Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[bkenobi]

All led strip lights I have seen are DC with a wall wart or some other type of transformer.  Therefor, if the one you have is te same, it will require a neutral line module.

[Backward Engineering]

My LED strip is powered by full wave rectified 115 vac.  The linearity of the dimming control is very different than that of an incandescent bulb.  Even with the neutral wire added to the module, a very low current will still pass through the load due to the zero crossing detection circuit of the module (350ua for the WS12A).  This may produce some low level lighting with the LED strip.  I believe this low level illumination when the dimmer is "OFF" could be fixed by installing a 0.1 uf mylar, 200volt capacitor connected between the load and neutral wires. This should result in less than 10 vac applied to the LED strip when the dimmer is "OFF".  The LED strip begins to increase illumination at around 50%.

The CREE 8.5 watt LED light bulb somehow avoids all of these problems and even works with all of my two wire X10 modules.  I would have expected them to be unstable when dimmed due to the lack of sufficient holding current in the module's triac.  Some day I want to look at the current and voltage on this bulb with my oscilloscope.  

I plan to install the WS12A with the neutral wire and the 0.1 uf capacitor, as described, across LED strip today.  I will post my results.

WS12A Schematic Attached

Don

[Brian H]

I have seen other good result posts for Cree LED Bulbs. With X10 and Insteon.
I have also seen a tear down the earlier Crees {not the TW series} with some data on the IC used for the dimming control.
I tried some on a test ZWave two wire dimmer switch and it was a 100% failure.

[Backward Engineering]

An update on my project to use a WS12A X10 wall module to control a 5 meter 115 V.A.C. LED strip.
Modified the WS12A adding the neutral wire and a .22uf cap across the load (I didn't have a .1 uf on hand).
I could control the LED strip from the push button on the WS12A with no problems.  Although I could turn the LEDs on via X10 command, once they were on, all other attempts to control the module via X10 commands failed.
The .22uf cap did allow the LEDs to fully extinguish when the module was switched off.
The LED strip is used in the living room as Home Theater accent lighting so the ability to dim the lighting is important.
I added a 20 watt ZENON lamp that is used to illuminate wall art to the load on the module.
This was enough load to solve the X10 command control issue.  This means that the .22uf cap was no longer necessary but I didn't feel like opening up the module again to remove it.

Don

[JeffVolp]

Modified the WS12A adding the neutral wire and a .22uf cap across the load (I didn't have a .1 uf on hand).
I could control the LED strip from the push button on the WS12A with no problems.  Although I could turn the LEDs on via X10 command, once they were on, all other attempts to control the module via X10 commands failed.

The .22uF capacitor represents just a 6 ohm load at 120KHz, which will severely load down X10 signals.  That is why the switch does not respond once it has been turned on.

Jeff

[Backward Engineering]

Hi Jeff,
Please look again at the schematic I posted.  The .22uf cap does not shunt the 120Khz signal.  The cap. is in parallel with the load (lamp) which is in series with the module and therefor does NOT kill the 120Khz signal.  On a 2 wire module (no neutral) the 120Khz signal arrives on the hot (black) wire and is coupled to the receiver via the .68uf capacitor and returns to neutral via the load (lamp) In a module modified with a neutral wire modification, the signal path returns via the white wire instead of through the load.  On an unmodified module with loads that present a high impedance, both the power supply current and the 120Khz signal current are limited and may prevent X10 commands from functioning properly.  The purpose of the .22uf cap on this module that I modified by adding a neutral wire was to shunt the current that passes through the LED strip light via the 330K resistor in the X10 module to detect the zero crossing.  The .22uf cap. prevents the LEDs from glowing when the module is "OFF".

My only concern in adding the .22uf cap. was the effect it might have on the triac's di/dt when triggered.

Also the .22uf cap. never loads the LINE to Neutral at zero crossing because the triac is never on during zero crossing.  The triac is gated several degrees after the zero crossing of the line voltage.

By the way, the module with the .22uf cap. and all other modules in the room work fine after adding the 20 watt load in parallel with the LED light strip.

Don

[JeffVolp]

I understand where the cap is.  Whether the cap loads down the X10 signal depends on whether the triac is conducting at that point.  One thing I did ignore was the series inductor.  47uH makes a series resonant circuit with a notch at about 50KHz.  Loading at 120KHz would be dependent on the Q of the network.  If you were using a relay wall switch, the capacitor would certainly attenuate the X10 signal level when on.

Out of curiosity I set up the circuit myself (without the LEDs) using both a wall switch and lamp module for comparison.  I connected a .22uF capacitor directly to the output, and monitored the line with both a scope and XTBM.  The problem was not decreased signal level, but a large spike when the triac turned on that set up ringing in the LC network.  It saturated the XTBM noise reading at .99Vpp when near full brightness on the dimmer, but dropped to the normal level of .02Vpp as the "light" was dimmed and the transient moved out of the X10 transmit window.  The first cycle of the transient was over 5Vpp on my scope.

Something else to consider is whether your light strip includes any "electronics".  One that I have just includes dropping resistors, but if yours has a switching regulator it could also be generating noise when on.

Jeff

[Backward Engineering]

I briefly considered the resonance of the 47uh inductor and the .22uf capacitor.  The impedance of the capacitor at 60 Hz was the factor that prevents the LEDs from glowing when the X10 module is "OFF".  Maybe I should have placed a resistor in series with the .22uf cap. to spoil the Q of the circuit and reduce the di/dt when the triac turns on.  I may modify a second WS12A module for additional bench testing.  Development testing in the living room is frowned on.

