Screen stays white


#1

Hi,

I have recently built a Zynthian box using parts mostly sourced externally, as well as the kit v2.1 case and all-in-one module PCB.

When connecting the screen through a standard computer HD flat ribbon cable, practically the same as what is available for purchase from the Zynthian store, the screen goes completely white and stays that way. However, when the Pi is connected directly to the screen without using a cable at all using the built in female header, the display is fine.

I have been testing the connections for multiple hours between the Pi, the cable and both headers of the screen, and there is no visible issue with the setup.

I am using the recommended screen, a PiScreen 3.5".

Any ideas?

Thanks.


#2

Hi @cv016!

Idea 1:
How long is the ribbon cable you are using? The PiScreen uses the SPI interface at a high speed, so a “too long” cable could add too much capacitance to the line and break the connection. FYI, zynthian’s ribbon bus cable have a total length of 15cm.

Idea 2:
Sorry, but … are you connecting it in the right position? There is a small “arrow” marking the pin 1 in the PCB’s side-row-pin connector. I know you saw that, but i had to explain it for future readers :wink:

Regards,


#3

Thanks for the fast reply,

The cable length is ~20cm, so although this is unlikely to be an issue I will be sure to try a shorter length. I have also tried using 26 female jumpers to connect the screen which worked fine, so could it be an issue with the thickness of the wire?

And yes, the pins are in the right places :wink:

Thanks.


#4

Although I’m not an expert in this kind of subjects, i will try to add 2 cents :wink:

About capacitance in flat cables, a typical (good!) value is about 45pF/m. Resistance is about 100Ohm/m.
So, for the official zynthian ribbon bus, we have:

C = 6.75 pF
R = 15 Ohm

Using this Low Pass Filter calculator:

http://sim.okawa-denshi.jp/en/CRtool.php

We have the cut frecuency at about 1GHz, quite far from the 32MHz used by the PiScreen SPI connection. Not problem with it.

Lets suposse we have a bad-quality cable (100pF/m), and quite long (40cm):

C = 40 pF
R = 40 Ohm

For that case, the cut freq is about 100MHz, what is too close to the 32MHz of the SPI bus freq. Not good and probably it wont work.

Kind Regards,


#5

Of course, resistance is also important. As i’ve explained, the wires can be modeled as a RC low-pass filter, and the cut freq. depends of C and R values.


#6

Hi @jofemodo,

Thanks for all this information, have come to the conclusion that a combination of a bad quality cable and cable length has created this issue. By shortening the cable, the screen works as it should.

Thank you everyone for your help.


#7

FWIW I think that the cause is the former more than the latter.
20cm is way less than the one I used form my first Zynthian tests, that was a 60cm IDE cable. The sound card was in the middle connector, and the display on the farthest connector from the rPI :slight_smile: