Two ATtiny406 dead within a month????

[satyamfifa]

Hello All,

I decided to do a project based on ATtiny406 and last month I ordered 5 prototypes. The ATtiny406 on the first prototype lasted just for about one week, after which it would no longer would program with error: "Framing Error", so I dismissed it as a fluke.

However, after one month the ATTiny406 on the second prototype also stopped programming with error "PDI physical timed out. (25)", of course changing to high voltage UPDI doesn't do anything. Although the microcontroller seems to function normally (UART and Timers are working but it just won't program).

So now I am puzzled I don't know what could be killing these ATTinys.

Also, it's worth mentioning that System VCC is 5.4V, although I don't know why this should be a problem.

Kind Regards,

Sparsh

Update: Three Chips are now dead

Update: more information: The first chip died within a week of testing as I am developing the firmware, the code is uploaded to chip several times, the first chip died with "framing error" and showed impedance of 13k ohm on the programming pin, the second chip lasted for over a month, I might have programmed it like 20 to 30 times at least until finally yesterday I could no longer program it all of a sudden and it gave the error "PDI physical timed out. (25)" but the impedance on the programming pin is normal about 47k ohms, the third chip died today it only lasted less than 24 hours, and I get the same error on this one “framing error” as with the first one, only again the impedance of the programming pin is low about 13k ohms,

Also, it's worth mentioning that at the time second one died I was continuously connecting and disconnecting the programmer as well connecting and disconnecting the power to the board.

Update: Almost Full schematic

ibb.co/09BQ8Vz

Update: Full schematic attached in PDF format

[amyk]

How many times have you been programming them?

[Psi]

How does the circuit get connected to a programmer when its time to get programmed?

Is it possible your programmer or device has some mains earth voltage leakage that is being shorted when the programmer is connected?
It may only be a tiny amount of current but if the voltage is high it may blow the input.

If the ground connection between the two is made first then it's fine, but if the programming pin connection is made first you can have issues.

Check the voltage on a DMM between the grounds of both systems. (AC and DC)

[pcprogrammer]

Also, it's worth mentioning that System VCC is 5.4V, although I don't know why this should be a problem.

Not criticizing, but why the 5.4V? Sure the ratings for the used chips are 5.5V, but it is pushing the limits, and most designs use 5V, or nowadays a lot is even 3.3V.

Also not saying that it is the root of the problem, but with this high supply and the protection diodes spikes might go over the absolute max rating of 6V and do some damage.

[Psi]

Also, it's worth mentioning that System VCC is 5.4V, although I don't know why this should be a problem.

Not criticizing, but why the 5.4V? Sure the ratings for the used chips are 5.5V, but it is pushing the limits, and most designs use 5V, or nowadays a lot is even 3.3V.

Also not saying that it is the root of the problem, but with this high supply and the protection diodes spikes might go over the absolute max rating of 6V and do some damage.

They are 5.5V max operating and 6V absolute max.
I wouldn't have expected 5.4 to cause any issues like this. (Assuming its stable 5.4 and not 5.4 with bursts up to 6V)

[Doctorandus_P]

All IC's are quite fragile and can easily get damaged by ESD or other over voltage conditions, such as for example a voltage spike from a relay without a flyback diode. IC's have internal ESD protection, but this is limited. Depending on circumstances (dry weather, sort of clothing) ESD discharges can be much greater than what the internal ESD protection of IC's can handle.

In my experience voltage regulators (such as LM7805, LM317, etc) are also not capable of stopping high frequency noise.
If you have for example a power supply with an toroidal transformer, then surges coupled in from the mains net can easily create voltage spikes on your power supply and damage your IC's. Adding a few ferrite cores is about the only thing that can stop such spikes.

But also:
I don't trust the Chinese garbage. I'm suspecting that a lot faulty products are being shipped  and those can have a life expectance of weeks or months.

[WattsThat]

Understand that posting schematic fragments will get you fewer and lower quality answers.

With schematic bits and pieces, it cannot be determined if any of the pins of the device are left floating? Which ones?

How is it programmed? Is the programmer operating at the same VCC?

Have you checked a faulty device for shorted pins? (Failed ESD diodes)

You’ve got a buck regulator supply in there with an unknown supply source, what do the startup/shutdown voltages look like? Staying within the limits of the device? Is it an automotive supply?

[satyamfifa]

Full schematic has been attached and the original post has been updated

[Siwastaja]

Have you looked at the V_BUCK_OUT with a scope, is it stable? Maybe it is oscillating?

[satyamfifa]

Yes that's exactly my next step would be, as soon as I get my hands on a scope. Although I am measuring AC with Fluke 17B MAX here and the AC is non existent like something in microvolts range

[cv007]

of course changing to high voltage UPDI doesn't do anything

That probably eliminates most programmers, but instead of trying to deduce what programmer you are using you may want to include that info. Also what application is controlling the programmer.

[Siwastaja]

Also make sure the programmer does not accidentally write some stupid fuses like the clock selection to "ext clock", that would be absolute classic. AVRs require an existing clock source to be programmable and you can permanently mess that up with fuses.

You can try to input a 1MHz square wave from function generator to the XTAL input (forgot if it was XTAL1 or XTAL2, stupid naming) and see if the device is now programmable again.

