HC(S)12 in the field...

[puppetmaster]

Hi,

it's not really a technical question, but I hope that's OK. I work as an embedded systems software engineer developing firmware for automotive sensors.

We are currently developing a new series of sensors based on Freescale S12G64 MCUs; the predecessors were based on a Microchip PIC18F4680 MCU.

My colleagues' opinions about all electrical issues are based on his experience with the named Microchip MCU. It's always like "We had issues with this so we found that solution and we're going to use it as it has been proved to be working just fine"... Well, I'm not one who says "throw away all experiences", but he's pretty stiff on that topic. You can forget to tell him that the newer chip might have better characteristics and so forth.

Especially when it comes to electrical noise which is a pretty relevant topic in the automotive field.

And that is mainly the cause for the question here... no one is willing to take some money, populate several prototypes and test the noise characteristics with different MCU clocks (with or without PLL, 4, 8, 16, 24 MHz)... We even got some noise charts from Freescale for this MCU, but no use -> "The results are for their test boards not ours"...

Therefore I'm stuck in the project I'm assigned to with a 8 MHz crystal and I'm not allowed to turn on the PLL which give me a bus clock of 4 MHz (which is 250ns per tick). The main reason for the 4 MHz is that the Microchip had issues with electrical noise while being clocked with 10 MHz (therefore failed some tests), so I get 4 MHz "and we'll pass every automotive noise test just fine".

It might not be the big deal for some, as the 4 MHz S12G still is approx. three times faster than the higher clocked Microchip MCU, but as we have to do a lot of safety checks (like mirror RAM, check RAM during runtime, check CAN, synchronise with second MCU etc..) this is pretty low end[..]

I'm more busy tweaking my OSEK/VDX implementation and communication stack for as less as possible delays than working on safety relevant issues[..]

Therefore I wanted to collect some "out-in-the-field experiences" from users of Freescale S12-Variants to get a simple overview of applications and noise characteristics. That's mainly because I CAN'T believe that a simple 24 MHz processor is not capable to pass automotive noise tests. (I was working for BMW and the MCUs I was working on were running with 120 to 180 MHz (depending on the ECU), and they care for electrical noise).

I hope some people are willing to share their experiences and I get some nice replies!

Thank you for reading that whole bunch of text!!

So here are my questions:

What derivative are you using?

In what kind of product(s) do you use the S12-derivative?

Which clocks are you using (with or without PLL) ?

Which field (automotive, aero, medical, consumer) is the target?

Did you pass critical noise tests for the target field? 

Did you have issues with passing those tests and how did you solve them?

If your product is used in the automotive field, did you have some "special issues" with the car voltage transients?!

[kef]

"and I'm not allowed to turn on the PLL"

It is odd. PLL on S12XE, S12G and others "fresh" S12 chips has spread spectrum modulation feature, which should help you passing EMC tests.

[puppetmaster]

Well, to make his long answer, which he gave me, short: "the PLL turns up the frequency and will cause more noise". That's his statement... One of the reasons why I'd like to have some in-the-fields experiences....

[RadekS · Accepted solution]

Your colleagues has in general partially true; higher frequency presents higher potential dangerous for EMC problem. But EMC test pass typically depends mainly on PCB layout, antennas, energy, currents, harmonics, power supply, blocking capacitors…

In general, higher frequencies require higher attention on these things.

I also believe (together with you) that bus frequency inside MCU cannot be taken as potential problem. More important are frequencies which are used in touch with external world (external antennas).

I work as technical support from S12 MCUs, so I don’t have so much direct experience from the field. But I never heard that bus frequency itself can cause directly any EMC issue.

In fact, S12G is available more than 2 years and this MCU is very popular. I was not able found any service request with EMC issue in case of this MCU family.

S12G family is suited for run from internal trimmed 1MHz RC oscillator (plus PLL). External oscillator is supported for application with higher demands on clock accuracy (like CAN communication).

I would like to recommend contact our support team (create service request) prior you starts with PCB layout. We can look at your scheme and check it for some obvious or potential issues.

There exist also some general PCB layout guidelines:

• Every supply pair must be decoupled by a ceramic capacitor connected as near as possible to the corresponding pins .
• Central point of the ground star should be the VSSX3 pin.
• Use low ohmic low inductance connections between VSS, VSSX1, VSSX2 and VSSX3.
• Keep traces of EXTAL, and XTAL as short as possible and occupied board area for capacitors and crystal as small as possible.
• Do not place other signals or supplies underneath area occupied by capacitors and crystal and the connection area to the MCU.
• Central power input should be fed in at the VDDA/VSSA pins.

Note: Yes, spread spectrum modulation feature can help you for pass over EMC test, but it can be taken as “some kind of cheating (but everybody use it)”. Moreover spread spectrum modulation is not ideal way for applications with critical time requirements.

I hope that information from me help you in constructive “fight” with your colleagues.

[puppetmaster]

Hi,

thanks for the answer :smileyhappy:. I didn't know about the PCB review possibility (I don't think our Freescale FAE talked about that the last time he visited us). I think I should relay this suggestion to the higher ups... At least I had a little success last week when I finally got an ear, during a meeting with a higher up, on the suggestion to make about seven prototypes to test the noise characteristics with different frequencies and two crystals (8 MHz and 16 MHz). Getting input from people who work with the stuff and have experience with it helps a lot :smileygrin:

Btw. it's less a fight rather a 'nice talk against a wall'; just the typical little "Software engineer vs. Electrical engineer" quarrels :smileywink: (though, I had a few years EE classes in my studies (and before))

Kind regards,

[RadekS]

Yes, Freescale FAE should have enough experience for helping you directly, but he is also able very easily create service request for technical support teams (plus other options). So, I would like to recommend contact appropriate Freescale FAE as first.

Customer’s satisfaction is our goal. :smileywink: