Side-by-side converters interfering with each other

Hello

We have two 7W Hysteretic Buck Converter LED drivers runing next to each other on a small PCB.
(LED load in each case = 24V, 330mA)
(Switching frequency ~ 200KHz for each LED Driver)

-Each LED Driver uses the LM3404 Hysteretic LED Driver IC.
**broken link removed**

-Each of these Bucks has a local 1uF ceramic input capacitor placed near to it.




V(in) = 30V = Secondary Rail of an offline flyback.
The flyback secondary capacitor is a 330uF Electrolytic and is about 4.5cm away from the two Buck LED Drivers.

Anyway, the problem is that the two LED Drivers, despite each having their own input capacitor, they pull some of their current from the neighbouring input capacitor aswell. (and probably pull some current from the 330uF ceramic capacitor aswell).

-Anyway, because i cannot have the individual input capacitors very near to BOTH of the converters, the pulling of some current from the neighbouring 1uF capacitor distorts the current sense signal at the other converter.
(due to the pulse currents flowing through the ground)
-The effect is worse for LED Channel 1 than for LED channel 2.

This results in distorted inductor and LED current

Here is the distorted inductor current in Channel 1:-
http://i46.tinypic.com/b7cumb.jpg

Here is a widened time-base view of the distorted inductor current:-
http://i45.tinypic.com/5ltd35.jpg

Do you know what i can do to stop these side-by-side LED Drivers from interfering with each other in this way.
?
The PCB is only small (3cm by 13cm) and there is not much , if any room for too many more components.

Hi,
What about some serial inductances/resistors pls to separate all PCB modules from the source, on that you has a "starform connection"?
K.

Talking of Star Grounding Schemes.................

Which one of these three Star Grounding Schemes do you believe is the best one?

ONE BRANCH STAR GROUNDING:
https://i50.tinypic.com/fvwa6g.jpg


TWO BRANCH STAR GROUNDING:
https://i50.tinypic.com/34e8d3s.jpg


INPUT CAPACITORS PLACED PHYSICALLY NEXT TO EACH OTHER:
https://i48.tinypic.com/2e36h3b.jpg




...we are currently using the TWO BRANCH STAR grounding scheme, however, U1 appears to also pull some of its supply current from C2, and this causes problems....
..because C2 is relatively distant from U1's current sense resistors, and the switching currents that return to C2 from the U1 converter cause the ground to bounce about....
...this is causing distortion of the LED current.

It is also a problem because both converters operate at nearly the same frequency, and every so often (about 200us or so) they both switch at the same time and this causes a periodic bouncing in the LED current....as i said about every 200us.

When one of the two converters is shut down.........the other one runs absolutely fine with no distortion of LED current whatsoever.

Hi,
My system blocks yet your pictures_sorry...
Only a short idea:Are your converters synchronisable pls?
-If you can have external "clock" do control in opposite phase the one and the other, as a push-pull system...
-Can you apply thicker cabling(or double/threefold wires (minimum) between the star points & the converters (&capacitors on LED-line)?
-I mean star points for both, GND & supply wires too, with (2 wires) sensing on the LED-line/Capacitor direct.
-Do you have 330uF ceramics?-or are their only 330nF pls?
K.

I agree with karesz, the "grounding schemes" are just as bad without some decoupling means.

Karesz:
I am afraid these two LED Drivers are not synchronizable.
The 330uF capacitor is exactly that, 330uF and not 330nF.
The two wire sensing for the current sense resistors sounds good, but then we have to fit in the extra wire, on our already overcrowded PCB.

FvM:
I take it that you mean that some series inductor beads are needed locally to the input supply of each LED Driver?

Incidentally, this PCB is about 3cm by 13cm and also contains the front end offline flyback which provides the secondary 30V rail.
-also, there is a 5V 300mW SMPS on the secondary which supplies 5V to a microcontroller.

The microcontroller will actually sense the average LED current by sensing the current sense resistor voltage via an RC filter.


It is getting difficult because the 2 LED Drivers, the 5V SMPS and the microcontroller all have their own ground attachments which obviously are unfortunately slightly distant from each other.........there is not enough PCB room to "star" each circuits ground back to the 330uF capacitor.

The PCB is only double sided.

I am just wondering which grounding scheme woudl be the best to go with?

Hi,
over the proposed serial inductance/resistor:
You can have nowadays small SMD components, i.e.0603 as a ~6A "ferrite bead", & has 1--300 Ohms Impedance in tenth of MHz range (with ca. 10 mili Ohm! DC-resistance), I think YOU HAVE SO MUCH PLACE on so a BIG PCB!
K.