Shop OBEX P1 Docs P2 Docs Learn Events
Somehow I managed to internally short Vss and Vdd pins on Prop — Parallax Forums

Somehow I managed to internally short Vss and Vdd pins on Prop

AndrejaKoAndrejaKo Posts: 24
edited 2012-06-11 02:12 in Propeller 1
Well I finally killed my first propeller and I'm trying to find if my theory on how it died is correct.

It seems that I forgot to power up the right-hand side power rail to which the DIP Propeller and the the 24LC256 were connected. It appears that the 24LC256 was pulling power from the Propeller power pins and I suspect that this is the reason why the Propeller died. Does it sound plausible?

I didn't notice the problem until I noticed that the Propeller is hot and it seems that in described setup it worked fine for several minutes since I checked the voltages at both sets of power pins and they were correct 3.3 V

Just to be thorough: the left power pins were connected to a LF33V regulator which was connected to a 7805 which was connected to a 7812 which was connected to a 12 V 10 A unregulated power supply with actual output voltage of 19 V. The power supply has 30 mF worth of capacitors at the output and I had an additional 2.2 mF capacitor at the breadboard. The regulators all had 1 μF input capacitors and 100 nF output capacitors and I had 100 nF decoupling capacitors at the Propeller power pins and at the 24LC254 power pins.

Comments

  • PJAllenPJAllen Banned Posts: 5,065
    edited 2012-06-10 13:23
    All Vcc and Vss pins must be connected (to +3V3 and GND, respectively), if that's what you're asking about.
  • AndrejaKoAndrejaKo Posts: 24
    edited 2012-06-10 13:27
    I'm aware of that. Looks like my original post seems a bit unreadable. I'll try to edit it and make it a bit clearer.


    What I'm asking is: Would pulling 400 μA from one side of the power pins be enough to cause an internal short between the Vdd and Vss on the Propeller?
  • Mike GreenMike Green Posts: 23,101
    edited 2012-06-10 13:58
    The short answer is no. Most overcurrent problems cause the conductors on the chip to melt, thus becoming open. I'm not saying it's impossible, just unlikely. The biggest problem caused by uneven power distribution across the chip is due to the Phase Locked Loop multiplexor (that chooses which clock multiplier to use) becoming damaged due to the voltage drops across the chip. As a result, the Propeller will identify itself and download programs, but they won't run because most programs use a crystal-based clock, but the identification and downloading functions use the RCFAST clock which is not affected.
  • RaymanRayman Posts: 14,827
    edited 2012-06-10 14:08
    A hot Prop means something was very wrong... Usually, it means the Prop chip is dead...
    There are a few ways of doing this... The usual is hooking up 5V supply to the Prop somewhere.
    It's hard to imagine that hooking up an EEPROM chip the wrong way would do that.
    I've had dead shorts on some Prop pins and had it survive.
    Maybe not hooking up all the Vdd and Vss pins would do this, but I haven't seen that.
    Putting 3.3V on Vss or Gnd on Vdd might heat of the Prop in a recoverable way, but I haven't tried that yet.
  • kwinnkwinn Posts: 8,697
    edited 2012-06-10 14:10
    The propeller chip getting hot indicates that there was quite a bit of current through the chip which was probably due to 5 or 12 volts being connected to it somehow. To make an educated guess as to how that happened would require an exact diagram of how everything was connected along with the sequence in which power was applied. Even with all that it would only be a guess.
  • jmgjmg Posts: 15,183
    edited 2012-06-10 14:56
    AndrejaKo wrote: »
    What I'm asking is: Would pulling 400 μA from one side of the power pins be enough to cause an internal short between the Vdd and Vss on the Propeller?

    No, 400uA flowing the right way (sourced from a IO) will not harm anything.

    With multiple supplies, there is another mechanism you need to watch for, which is Latch-Up.
    On some data sheets you will see this given as a Current spec.

    Most CMOS devices parasitic clamp diodes, are actually lateral transistors - so the PN clamp to Vcc, is the E-B junction of a PNP, with the
    collector laterally located. Similar for the Vss clamp.

    These lateral current flows mean you have a low gain SCR built in, just waiting to be triggered !!.

    Trigger is usually via excessive clamp diode current, and good devices need > 200mA to fire this mechanism.
    Sometimes ESD events can cause this, which can include long cables with significant floating capacitance being plugged in while powered.
    TV monitors are a good example - if double insulated, they have no inbuilt capacitive discharge paths.

    Once the parasitic SCR fires, it crowbars the power supply to a SCR drop, and depending on the powers involved, can be fatal. (classic HOT device)

    The holding current of such SCRs can be surprisingly good - tests I did some years back on a Low-current limited supply, found under 10mA was needed before the SCR released on the CMOS devices I tested, and some needed almost 1A to trigger.

    The Prop data sheet does not seem to spec this ? - but it is easy to test, the current pulse width can be << 1ms.

    Is you have multiple rails, and power the Prop _last_, the other rail can supply current, or plugging in physically large loads etc.
  • Duane DegnDuane Degn Posts: 10,588
    edited 2012-06-10 15:12
    I second Mike's suspicion. I remember Chip saying something about the PLLs not tollerating current across the Vdd pins (from one Vdd pin to the other Vdd pin).

    So hooking up only one Vdd to the power supply and then the other to the EEPROM would cause current to flow across the two Vdd pins.

    Though I admit, I don't think this explains the heat.
  • AndrejaKoAndrejaKo Posts: 24
    edited 2012-06-10 15:42
    Thanks a lot for interesting answers.
    The resistance between any of the Vdd/Vss pins is around 9 Ω, so there was definitely some current going on.

    As for the latchup part, if I understand the explanation correctly, I'd need to have current going through one of the non-power pins, right? When this happened, I only had the Propeller connected to the EEPROM.

    I'll post the exact diagram in the morning, since it's 0:41 here right now.
  • jmgjmg Posts: 15,183
    edited 2012-06-10 15:58
    AndrejaKo wrote: »
    As for the latchup part, if I understand the explanation correctly, I'd need to have current going through one of the non-power pins, right?

    Usually it is tested as current injection into IO pins, tho you can also trigger this with a supply voltage spike.
    Were you connecting/disconnecting anything, or powering any other equipment on/off ?
  • Mark_TMark_T Posts: 1,981
    edited 2012-06-11 02:12
    My advice for building a new circuit is don't apply power till you've used a multimeter to test that the power rails are not shorted and than each device is getting its power and ground on the correct pins. It saves you from little accidents like this. Double checking every connection is the ideal, but if you check the power a lot of the flaws that fry devices will be found.

    I also use a bench power supply with adjustable current limit and set the limit low on first test, ready to kill the power if too much current seems to flow.

    And for this incident I'd tend to agree that SCR latchup is likely - the chip getting hot is a classic sign. I've had an ADC chip do this at swichon for some mysterious region (probably spike on an input), but it survived after power cycling. Protection against input over- and under-voltage can be done with schottky diodes in parallel with the on-chip protection diodes (schottky diodes have lower forward voltage and prevent the on-chip diodes conducting)
Sign In or Register to comment.