How sensitive is the Prop chip PLL crystal input to damage
jon123456789
Posts: 11
Hi,
just after others experiences with this.
I've been using my prop in a breadboard installation using a 3.3v regulator fed from a 12v bench top psu with no caps etc, using wire links between various other ic's and Inputs / Outputs to generally have a play around and learn spin. All ok so far.
I've then built a more permenant prototype of my circuit on copper strip veroboard (before I went on to get it done properly as an etched solution) basically links some volt free digital inputs to 8 pins and connects to a MAX232 (external caps) and onboard EEPROM and 3.3v and 5v regs with the necessary tant caps to stabilise the supply which I use as a link between digital IO and a SMS modem to send text messages from my alarm system and after a few hours of use (including power cycling the PSU) the program fails to run. A bit of diagnostics later reveals that the prob is no longer·responding to the crystal input or the PLL multiplier. All other pins are ok and it will run the exersise 1 spin code to simply blink an led on any pins and the eeprom works.
Basically the crystal input is stuffed. The crystal is ok (tested on another circuit)
I've bought another prop, and the same happened.
I've checked for shorts / open circuits / voltage levels etc and all I can assume is that somehow it is getting spiked from caps / psu and breaking something internal.
What do others do to protect their props from situations like this? Zeners? Bigger caps? different PSU? spacings around the crystal? Other?
Has anyone else found a reason that the crystal input / PLL multipler part of the prop is most sensitive to damage and especially what would you do to protect it?
My previous breadboard experience of no in circuit protection, no psu caps,·man handling chips in / out gave me the impression the chip was almost indestructable but obviously (and understandably) not!
Cheers
Jon
just after others experiences with this.
I've been using my prop in a breadboard installation using a 3.3v regulator fed from a 12v bench top psu with no caps etc, using wire links between various other ic's and Inputs / Outputs to generally have a play around and learn spin. All ok so far.
I've then built a more permenant prototype of my circuit on copper strip veroboard (before I went on to get it done properly as an etched solution) basically links some volt free digital inputs to 8 pins and connects to a MAX232 (external caps) and onboard EEPROM and 3.3v and 5v regs with the necessary tant caps to stabilise the supply which I use as a link between digital IO and a SMS modem to send text messages from my alarm system and after a few hours of use (including power cycling the PSU) the program fails to run. A bit of diagnostics later reveals that the prob is no longer·responding to the crystal input or the PLL multiplier. All other pins are ok and it will run the exersise 1 spin code to simply blink an led on any pins and the eeprom works.
Basically the crystal input is stuffed. The crystal is ok (tested on another circuit)
I've bought another prop, and the same happened.
I've checked for shorts / open circuits / voltage levels etc and all I can assume is that somehow it is getting spiked from caps / psu and breaking something internal.
What do others do to protect their props from situations like this? Zeners? Bigger caps? different PSU? spacings around the crystal? Other?
Has anyone else found a reason that the crystal input / PLL multipler part of the prop is most sensitive to damage and especially what would you do to protect it?
My previous breadboard experience of no in circuit protection, no psu caps,·man handling chips in / out gave me the impression the chip was almost indestructable but obviously (and understandably) not!
Cheers
Jon
Comments
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Visit some of my articles at Propeller Wiki:
MATH on the propeller propeller.wikispaces.com/MATH
pPropQL: propeller.wikispaces.com/pPropQL
pPropQL020: propeller.wikispaces.com/pPropQL020
OMU for the pPropQL/020 propeller.wikispaces.com/OMU
You will probably find that you did not link the ground pins of the Prop together directly (as well as the supply). Do not assume that by connecting ground to a ground pin that that is all that is needed, link them all directly. I never run any power components off the same voltage as the Prop, regulators are cheap.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
*Peter*
I'm convinced looking at the schamatic and layout that it has to do with spikes from the PSU when power cycling, it is just strange that the exact same circuit (although a different layout) has worked fine on a breadboard. Its almost like the caps made it worse!
All ground pins were directly linked under the dip socket.
I've ordered a load of 0.1uf, 1uf and 10uf tant caps and will insert them as per the demo board. I'll also stick the caps nearer to the prop VDD pins to gnd.
Still bemused as to why this bit of the prop breaks first!
Good job I bought a few of these chips to have a 'play' with!
I am using the same 3.3V reg (via a 5V reg, from a 12V DC supply) as the demo board. What do others use for a robust / safe / clean 3.3V supply?
Cheers
Jon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Visit some of my articles at Propeller Wiki:
MATH on the propeller propeller.wikispaces.com/MATH
pPropQL: propeller.wikispaces.com/pPropQL
pPropQL020: propeller.wikispaces.com/pPropQL020
OMU for the pPropQL/020 propeller.wikispaces.com/OMU
It's more likely to be caused by not connecting all the Vdd pins. Inadequate decoupling can cause other problems, of course.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Don't assume anything is in right or is the right component, go back over your proto component by component.
