Welcome to Laser Pointer Forums - discuss green laser pointers, blue laser pointers, and all types of lasers

LPF Donation via Stripe | LPF Donation - Other Methods

Links below open in new window

ArcticMyst Security by Avery

Rethinking our approach to the DDL?

rhd

0
Joined
Dec 7, 2010
Messages
8,475
Points
0
Something occurred to me this evening -

I was testing out the MIC29312 IC in a DDL-like configuration. Jufran88 was ambitious enough to put together a PCB based on my suggestion that it would be a good IC to test out. The benefit to this IC, in theory, being that it has an insanely low dropout of 0.35V at 2A current. I won't jump too far into my findings with that IC in the end (Although Jufran88 - your PCB design works great now! The setup just doesn't give us the low dropout we were hoping for). Rather, this thread is really about the fundamental limitation we're encountering in ALL implementations of DDL drivers.

Specifically, DDL drivers use voltage regulators like the LM317, LM1117, and more recently the 1084, 1085. Now, some even fancier ICs like the MIC29312 are starting to poke their heads up. However, the design of the circuit itself is the new limiting factor.

Why? Because our DDL circuit relies on the fact that these voltage regulators work by maintaining a drop of 1.25V across two of their pins. As we change the resistor value between those two pins, the regulator needs to pump more (or less) current across the resistor, in order to engage Ohm's law in a way that produces a 1.25V drop. By that process, the current moves (in an LM317 setup) out from the "out" pin, across the resistor, dropping 1.25V, at which point the drop is measured at the "Adj" pin, and the current then continues on to our laser diode. The fundamental limitation, is that this 1.25V drop must ALWAYS occur, and will (in a DDL circuit) do so in the path of current that flows to our diode. In other words, our DDL circuits will never drop LESS than 1.25V, even with an imaginary IC that could achieve 0V dropout.

So, what's the solution?

The solution I would like to toss out for discussion is not a perfect one. However, short of finding an IC with a lower-than-1.25V reference voltage, this may be out next best shot at improving the DIY linear driver space. For some time, users here have been building 405s, 445s, and even 650s with cheap LED drivers. Yet cheap LED drivers aren't typically constant current sources, they're generally constant voltage sources.

There are drawbacks to a constant voltage circuit, and instinctively it feels like the wrong approach to take. Yet there is a fair amount of evidence on these forums of constant voltage circuits working. So why not use the same ICs we've been building into DDL drivers, and use them in their intended capacity as constant voltage sources? The benefit here being that we could actually begin to unlock the incredibly low ~0.3V dropout of newer LDO/ULDO drivers.

The result might finally be a DIY linear circuit that can run a RED from a single lithium ion, or 445 builds that continue to provide their output at lower levels of battery charge.
 
Last edited:





Joined
Jan 14, 2011
Messages
3,816
Points
63
Hm. Good thinking. I think I will begin work on that MIC29312 as a voltage regulator.

However, one thing - how do we test the current output? I would assume it would output the current that the diode would draw at said Vf... but does that mean that we have to start thinking of "power input" into our diodes as voltage and not current? That will completely change up the game plan.

And, as we have seen before, voltage regulating *can* be more dangerous because a minor change in voltage will result in a huge change in current....

Further, what about the heat issues?

I guess if we are going to use a voltage regulator, we should probably keep it a "relatively" low input current/voltage so that, even if it heats up and draws more current, it will still be safe.
 

rhd

0
Joined
Dec 7, 2010
Messages
8,475
Points
0
That's the tricky part - we would need to plan around the Voltage a diode pulls at a particular current level. So it would require a bit more precision. 4.7V could be disastrous for a 445 whereas 4.5V might be perfect.

It might even vary per diode, so that you'd want to take a constant current driver at test your diode's Vf at a particular current level, and then configure your constant voltage driver to supply exactly that Vf.

It would certainly be a more involved process than building a constant current linear driver. At the same time, for some builds where a DDL driver struggles (single cell red), it might be worth the effort.
 