[MK14]

Yes that's exactly my next step would be, as soon as I get my hands on a scope. Although I am measuring AC with Fluke 17B MAX here and the AC is non existent like something in microvolts range

A quick look at the specifications (couldn't find max model via fluke, I'm not sure if it matters), seems to indicate that the AC voltage measurements, only go up to around 500 Hz.

https://www.fluke.com/en-id/product/electrical-testing/digital-multimeters/fluke-17b-plus

Ok, I managed to find the MAX model in the end, it seems rather similar:
https://dam-assets.fluke.com/s3fs-public/15BMAX__umeng0000.pdf?bJYgDKSn8bhq5XVTwS3aFVT8ehzxQMVw

Perhaps there is noise/transients (e.g. on the power line), exceeding the absolute maximum voltage for the IC, which seems to be 6 Volts.  As that, especially over time, could in theory, damage/destroy the IC.  A multimeter, can't easily pick up very short transients. Or even oscillations on the power line, as already mentioned.

A scope would be much better, at seeing what is really going on.

[satyamfifa]

How does the circuit get connected to a programmer when its time to get programmed?

Is it possible your programmer or device has some mains earth voltage leakage that is being shorted when the programmer is connected?
It may only be a tiny amount of current but if the voltage is high it may blow the input.

If the ground connection between the two is made first then it's fine, but if the programming pin connection is made first you can have issues.

Check the voltage on a DMM between the grounds of both systems. (AC and DC)

I indeed had a floating ground on the prototype, I measured the voltage between two grounds and it read up to 8V RMS, now I have grounded everything together

[cv007]

AVRs require an existing clock source to be programmable and you can permanently mess that up with fuses.

This is the newer avr series 0/1, you will always have a clock. The only thing you can mess up in fuses is the updi/reset pin config for the mcu's that do not have a dedicated updi pin (mega0 series has a dedicated updi pin, for example). The mplabx ide for example will not let you change this fuse bit if your programmer is not capable of producing the 12v pulse to override/enable the updi pin, to prevent you from locking yourself out. If you have a hv capable programmer as it appears in this case (but not specified), it does not matter.

[satyamfifa]

I am using PICkit4 with MPLABX

[Psi]

I indeed had a floating ground on the prototype, I measured the voltage between two grounds and it read up to 8V RMS, now I have grounded everything together

8 volts of floating potential between them is likely not a problem.
It's when you see 50-300V that you have issues.

But I'm confused when you say  "now I have grounded everything together"
They always needed to be grounded together to function, so i'm not sure what you did or what you were doing previously?
If you had previosuly been using mains earth as "the ground" between circuit and programmer then that was probably your problem.
Any mains spike could damage your circuit.

[satyamfifa]

The thing is is the programmer and the target does not already share a ground, and if the programming pin makes a contact before the ground pin does, and potentially the programming pin could be damaged. I did not pay attention to the proper grounding before.

[Psi]

The thing is is the programmer and the target does not already share a ground, and if the programming pin makes a contact before the ground pin does, and potentially the programming pin could be damaged. I did not pay attention to the proper grounding before.

Yep, that's the key,  ideally you use a connector that is designed to connect ground first. (It's a thing, usually the ground pin is longer).
Or you add some voltage spike protection on all the inputs that can clamp any spike for the time required to fully connect GND and everything .

[pcprogrammer]

The thing is is the programmer and the target does not already share a ground, and if the programming pin makes a contact before the ground pin does, and potentially the programming pin could be damaged. I did not pay attention to the proper grounding before.

Only an issue when done "hot".

The way I do these things is to first disconnect the programmer from the PC, and the target from its power source. Make the connection between the programmer and the target. Connect the programmer to the PC and then power up the target.

[bidrohini]

Very disappointing, no doubt.

[PinheadBE]

A bit off topic:
If the goal of D24 is a reverse-polarity protection, I should have placed it AFTER the polyfuse

Just my 2c...

[PCB.Wiz]

...
So now I am puzzled I don't know what could be killing these ATTinys.
Also, it's worth mentioning that System VCC is 5.4V, although I don't know why this should be a problem.
..
... Update: Three Chips are now dead
..
Also, it's worth mentioning that at the time second one died I was continuously connecting and disconnecting the programmer as well connecting and disconnecting the power to the board.

The OP may have solved this already, but I'll add that connecting and disconnecting the power to the board. needs care.

A couple of years back we had a few Atmel parts fail on board final test, and we tracked that to the live power connection of a bench power supply.

We added a wirewound 2r2 series R, and no more problems.

It seems the abrupt connection of an already charged power supply, gives very high dV/dT to the wiring and PCB and that was enough to literally  spike things.
Normal power on/off does not have the same extreme rise times.

Some useful waveforms showing hot-swap / live-connect effects are here :
https://www.analog.com/media/en/technical-documentation/application-notes/an88f.pdf
and here
https://www.ti.com/lit/pdf/slva703

[Psi]

Another thing to watch is the location of the capacitor on AVCC pin, especially if using an inductor to feed VCC to AVCC.

If you feed power to the AVCC pin through the inductor from VCC but then you accidently put the cap to GND on the VCC side of the inductor rather than the AVCC side (where it belongs), then you can run into MCU frying problems.

MCUs draw current on some IO pins through the AVCC pin rather than VCC. (Check DS to see which pin get power from what).

If you have the cap in the wrong place and then do like 20mA fast PWM on a IO that draws power from AVCC then you're pumping high frequency current through an inductor and subjecting your AVCC pin to high voltage pulses above 6V.
With the cap on the wrong side of the inductor there's nothing to absorb these spikes on the AVCC Pin.
So the MCU wont last long like this before you fry the AVCC protection diodes and then fry the chip itself.

[PCB.Wiz]

Also, it's worth mentioning that at the time second one died I was continuously connecting and disconnecting the programmer as well connecting and disconnecting the power to the board.

Live/HOT plugging is a known risk. The cable inductance and low impedance ceramic caps, can give spikes that damage semiconductors.