FT811: 33MHz pixelclock?

[Pinguin66]

Hi there,
I'm using a FT811 based 5" display (800x480) which is running fine using GD2 library.

Problem is, that I want to connect the FT811 pcb to another display (OEM display in my car stereo, 6,2" 800x480) that needs a pixelclock of 33MHz.
The FT811 systemclock is set to 60MHz, devided by 2 = 30MHz, with this my car stereo display shows no picture.
There is no datasheet for the display, but the display is connectet via serialized LVDS to the pcb of the car stereo and uses an MAX9248 deserializer to drive the display panel. Here I can measure the pixelclock of 33MHz. No way to change this.

So, I need to find a way to force the FT811 to generate 33MHz pixelclock. But how?

Is it possible to use another crystal for the FT811? 12MHz is default (12*5=60MHz). A crystal with 13,2MHz would be perfect (13,2*5=66MHz, devided by 2 = 33MHz). Is this possible?

Many thanks!

[jamesbowman]

Yes, I've run the 811 with a 66 MHz crystal. Works just as you described, you can get a 33 MHz pixel clock.

[Pinguin66]

Thank you for the answer James. But did you use a 66MHz crystal or a 13.2MHz crystal as I wrote?

[Pinguin66]

What if I just use a 33MHz crystal?

[jamesbowman]

I actually used a 13 MHz crystal to get a 65MHz final clock rate.
So a 13.2 MHz crystal should also work.

AFAIK the input signal must be in the ~12 MHz range, so a 33 MHz crystal seems unlikely to work.

[Pinguin66]

Hmm, a 13.2MHz crystal is hard to find.

I switch to use the internal oscillator. With REG_TRIM I will try to get ~11MHz out of this oscillator, then set CLKSEL to 3, that gives 33MHz and hopefully the rest is working correctly. Because, reading the datasheet, FT81x only supports 24, 36, 48 and 60MHz.

I will post my results on weekend or so.

[jamesbowman]

Right. There is an input clock in the ~12 MHz. And there is a PLL multipler that can be set to 2,3,4, or 5.

Maybe a 16.5 MHz crystal, with the multipler set to X4.

That frequency seems to be available.

[Pinguin66]

Its working as I wrote now!
Using the internal oscillator, trimming it to 11MHz and multiply with 3.

I get now 99% accurate 33MHz :-)