Joined
Jan 14, 2011
Messages
3,816
Points
63
Well, I actually just recently bought some copper clad, and I actually happen to have some ULDO regulators on hand, and I have a crap ton of nice new 1W resistors... so I can actually experiment with this in the coming week. I may even in fact have an MIC.

I will see if I can work out some good designs and test it out.

But, this also means our test-loads that we usually use won't work anymore.
 
Joined
Oct 26, 2007
Messages
5,438
Points
83
What makes you think that these voltage regulators are not intended for other uses such as constant current sources? The only thing they regulate is the voltage between their output and the ADJ/GND pin, enabling all kinds of functional circuits. Even the "DDL" driver is just an adaptation of the constant current example in the National LM317 datasheet. It's like saying that op-amps are only intended to be used for negative feedback.

The reason we use the DDL circuit is because it is reliable, easy to build, and easy to analyze and adapt. All you tell the person using it is to apply Ohm's law to the constant voltage differential the voltage regulator generates. The voltage regulator provides a consistent, isolated voltage that most people don't have access to without specialized equipment like power sources, or relying on batteries. Once the circuit is built, the laser diode, barring heat damage or bad soldering, is essentially shielded from current spikes or other problems stemming from battery hookups or other problems. It's really a great circuit for beginners and regulars alike.

There are other circuits of course. We could use MOSFET-BJT/op-amp current sinks, or maybe even try to make a DIY boost circuit, but sourcing the parts is a pain for people, such as finding precise current sense resistors, and even MOSFETs are hard for people to find. Plus, with the high current requirements of our lasers, some circuits, such as boost-buck circuits are hard to construct by hand, and need specialized SMD chips, inductors, and other exotics to get working. That's why people pay $20-$30 for Dr. Lava's drivers.

Probably a good step up from a DDL circuit is to make your own MOSFET-based current sink. That way you're only sucking up 1-2V below your laser diode, and you can regulate far more current than an LM317.

A more ideal circuit would be a boost-buck circuit, allowing lower-voltages to power higher-voltage devices. However, as noted before, commercial chips for such power sources are usually SMD varieties, require inductors, capacitors and transistors, and must be carefully constructed to ensure they can oscillate at the correct frequency. Maybe a "Joule thief" or flyback converter could work, but they're usually too low current to work for lasers.
 

rhd

0
Joined
Dec 7, 2010
Messages
8,475
Points
0
What makes you think that these voltage regulators are not intended for other uses such as constant current sources?

- Nothing. My OP wasn't suggesting that using them as constant current sources was in any way innappropriate. I still use DDL style constant current drivers made from voltage regulators all the time. They're great. But using them in that configuration will always drop the Vref voltage, and that's what I'm interested in moving beyond.

There are other circuits of course.

- MOSFETS, boost, joule-thieves, and boost-bucks are much more complicated than a DDL, and likely more complicated than the voltage regulating approach I put forward for discussion here. Those that rely on precise frequencies can even be impacted by the spacing of traces and the layout of components on a PCB. Plus, many here are familiar with voltage regulator ICs, and dare I say are likely to have a bunch on hand.

Probably a good step up from a DDL circuit is to make your own MOSFET-based current sink. That way you're only sucking up 1-2V below your laser diode, and you can regulate far more current than an LM317.

- Sucking up 1-2V below our diode's Vf is entirely the issue I was addressing in my OP. The objective is to avoid sucking up 1-2V below our diode's Vf. If all we wanted was to better the LM317, then an LM1117 would do the trick, and we've been there, we've got down that road. For higher current with the LM1117's low dropout, we can use a 1084 or 1085, again, sucking 1-2V below our diode's Vf (probably closer to 2V)

The objective, is to find a DIY circuit that can move beyond the 1.25V theoretical minimum dropout of our current DDL design, while remaining relatively doable for a hobbyist without PCB manufacturing capabilities.
 
Last edited:
Joined
Jan 14, 2011
Messages
3,816
Points
63
Hmm. I just checked out the MICs datasheet, and it says it can regulate down to at least 1.25V... that makes me suspect that that is still the lower limit on voltage drop. It's much better than 2.5V or so from current sources, but it's still quite a bit of overhead.

