SCL stuck at 0 on GLK12232-25-SM

[Michael Zoran]

I bought a new GLK12232-25-SM and I desoldered the RS-232 pads and soldered the I2C pads. When I use the display, I can't signal a start condition. I checked the display with a multimeter, and the SCL pin on the display is always pulling the bus low. Is this a defect in the display, or am I doing something wrong?

Interesting, the display had a little dab of solder on the I2C pads when I took it out of the box but not enough to connect the pads. Was the display already used before I got it.

[Michael Zoran]

I did some more experimentation with my multimeter and I can see that the SDA/SCK pins are connected to the correct pins on the PIC16F877A that is in the unit. But I am also seeing that the SCK pin is also connected to pin #13 of the RS-232 level converter on the board. Is this correct?

I read somewhere on this forum about people disconnecting some of the pins of the RS-232 chip to completely disconnect it, but I don't want to do something this drastic unless I have to so that I can still send the unit back to Jameco if necessary.

Thanks.

[Michael Zoran]

I became brave tonight and pulled pin #13 on the RS-232 chip and this brought the SCL pin up to 2.3V instead of 1.2V. I also replaced the pullup resistors with 4.7K instead of 10K. This further brought up the voltage to 3.3V which appears to be high enough to get things to work.

I plan to put other chips on the I2C bus including another PIC chip and some serial flash ram chips. I'm alittle concerned about the bus not going up to the full 5V. How can I further modify the board to kill off the remaining pulldown?

[Michael Zoran]

One more quick question. I'm not trying to slander the product or Matrix Orbital, just trying to understand and make sure I'm not doing something very wrong.

Now that I have things sort of working, I'm finding that the display is very slow at processing input. I need to have a 4ms delay between characters otherwise the display drops characters. I think this means the true speed of the device is really only 2500bps even though the device can handle port settings of 115000bps or 100K for I2C.

Is it possible to optimize the font for faster speed? I'm trying to achieve 10 updates of the display per second for the embedded device I'm building.

Thanks.

[Aniso]

I'm going to start real slow and take it bit by bit because I am a little rusty.

First, the unit was made by a robotic machine and those pads are given a little dab of solder paste. It is very improbable (I want to say impossible) that it is "used".

I don't know the details of the board right now, but before I look at it, can you please confirm which board revision you have?

The communications speed is a little tricky. I would like to refer you to something in the manual (if it is still there - I am trying to load it now, but I am working with a POOR Internet connection), but the general gist of it is thus:
It really is full speed (400kbps), but between each character the processor needs to do something with the received data. As opposed to a strictly hardware device (like a memory chip) which can usually sort of do everything at once, the limitations of a (inexpensive) processor means that it has to stop taking data while it does something with old data. Sure, we buffer it (more than a 100 deep, I think), but even storing it to the buffer takes time. I believe (and I still need to confirm this) that the time needed to store a byte is equivalent to a byte-rate of data incoming at 50kbps. So, if you don't want to format your communications to transmit bytes at full speed and then wait between bytes, you can talk at a consistent 50kbps and all should be fine.

Now, you seem to be telling me that it won't do even that so we need to talk more, but I hope that this explaination helps.

[Michael Zoran]

I have version 1.4 of the board.

Can you answer the question about why the SCL pin is being pulled so low? I'm alittle concerned this may affect other components on the I2C bus.

I don't believe the board is capable of doing 50000bps. I need 4a ms delay between characters as 3ms is not long enough. This is 1000/4 = 2500bps much slower then 50kpbs. Also, I don't believe the board is using the full 92byte buffer for the I2C connection. I can see buffer overrunning in as few as 10 characters.

[Tom]

There is also a smt resistor between the crystal and the rs-232 chip that is pulling the signal lower too. This resistor is a pull down resistor that does not need to be there. Try taking that off, and see if this improves anything for you.

Best Regards,

[Henry]

Interesting, the display had a little dab of solder on the I2C pads when I took it out of the box but not enough to connect the pads. Was the display already used before I got it.

As Aniso said, the unit is manufactured by a robot and when the solder paste is applied to the PCB, all exposed pads get some solder, whether or not there is a component placed there.