ColdFire V2 - electrical guide ?

[SamChi]

Hello.

I'm building an input-output board based on MCF52259CAG80 (that's the 144 LQFP one). You may find some information about it here: http://labserver.uv.es/ioboard/ .

I'm still not sure how to setup the clocking. I want working USB (as device only), MCU clocked near its maximum frequency and possibility to use UARTs at standard RS-232 frequencies. My guess it that there is no other way than to connect 48MHz source to USB_ALTCLK, use some crystal multiple of RS freqs and multiply it with PLL to drive internal clock near 80MHz. But I don't like the idea of designated USB clock while I could just connect the XTAL directly. Clocking UARTs from timer inputs isn't really interesting either, because I won't be able to divide that frequency.

My other concern are the connections through resistors. M52259EVB uses resistors on almost all the signals, to reduce EMI as it says. Should I really take EMI into account ? On the other hand, M52259DEMOKIT uses resistors on clock pins only, probably to make sure no clock edge is detected while not all the signals are ready, but I couldn't find any information about it. Is my reasoning right? That's where a decent electrical guide would be helpful.

Thanks, in advice, for any usefull response.

[mjbcswitzerland]

Hi

1) When using USB it is important to use a 48MHz crystal (or 48MHz clock source). This is the only way to achieve the USB jitter specifications. There is no reason to use USB_ALTCLK since the main crystal clock is adequate.

2)  The system clock can be set to 80MHz (PLL) by using DIV = 6, MUL = 1.

3) When using UARTs the 115200 Baud that can be obtained from the internal Baud-rate generators is 113'636 (-1.5%). If this accuracy is not adequate a system clock setting of 76.8MHz [DIV = 5, MUL = 8] allows 114'285Baud (-0.8%) to be achieved

4) It is very important to put 100nF and 10nF decoupling capacitors in parallel as close as possible to the PLL power supply pins to achieve reliable PLL operation.

5) It is advisable to put resistors in series with port lines - especially if they have long connections on the board.  Even of the port is not switching, the line can carry high frequencies originating from within the processor. If the port is not driving current, resistances of 1k close to the chip stops potential EMI.

Regards

Mark

www.uTasker.com
- OS, TCP/IP stack, USB, device drivers and simulator for M5221X, M5222X, M5223X, M5225X. One package does them all - "Embedding it better..."

Message Edited by mjbcswitzerland on 2009-03-21 10:44 AM

[SamChi]

Thanks a lot.

Looks like the 80MHz frequency will give me baudrate close enough to the standard values.

But EMI resistors of 1k seem like a very large value. I'm going to put the 22 Ohm chips, like EVB does, it won't be a problem to change them to larger values if needed.

[mjbcswitzerland]

Hi

As I mentioned, when the signals are used for simple switching operations (often the case with GPIO lines) and are not delivering current, a large serial resistance will not affect operation but will give high EMI filtering levels.

For clock signals the maximum resistance will depend on the frequency and the board capacitance. 22R may also be suitable for matching the line impedance too.

Of course foreseeing a resistor in the lines allows flexibility for all cases.

Regards

Mark