LED Tape - Varying Brightness From Different Modules.

[Dan F]

My LED tape isn't dimming properly and below 45% becomes patchy.  It's not an installation or voltage drop issue and so I'm suspecting it's something to do with some kind of incompatibility between the waveform produced by the EldoLED LinearDrivee constant-voltage driver and the on-tape "current stabilization".  But I can't work out what the on-tape IC is to find a datasheet.  Any ideas?

 

This is the tape: https://www.holectron.com/products/flexible-led-strips/tunable-white-flexible-led-strips-nichia/tw-70-d4/

This is the driver: LINEARdrive 200D-D2Z2C

 

[jack]

Can't help you with the on-tape stuff, but I don't suppose the PWM rate of the driver is adjustable via the Eldoled programming interface?

[Dan F]

21 minutes ago, jack said:

Can't help you with the on-tape stuff, but I don't suppose the PWM rate of the driver is adjustable via the Eldoled programming interface?

No, it's not adjustable.  The weird thing is with a Meanwell PWM driver dimming is perfectly consistent (if I hook up just one colour LED), the issue is only with the EldoLED driver!   The reason I'm using these EldoLED drivers is because: i) the tape is tunable white and there are a limited number of drivers available with a DALI DT8 interface, ii) EldoLED are supposed to be good.   I think both Meanwell driver is 2.5kHz PWM, I think the EldoLED is similar but they don't have it in their datasheet.

[Temp]

Possibly..

https://www.farnell.com/datasheets/3204748.pdf

 

 

[Temp]

I had a quick look at the data sheet for the driver LINEARdrive 200D-D2Z2C and the default appears to be a log curve...

 

 

 

So below 45% the output is going to be just a few percent.  Does the Meanwell go that low?

 

 

 

[Radian]

LEDs with "Stabilized current design" will attempt to maintain constant power in the LEDs over a range of different voltages (within limits)

"LINEARdrive driver output voltage 12 - 28V" might suggest a range of different output voltages - so dimming will only occur outside the LED tape's limits, and then the dimming will be patchy as the current is no longer equalized in LED strings.

Maybe.

Unusual driver topology if it really is linear and not PWM (although it could be filtered PWM)

[Radian]

7 hours ago, Dan F said:

But I can't work out what the on-tape IC is to find a datasheet.  Any ideas?

No active components that I can see surrounding the IC so probably an integrated back-to-back transistor pair configured as a constant current source.

[Dan F]

3 hours ago, Temp said:

Possibly..

https://www.farnell.com/datasheets/3204748.pdf

 

Yeah, good find.  Although given the "IN" marking (Nov 21) and it's location, I think it may be the Diodes Inc version: https://www.diodes.com/assets/Datasheets/BCR430UW6.pdf.

 

[Dan F]

3 hours ago, Temp said:

I had a quick look at the data sheet for the driver LINEARdrive 200D-D2Z2C and the default appears to be a log curve...

 

 

 

So below 45% the output is going to be just a few percent.  Does the Meanwell go that low?

 

Yes, it uses Log curve.

DALI defines 45% as about 2.2%.  The Meanwell driver goes down to 0.2% and the EldoLED down to 0.1%.  Of course, while 2% seems low, from a perceived light perspective it's still quite bright and not ideal for this weird dimming behaviour as observed to happen.

[Dan F]

1 hour ago, Radian said:

LEDs with "Stabilized current design" will attempt to maintain constant power in the LEDs over a range of different voltages (within limits)

"LINEARdrive driver output voltage 12 - 28V" might suggest a range of different output voltages - so dimming will only occur outside the LED tape's limits, and then the dimming will be patchy as the current is no longer equalized in LED strings.

Maybe.

Unusual driver topology if it really is linear and not PWM (although it could be filtered PWM)

 

The LINEARdrive is used with a 24v Meanwell power supply, so the input is fixed at 24v

The "Stabilized current design" means that each LED module (10cm with 7 LEDs) has its own driver IC which ensures constant current.

The dimming is via PWM from LinearDrive.  The on-tape IC output, in theory, also switches on/off based on PWM input.

 

Looks like it's explained in the IC "application notes":  https://www.diodes.com/assets/App-Note-Files/BCR430UW6-App-Note.pdf

 

There seem to be some caveats in there about the "minimum duty cycle" and "rise time".  Wonder if this might explain why dimming is perfect with a Meanwell driver but results in patchiness with the Eldoled.

 

[Temp]

5 hours ago, Dan F said:

There seem to be some caveats in there about the "minimum duty cycle" and "rise time".  Wonder if this might explain why dimming is perfect with a Meanwell driver but results in patchiness with the Eldoled.

 

Would be interesting to see the waveforms of both drivers. 

 

The rise and fall time will effect how much RF interference they generate and might be deliberately limited?

