How does mosfet linear regulator work

[vicv]

I'm talking about the convoy 5A driver 12 mode/4 mode driver (I believe their ramping driver works in a similar fashion, just different control). I understand how buck drivers work. And 7135 drivers. And even DD fet drivers. But from what I understand, this driver regulates current without pwm by changing the gate voltage. That would mean it's limiting how much the mosfet is closing…..wouldn't this burn it out very quickly? I thought mosfets either liked being 100% on or 100% off. I know it's not cycling it because that would be pwm. But apparently this driver uses analog current regulation.

 

[PhotonWrangler]

I'm talking about the convoy 5A driver 12 mode/4 mode driver (I believe their ramping driver works in a similar fashion, just different control). I understand how buck drivers work. And 7135 drivers. And even DD fet drivers. But from what I understand, this driver regulates current without pwm by changing the gate voltage. That would mean it's limiting how much the mosfet is closing…..wouldn't this burn it out very quickly? I thought mosfets either liked being 100% on or 100% off. I know it's not cycling it because that would be pwm. But apparently this driver uses analog current regulation.

MOSFETs can operate quite nicely in the linear region of their transfer curve as long as they're biased properly. I suspect the design would be similar to the classic 7805 linear regulator. Inexpensive but not nearly as efficient as a switching regulator, and probably generates way more heat.

 

[vicv]

Ya so I guess it does limit current in a similar way to a 7135. By varying its resistance to burn off voltage. This is just able to be controlled. Whereas a 7135 allows 350ma only. And on a 7135 driver, all modes besides high are pwm. Which I don't care about but I guess as the battery gets low and can no longer sustain high, the lower modes aren't regulated either as it's still switching high on/off rapidly. Whereas the mosfet just closes less.

 

[Dave_H]

MOSFET regulator operates similar to a bipolar (NPN or PNP). It is in series with the load, operating as a variable resistor which is continuously adjustable, to maintain set output voltage (in voltage mode). Current-mode e.g. AM7135 etc. just measures output current and feeds back to maintain set current.

Linear can be just as efficient as switching, or at least reasonable, if the input to output voltage difference is small.

It appears the way a lot of cheap multi-level flashlights work, highest current is set by the regulator but from there down reduced by PWM. What I don't get is how some of these cheap 3AAA zoomies get away with regulator of such small size (SOT-23) when on high mode. I measured 1.2A using a 4v Li-ion cell. If the LED vf is around 3v, that's over 1W of heat in the little chip.

Dave

 

[PhotonWrangler]

I've seen 78L05s in operation that were running too hot to touch. I think those top out around 300ma.

 

[vicv]

So another thing I came up with which I don't understand. We have these linear drivers. I've seen them in 5a, 6a, 8a, 9a, 12a. Emisar/noctigon uses them with fets added for turbo. But why add a fet when it has one already? Or why limit these other drivers to a set current. And the higher current ones are generally on bigger boards. But there's 15mm fet drivers.

If the fet is being used in a linear way, I.e different levels of closed, why can't the same fet be closed 100% for a turbo?

 

[aznsx]

I'm stayin' out of this one, except to offer an important word of caution to keep in mind. Do not make the mistake of assuming that because you see a particular design implemented in certain lights being sold that it is a proper or wise design, or complies with accepted design 'best practice'. I know for certain there are shoddy designs out there in all various respects (engineering / circuit design / mechanical design / manufacturing quality, etc.), so don't assume something is necessarily of good design just because it is being sold. What you might be looking at may in some cases be an example of the above. This may be obvious to you, but wanted to remind you of these facts. Good, quality engineering / design / manufacturing costs money, and not all manufacturers make that investment, because it doesn't help their profit margins. No, I am not 'naming names'. I'm not very smart, but I'm smarter than that;-)

As you were. AZ out.

 

[vicv]

I have no idea what you're saying. But thank you. There's nothing controversial here, just talking about driver design

 

[aznsx]

I'm not very smart

See, I told ya. I don't write very well, either

 

[alpg88]

they work very quietly in mysterious ways

 

[PhotonWrangler]

I'm stayin' out of this one, except to offer an important word of caution to keep in mind. Do not make the mistake of assuming that because you see a particular design implemented in certain lights being sold that it is a proper or wise design, or complies with accepted design 'best practice'. I know for certain there are shoddy designs out there in all various respects (engineering / circuit design / mechanical design / manufacturing quality, etc.), so don't assume something is necessarily of good design just because it is being sold. What you might be looking at may in some cases be an example of the above. This may be obvious to you, but wanted to remind you of these facts. Good, quality engineering / design / manufacturing costs money, and not all manufacturers make that investment, because it doesn't help their profit margins. No, I am not 'naming names'. I'm not very smart, but I'm smarter than that;-)

As you were. AZ out.

