PDA

View Full Version : Strange failure mode - anyone seen anything like this?



CarlosFandango
03-17-2009, 08:05 PM
Hi All,

I've been away a while working on other things, but got a chance to work on a prop based project recently (museum exhibit). Now this has arrived back on my doorstep because it's broken...! I used a prop protoboard in this design, having a look at it this morning, it will load / verify both RAM and EEPROM no problem - but subsequently doesn't seem to want to do anything at all. Won't even flash an LED. So it's not booting up it seems - I checked things like the reset line and the integrity of the crystal, but all seems to be in order.

Has anyone seen a failure mode like this before by any chance? It might help me figure out what went wrong in the first place.

-CF

Peter Jakacki
03-17-2009, 08:09 PM
As has been reported many times in this forum this sounds like the classic PLL failure. Try running at a slower clock-rate to confirm. The boot procedure uses the RCFAST mode so is not affected by the PLL failure.

From memory the failure usually comes about because of improper grounding plus switching heavy or inductive loads.

*Peter*

CarlosFandango
03-17-2009, 08:58 PM
Peter said...
As has been reported many times in this forum this sounds like the classic PLL failure. Try running at a slower clock-rate to confirm. The boot procedure uses the RCFAST mode so is not affected by the PLL failure.


Yep, that looks like the problem. Great..

Well now, I'm quite aware of how ground loops and spikes can cause trouble. This prop controls a set of relays through standard npn transistor drivers; all relays are provided with clamping diodes, of course. The other ports on the prop go to serial ports or are inputs switched by phototransistors. As far as I can see, there's nothing in this design that could cause a problem of this sort, the unit has been operating continuously for about three weeks, yet this has in fact happened! I've been reading around the forum on PLL failure and as you say there's a fair bit of info there.

Obviously I need to look at my design more closely... but I'm still confused

-CF

cocokiwi
03-17-2009, 09:19 PM
CarlosFandango said...

Peter said...
As has been reported many times in this forum this sounds like the classic PLL failure. Try running at a slower clock-rate to confirm. The boot procedure uses the RCFAST mode so is not affected by the PLL failure.


Yep, that looks like the problem. Great..

Well now, I'm quite aware of how ground loops and spikes can cause trouble. This prop controls a set of relays through standard npn transistor drivers; all relays are provided with clamping diodes, of course. The other ports on the prop go to serial ports or are inputs switched by phototransistors. As far as I can see, there's nothing in this design that could cause a problem of this sort, the unit has been operating continuously for about three weeks, yet this has in fact happened! I've been reading around the forum on PLL failure and as you say there's a fair bit of info there.

Obviously I need to look at my design more closely... but I'm still confused

-CF

are the relay drivers run on the same power that the prop is on? If so I think that is what did it,I think more than one relay kicked in at the same time causing an inductive pulse that killed it,my thoughts!
if only 2 cents worth.(grin)run the relay power on it's own supply link so any load spikes don't kill the prop.

Cheers Dennis

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
http://people.delphiforums.com/cocokiwi/Image/picture.jpg

Peter Jakacki
03-17-2009, 09:27 PM
A great deal of these failures seem to take place on the protoboard (but not exclusively). It is perhaps the prototyping area that has all manner of indignities foisted upon it all in the name of prototyping naturally enough. I don't know if the Propeller shares it's ground trace with other devices that can be attached but it would be a cause for concern. In my design methodology I opt for a small 3.3V regulator close to the Propeller and power only the essentials such as the EEPROM and coms perhaps. Also I use low ESR tantalums on the output of the regulators leaving any electrolytics for input filtering only. The LDO regulators are slower than their non-LDO cousins so the faster low-ESR tantalums help to smooth things out and certainly relays and inductive loads (with their current lagging voltage thing for one) would tend to upset them. The regulators could respond slowly and then overshoot. Perhaps a fast 3.6V clamp across the 3.3V might help.

The protoboard is just that, it can't take everything into account so some thought needs to go into how the loads are connected and powered. I wouldn't use the +5V for the relays or any non-logic load as the CPU depends upon a clean supply.

Carlos, would you happen to have a diagram or photos of your layout because it would be worthwhile gathering information about these board failures.

Is there an any diagram showing the copper layers on the protoboard, anyone?

*Peter*

kwinn
03-17-2009, 09:47 PM
If at all possible I try to use an opto isolator and separate power supply to drive relays. Transistor or darlington output for DC, triac for AC. In a lot of cases the darlington and triac output can drive the coil directly.

CarlosFandango
03-17-2009, 09:58 PM
Peter said...
Carlos, would you happen to have a diagram or photos of your layout because it would be worthwhile gathering information about these board failures.


Diagrams attached (relevant parts). The processor board is a proto board, I just reproduced the essential parts of the diagram here.

I'm afraid these diagrams may not be too clear as there at first glance does not seem to be any direct way of connecting them. That's because there is a loom which goes from the 25 way plug on the processor board and the 15 way plug on the relay board, and this isn't shown here.

Also, the relay board has a power plug supplying the relays separately. Nothing is supplied directly from the proto board.

I'm thinking that I am going to rebuild this processor board separately using a 40 dip prop, that way I can swap the chip out if it goes bang again.


cocokiwi said...
are the relay drivers run on the same power that the prop is on? If so I think that is what did it,I think more than one relay kicked in at the same time causing an inductive pulse that killed it,my thoughts!


No they're not, but I'm thinking something similar myself...