser, tunable in both intensity and WL. I'm pretty sure I could figure a way to program the Arduino to change WL while maintaining average power (decreasing duty cycle and increasing current).
I could include an LCD to view approximate current, power and WL.

 

[Benm]

Well, this would be the way to find out indeed.

If the effect is thermal there should be no difference between say running 25% of the time at 100% current vs 100% current 25% of the time.

I know for a fact that the longer wavelength with increasing temperature is purely thermal for led and IR diodes (figured that out due to poor heatsinking at times).

For blue diodes it -could- be depending on current and the effect would be reversed as well: higher current should produce a shorter wavelength rather than a longer one as observed in the thermal effects.

This effect can occur in blue LED's, but i'm not sure it would hold true for laser diodes.

The downside is that you can't really test it: laser diodes will suffer optical failure if driven at high current for a short amount of time. You cannot simply run a blue laser diode at double the rated current half of the time, of more extremely 10 times the current 10% of the time.

LED's are often quite okay with that sort of overdrive if you do it fast enough since the failure mode is mostly thermal, allowing things like multiplexed displays with leds running peak currents far over the maximum rating for continous operation.

 

[Atomicrox]

I don't think the effect is reversed. See the tests here: http://laserpointerforums.com/f65/nubm07e-465nm-2-9w-diode-test-hitting-470nm-94437.html
Should be similar with the NDB7675.

As long as it's below the point where it dies when overdriven in CW mode it shouldn't be more dangerous if it's pulsed, right? With the NDB7675 I think that's around 2.9A.