EDIT: So I think we may run into a problem without delving into bucks or boosts... apparently, the band gap voltage of silicon *is* 1.25V. So, as far as I understand, that means that any "reference voltage" used in an IC made of silicon (like, every IC), we will be stuck with that reference of 1.25V.
 
Last edited:
Joined
Sep 20, 2008
Messages
17,622
Points
113
What makes you think that these voltage regulators are not intended for other uses such as constant current sources? The only thing they regulate is the voltage between their output and the ADJ/GND pin, enabling all kinds of functional circuits. Even the "DDL" driver is just an adaptation of the constant current example in the National LM317 datasheet. It's like saying that op-amps are only intended to be used for negative feedback.

The reason we use the DDL circuit is because it is reliable, easy to build, and easy to analyze and adapt. All you tell the person using it is to apply Ohm's law to the constant voltage differential the voltage regulator generates. The voltage regulator provides a consistent, isolated voltage that most people don't have access to without specialized equipment like power sources, or relying on batteries. Once the circuit is built, the laser diode, barring heat damage or bad soldering, is essentially shielded from current spikes or other problems stemming from battery hookups or other problems. It's really a great circuit for beginners and regulars alike.

There are other circuits of course. We could use MOSFET-BJT/op-amp current sinks, or maybe even try to make a DIY boost circuit, but sourcing the parts is a pain for people, such as finding precise current sense resistors, and even MOSFETs are hard for people to find. Plus, with the high current requirements of our lasers, some circuits, such as boost-buck circuits are hard to construct by hand, and need specialized SMD chips, inductors, and other exotics to get working. That's why people pay $20-$30 for Dr. Lava's drivers.

Probably a good step up from a DDL circuit is to make your own MOSFET-based current sink. That way you're only sucking up 1-2V below your laser diode, and you can regulate far more current than an LM317.

A more ideal circuit would be a boost-buck circuit, allowing lower-voltages to power higher-voltage devices. However, as noted before, commercial chips for such power sources are usually SMD varieties, require inductors, capacitors and transistors, and must be carefully constructed to ensure they can oscillate at the correct frequency. Maybe a "Joule thief" or flyback converter could work, but they're usually too low current to work for lasers.

Although I find the search for the hobbyists Ultimate Low Dropout
Regulator a good thing....
It is nice to see that someone else smells today's reality...;)


@ rhd... that doesn't mean you should stop looking... It may possibly
be out there...


Jerry
 
Last edited:

anselm

0
Joined
Nov 22, 2010
Messages
2,448
Points
0
I've been considering trying a constant voltage linear driver using a "low dropout" IC,
like LM1117 or 1085, for the single special case of driving the red LCC/LOC with a single LiIon.

If you notice, for the standard constant voltage circuit of these linear ICs, there's no
resistor on the output pin!
That right there tell's me we can get a lower total driver voltage dropout compared
to the constant current approach.

I haven't done it yet simply out of laziness of figuring out what resistor value to use on the
adjust pin,
you know, like that 1.25V/current formula...:rolleyes:

I have thought about it setting the voltage like so:
Use a known driver you can adjust for a precise current, the one you'd like the diode
to work at.
Run the diode for a while until it gets warm and stabilizes it's temperature.
At this point (actually, might as well keep reading it at different times as it warms up),
measure voltage between the anode and cathode. This will be your target when
setting up the constant voltage source. When the diode is cold, it will a have a higher Vf
for a given current, so in practice in the final build, the diode should start off at only partial power
and grow in intensity to full power as it reaches operating temperature.

You really want the diode to be heatsinked properly to prevent thermal run-away,
but you should have noticed and figured that out by the time you took your Vf measurements
at different temperatures on your constant current setup before.


All in all, a bit more involved (diode per diode setting) and risky, but sounds worth a
quick soldering and measuring session...