[Temp]

5 hours ago, Dan F said:

 

Looks like it's explained in the IC "application notes":  https://www.diodes.com/assets/App-Note-Files/BCR430UW6-App-Note.pdf

 

Quite interesting read. Looks like they have to use variable frequency to meet the IEEE safe operating area, something I hadn't thought of.

 

There is also a note on page 8 that suggest too fast a rise time might cause the regulator FET to turn on before the stabilisation circuit is working. This would possibly cause some LED sections to be too bright at low PWM. If that's what's happening then adding capacitance to the output of the PWM driver to reduce the rise time might work. But what value would be required would be hard to determine an might have unwanted side effects like heating the driver.

[Dan F]

3 hours ago, Temp said:

 

Quite interesting read. Looks like they have to use variable frequency to meet the IEEE safe operating area, something I hadn't thought of.

 

There is also a note on page 8 that suggest too fast a rise time might cause the regulator FET to turn on before the stabilisation circuit is working. This would possibly cause some LED sections to be too bright at low PWM. If that's what's happening then adding capacitance to the output of the PWM driver to reduce the rise time might work. But what value would be required would be hard to determine an might have unwanted side effects like heating the driver.

 

Yeah, I don't know if what I'm seeing is some brighter modules, some duller modules or a mixture.   I assume the rise time is the same regardless of the duty cycle,  but this side-effect only seems to be happening when the high is < 10uS   (assuming 25 kHz PWM)

 

Here's where it gets slightly more interesting.   I happen to have a 5m length of the previous generation of this tape which uses the 420 (https://www.diodes.com/assets/Datasheets/BCR420UFD-BCR421UFD2.pdf) instead of the 430, and this appears to work fine.  If you look at the 420 datasheet it talks about dimming down to 0.5% with 25kHz PWM which implies it deals with highs down to 0.2uS.  On the other hand, the 430 datasheet seems to recommend reducing the frequency to achieve dimming. 

 

Meanwell + 420  OK

Meanwell + 430  OK

EldoLED   + 420  OK

EldoLED   + 430  BAD

 

 

[Dan F]

3 hours ago, Temp said:

Would be interesting to see the waveforms of both drivers. 

Yes, definitely.  I've borrowed a cheap scope, so will be interesting to see what they both produce.

 

The reality is that the LED tape supplier should supply compatible tape/drivers (which doesn't seem to be the case), and given they supposedly have 5yr warranty, I'd like to think they can resolve this, especially given I must have a good 50-100m of this tape in total.  That said, the diagnostics is very interesting, and if I know what the issue is it makes it easier to push for a solution and not get fobbed off 🙂

[Temp]

10 hours ago, Dan F said:

I assume the rise time is the same regardless of the duty cycle,  but this side-effect only seems to be happening when the high is < 10uS   (assuming 25 kHz PWM)

 

If I've understood the note correctly... if the rise time is too fast the first few uS (?) of each pulse will be full current (eg not regulated). At max brightness those few uS would be an insignificant percentage of the ON time, but when dimmed they will be a much greater percentage of the ON time and therefore much more noticeable. 

 

Try increasing the brightness very slowly in small steps. If its as I think I would expect them to gradually become the same brightness. If there is an obvious/sudden point where they switch from different to the same then its probably something else.

[Radian]

1 hour ago, Temp said:

If I've understood the note correctly... if the rise time is too fast the first few uS (?) of each pulse will be full current (eg not regulated). At max brightness those few uS would be an insignificant percentage of the ON time, but when dimmed they will be a much greater percentage of the ON time and therefore much more noticeable.

Yes, the datasheet specifically states that the drain-gate capacitance of the switching FET will cause it to turn fully on before the control circuitry has stabilized. A series inductor could be used In order to slow the rising edge to greater than 5us and not load the PWM driver (rather than using a parallel capacitor).

[Dan F]

1 hour ago, Temp said:

 

If I've understood the note correctly... if the rise time is too fast the first few uS (?) of each pulse will be full current (eg not regulated). At max brightness those few uS would be an insignificant percentage of the ON time, but when dimmed they will be a much greater percentage of the ON time and therefore much more noticeable. 

 

It may be this, but how would you explain it affecting some modules more than others?    Also, my initial tests with the cheap scope seem to show that the Meanwell rise time is just as fast..   (let me post some screenshots)

 

1 hour ago, Temp said:

Try increasing the brightness very slowly in small steps. If its as I think I would expect them to gradually become the same brightness. If there is an obvious/sudden point where they switch from different to the same then its probably something else.

 

They all become the same brightness at around 2% (45% perceived), there isn't any sudden jump.   My feeling is that generally modules are dimmer than they should be, rather than brighter, but it's hard to tell as I don't have two tapes sides by side with the different drivers.

[Dan F]

EldoLED @ 1% output (34% perceived)

 

 

EldoLED @ 5% output (57% perceived)

 

Edited May 24, 2023 by Dan F

[Dan F]

Meanwell @ 1% output (34% perceived)

 

 

- Meanwell is 2.5kHz and EldoLED 2.4kW. 

