Trying to control Philips backlight manually with PWM — brightness increases with frequency...?

[ohren]

I raised the question of whether PWM on the power board "dim" pin would be a reasonable assumption for controlling the backlight of a Philips TV (55put6101/12): https://www.badcaps.net/forum/troubl...d-tv-backlight. I didn't receive any insights there so I went ahead and tried it.

The setup is basically:

Rpi3 hardware PWM → 220 Ohm → NPN transistor → dim pin

Transistor is switching the 3.5 V supply on the power board.


The brightness and power draw increases significantly with frequency, so I'm sure I'm missing something. Here's my measured power draw:

• PSU on, backlight off: 7 W

1 % duty [Hz] [W]:

• 15 10
• 30 11
• 60 13
• 120 17
• 240 27
• 480 46
• 600 55

Same frequencies, 25 % duty [W]:

• 39
• 40
• 42
• 45
• 52
• 65
• 71

50 % duty [W]:

• 68
• 71
• 72
• 75
• 82
• 94
• Didn't measure 600 Hz

What's going on here? Is PWM the wrong assumption?


PS. I don't own an oscilloscope unfortunately, so I'm relying on the duty and frequency values used in code. Since the values aren't extreme I'm assuming they're reasonably accurate.

[R_J]

From a led driver datasheet: A typical frequency range for the (PWM) input is from few hertz to 50 kHz. The maximum dimming frequency is limited by the minimum rise/fall time of the current which is around 4 µs each.

[Diah]

in other word if you want to control Dim pins... you need to use ardunino fed fromsame source 3.3V and write yours script steps Freq as you wish.

[ohren]

Ok, thanks, but I'm not sure I understand!

I'm already using frequencies between a few Hz and 50 kHz in my tests. I don't understand why brightness increases with frequency in my test when using identical duty percentage.

Why do I need to use Arduino? Is a Pi not sufficient? Its PWM at least worked well enough when testing with a single LED when using a 1000 ns period.

Why do I need to use 3.3 V instead of the power board's existing 3.5? Does it really matter?

I'm not quite sure what you mean when you say "write yours script steps Freq as you wish" — do you mean that I can use whatever frequency I like?


Anyway, I'm still not sure if PWM is entirely the right assumption considering it doesn't behave as expected. I expected brightness and power draw to be roughly the same when I varied the frequency and kept duty the same. It didn't. What kind of signal do these power boards usually expect from the logic board?

[Diah]

you captured the duty W against the Hz, if you monitored the currents you will see why duty increase by Hz up.
Rpi PWM also responsible for this currents behavior, on other hand Arduino will pules only Hz depending on the steppers code you write.
but as R_J said.. you are not free to pules any freq you like, there are limitation, best help for yours project "forum.arduino.cc"
3.5V for me are high to control DIM., and could be reason too.. as its original circuit max 3.3 - 0.5 V

[ohren]

Thanks for your reply!

if you monitored the currents you will see why duty increase by Hz up

Ok, so I take it this means that I need an oscilloscope if I want to know why this happens? That's a shame, I was hoping it'd be some common phenomenon.

Regarding the Pi being responsible for this behavior, I'll have to take your word for it since I don't know what's going on. I'll investigate that path, thanks.

I understand what R_J wrote, but I don't understand what they're hinting at by mentioning it, considering that I'm well within the mentioned frequency range in my test — I just don't know how to use that info to explain the observed behavior.


Thank you for the forum recommendation and suggestion about the voltage being too high. I'll investigate those points.

[R_J]

What is the frequency range you are using? looks like 15Hz to 600Hz to me looking at your post, try 25,000Hz and see what happens

[Diah]

[QUOTE=R_J;n3522935]What is the frequency rang/QUOTE]

with Rpi3 will be allowed only KHz., no way for low Hz

[ohren]

Yes, 15 to 600 Hz was the test range.

try 25,000Hz and see what happens

Really? I did search for what TVs and displays use in their backlights and it was suggested that many of them used way lower frequencies even than the 10 kHz generally accepted as sufficient for indoor lighting, i.e. around 2000-6000 Hz, or even lower for "strobing" displays which use the same frequency as the video signal (if I understood that concept correctly).

