|
|
View previous topic :: View next topic |
Author |
Message |
hello188
Joined: 02 Jun 2010 Posts: 74
|
GPIO on PIC undershooting |
Posted: Sun Sep 22, 2013 9:43 pm |
|
|
Hi, I was troubleshooting SPI /SS line and noticed that the output undershoots about 500mV(to -500mV), for about 5ns and returns back to 0.
In datasheet, the absolute maximum rating for the pin is supposed to be only -0.3V.
So far, I am debugging with ICD3 and the chip seems to work fine.
However, would it cause problem later on?
What are the possible reasons for undershooting?
Thank you |
|
|
FvM
Joined: 27 Aug 2008 Posts: 2337 Location: Germany
|
|
Posted: Mon Sep 23, 2013 12:39 am |
|
|
The maximum voltage rating should be read in conjunction with the maximum clamp current specification
of +/- 20 mA. As long it's not exceeded, there should be no risk for the chip.
An over/undershoot usually happens if an unterminated PCB line or cable of some length is driven with fast
rise time and high current strength. The best way to reduce is to insert a series resistor (e.g. 30 - 100 ohm)
at the driver side (source side impedance matching) or reduce the drive strength, if it's programmable. |
|
|
Ttelmah
Joined: 11 Mar 2010 Posts: 19553
|
|
Posted: Mon Sep 23, 2013 12:43 am |
|
|
First start by making sure your scope probe is properly compensated.
Unless this is correctly adjusted, it can show over/undershoot where none exists. In fact the undershot can be caused by the probe itself in this case...
Second then look at where the scope is grounded, relative to the chip. It is easy to 'see' undershoot, when what is actually happening, is that the point the scope is grounded to, is rising.
In real circuits, a good ground plane, is the first step in controlling this.
A tiny series resistor in the line will fractionally slow the rising/falling edges and reduce overshoot. Perhaps 10R.
Basically, any signal line, carrying a high frequency signal, is a 'transmission line'. You will get undershoot, when the line being driven is not terminated to have a good match. The signal travels down the line, bounces back off the unterminated end, and arrives back at the driver a few pSec (depending on the length of the wire), after if went low, and pulls the line down.
So, things you can do:
1) Add termination to the line.
2) Make sure you have good decoupling. If this is inadequate it can again give 'apparent' undershoot.
3) Ground layout. Make sure everything is grounding back to the same point.
4) Slow the signal edges (this is part of what the series resistor does). Adding a (tiny) amount of capacitance to the line, can also be done.
Best Wishes |
|
|
hello188
Joined: 02 Jun 2010 Posts: 74
|
|
Posted: Mon Sep 23, 2013 8:20 pm |
|
|
Thank you.
I redid the auto-compensation on my tektronix oscilloscope, and still shows
the undershoot. There is overshoot too, but it's not as bad as the undershoot
So, basically, I have to control the slew rate of the driving pin's.
Unfortunately, I don't think PIC18F 's don't have that option.
Do I have to add a series resistor pin right after the driving pin's? |
|
|
RF_Developer
Joined: 07 Feb 2011 Posts: 839
|
|
Posted: Tue Sep 24, 2013 1:42 am |
|
|
I suspect you have a basic misunderstanding of the significance of the data presented in the datasheet.
The absolute ratings related to what can be applied from an external source, and in a sense are macro-scale ratings in terms of time, i.e. you will probably get away with applying a higher voltage than an abs max if its only for a very short time.
Overshoot and undershoot - essentially the same thing, just in after different direction of change of signal - are normal and almost impossible to eliminate. The fact that an output from the PIC overshoots to beyond abs max is not really a big issue: its momentary and the energy is low. It's not great, but its almost impossible to get rid of. The theory is "simple": match the transmission line and load to the impedance of the output and the overshoot will pretty much go away. However the output impedance is not constant and not defined, therefore you cannot get it "right", just "close enough" at best. PICs are also hardly the fastest devices in our increasingly high speed digital world. PICs clock rates are hundreds of times slower than many other processors (tens of MHz v several GHz) and line matching for PICs is generally a non-issue. Yes, you'll get over and undershoot, which if you're really worried about it, a series termination resistor typically a few tens of ohms tops and half-decent PCB layout, will largely tame, but that's quite normal and nothing to worry about and I doubt few, if any PIC users bother, except in a very few exceptional cases.
There certainly are abs max voltage rating problems with PICs and other processors, but overshoot is generally not it: there are bigger fish to fry. What IS a real problem is things like analogue inputs being driven beyond Vdd/Vss by op-amps powered from higher voltage rails, e.g. +/-12V when the Vdd of the PIC is 5V, or a PIC running at 3.3V being attached to peripherals running at 5V. |
|
|
Ttelmah
Joined: 11 Mar 2010 Posts: 19553
|
|
Posted: Tue Sep 24, 2013 3:00 am |
|
|
Also, if this is a pin used to drive a 'select', then you might want to consider adding a pull up resistor at the opposite end of the line. Has the advantage of meaning the end is driving a slightly lower impedance, and ensures that the line is pulled high when the PIC has not yet woken up.
Best Wishes |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
Powered by phpBB © 2001, 2005 phpBB Group
|