crazy boost converter for LED use is too noisy!

[Joerg]

The pulsing in this case is because the camera does not take images
continously anyway.

The design I started with was the same as in the Maxim application you
suggested - and the problem was partly the low feedback voltage, which
I easily raised to 1V, but also partly (and mostly) the inherent noise-
generating capability of switching a 4-5 amp current at 1 mhz through
a large inductor on a PCB... that will never be silent, it seems.

Consider a toroid. As they say, "toroids don't talk much".

This is interesting. I looked into the current regulating diodes but
most off-the-shelf parts seemt o be in the < 50 mA range while I need
500 mA. Should I "roll my own" with a JFET then perhaps ? And spice
simulate it..

(The jfet is a feedback circuit as well of course, albeit with a
shorter roundtrip maybe than an op-amp and a BJT....)

Very tolerance prone. And expensive if you want to by one at a desired
current. IOW those have become boutique parts.

 

[[email protected]]

Consider a toroid. As they say, "toroids don't talk much".




Very tolerance prone. And expensive if you want to by one at a desired
current. IOW those have become boutique parts.

It's been my experience that the best way to generate a well defined
current pulse is to use a current steering scheme, i.e. long tail
pair. Since you do laser circuits, do you have a better idea?

 

[Joerg]

It's been my experience that the best way to generate a well defined
current pulse is to use a current steering scheme, i.e. long tail
pair. Since you do laser circuits, do you have a better idea?

With lasers it's usually done this way: A current source dumps a
constant current into the laser diode. Then an RF transistor switches
across the laser diode, robbing part or all of that current, depending
on what you have to build.

 

[[email protected]]

With lasers it's usually done this way: A current source dumps a
constant current into the laser diode. Then an RF transistor switches
across the laser diode, robbing part or all of that current, depending
on what you have to build.

I can see that scheme working well too. The idea behind both schemes
is to keep the current source running and gate the current as it is
needed. Trying to create a fixed current or regulate one on the fly
requires feedback, and thus has all the associated problems such as
settling time, overshoot, etc.

However, I don't think such a circuit is what the original poster
wanted.

 

[BW]

I can see that scheme working well too. The idea behind both schemes
is to keep the current source running and gate the current as it is
needed. Trying to create a fixed current or regulate one on the fly
requires feedback, and thus has all the associated problems such as
settling time, overshoot, etc.

However, I don't think such a circuit is what the original poster
wanted.

Hi,

yes, my pulses are of the order 0.3-1ms so a feedback loop (including
a boost converter in current regulation) actually has time to regulate
with minor overshoot. It is easy to design a "conservative" circuit
with a separate op-amp + BJT and sense resistor for each LED string,
providing good regulation, I've done this and measured and it works
well (without the super-capacitor-noise the 80V boost generated .
But it takes up too much PCB space really, perhaps the best solution
is to simply try to optimize the size of the component selections
instead of the circuit topology.. choosing the smallest possible power
ratings etc. Also I can mitigate it by boosting 12->24V for example
and have half the number of strings, so I have plenty of suggestions
and ideas to go forward with now thanks!

Nevertheless it is interesting to read about the laser and other short-
pulse scenarios! If the pulses had been in the microsecond range,
something like that would have been necessary definitely..

Best regards,
Bjorn

 

[[email protected]]

Hi,

yes, my pulses are of the order 0.3-1ms so a feedback loop (including
a boost converter in current regulation) actually has time to regulate
with minor overshoot. It is easy to design a "conservative" circuit
with a separate op-amp + BJT and sense resistor for each LED string,
providing good regulation, I've done this and measured and it works
well (without the super-capacitor-noise the 80V boost generated .
But it takes up too much PCB space really, perhaps the best solution
is to simply try to optimize the size of the component selections
instead of the circuit topology.. choosing the smallest possible power
ratings etc. Also I can mitigate it by boosting 12->24V for example
and have half the number of strings, so I have plenty of suggestions
and ideas to go forward with now thanks!

Nevertheless it is interesting to read about the laser and other short-
pulse scenarios! If the pulses had been in the microsecond range,
something like that would have been necessary definitely..

Best regards,
Bjorn

Why can't you just charge the inductor, then dump. That is, make it a
one-shot. You have at least 100ms to charge the inductor. The current
pulse would not be flat of course. It would fall off the cliff from a
peak. You would probably want to diode clamp the string to make sure
you don't reverse bias the LEDs.

 

[BW]

Why can't you just charge the inductor, then dump. That is, make it a
one-shot. You have at least 100ms to charge the inductor. The current
pulse would not be flat of course. It would fall off the cliff from a
peak. You would probably want to diode clamp the string to make sure
you don't reverse bias the LEDs.

You mean like a buck or boost converter without the output capacitor,
with the load directly on the inductor output and another FET for
pulse-shaping in series with the LED string ? And then skip output
current regulation, but instead make sure to switch when the first
phase of current through the inductor reaches the desired peak ?

So when the inductor current reaches say 0.6 A, you turn off the
switch FET and turn on the output FET for the desired time, then the
inductor will "discharge" a shape with peak at 0.6 and below - the
shape is depending on the inductance of L.

Would this scheme work equally well with a buck or boost topology at
the switch-side ? The important thing here seems to be the measurement
of the current through the inductor to cap the output current.

How would the slow charging affect the power-efficiency btw (actually
my pulses are more like 15 ms apart than 100 ms)? Time for some
LTspice simulation maybe..

/Bjorn

 

[[email protected]]

You mean like a buck or boost converter without the output capacitor,
with the load directly on the inductor output and another FET for
pulse-shaping in series with the LED string ? And then skip output
current regulation, but instead make sure to switch when the first
phase of current through the inductor reaches the desired peak ?

So when the inductor current reaches say 0.6 A, you turn off the
switch FET and turn on the output FET for the desired time, then the
inductor will "discharge" a shape with peak at 0.6 and below - the
shape is depending on the inductance of L.

Would this scheme work equally well with a buck or boost topology at
the switch-side ? The important thing here seems to be the measurement
of the current through the inductor to cap the output current.

How would the slow charging affect the power-efficiency btw (actually
my pulses are more like 15 ms apart than 100 ms)? Time for some
LTspice simulation maybe..

/Bjorn

15ms apart is not a camera flash. The camera lag is closer to 100ms.
It would really help if you wrote a specification.

Use the boost configuration. I don't see the need for two fets. Just
the standard one FET and a fast recovery diode.