I do like the idea of using power transitors or MOSFETS, because it seems to me you can get away with a smaller
and less current capable components to drive the gate, no?
Driving that gate though...... there's another gaping hole in my (very limited anyway) knowledge in electronics.
 
Last edited:
Joined
Sep 12, 2007
Messages
9,399
Points
113
Constant voltage has already been done by a few people. It is more risky for obvious reasons, but not more risky than say... running diodes at twice their rated current or in heatsinks without fins :shhh:

Here's how I would do it. They're linear devices, so Iin = Iout. Measure the current on the input side. It requires fine-tuning, so resistors-only won't work. You need a resistor and a pot. Set the voltage to a low value and turn it up until the current reads a little under what you want it to. 10% maybe? This allows headroom for when the diode's voltage-drop lowers when it heats up to draw more current.
 
Joined
May 25, 2010
Messages
3,655
Points
0
Here a video i seen a wail ago of I think Dont laz me bro testing this.


Edit:

Not sure why this part dint post.

But i also said that "This sounds like a cool idea. I would not seeing more of this.:)"
 
Last edited:

rhd

0
Joined
Dec 7, 2010
Messages
8,475
Points
0
All in all, a bit more involved (diode per diode setting) and risky, but sounds worth a quick soldering and measuring session...

Constant voltage has already been done by a few people. It is more risky for obvious reasons, but not more risky than say... running diodes at twice their rated current or in heatsinks without fins :shhh:

Here's how I would do it. They're linear devices, so Iin = Iout. Measure the current on the input side. It requires fine-tuning, so resistors-only won't work. You need a resistor and a pot. Set the voltage to a low value and turn it up until the current reads a little under what you want it to. 10% maybe? This allows headroom for when the diode's voltage-drop lowers when it heats up to draw more current.

A bit more work - but I think the payoffs might make it a worthwhile bit of work to undertake. It would still be a require a level of experience lower than that required to build something with coils, MOSFETS, etc.

Anselm: You hit the nail on the head with the resistor not being in the output path when we use these ICs as Voltage regs. That's the key here - no 1.25V reference drop. Effectively, we'd be left with the voltage drops on the spec sheets, which can be quite low. I've seen some below 0.3V, and this particular MIC mentioned above is only about 0.35.
 
Joined
Oct 26, 2007
Messages
5,438
Points
83
- Nothing. My OP wasn't suggesting that using them as constant current sources was in any way innappropriate. I still use DDL style constant current drivers made from voltage regulators all the time. They're great. But using them in that configuration will always drop the Vref voltage, and that's what I'm interested in moving beyond.

I'm just responding to your claims that these voltage regulators are "intended" for voltage regulation, which clearly they're not limited to, and that it is a "wrong" approach to current regulation, which again is not true.

- MOSFETS, boost, joule-thieves, and boost-bucks are much more complicated than a DDL, and likely more complicated than the voltage regulating approach I put forward for discussion here. Those that rely on precise frequencies can even be impacted by the spacing of traces and the layout of components on a PCB. Plus, many here are familiar with voltage regulator ICs, and dare I say are likely to have a bunch on hand.

Then I'm confused by what you're actually looking for. First you want to abandon the voltage regulator-based constant current ("DDL") circuit, and now everything else is too complicated? You can't have your cake and eat it too.

The objective, is to find a DIY circuit that can move beyond the 1.25V theoretical minimum dropout of our current DDL design, while remaining relatively doable for a hobbyist without PCB manufacturing capabilities.

But are we really rethinking our approach to the DDL circuit at all? It's the same thing, only different Vref and Vdo.

If you want to rethink the current regulation, you're going to have to rethink the circuit, and people are going to have to do more than just apply Ohm's law to a Vref. If building something like a BJT-Mosfet current sink is too complicated for people, maybe they really shouldn't be in the laser making hobby to begin with.
 

rhd

0
Joined
Dec 7, 2010
Messages
8,475
Points
0
I'm just responding to your claims that these voltage regulators are "intended" for voltage regulation, which clearly they're not limited to, and that it is a "wrong" approach to current regulation, which again is not true.

