Posted on February 11, 2017 at 20:54
Some more information regarding this problem:
Document http://www.st.com/resource/en/user_manual/dm00321394.pdf confirms that the culprit is indeed the display backlight PSU. It supplies 25.6V.
Here's a photograph of the circuit:
- http://www.st.com/en/power-management/stld40d.html
- http://www.st.com/content/ccc/resource/technical/document/datasheet/5b/92/e8/f8/41/e5/43/59/CD00098098.pdf/files/CD00098098.pdf/jcr:content/translations/en.CD00098098.pdf
The maximum on time is 4 us so the switching frequency is clearly not the direct cause of the 8 kHz whistle. However, the brightness is controlled by applying a PWM signal to the enable (EN) pin of the above chip. This is doubtless the cause of the problem.
Page 9 of document http://www.st.com/resource/en/user_manual/dm00321394.pdf shows the display backlight circuitry.
It shows the EN pin as either being connected to BL_CTRL via R4, or being connected to CABC, which is an output (I think) of the LCD display module, via R1 and CN2. It also shows EN as being connected to 3.3V via R5 (4K7).
On my display board, shown above, R1 is present and R4 and R5 are missing. The EN signal for the PSU chip therefore comes from the LCD module. Had R4 been fitted instead of R1, EN would have been connected to pin 53 of the DSI LCD connector, which is the LCD_BL_CTRL (LCD backlight control) signal on the main board, which in turn connects to port PI14 of the MCU (pin 81). In this case, the 8 kHz problem could have been solved by reprogramming the PI14 output.
When the demonstration software starts, with R1 present, the EN signal is a PWM signal of just over 8 kHz as expected. Initially it has a 50% duty cycle. If the screen brightness is increased to 100% using TouchGFX => External hardware => Screen brightness (swipe left), the EN signal changes to a duty cycle of 100%. (Incidentally, setting screen brightness to 0% results in a 20% duty cycle). With screen brightness set to 100%, and hence with a 100% duty cycle, the 8 kHz noise initially fades and then ceases.
Candidate solutions:
- Move the zero ohm resistor from position R1 to R5, thereby connecting EN to 3.3V. This would result in maximum brightness and no 8 kHz audible noise.
- Find out the method that TouchGFX uses to control the brightness via the LCD module, and use it to set the brightness to 100%.
- Move the zero ohm resistor from position R1 to R4, fit a 4K7 (or preferably larger to reduce power consumption when the backlight is off) in position R5, and control the brightness using PI14. (R5 is so that the screen lights even when the MCU pin is not set as an output).
- Reprogram the LCD unit to output a much higher frequency CABC. However, it may be that the output frequency is hardwired, in which case there's no solution to be found there. So far I've not found any relevant information.
- Find a replacement for L2 etc that does not emit sound.
I've done option (1). It cures the whistle, avoids software compatibility problems, and solves the problem of the brightness control PWM (of whatever frequency) appearing on the DAC output.
Option (2) may achieve the same as (1) more conveniently (provided it can also set the brightness to 0%). However, I've not found any information on the method involved.
I hope this helps.
JB.