Since I haven't reverse engineered the power board and don't know exactly what the dim pin is connected to, I didn't dare ask anything close to 25 kHz from it. I guess I'll try it later and see.

with Rpi3 will be allowed only KHz., no way for low Hz

I don't follow. Even if Rpi3 hardware PWM couldn't do sub-kHz (which to my eyes it appears to be able to do just fine, but granted, my eyes can't tell the quality of the signal) it would still be trivial to do low frequency PWM manually in code, no? Sure, it's susceptible to preemption in user space, but on average that should even out and still not result in a higher duty cycle / power draw.

I've not used its PWM extensively so I'm just reporting what I've seen with my eyes when testing both with an LED and with the backlight. 15 / 24 / 30 Hz appears to work, and is truly headache inducing.

If you have more info on why Rpi3 can't do it then I'd be very interested to hear.

[R_J]

I could not find a datasheet on the TPV101AD (guessing this is the ic on your board) but this is similar ic datasheet for a led driver with dimming
I was just going by a couple datasheets for these ic's that control the dimming with pwm,

I could be wrong about their operation.
I just checked the datasheet for a LD7890GS and it uses a pwm freq of 100Hz ~ 800Hz

[ohren]

I just tried 1 and 2 kHz at 1% duty and the power draw just kept climbing — 122 watts at 2 kHz, which is 1.2 W per diode. I've no guidelines with regards to how much heat they can dissipate, or survive, so I think I'll not push them further for the time being.

Nice of you to look it up and sharing the pdf! Unfortunately it looks like the two drivers are under soldered down heat sinks so I couldn't confirm. All I could determine was that they're eight-legged and that the dim pin is connected directly to the two drivers' pin 1. I'll take a look at the data sheet.

(The reason I didn't go much higher than 600 Hz was that the power draw approached or exceeded the max draw seen when the logic board was in charge)

[ohren]

I think you might be right in your identification of the driver.

If the datasheet describes a similar driver, then it confirms the behaviour that I fail to observe:

The LED switch is driven by a PWM signal applied to the control pin (CTRL), which directly controls the LED brightness. With this control method the LED brightness depends on the PWM duty cycle only and is independent of the PWM frequency and amplitude

It's starting to seem difficult to understand this without a scope...

[R_J]

Download the service manual, it has the schematic

[stj]

PWM uses a fixed frequency,
you ajust the amount of high-time vs low-time in the pulse.
hence "Pulse WIDTH Modulation"
useable frequency range will be in the controller datasheet.

[ohren]

Download the service manual, it has the schematic

Oh yeah, you are right — TPV101AD.

I'm always taken by surprise whenever modern products have service manuals, and discovering that not everyone has gone full Apple. Maybe I've been brainwashed. Thanks for the reminder.

Anyway, I didn't have more luck finding the data sheet, and instead mostly found more threads on here that were also looking for it. Even so it might still not have explained what I'm seeing, so I'll have to conclude that I'm blind without a scope.

[ohren]

I solved it. Turns out that removing the transistor and connecting the Pi directly to the dim pin gets rid of the unexpected behavior; I can now change frequency without changing the brightness.

Since I don't know, and since I'm not an EE, I'll assume that it was my use of the transistor that was somehow improper (it was included in a generic Amazon/aliexpress Pi kit, marked "8050"). I just tested my circuit wit a single LED before hooking it up to the backlight. Don't know why the transistor appears to have changed the effective duty cycle of the signal . Should I have used a pull-down resistor between it and the dim pin? Yeah I'm lost :-)

I guess my next challenge is to find out what I did wrong (haven't really used transistors myself before so I'm happy it at least worked somewhat).

Theories welcome, otherwise I'll just ask in a more appropriate place.