Pins shorted on ATM324P (AVCC and GND)

Go To Last Post
11 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hey all, just finished debugging an interesting problem.

I was trying to program a chip in a custom test fixture I created for the purpose. My programmer would not recognize the chip.

I was checking for shorts with my multimeter on the 200 ohm setting. After some time I realized the resistance might be much higher if it was a small short. I set my meter to the 2 megaohm setting. With the chip in the board, I found that VCC and GND were shorted.

I removed the chip just to be sure it was the board, but then I noticed that the board was fine. Everything checked out.

I looked over at the chip, and realized the problem. I checked the chip out of the fixture, and measured 55.5k-ohm across pin 30 (AVCC) and pin 31 (GND).

I quickly tested my fixture with a spare ATM324P that I had (in case something like this happened), and the pins were not shorted, and the fixture worked fine.

Now the absolutely perplexing part is that the "bad" chip CAN be programmed, but only on a breadboard. The only difference between my fixture and the breadboard is that the breadboard has 0.1uF decoupling caps across VCC and GND, and a 47uF electro across the main supply. I don't see how capacitors would mitigate the effect of a short circuit.

The current between pin 30 and 31 would only be (5V / 55.5kOhm = 90uA), which is almost negligable.

I'm also wondering what might have caused this kind of a short to occur. What's really strange is that although the bad chip can't be programmed in a properly wired test fixture, it will still run its program properly.

My mind is blown. I have no idea what's going on.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It may not be unusual to read something other than full-open across micro pins. Especially with a cheap meter there may be sneak paths in the chip that the metering current is "exploring". I guess take a virgin/spare or a few and perform the same test on them when not inserted into anything.

Re "how can this happen"? You mention/imply moving the micros. Perhaps it was inserted backwards at some point.

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I've measured the 'resistance' between power and ground on several 'good' atmega processors using my DVM in the 'diode' position. Here the meter is actually measuring the forward drop across the diode (put an Si diode across the probes and read about .7 volts, .2 with a Ge diode). In my case I read an open circuit across the processor power and ground in the forward direction and about 500mv across it in the REVERSE direction! And this was with a GOOD WORKING chip! I'm guessing the reverse reading was across a protection diode on the substrate that is reverse biased under normal conditions. Also my ohmmeter probably doesn't have enough output under test to get the chip biased into conduction in the forward direction. I'd take any ohmmeter readings on the chip itself with a grain of salt.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I would hardly call 55K a "short", even if the reading is mostly boooogus.

Think of what 55K means in a 5V system. Ohms law gives you 100ua, more or less. That is a short circuit? I assert that a "reality check" could well be in order.
If

Quote:
My mind is blown
, then a reality check is an absolute necessity!

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
the breadboard is that the breadboard has 0.1uF decoupling caps across VCC and GND,
Do you mean that your fixture does NOT have any bypass caps??? :shock:

John Samperi

Ampertronics Pty. Ltd.

www.ampertronics.com.au

* Electronic Design * Custom Products * Contract Assembly

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

What happens if you swap the connections on your multi meter?
I think you have measured the protection diodes inside the controller and not a short. A normal multimeter will either send a constant current or voltage through its leads and then measure the voltage or current. If you made VSS high there will be 2 diode junctions to VDD and the multimeter will give that a resistive reading as a current is flowing.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I've had moments where I've forgotten to connect AVCC. I was horribly confused as well at first why I could program it but it acted weird. Most AVRs with an ADC use AVCC as the power rail for at least one port of I/O. If it's not connected, but Vcc is, you can get some odd results.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

js wrote:
Quote:
the breadboard is that the breadboard has 0.1uF decoupling caps across VCC and GND,
Do you mean that your fixture does NOT have any bypass caps??? :shock:

I was under the impression that chips program at a lower speed than they operate at. I didn't think they'd be necessary. I can successfully program an ATTiny84 in a similar fixture without bypass caps.

Does an AVR program at a higher speed when you increase the clock speed via the fuse bits?

My main problem is that in the fixture, AVRDUDE simply refuses to acknowledge my device, or it can't connect with it.

so far the only obvious issue I found was the connection between GND and AVCC. I'll try installing some bypass caps today and see if this makes a difference.

Thanks for all the advice and tips.

@ka7ehk: yes, I suppose 55k isn't a short. I used the terminology quite loosely. I had confused continuity between VCC and GND as a short without interpreting the meaning of such a high resistance value. Good call.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Always use the decoupling caps even at 0.00001 Hz. If you could successfully complete making such a foolish thing as programming the AVR without a decoupling cap, this doesn't mean it was not a mistake.

Never try to evaluate the resistance of pure digital non-powered CMOS chips. There are no resistors, just diodes - you can measure the forward diode voltage but measuring the resistance makes no sense and can vary from almost 0 to infinity depending on the meter's probe voltage applied to the pins (different meters have very different voltage values which additionally varies with meter's range switching or autoranging process).

Warning: Grumpy Old Chuff. Reading this post may severely damage your mental health.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

MBedder wrote:
Always use the decoupling caps even at 0.00001 Hz. If you could successfully complete making such a foolish thing as programming the AVR without a decoupling cap, this doesn't mean it was not a mistake.

Never try to evaluate the resistance of pure digital non-powered CMOS chips. There are no resistors, just diodes - you can measure the forward diode voltage but measuring the resistance makes no sense and can vary from almost 0 to infinity depending on the meter's probe voltage applied to the pins (different meters have very different voltage values which additionally varies with meter's range switching or autoranging process).

would this mean that the internal pull-ups are implemented with diodes? I know this doesn't apply to the pins in question, but you have me wondering now (if CMOS don't have resistors).

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

There are no resistors - the pullups are implemented with a depletion mode MOSFET structures which are active only when the chip is powered. When one is applying the ohmmeter's interprobe voltage to the unpowered chip, the pullups may get activated or not depending on this voltage value and a meter's current sourcing value. The value of a "resistance" measured that way is just a nonsense - it may for example reflect the on-chip peripherals operating current if the voltage was high enough to provide a parasitic power to the rest of a chip via the port protection diodes.

Warning: Grumpy Old Chuff. Reading this post may severely damage your mental health.