Does the 5225x feature 5V tolerant inputs?

[Kopone]

Hi,

I've come across a Freescale document from the FTF2009 by Wang Hao where it says (on page 4) the 5225x GPIOs were 5V tolerant.

Found it here

However I can't seem to find any confirmation on this in the Data Sheet or Reference Manual (says 4V max. - which would be ok if there were clamping diodes to Vcc taking the remaining volt).

So yeah, before I toast my development board I just wanted to make sure I can blame it on someone else, haha. No, really, did I just overlook relevant parts in the DS or is the info wrong?

Regards,

 Sven

[JimDon]

The devil is in the details.

The data sheet says:

1. Digital input voltage .VIN –0.3 to +4.0 V

2. With this foot note.

Input must be current limited to the IDD value specified. To determine the value of the required current-limiting resistor, calculate resistance values for positive and negative clamp voltages, then use the larger of the two values.

3. Instantaneous maximum curren tSingle pin limit (applies to all pins). IDD = 25 mA

4. Footnote:

All functional non-supply pins are internally clamped to VSS and VDD..

The problem  with 2 is that for one thing, the measured clamp voltage at a few mA is 4 V (VDD=3.3V), not VDD, combined with the statement by DrSeuss that the current must be limited to <2mA (please, where did you get that???).

You see  this leads to a series resistor that is too  high and cause problems in that weaker drivers can  not pull the input as low, and this both reduces noise margins and smears the pulse due to capacitance combined with the series resistor (creates an un wanted low pass filter).

So, again, where on earth did you come up with <2mA -while I am  not suggesting 25 mA is reasonable, nether is <2 ma.

(In practice, I have had no issues connecting inputs to the 5v output of a chip that limits the output current to 3.5 mA, nor should I expect to, based on the published specifications.). If you got from the "DC Electrical Specifications" "Output high voltage (all input/output and all output pins)" then I say you mis-read it, as it is not a limit. If you  got from some internal secret source, then say so.

[TomE]

Use (or abuse) of "Injection Current" is very common in Automotive circuits where every fraction of a cent counts.

it is very common to connect 12V signals (door switches, brake and park light power levels, Ignition etc) directly to microprocessor inputs through a 100k resistor to the pin. There's usually another resistor from that pin to ground, but that isn't to drop the voltage as much as it is selected to set the on/off switching level. Or a 5V micro with a 2.5V threshold the combination of 100k and 100k is common and gives a 5V threshold. 100/68k gives 6.2V. Add 10n or 100n capacitors to the pin for low-pass, ESD and EMC reasons and one to the other side of the first 100k for the ESD and EMC as well. Why as high as100k? You can get 50V sustained and 200V spikes in automotive electronics.

The 52259 Data Sheet says the "Instantaneous maximum current Single pin limit (applies to all pins)4, 5" is 25mA.

It also warns that if the design is using a lot of "injection current" then it has to go somewhere. If the rest of the load on the Vdd supply doesn't take all that current then there has to be something connected to the Vdd rail to make sure it doesn't go high. Try using a Zener. I dare you. It doesn't work.

It also details that the ADC inputs have "Input injection current, per pin" of 3mA  with the note "Current that can be injected or sourced from an unselected ADC signal input without impacting the performance of the ADC".

Compare this with the MCF52235 Data Sheet. It has the following to say:

DC injection current 3, 5, 6, 7
     VNEGCLAMP =VSS– 0.3 V, VPOSCLAMP = VDD + 0.3
  Single pin limit                                    -1.0 to 1.0 mA
  Total MCU limit, Includes sum of all stressed pins  -10 to 10 mA

This data sheet has been better written than the MCF52259 one as it lists the clamp voltages and details single pins and the total current, where the MCF52259 one is a bit ambiguous.

> DrSeuss that the current must be limited to <2mA (please, where did you get that???).

