LCD clock output looks wrong

[dave-at-axon]

Does anyone know what the LCD CLOCK output should look like?

I hooked up a scope today to check the timing on the CLOCK and although it was 30ns indicating 33MHz, I was surprised to find that it was not a square wave as expected. It could indicate why I am seeing some issues and small artifacts on the display, especially when running on battery power.

The attached image shows a capture of the clock input to the LCD. The voltage swing is 1 to 1.8 volts. Looking at the LCD timing this does not look correct. The timing shows the clock going from 0 to 3.3V. At first I thought it was the scope probe but it is rated for 200Mhz and I checked with a 50Mhz square wave from a signal generator and it looks clean and nice and square. It is also the fact the voltage swing looks wrong. Looking at the other timing signals they are all approx 0 to 3.3V swing.

I have been going through the datasheet for the A20 and can't find anything that references the clock output.

I was wondering if anyone else could check the clock on their LCD and see if the same.

[blejku]

On your graph you can not see too much. Stretch it a little , change the time base . 100 nS This whole 10Mhz/p.
What's the oscilloscope ?

[dave-at-axon]

Try this. I changed the timebase to 10ns.

I can't understand why this is a sine wave when I was expecting a square wave. Where exactly does the input timing detect the falling edge of the clock as this is where the LCD samples the data input?

The scope is a Siglent SHS810, 100Mhz 1Gs/s handheld unit.

[dave-at-axon]

As a matter of interest, I decided to check another board I had with an LP1788 micro and it has a similar clock output so this must be correct.

[dave-at-axon]

I've just been going through the datasheet for the LCD and the electrical characteristics show that for the input signals, including the DCLK, that the max voltage for a low level is 0.3 x Vcc which would be 0.99 V and yet the clock on the scope is around 1V min swing. No I realise that inputting a sine wave signal into an input that expects a square wave type will work when the voltages meet the HIGH and LOW min/max levels but this seems way too close to get a valid LOW signal input and due to it being a sine wave, is right at the lowest point in the wave form.

[JohnS]

If you want sharper edges maybe you can set the drive strength higher.

You'll make more EMI / RFI and may suffer crosstalk of course.

John

[dave-at-axon]

Hi John,

I don't see any option in the A20 to set the drive strength when the output is set for the LCD clock.

It's not so much the sharper edges I was concerned about, it's more the voltage levels. According to the LCD datasheet the clock input low limit is 0.3 x Vdd which would be around 0.9V at 3.3V. This is just on the limit of the clock signal I am seeing.

[JohnS]

Do the Olimex ones have similar levels?

I'd suggest looking at ones used in typical tablets but they can be tough to identify never mind find data sheets!

John

[dave-at-axon]

Hi John,

I found the drive level you indicated. There is 4 choices, 10, 20, 30 and 40mA with 20mA being the default on power up.

I doubt increasing this would help as the current drawn by the LCD input is very small.

I checked the clock on an LPC1788 and it too was offset. All the other signals are nice 0 to 3.3V levels.

I've posted a question on this to a LinkedIn group to see if any of the boffins their know how this all works. It's all new to me as it all just worked before. This time is loses sync when I am on battery power and the screen flashes. As soon as you plug in DC power it is rock solid. There is hardly any drop in the 3.3V rail so my only conclusion was the clock levels where right on the limit.

I'll post back my findings as they may benefit a few others as I've seen timing issues in the past on the forum.

[JohnS]

OK, I was hoping it would help.  It does with normal digital outputs - it basically connects more MOSFETs in parallel so you get rapid current rush and faster edges.

Those levels do look likely to be bad news, sadly.  Maybe you could tweak them a bit - there are various techniques but I'm the wrong person to advise on them!  About all I'd manage would be a buffer or the like of a family with better levels.  You'd go from sine wave to square wave more than likely, too.

John

[dave-at-axon]

Hey John,

It looks like the issue is more to do with the battery voltage dropping too low for the 3.3V regulator to maintain a stable output. The LCD is fine at higher voltages but once the level drops to below 3.5 I start get timing issues and see lots of white flashing on the LCD. Plug back in the USB or apply DC power and it goes back to solid. Voltages higher than 3.6V seem to be good and the voltage drop is negligible on the 3.3V rail.

The regulator that is used on the A20 is a buck type and I did comment on this in the past. It should really be a buck/boost so that it can still provide 3.3V even if the battery voltage is below 3.3V

[JohnS]

They maybe thought people would power off the LCD if on battery.

Could you use a SOM with battery & power external so you can change to that kind of approach?

Or Olimex might make you the kind of board you want but MOQ may be 500 or so.

John

[dave-at-axon]

Nice idea to get a custom board but my systems are all very low volume and would not be able to absorb an MOQ of 500.

I am considering a rework just now and get a custom board made in-house based on the A20 and will purchase the parts from Olimex so they still get revenue from it.

I am busy building up the schematic and library of parts just now. It's quite a bit of work. I can remove all the stuff I don't use such as HDMI, SATA etc.

[JohnS]

I suspected MOQ 500 was optimistic, sorry!

Hope it goes OK.

John