Excessive IO ripple on microcontroller output pins

snow

Hi
I am currently testing a new designed PCB running a ATMega234pb. I have found that multiple IO pins seem to have a high ripple voltage when switched on and even when switched off.
When pin are high the ripple is 70mv and low 20mv. Has a frequency of around 4mhz.

Design notes are as follows

1. 5v Rail plane 8mv pk-pk
2. Has 100n and 2.2uf caps on Vcc for decoupling
3. Micro running a external 8mhz crystal
4. Pin currently testing PB.4 , PA.3 PD.4, PD.5

Queries
These pins are part on the Peripheral Touch Controller (PTC), but I'm just using them as a enable pin for a Op-amp and switch a transistor for an LCD backlight.
I don't have any success with setting the PTC off with PRR2.3 = 1 to stop the ripple.

The only way to stop the ripple is to put the micro into power save mode ( CONFIG POWERMODE = ADCNOISE) or by adding a 100nf cap to that io line

Attached are the scope readings

EDC · Joined: 26 Mar 2014 Posts: 1142

Check if this is not a "signal ringing" or "circuit ringing". There is some tutorials on YT. Someday I saw a very good one but now I culd not find it to link here.

snow · Posted: Mon Jul 21, 2025 10:25 am Post subject:

Hi Team,
Thanks for the input so far.

Would you consider 70 mV peak-to-peak ripple excessive? My scope only samples at 1 GHz, so the actual ripple could be higher.

Today I removed the external 8 MHz crystal and switched to the internal oscillator — unfortunately, that made no difference.

The issue first showed up after enabling an op-amp, which feeds into a 12-bit ADC. The ADC output showed ringing, which led me to start probing around and uncover this ripple.

I’m not 100% sure if the port is correctly set as an output. It’s declared as one in code, but I’m unsure whether the Peripheral Touch Controller (PTC) is still active and clocking the pins.

According to the datasheet:

Page 39, Section 10.1.3

Page 62, Section 12.11.5

These mention PRR2 bit 3 (PTC disable). I’ve tried clearing this bit (i.e., setting PRR2.3 = 0), but how can I actually verify that the PTC module is completely shut down?

Also, I’d expect the pin ripple to match or slightly exceed the Vcc ripple, which is ~8 mV peak-to-peak. For reference, I have 100 nF and 2.2 µF decoupling caps on the Vcc and AVcc pins. Interestingly, placing a 100 nF cap directly on the affected pin reduces the ripple to 20–30 mV — but from what I’ve read, that’s not really a recommended solution.

Cheers,
Snow

JC · Joined: 15 Dec 2007 Posts: 630 Location: Cleveland, OH

What are you using for your Ground Lead / Connection for your O'scope probe?

When you want to measure small signals on the PCB, you should be using the little 1/2 cm wire on the tip of the probe, and not the 6 inch long wire with an alligator clip on it.

If you are using the (usual?) 6 inch wire, move it around, use different ground points and have different orientations of the wire "Loop", and you might well see the amplitude of the signal change significantly, (meaning that the signal is largely an artifact, and not real).

JC

snow · Posted: Mon Jul 21, 2025 9:33 pm Post subject:

HI JC.
Yes i a using a probe earth spring at the Gnd connection. With more luck than design I was able to test the VCC pin and the IO pins with the same GND connection point..
Also i made a typo on the previous message. Setting the PRR2.3=1 disables the PTC module. I have tried setting and clearing the register to compare the difference
but this does not make any difference.

This is my first time using this chipset ATMega324pb . I usually go with the ATMega328pb... Could this just be what the chip set does???

Cheers
Snow

snow · Posted: Wed Jul 23, 2025 5:37 am Post subject:

Hi Team,

I'm trying to read and write to the PRR2 register (address 0x66), but running into some issues.

For example:

PRR2.3 = 1 ' This line compiles, but doesn't seem to do anything
Setreg R66, 8 ' This doesn't compile
Temp = PRR2.3 ' This also doesn't compile
Also, I'm curious how BASCOM performs a fuse and lock bit read under the hood. Specifically:

For reading something like the Fuse Low Byte (at address 0x63), which holds clock setup bits, I tried:

Var = GetReg(R63) ' Doesn't compile
Var = GetReg(63) ' Also fails
Am I missing something about how GETREG or low-level register access is meant to work?

Lastly, someone suggested probing the pin that's showing a clock signal — and it does appear to be coupled. I've attached the scope results showing the behavior if that helps clarify the issue.

Thanks in advance for any insights![/code]

albertsm · Joined: 09 Apr 2004 Posts: 6209 Location: Holland

getreg/setreg only work on registers r0-r31
but you can also use the registers direct since they are exposed to bascom :
somevar=r0 ' assign a variable with register R0

fuses reading and writing is explained in the memory programming topic in the datasheet. but this only works with external programming.
some processors allow to read signatures or fuses.
have a look at mcs.lib _readsig and the readsig() function.

PRRx registers are IO registers that work like any other IO register. but you need to check the datasheet to see if they are read only and what you need to write.