I think you've completely mis-understood, or perhaps misread this thread and the OP. You put the word "wrong" in quotation marks, but I'm wondering what comment of mine you're quoting? I haven't made ANY claims that using these ICs as current sources is the "wrong" approach to current regulation. The closest I came to that type of comment was actually with respect to using them as voltage sources when I said: "There are drawbacks to a constant voltage circuit, and instinctively it feels like the wrong approach to take. "

If you've been paying attention to this thread, I think you'll notice that you already made a similar comment, to which I reaffirmed the position that using these ICs as current sources is A-OK, when I replied with: "My OP wasn't suggesting that using them as constant current sources was in any way innappropriate. " I don't mean to insult here, but this has been covered, and addressed. DDL circuits are great current sources. I don't need to be won over on that point :)

Then I'm confused by what you're actually looking for. First you want to abandon the voltage regulator-based constant current ("DDL") circuit, and now everything else is too complicated? You can't have your cake and eat it too.

Again, I think you've missed the point of this whole OP. There's more than just the "voltage regulator-based constant current ('DDL') circuit", or "MOSFETS, boost, joule-thieves, and boost-bucks". Even if I have "abandoned" those (which is silly, I haven't, its just not what I was exploring here), that doesn't rule out every possible alternative. What's left to talk about?

Voltage regulator-based constant voltage circuits!

LOL, that's what this whole thread has been about ;) IE, the potential to use essentially the same components we're used to using for DDLs, and build a circuit that wouldn't require us to drop 1.25V through a resistor.

But are we really rethinking our approach to the DDL circuit at all? It's the same thing, only different Vref and Vdo.

No, it's not the same. I'm not sure if you've scoped out the spec-sheets for our common voltage regulator ICs, but using them as a constant-voltage source implies a different circuit. I suppose it's the same thing in the sense that they're both circuits. But then I may as well rename this thread "rethinking our approach to the Intel 8086 CPU" ;)

If you want to rethink the current regulation, you're going to have to rethink the circuit, and people are going to have to do more than just apply Ohm's law to a Vref. If building something like a BJT-Mosfet current sink is too complicated for people, maybe they really shouldn't be in the laser making hobby to begin with.

I half agree with you here. The approach I outlined here DOES require us to do more than just apply Ohm's law to a Vref. We've had a few constructive suggestions as to how to accomplish what I outlined in my OP, and those suggestions absolutely require more work than a DDL.

Having said that, I totally disagree with you suggestion that any of us who can't building a BJT-Mosfet current sink shouldn't be in the laser making hobby to begin with. That's ridiculous. I can't build a BJT-Mosfet current sink. I shouldn't be in the laser hobby?

I may one day learn how to build a BJT-Mosfet current heatsink. But if I do, it would only be because this hobby enticed me to learn how. In the mean time, some of us with less electronics component experience need to learn a little bit at a time. I'm comfortable pushing my knowledge past the relatively simply DDL circuit, but for me at least, I've got a ways to go before I can painlessly jump from DIY-ing a DDL circuit, to a joule-thief, to a boost-buck, during a casual Sunday afternoon of electronics-ing ;)
 
Last edited:

anselm

0
Joined
Nov 22, 2010
Messages
2,448
Points
0
I haven't done it yet simply out of laziness of figuring out what resistor value to use on the
adjust pin,
you know, like that 1.25V/current formula...
OK, so I read the LM317 PDF a little and here is the formula

Vout = Vref(1+R2/R1)+ Iadj*R2
Vref is 1.25V and Iadj is so small (microAmperes?) that we can ignore it if we pick large enough R1 and R2, so:

Vout=1.25(1+R2/R1)

Where R1 is between ADJ and OUT and R2 is between ADJ and GND.
That's it.
So you'll want R2/R1 to be somewhere between 1.4 and 1.8 as a broad approximation.
For the LOC, that is.

I'm assuming this is correct for all the other pin compatible 1.25 Vref linear ICs.
 
Last edited:




Top