[FEELER] Cheap, small Laser diode Drivers

[rhd]

There is no ripple on the output, that is the point of the output capacitor.

There is no crime in New York, that is the point of the police force

....wouldn't it be great if life and logic worked that way? It sure would be simple.....

Unfortunately, it's not. If we employed your logic, every switching regular used on this forum has zero ripple (because I promise you that they're all using output capacitors)

An output capacitor on a switching regulator doesn't remove the switching ripple, it smooths it, but not completely. There are some *decent* formulas you can employ to estimate the residual output ripple (we talk about some of them in the CCBost thread), but they're absolutely no substitute for proper scoping of the output. The residual output ripple can be substantial, and it can easily be enough to start killing people's diodes. IMO, you shouldn't be selling new drivers until you've had them properly scoped.

Frankly, nobody cares about the scoping of an intermediate stage in your driver (except perhaps to the extent that a bad result early on could be indicative of overall dirty output). A scope on the driver's final output, when the driver is connected to an appropriate test load with characteristics of the LD you'd actually be driving, is what you absolutely positively need to show, IMO.

I STRONGLY suggest not selling these drivers until more comprehensive output scoping has been conducted

 

[amkdeath]

There is no crime in New York, that is the point of the police force

....wouldn't it be great if life and logic worked that way? It sure would be simple.....

Unfortunately, it's not. If we employed your logic, every switching regular used on this forum has zero ripple (because I promise you that they're all using output capacitors)

An output capacitor on a switching regulator doesn't remove the switching ripple, it smooths it, but not completely. There are some *decent* formulas you can employ to estimate the residual output ripple (we talk about some of them in the CCBost thread), but they're absolutely no substitute for proper scoping of the output. The residual output ripple can be substantial, and it can easily be enough to start killing people's diodes. IMO, you shouldn't be selling new drivers until you've had them properly scoped.

Frankly, nobody cares about the scoping of an intermediate stage in your driver (except perhaps to the extent that a bad result early on could be indicative of overall dirty output). A scope on the driver's final output, when the driver is connected to an appropriate test load with characteristics of the LD you'd actually be driving, is what you absolutely positively need to show, IMO.

I STRONGLY suggest not selling these drivers until more comprehensive output scoping has been conducted

I'll post the DC line scopes on monday. The "residual output ripple" is as calculated, in the post on the top of the previous page.

That being said, I'm sad to see that you find no value in the oscilloscope readings from the main switching diode. Those are not readings from some "intermediate stage", they are the readings from the ONLY stage, the one that allows for this electrical magic to happen in the first place. take a look here

When the lm3410 switches on, the voltage on that inductor/diode drops, because the coupling capacitor is discharging into the inductor. When the switch opens, Current from the inductors then drives the load, and that's where the ripples on the output come from. That's what's being shown on the scope readings I posted. The inductors rebound back, and force that current back into the laser diode, and naturally there are some self resonances involved with these components. Poor design can also add more weird frequencies into your transient response. Whatever is not filtered out by spacing components close to each other, etc, etc, the final capacitor (capacitor C2 in the wikipedia article) absorbs.

Which brings me to my next point: I wasn't referencing any old random output capacitor

To quote from the datasheet[1]: "The output ripple of the converter is a function of the [output] capacitor’s reactance and its equivalent series resistance (ESR)"

And then it goes on to remind you that MLCCapacitors (which I used) have virtually no ESR, and so the majority of the output ripple is actually just the capacitative ripple of the output capacitor. As you can see in the equation above, this value is affected by output voltage, Duty Cycle, Switching Frequency, and Load Impedance. The first and last of those parameters are set by the diode or load that you are using, and we have little control over that. The switching frequency is a constant set by the switching chip (1.6MHz), which leaves the duty cycle.

The Duty cycle you can measure using an oscilloscope (see pictures posted at the top of page three) or it can be approximated like so:

given the efficiency, which can be measured by recording input and output voltages/currents (P=IV) and analyzing the differences (provided on top of previous page)

The inductor current ripple (can be calculated exactly from scopes posted on previous page) (Delta iL) follows this overall pattern because of the switching:

This value can be approximated from the formulas above (posted on previous page)

You see, you are correct. As with anything engineering related, there are some *decent* equations that approximate a real world situation, however, It's only when those calculations are combined with accurate empirical measurements that we start to leave the realm of *decent* and start to enter the world of accurate design. The only question is, which measurement is going to give us the most information...

