MOSFET-based LM334 regulator

jamaica

Newly Enlightened
Joined
Mar 23, 2003
Messages
101
Location
HB, SoCal
Welcome goodyear99,
The MIC2171 you've selected looks like a decent boost converter for ordinary uses, but it's slow speed (100KHz) will make for big parts, it's saturation voltage of 0.8v makes it kinda lossy, and you'll have to wind a custom transformer to use it as a downconverter. Other switcher parts would be much simpler to apply.

Your question deserves its own thread, especially since the discussion here is about linear (not switching) regulators.

Best,
Jamaica
 

goodyear99

Newly Enlightened
Joined
Mar 6, 2005
Messages
3
Thanks Jamaica,

I hadnt considered the switching frequency, and had even less idea about the possible efficiency.

Ill look around on these threads a little more... Seems like its hard to find a circuit that will fit my needs, since most drivers are designed for lower current leds.

Thanks again,

Goodyear99
 

wquiles

Flashaholic
Joined
Jan 10, 2005
Messages
8,459
Location
Texas, USA, Earth
Thank you all for the great insights into the LDO designs. I am a EE but incredibly rusty from 10 years of not applying EE in my daily work but reading this post does remind me of things learned a long time ago /ubbthreads/images/graemlins/smile.gif

Will
 

MrAl

Flashlight Enthusiast
Joined
Sep 9, 2001
Messages
3,144
Location
New Jersey
Hi there Will,

Stick around ... we end up talking about this stuff all
the time /ubbthreads/images/graemlins/smile.gif
If you're looking to refresh there's always the EE course
in the sticky thread too. It's basic circuit theory.


Take care,
Al
 

Doug S

Flashlight Enthusiast
Joined
Jun 20, 2002
Messages
2,712
Location
Chickamauga Georgia
Darin, I note in your revised circuit that the time constant of the lower resistor times the capacitor value is 4.7K X 2.2uF = 10msec. See my post #4 in this thread where I noted that I found that 10K X 1.5uF = 15msec was the lower bound for consistant stability. I believe that you will find that the ST part will be fine too if you up the time constant of the above components to 15msec or greater.

evan9162 said:
Update:

I was testing two of these regulators, and noticed that one had totally flat output across a wide input voltage range, while with the other, the current decreased slightly as input voltage was increased.

I noticed that the flatter output version was using a National Semiconductor '334, while the decreasing one used an ST Micro '334.

I re-tested the scenario on my test board. Driving a Lux3 at 920mA, the National '334 kept the output steady to an input voltage of 6.7V (the upper limit of my voltage regulator).

The ST '334 started at 920mA right at regulation, but dropped down to 850mA at 6.7V in. The current dropped steadily with input voltage. I suspect the circuit is oscillating slightly with the ST part in place.

So it looks like the National Semiconductor part is the better one for this circuit, as it keeps output flat with rising input voltage.
 

evan9162

Flashlight Enthusiast
Joined
Apr 18, 2002
Messages
2,639
Location
Boise, ID
Yeah, I was experimenting with the ST part on my prototype board, and putting another capacitor in parallel with the existing one smoothed out the ST part and brought it back into proper regulation. It is nice to know the proper time constant though - thanks for that info.

I actually haven't built one of these in over a year, but I still have loads of parts to do so...I moved into a different house that's required some maintenance the former owners never did (and some that they tried to do but failed at miserably), so that's been occupying most of my time - but I digress..
 

MrAl

Flashlight Enthusiast
Joined
Sep 9, 2001
Messages
3,144
Location
New Jersey
Hi again,

Wow, quite an old thread now :)

It's been a while since i looked at this kind of circuit, but at first glance i can
say that it might be possible to use
a much smaller cap to stabilize the circuit if the cap is connected from the
gate to the drain (gate to anode of the LED) instead of gate to source.
Im guessing 0.01uf would be a good place to start with this value.
The reason for this possibility is because of the "Miller Effect", which comes up
when a cap is connected like this, and basically it amplifies the effect of the
capacitance so it looks much larger to the overall circuit operation. In the case
of 0.01uf, it might look like a 1uf or 10uf or even 100uf connected gate to
source, even though it will be connected gate to drain.

Of course i am assuming that the circuit works in the first place according
to how a linear circuit should operate and it's not operating by some
unlucky hookup where it looks like it's working over a short range of
some parameter but then fails badly for some other reason.

I'll try to take another look at this circuit a bit later, but if anyone wants to
try this simple modification please let us know how it works out.
Of course you must have a way to check for oscillations, either high or
low frequency as that is one important aspect of this kind of design.
 
Last edited:

Doug S

Flashlight Enthusiast
Joined
Jun 20, 2002
Messages
2,712
Location
Chickamauga Georgia
Hi Al, it has been a long time. With your suggestion this thread has come full circle. As originally posted several years back, Darin was doing what you are now suggesting. I dissuaded him from locating the capacitor there with the arguement in post #4 that under some conditions this will subject the LED to a brief high current pulse on powerup. This is due to the initial voltage across the capacitor being zero and it takes time to charge to the value required for regulation. This is only a potential problem where the difference between the supply voltage and that required for the LED string is greater than the Mosfet gate-source turn on voltage.
 
Top