The LED strip is made up of 1 meter sections powered from a small adapter inline with the power cord.  I don't know if the adapter has switching electronics inside.  I suspect it is nothing more than a bridge rectifier.  I need to open it up to find out.

The LED strip does flicker a bit when dimmed to the lower light levels.  I have considered designing my own dimmer that would respond to the X10 controls and control the LEDs by way of a PWM MOSFET switching regulator running at around 10-20 Khz.  One way would be to use an X10 module chip and integrate the phase delay output and use this variable to control the PWM.  This could eliminate any 120 Hz component and/or flickering from the light output.

LED strip lighting seams to be the way forward in efficient lighting.  I was at a new Kaiser Permanente facility today and the building looks to be 100% LED lighting, mostly strip lights.

[Backward Engineering]

I opened the adapter on the LED strip light.  As I expected it is nothing more than a bridge rectifier.

I removed the .22uf cap. from the WS12A that I had previously connected across the load.  The X10 module control did not change.
Without the 20 watt zenon lamp, the module would respond to the X10 ON command but would not respond to an OFF or DIM command.
With the 20 watt connected in parallel with the LED strip light, all X10 commands work fine.

I'm thinking that the low current through the triac is resulting in zero crossing timing errors in the chip when the triac turns off early at the end of each half cycle due to a lack of holding current in the triac.  I will see what happens if I move the 330K resistor, connected to the triac/inductor to neutral thereby providing a zero crossing signal regardless of the state of the triac.

The mystery is, what does CREE do to avoid the same problem with their 115 volt 8.5 watt LED lamps?
Maybe I need to open up one of the CREE lamps to see what chip they use.
Any tear-down information on the CREE product would be appreciated.

Don

[Brian H]

http://www.designingwithleds.com/cree-60w-led-replacement-bulb-review-and-tear-down/
http://www.designingwithleds.com/qa-with-cree-about-60w-replacement-led-bulb/
https://www.youtube.com/watch?v=n6DDFRBrSas&list=UUqp2_p4YjtaTKiHuNZv0mAQ&index=49

Yes the baking the bulb mentioned in the first link does work.  

I have not seen a tear down of the newer 8.5 Watt and 13.5 Watt TW Cree LED Bulbs.

The 5000K 60 watt ones are similar but they get the 800 Lumens from one row of LEDs against two rows in the 2700K ones.
The original 40 watt 2700K also used one row of LEDs.

[Backward Engineering]

PROBLEM SOLVED:  DIMMING LED LIGHT STRIP WITH A WS12A

Adding a 120K resistor in parallel with the LED on my WS12A module that I modified with the neutral wire solved the problem.

It looks like my theory on zero crossing detection errors was correct.
I first tried connecting the 330K resistor to neutral.  That did not work at all.  The module would not turn on from either the push button or via X10 commands.  I then reconnected the 330K resistor to it's original configuration and added a 120K 1/2 watt resistor from neutral to the blue wire. This is the same location where I had previously connected the .22uf cap.  This, my latest configuration, does not have the .22uf cap.

The problem occurs when the load is non-linear, meaning that the load does not draw any current at lower voltages found near zero crossing.
The triac stops conducting due to the current being below the holding current.  This leaves the zero crossing sense resistor connected to a floating connection and without the zero crossing timing applied to the chip, the control chip fails to detect the X10 commands.

With the 120K resistor connected, no additional loading is required.  The LED strip light is now controllable via X10 commands.
The maximum power dissipation of the 120K resistor is 1/8 watt, however a 1/2 resistor is recommended due to the voltage rating of resistors.
I believe this explains why the CREE LED lamps work.  However, I have found that the CREE will even work with the 2 wire modules.

I will update the WS12A schematic that I posted with this latest configuration and re-post it.

Don

[Brian H]

Glad you found a solution to the problem.
I have had good luck with Cree bulbs and X10 modules also.

[Backward Engineering]

Here is the revised schematic of my modified WS12A module.
Note that the modification may not work on shorter length light strips (less than 5 meters).
The concern is with the triac's holding current.

Don

[Brian H]

Where the Cree Teardown links I found, any help?

I know they give an IC part number for the driver and Power FET used.

From what I can tell. The TW versions look basically the same as the originals construction wise. Probably different LEDs, components to get the proper drive current and the rare earth element in the glass shell.

If you decide to take one apart. The bake in the oven thing should work.
I would also like to know what you found if the disassembly is done by you.

[Backward Engineering]

Thanks for the teardown links.  They have a lot of good information on the CREE lamps and the technology in general.
The lamps are designed to provide a constant level of light over a wide range of voltages yet respond to the dimmers.

On the other hand, my LED light strip  does not have any electronics other than a bridge rectifier, some series resistors, and the LEDs.
The lack of any switch mode electronics means that it will not radiate RF interference.  Of coarse the power factor of the CREE lamps
is probably better than that of the LED strip light which draws all of the load current during the peak period on the sine wave.
As soon as any light is dimmed by phase delay, as in the X10 dimmers, the power factor will get worse.

I have a Kill-A-Watt meter that I could use to confirm this, however my main interest is in saving on the light bill while maintaining the
ability to use my X10 system.

Don