- That last 2.0ms plot isn't showing reality, as the Meanwell waveform is very consistent.

- Meanwell actually seems to have a faster rise time, but it performs flawlessly vs. the EldoLED.

 

Looks like the issue may be more to do with the fact that pulses don't seem to be consistently 24v with the EldoLED.  If this is deliberate, a side-effect of the IC or some other component on the tape or some other issue I have no idea!

 

[Radian]

Rise times are both well under 5us EldoLED is the slower at around 1us and the Meanwell is probably around 100ns so both would shoot-through the mosfet. So both will be unregulated for a brief period. I don't see any great inconsistency in the EldoLED 24V. What does differ is the voltage decay when the driver transistors are off. The LED's will stop conducting when the voltage falls below their forward voltage (looks to be around 15V) so the Meanwell output staying high for around 200us suggests some additional driver output capacitance above the (apparently non-existent) EldoLED. Not sure why this would result in a visible difference becuase the LEDs are fully off in this period anyway, Very puzzling.

[Dan F]

40 minutes ago, Radian said:

Rise times are both well under 5us EldoLED is the slower at around 1us and the Meanwell is probably around 100ns so both would shoot-through the mosfet. So both will be unregulated for a brief period. I don't see any great inconsistency in the EldoLED 24V. What does differ is the voltage decay when the driver transistors are off. The LED's will stop conducting when the voltage falls below their forward voltage (looks to be around 15V) so the Meanwell output staying high for around 200us suggests some additional driver output capacitance above the (apparently non-existent) EldoLED. Not sure why this would result in a visible difference becuase the LEDs are fully off in this period anyway, Very puzzling.

 

From my (more novice) perspective, Meanwell has a consistent 24v pulse which will ensure that the LED forward voltage will always be high enough to be "on". 

But, the EldoLED shows peaks of mixed voltages at a low-duty cycle. (I can see 5v, 10v, 17v and 19v peaks in there).

 

The 7 LEDs per module have a forward voltage of around 2.9v x 7 = 20.3v.  So, my thinking is that any peaks below 20v will mean the LEDs won't come on at all, or will only partially come on.

This would make sense, given light output is less than expected at these levels.  But why this would impact different modules differently I don't know! Any why the EldoLED driver does this is even more confusing.

[Radian]

1 hour ago, Dan F said:

But, the EldoLED shows peaks of mixed voltages at a low-duty cycle. (I can see 5v, 10v, 17v and 19v peaks in there).

 

 

That's a sampling artifact of the scope as you increase the timebase. I'm so used to it on my DSO that I automatically discount it. Notice that it also shows to a lesser extent on the Meanwell trace at 125kSa/s

[Dan F]

2 minutes ago, Radian said:

 

That's a sampling artifact of the scope as you increase the timebase. I'm so used to it on my DSO that I automatically discount it. Notice that it also shows to a lesser extent on the Meanwell trace at 125kSa/s

 

I'm talking about this one.  I can see that there are sampling issues at 125kSa/s, but how can you tell that the smaller peaks here are phantom and that signal only has 24v peaks? 

 

If I switch from "auto" to "once" and re-trigger multiples times, each time it triggers I see different peaks, which match this plot.

 

 

 

[Radian]

39 minutes ago, Dan F said:

I'm talking about this one.

Sorry, I missed that! It's really odd isn't it...

Implies some kind of filter?  "LINEARdrive" sounds like it might be something in addition to PWM but there's not much you can do to get a lossless volt-drop like that without maybe cycling the on time - sort of 'modulating the modulation'

The clue would be how much time it spends at each different step and if there's a pattern to it which will be hard to see without a bigger storage capacity.

[Dan F]

1 hour ago, Radian said:

Sorry, I missed that! It's really odd isn't it...

Implies some kind of filter?  "LINEARdrive" sounds like it might be something in addition to PWM but there's not much you can do to get a lossless volt-drop like that without maybe cycling the on time - sort of 'modulating the modulation'

The clue would be how much time it spends at each different step and if there's a pattern to it which will be hard to see without a bigger storage capacity.

 

Yes, it's odd.  It's definitely related though, as this phenomenon goes away when the light output becomes consistent.

 

LinearDrive is just a brand name for their CV driver.  They use a specific PWM frequency which is a multiple of 25, 30. 50 and 60Hz (for TV compatibility in theory) and from what I've seen at higher outputs the signal isn't pure PWM (bunched pulses instead of evenly spaced), but I'm told by their rep that they do not reduce the voltage at lower outputs (that's not what I'm seeing though!) 

 

If the voltage is reduced though,  I've no idea why the LED tape with another LED driver IC (420 instead of 430) is absolutely fine though!   Unless it's something to do with other components on the tape/IC somehow impacting the driver output.   

 

 

Edited May 25, 2023 by Dan F