I once saw a piece of expensive broadcast equipment that had a 5mm LED power indicator soldered across the 5v supply with no current limiting resistor. It was effectively a NED (noise emitting diode). It only worked once. Briefly.

 

[vicv]

A lot of flashlights use a mosfet that directly puts the emitter is series with the cell. Like a lot of them. This isn't a new idea

 

[Dave_H]

I'm talking about the convoy 5A driver 12 mode/4 mode driver (I believe their ramping driver works in a similar fashion, just different control). I understand how buck drivers work. And 7135 drivers. And even DD fet drivers. But from what I understand, this driver regulates current without pwm by changing the gate voltage. That would mean it's limiting how much the mosfet is closing…..wouldn't this burn it out very quickly? I thought mosfets either liked being 100% on or 100% off. I know it's not cycling it because that would be pwm. But apparently this driver uses analog current regulation.

A switching-type CC driver will adjust its output voltage to the point where the set amount of current is flowing through the load i.e. LED(s). It can do this at high efficiency. The MOSFET(s) are being switched on/off completely, but it's a bit more complex than this.

Dave

 

[vicv]

I'm not talking about a switching regulator. I'm talking about a linear regulator.

 

[Dave_H]

I'm not talking about a switching regulator. I'm talking about a linear regulator.

In the linear case yes, on-resistance varied by gate voltage, power loss and temperature of the MOSFET has to be managed. Therein lies the difference between good designs and poor.marginal ones.

Self-destruct temperature for silicon is +175C but for good operation and long life and other reasons should be kept well below this.

Dave

 

[Dave_H]

So another thing I came up with which I don't understand. We have these linear drivers. I've seen them in 5a, 6a, 8a, 9a, 12a. Emisar/noctigon uses them with fets added for turbo. But why add a fet when it has one already? Or why limit these other drivers to a set current. And the higher current ones are generally on bigger boards. But there's 15mm fet drivers.

If the fet is being used in a linear way, I.e different levels of closed, why can't the same fet be closed 100% for a turbo?

MOSFET conduction loss i.e. when fully switched on is proportional to current squared times "on-resistance". Paralleling MOSFETS lowers the overall resistance, and therefore losses, especially at higher currents and/or for less-good MOSFETs having higher on-resistance. It spreads out the heat for better thermal management.

On-resistance increases with temperature, in turn increasing losses.

Dave

 

[vicv]

In the linear case yes, on-resistance varied by gate voltage, power loss and temperature of the MOSFET has to be managed. Therein lies the difference between good designs and poor.marginal ones.

Self-destruct temperature for silicon is +175C but for good operation and long life and other reasons should be kept well below this.

Dave

I would not call a linear regular to be a poor design. It's a little less efficient than a switching regulator, but a buck driver is not a very good option for a 3 V LED and a single lithium ion cell. Look how short of a time M61s are under regulation. Buck drivers are great with 2 cells. And two primary cells. But not a single. Too much voltage drop in the electronics

 

[Dave_H]

I did not say that linear designs are necessarily poor, some may be. Linear control with single Li-ion cell driving 3v LED can be reasonably efficient, and I agree it is simpler.

Dave

 

[vicv]

I did not say that linear designs are necessarily poor, some may be. Linear control with single Li-ion cell driving 3v LED can be reasonably efficient, and I agree it is simpler.

Dave

I was just responding to your "poor/marginal designs". It's all good. Each driver style has pros and cons. Actually with the newer LEDs having such low vf, buck drivers may be a good option. Only issue is the size to fit the coils

 

[aznsx]

I once saw a piece of expensive broadcast equipment that had a 5mm LED power indicator soldered across the 5v supply with no current limiting resistor. It was effectively a NED (noise emitting diode). It only worked once. Briefly.

You must have been tuned to my frequency. You must be one of those danged old sparky guys.
BTW, I share your love of high catwalks in dark places :-(
I also still dig your thumbnail, so I hope you don't change it. It makes me smile.