Next message: markhoenke: "[world-cruising] Re: [World Cruising] LED cabin lights reply... bit windy (WARNING: may be too much info for some)"
--- In , Ken James <nautikat at dot dot dot > wrote:
>
> Ken:
>
> Have you noticed and low level frequency interferences with
> electronics from the led drivers, especially the dimming drivers?
>
>
> Reply;
>
> This might be of some interest to those who want to design their
own LED
> lights using any type of 'switch mode' design, but most will prob.
want
> to hit [DELET] <G>
>
>
>
>
> Although a Pulse Width Modulation (PWM) type LED light 'driver' or
power
> supply is usually not as prone to RFI as a 'dc-dc switch mode' type
can
> be, to my chagrin we discovered that our very first LED lights
based on
> a PWM type of design over a decade ago, the MKI design, DID exhibit
some
> RFI at times.(GROAN!) This was a problem of course, and I had to
> replace those lights with an improved design when the problem
showed up
> (mainly when the light was mounted in the 'near field' of a
sensitive
> antenna, but sometimes it could be even a few feet away from any
antenna
> and still cause problems).
>
> The follow on MKII design solved the RFI problem by incorporating
> blocking filters in the power circuit, but that cost some
efficiency and
> made extra heat, so was what I considered a less than optimum
solution.
> Not only that, but one part of the filter had an annoying habit of
> breaking loose from its solder joints once in a while causing the
light
> to go dim or go out.
>
> The next design, the MKIII, addressed the challenges in a number of
ways.
> The previous power filter was dropped from use and the control IC
was
> replaced with a different control IC with much better wave shaping
> properties and a lower voltage output signal (less output voltage
means
> less possible energy for RFI), and suppression of any unwanted
> frequencies at the LEDs power switching transistors 'gate' was
attained
> (the power switching transistor being a MOSFET type it has a
Source,
> Gate, and Drain un-like 'old fashioned' bi-polar transistors with
their
> Emitter, Base, and Collector, and this MOSFET also has an
equivalent
> "on" resistance of .032 ohms at a max of 1.5 amps!), producing a
power
> output to the LEDs thoroughly free of any freqs. that might cause
RFI
> (the LED power switching transistor is what turns the power to the
LEDs
> off and on in a Pulse Width Modulation (PWM) circuit, or in my
Variable
> Pulse Width (VPW) type of PWM design).
>
> This suppression of any unwanted signals or freqs. at the switching
> transistors 'gate' was achieved through a number of simple but
efficient
> methods, one of which was an optical isolator/filter that was added
in
> the switching transistors 'gate' circuit that COULD NOT pass any
radio
> freq. signals from previous elements or from outside the light
> (externally generated interference) to the switching
transistors 'gate',
> yet this filter DID pass with almost no measurable degradation the
> switching pulse signal, and so a 'clean' square wave pulse signal
was
> imposed onto the power output of the switching transistor
(the 'gate' is
> what turns the switching transistor off and on).
>
> Dampening or 'attenuation' below a certain threshold or 'cut-off'
of any
> high freqs. that were imposed on the 'gate' signal after this
filter and
> before the 'gate' itself to levels that were 'clamped' at
the 'gate'
> below those that could possibly activate the transistor was used to
> further assure no outside influence could effect the switching
> transistors power output to the LEDs, although the copper PCB trace
from
> the filters output to the switching transistors 'gate' was kept
very
> short in any case.
>
> Also a different pulse switching frequency from that of previous
models
> was created by the control IC, a freq. was selected that by itself
could
> not possibly generate any RFI in the circuit (the pulse switching
> freq.used in the MKIII was way to low a freq. to make RFI by itself
in
> any case) or make any harmonics or 'feedback' that might do so, and
> better copper PCB trace designs to the LEDs were used so there was
very
> little chance any RFI could be induced from outside were used, and
a
> different day-sensor shut off control method was used to isolate
and
> buffer the day-sensor input to the circuit in order to totally
eliminate
> the possibility of any transient voltage spikes becoming amplified
and
> 'fed back' into the control IC.
>
> Also all other inputs to the control IC were now 'buffered'
starting
> with this design, making the circuit more robust and reliable.
>
> The present MKIV design follows this even further by refining the
> switching frequency even more, and it uses a better and more
efficient
> (every milli-amp counts!) power supply to the control IC that has
a
> superior and very stable output voltage and current. With this
power to
> the control IC further filtered to almost zero 'ripple' under any
and
> all conditions, and because the power supply now reliably switches
off
> only AFTER the control IC goes into low voltage 'hibernation' in
the
> event of a low voltage situation or 'brownout' and switches on
BEFORE
> the control IC tries to come out of 'hibernation' as the voltage
rises,
> the circuit achieves increased stability overall making the
possibility
> of issues such as low voltage 'latch up ' become insignificant.
('Latch
> up' is where the control IC sometimes 'freezes up' at very low
voltages
> and does not re-activate properly when the voltage rises later).
>
> And with a better power supply the control IC has even better wave
> shaping output performance, and cleaner more accurate output
waveforms
> mean better efficiency (OK it is only a 1% difference or so,
but...). A
> better PCB 'ground plane' was added to further assure stability and
> produce a more efficient thermal pathway as well.
> The MKIV design also incorporates an active thermal sensor that
supplies
> feedback to the control IC to protect the LED array (temp
management is
> critical with these super LEDs!) from abnormally high environmental
> conditions by reducing power to them an appropriate amount when it
> senses the LED array is too hot and the LEDs should be de-rated
('turned
> down') in accordance with the LED mfg's instructions and data to
insure
> the LEDs won't overheat and grow prematurely dim.
>
> The dimming control makes no difference in RFI or EMI generation
with
> this design since all it does is lessen the max pulse width
(within
> certain parameters) that would be applied to the selected LEDs
under the
> given conditions if they were not operated in a 'dimming' mode.
>
>
>
>
>
> The result of this design is NO radio frequency interference or
> electro-magnetic interference at all, unlike poorly designed PWM
or
> dc-dc switch mode designs (and yes there are some out there, just
as
> there are still some cheap power inverters that make a lot of RFI
> 'noise' ).
>
>
>
>
>
> For anyone that wants it, I would be willing to send an audio file
I
> have of a MKIV light being held right next to the antenna of a hand
held
> ham radio set at #1 squelch level (using zero squelch results in a
lot
> of static no matter what across most bands). The radio is cycled
through
> its many bands, and the light is repeatedly turned on and off as it
is,
> with the result that no difference can be heard from when the light
is
> on to when it is off, a finding backed up by measurements I have
taken
> that detected no RFI.
>
> Can't hear it, can't measure it, because it's just not there!
>
>
>
> So in sum, to prevent RFI or EMI with this type of design, select
your
> power switch freq. carefully, use very 'clean' control IC power,
get a
> good sharp pulsed square wave signal output from the control IC,
buffer
> and/or de-couple all inputs, filter, suppress, and keep the signal
> traces short and the power paths robust.-Ken
>
Mark on SV Perseverance
we intend to replace the cabin and nav lighting with LED. my web
searches so far have led me to http://www.bebi-
electronics.com/serving.html They have voltage regulation built into
the bulb aray and offer lights for 12 and 24 VDC systems that are
tollearant of the variation in voltage between charging and running.
the AQUA SIGNAL lights on my boat are identical to the examples they
show. Our system is all 24 VDC.
