Maker Pro
Maker Pro

Op-amp/feedback selection: Constant current LED driver using switching regulator IC

J

Joel Kolstad

Jan 1, 1970
0
I'm attempting to design a small, constant current LED driver using a Linear
Technologies LTC1174HV switching power supplied controller IC. I'm using a
Luxeon LED that can take up to 375mA of drive current and attempting to use
this as a reading light at a cabin that has a 12V solar power system --
hence the desire for high efficiency with the switching regulator. I have a
question about which op-amp I should be able to use and if I might need
additional feedback compensation.

The schematic is here: http://oregonstate.edu/~kolstadj/CCLED.gif

The op-amp is used to allow the variable current selection as well as to
reduce the power dissapated by the shunt resistors that measure the current
through the LED (the LTC1174 uses a reference voltage of 1.25V). I first
attempted to use an LT1490 op-amp because it boasts rail to rail output and
very low power consumption. Unfortunately, it didn't work so well, with the
switching output of the regulator pretty wildly jumping around -- the
current through the LED tended to jump around wildly (I initially used a 10
ohm power resistor in place of the LED for testing -- the LED is rather
spendy!). I therefore decided to just build a constant voltage power supply
using the LT1490 as a unity gain follower coming from a resistive devider --
this didn't work either! I changed the LT1490 for an LM358 (that I had
lying around) and finally did achieve a nice, constant square wave output at
the LTC1174's 'switch' pin. At this point I began to wonder what the
problem with the LT1490 might be... the only thing I can think of is that it
has a UGBW of 200kHz. I didn't think this would be a problem, since I'm not
interested in the transient response of this circuit as the load will always
be the same... but I'm now thinking that an op-amp that slow perhaps allows
the LTC1174 to initially overshoot its target... followed by
undershooting... etc., and therefore the entire system is unstable. (The
LTC1174 is a 'constant off-time' switcher with a 4us off-time. It
automatically switches to a voltage-driven hysteretic mode -- 'burst mode'
well under light loads.)

Anyway, thinking I had the problem identified, I went back to the constant
current configuration as shown in the schematic. With the LM358, it "pretty
much" works in that the switch pin runs for awhile... then stops... and then
runs again. The current through the LED is relatively constant, so I think
what's happening is that the system still overshoots... the LTC1174 waits
for the voltage to drop to something 'reasonable' again... and then the
cycle repeats. Under heavier loads, there's a combination of the 'several
switching cycles-brief pause' as well as longer pauses inbetween these.

My feeling is that the LTC1174 is getting confused between whether it should
be in 'burst mode' or its regular mode and this makes the entire circuit
somewhat less than optimally. However -- with the LM356 -- it does seem to
be workable.

I've stuck some waveforms up as well:

http://oregonstate.edu/~kolstadj/TEK00000.BMP -- This is under a constant
current of ~75mA. Notice the switch pin is on twce in rapid succession (no
4us gap here!), turns off for awhile, the current through the catch diode
stops about 7us later, and then everything 'sleeps' for about 35us before
the process repeats.
http://oregonstate.edu/~kolstadj/TEK00001.BMP-- This is a constant current
of ~335mA. Similar to the above, but there are some really short spiky
turn-ons of the switch as well. Strange...

Anyway, I'm wondering... what would a good op-amp be to use? Is it the lack
of speed on the LT1490's part that caused the initial troubles? I'm very
much wanting a single supply op-amp. Its output doesn't need to go all the
way down to 0V, as it'll normally be sitting right around 1.25V (the
LTC1174's reference voltage).

Is there any easy way to introduce some filtering somewhere to get the
switch output to be a 'nice' square wave? Should I even bother trying?
Putting a capacitor across the op-amp's feedback resistor to drop the high
frequency response doesn't really improve things much, nor does placing one
from the op-amp's output to ground.

Thanks in advance for any assistance!

---Joel Kolstad
 
G

GPG

Jan 1, 1970
0
Try shifting filter to output of amp









|\|
-|-\ ___
| >--|___|| Pin1
-|+/ ---
|/| ---
|
created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de
 
G

GPG

Jan 1, 1970
0
Try shifting filter to output of amp









|\|
-|-\ ___
| >--|___|| Pin1
-|+/ ---
|/| ---
|
created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

The led does not need cap filter, from look at sheet pin1 does (close to chip)
Let me know
 
Top