Either way, I appreciate your strong suggestion. You may be right, these might not be safe to use at all. According to my comprehensive calculations, measurements, and observations however, they seem to be working fine. If (more like when) they do end up malfunctioning, we will adjust, and make an even better driver from there (and make sure everyone is happy, of course).

Best,

Amk

 

[rhd]

You may be right, these might not be safe to use at all. According to my comprehensive calculations, measurements, and observations however, they seem to be working fine. If (more like when) they do end up malfunctioning, we will adjust, and make an even better driver from there (and make sure everyone is happy, of course).

I'm not actually saying they're unsafe. I'm really only saying that it's unknown. They may be perfectly safe, they may not be at all, or they may fall somewhere in between.

 

[amkdeath]

Here are the scopes from the output.

On the VERSION 1: BOOST board, there was very minuscule (basically nonexistent) ripple

On the SEPIC version, there was slightly larger ripple on the switches, but it didn't kill the diode. Still, seemed a bit excessive. See below. I am about 80% certain that this is because on the current SEPIC board, the two inductors and the output cap are waaaay to far away from the chip, and the lack of a ground plane is causing a voltage drop between the two ends of the ground trace.

So with that, I've finished testing round one beta boards. I have a couple dozen boards in total, if anyone wants to purchase those for use. They work, but I wouldn't put my seal of approval on them yet. All proceeds from these sales will go directly into more boards and more testing.

Also, I just finished LED'ing my last working diode, a nifty little Red open can I got as a sample from china a few years back. The bastard took ~780mA of current for a good minute and a half without making a sound, and then it died. Totally didn't realize I had the wrong driver hooked up... haha.

Round two boards:
4 Layer Board design
added ground plane
added more damping on output cap
added CLC filter on output, poles about a decade away from the 100MHz transients (should give about 40dB down damping on those nasty ripples)











Click the images to enlarge. Blue is SEPIC, Yellow is BOOST.

Best,

amk


P.S. Anyone know what the absolute cheapest diode I can purchase atm is?

 

[rhd]

View image: IMAG0276

Isn't that 400 mV of ripple on the boost?

EDIT: Also, if I'm understanding this correctly, you're powering the boost driver with 4.7V input ?

http://postimg.org/image/sfv8ljn1p/

That's not a meaningful measurement to take if you're trying to determine the ripple your driver will create real-world. A boost driver would be expected to produce much less ripple when the input/output differential is small. The ripple at 3.7V input would be expected to be much more sizable than the ripple at 4.7V input, assuming a consistent boosted output of, say 5.25V, etc.

 

[amkdeath]

View image: IMAG0276

Isn't that 400 mV of ripple on the boost?

EDIT: Also, if I'm understanding this correctly, you're powering the boost driver with 4.7V input ?

View image: IMAG0275

That's not a meaningful measurement to take if you're trying to determine the ripple your driver will create real-world. A boost driver would be expected to produce much less ripple when the input/output differential is small. The ripple at 3.7V input would be expected to be much more sizable than the ripple at 4.7V input, assuming a consistent boosted output of, say 5.25V, etc.

No, that's the Max/Min from previous measurements, it holds the value. I never reset it. It's a straight line, the next picture is that same signal zoomed all the way in (about 40mV ripple).

Both of the measurements were taken at 4.7V, that's the experimental 780mA version. It doesn't run well under 4V, which is why it's not available for sale. Everything is the same, except for the set resistor.

These are the test results from this batch. I will not be doing any more testing until the revised boards come in.

Best,

amk

 

[rhd]

No, that's the Max/Min from previous measurements, it holds the value. I never reset it. It's a straight line, the next picture is that same signal zoomed all the way in (about 40mV ripple).

Both of the measurements were taken at 4.7V, that's the experimental 780mA version. It doesn't run well under 4V, which is why it's not available for sale. Everything is the same, except for the set resistor.

These are the test results from this batch. I will not be doing any more testing until the revised boards come in.

Best,

amk

But the line is 400 mV wide.....

Also, if your driver input is 4.7V, and the output is not even 5V, then you're barely boosting, and the ripple won't occur.

Ripple is largely influenced by the differential between VIN and VOUT. So in order to understand whether a switching driver is safe, you need measurements taken in real world conditions. IE, input of 3.7 V (ideally a bit less) to simulate a lithium ion supply, and output of around 5V, to simulate the laser diode (assuming a diode like a 445, with a roughly 5V forward voltage). That's the scenario you need to scope the output of (though you can use a test load rather than an LD of course.