The other thing to look out for are any +5V inputs. I noticed you are using a +5V rail and a MAX232 so how are the digital outputs from that tied back to the Prop? Through resistors I hope.
These Prop chips are rugged but stress anything beyond it's ratings and it will break, preferably in a predictable fashion. No use trying to protect the PLL as something else will probably go anyway because the chip is stressed, better off not stressing it and find out what's wrong with your circuit rather than picking at the Prop.
@Jon, photos are everything, I'm sure we'll take one look at the photo and go "aha!"
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
*Peter*
That's what I was getting at. I wouldn't expect you to miss a .1uf decoupling cap and have the chip's PLL burn up. I would think there would need to be more at work than that. I know the regulators can throw fits without proper output caps so that would make some sense. For me, if that board had eaten two fairly expensive and now difficult to replace chips I would be looking to start over and reverify everything.
I've rebuilt using the Tant caps at all places shown on the demo board (so 12V input, and both 5V and 3.3V outputs and directly between VDD and VSS as close to the chip as possible) and so far 12hrs running ok (thats 11hrs and 57 minutes more than before for 2 x chips!). That is all I've changed.
Serial coms and EEPROM are all ok.
All VDD and VSS pins were correctly connected.
1) Stupid thing I did: I used Polyester film caps by mistake, not tants. Correct values, but wrong type of cap. I did not fully appreciate the reason to use one type over the other and I had the poly films to hand. Any words of wisdom on this?
2) Not so stupid thing I've done no2 (but will rectify), you are correct I have directly connected the Max232 to the Prop in the belief that both were TTL spec. I have NO resistors. I'll do a bit of digging around and see what will be sensible to use unless any one has a suggestion? 1Kohm? At the rate of 1 chip per day I want to do what ever is sensible to protect my circuit!
3) Interesting point I am using some reed 12V relays to interface the signals from my alarm panel into the Prop as digital inputs which are intended to be initiated from a completly seperate psu (they do have internal coil diodes). I was testing these using the same 12V bench PSU. I discovered one of these relays had shorted and effectivelly dragged the PSU down when it was initiated, this would have given the regs a kick and coupled with the wrong / missing caps 'could' have caused the damage?
I found this out by putting the example 1 'led blinker' code in the EEPROM and then putting my code only into RAM. This is interesting as it will run the RAM code until the unit resets and then it immediatly shows that the prop has reset as it runs the code in EEPROM. So in my case initiating the relay caused the prop to reset as the psu voltage dipped.
To be honest the circuit is very basic (read as not much to go wrong!). I'll try and get a piccy and up load for amusment sak
Good job I'm still at the prototype stage so I can insert all these 'protective devices'.
Some good learning for me out of this one.
Cheers
Jon
·
There's a long thread in one of the "sticky" threads (Propeller: Getting Started and Key Thread Index) on this subject. The Propeller I/O pins have substrate diodes designed to conduct (to the supply rails) when an input goes above 3.9V or below -0.6V (one diode drop above Vdd or below Vss). These diodes are rated at 500uA, so any input resistor should limit the current to that. For 5V, that would be at least 2K.
Any "noise" spikes on either supply line could damage the Prop. In particular, positive spikes on Vdd could easily exceed 4.0V, the point at which damage can occur (by puncturing gate oxide). Negative spikes on Vss typically due to ground loops can also cause the Vdd-Vss voltage to exceed 4V which would also begin to damage gate oxide layers. Positive and negative spikes on I/O pins can also drive current into the chip supply rails and, due to the resistance of the supply lines on-chip, can also cause voltage differences to occur between sections of the chip.
Despite all this, the Propeller is actually pretty robust as these things go. Unlike a lot of microcontroller I/O circuitry, the Prop appears to be immune to I/O short circuits. The datasheet puts a maximum current limit on the I/O pins, but testing has not produced any failures.
Propeller based systems this year and none of them have been returned for an electrical
problem.
The trick is in designing the power supplies properly and good PC board layout. You need
to pay very close attention to 5V interfaces and there protection resistors for the reasons
that Mike explained above.
Russ
You should have used ceramic capacitors instead of tantalums, they are cheaper and smaller. Use a MAX3232 instead of the MAX232.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
As a word of warning, I can also recommend against inserting tant caps the wrong way round late a night to get a circuit working and then leaving it to 'test' overnight when it seems to be ok.
Result, a nice burnt patch on a (thankfully) none flammable table as one of the little blightlers acted like a time bomb fuse and must have popped many hours after the supply was turned on!
A timly reminder that I really need to pay more attention to this!
PS, another blown chip!
Jon
Cheers
Jon
thanks, will do.
Jon