That was in response to a quesion on the MCF5223x which gives a per-pin limit of 1mA and not the MCF5225x which looks to be 25mA for a single pin or 25mA total for all pins. So DrSeuss was a little high, but he probably assumed design engineers would read the data sheet for the design value, rather than listening to "A Random Poster" or "Another Random Poster".

> smears the pulse due to capacitance combined with the series resistor (creates an un wanted low pass filter).

Of course you have to properly evaluate what a resistor plus a few picofarads (7pF for the pin plus a bit for the track) of stray capacitance will do to your signal. Following the data sheet instead of hand-waving will give a resistor value of (5V-3.6V) = 1.4k per milliamp of injection current. That's within spec so we'll assume 1.4k. 1.4k and 10pF would give a 3dB rolloff to frequencies above 71MHz or so, and since these parts only run at 80MHz there probably aren't many signals that high you'd want to feed into them. You'd probably want to design in more low pass filtering, especially if the signals are originating "off board' at all.

> You see  this leads to a series resistor that is too  high and cause problems in that weaker drivers can  not pull the input as low,

I have no idea what JimDon means there. Drivers only have to pull down the "input leakage current" of the pin which is 1uA. That would develop 1.4mV across a 1.4k series resistor. The "weakness" or otherwise of the driver shouldn't matter in the slightest here. It might be a problem if trying to use a GPIO pin in bidirectional mode, but a bit of sensible design and checking should lead to a workable solution.

> while I am  not suggesting 25 mA is reasonable

25mA is the design maximum. Anything below that meets specs. If the design input is "12v with occasional 200V spikes" like it is in all cars then the design point would be 25mA at 200V or 8k for one pin, 16k for 2 pins, 80k for 10pins and so on.

 

Tom (The Original Random Poster, Jim :smileyhappy:

[JimDon]

Get out of your armchair and into the lab and you will see how that will not work.

[J2MEJediMaster]

Guys, you have both made valuable contributions to this forum, and it is much appreciated by Freescale. Unfortunately, some of your most recent posts are starting to get a little personal.

Please do not make me take out my moderator's wand and use it.

---Tom

[scifi]

Hi Sven,

Since you have the dev board handy, why don't you just tie a pin to 5V through a 1 kOhm resistor and see what happens?

My guess is the pins are not 5V-tolerant.

Regards,

- mike

[mjbcswitzerland · Accepted solution]

Hi

I think that DrSeuss gave a good answer to this question here: https://community.freescale.com/message/20209#20209

Regards

Mark

www.uTasker.com

[Kopone]

Thanks guys,

while this is all useful and valid information, it is a tad to vague to use it as a basis for a production run.

I meanwhile found the section in the datasheet where it is mentioned that all IOs are clamped to the power rails. However there is no hint about what constant currents are allowed across those diodes. Though even if the post from DrSeuss would apply to the 5225x too, I can hardly quote that as a reference when the lot starts failing in a year So I'll have to add a level converter or a simple voltage divider, no problem. (I was just hoping I overlooked in the Spec Sheets since it was explicitely stated in that presentation doc).

Anyway, I can use the info for devboard hookup and get along with a simple resistor here, that's great already!

Thanks,

 Sven

[scifi]

---

Kopone wrote:

...

However there is no hint about what constant currents are allowed across those diodes.

...

---

Well, it's hardly surprising since overdriven inputs are normally seen as a stress to the device, so why test this as a normal usage pattern? You'd also want to limit the total injected current on all pins since it ends up on the supply rail, and some MCU manufacturers do quote such a limit.

Common sense and experience suggest that if you limit the current to 1 mA per pin everything should be fine.

[Kopone]

Thanks for the input, Mike.

I havent faced this problem before (always lucky enough to work with matching voltage devices) so I am not familiar with the details at all. Now that you mention it, of course all the current had to go somewhere...eventually the LDO had to sink it, I had to check with the supply guy if it even could or if there's a diode across or something. Learn something new every day,

thank you,

 Sven