Power Supply Design - Input Protection Scheme

[Fred]

Thanks! Do you have any thoughts/feedback on my dual Zener plus current limit idea?

[TonyS]

I think there may have another related problem.

When I posted this -

... but then realized that because there are seperate 5V supplies, the differenct output voltages might be far enough apart to cause the clamp diodes to start conducting even with no fault.

it was in regards to a solution that I proposed (clamping to the 5vdc-cpu bus).

With two different 5V supplies, it appears that this issue already exists.

If the 5vdc-analog supply is greater than the 5vdc-cpu supply and if the difference is greater than the difference between the forward drop of the external clamp and the analog input's internal clamp (which is connected to 5vdc-cpu) it will be the internal clamp conducting and not the external clamp.

Thanks,
Huff

[Fred]

I don't understand your description, or, you don't understand the topology on the Jag. Please clarify.

[TonyS]

Not trying to be funny, just concise this time : )

Assumptions -
uC analog input pin has an internal clamp diode to 5vdc-cpu.
External schottky diode clamped to 5vdc-analog.
LM2937 = 4.75 to 5.25V

If (IntClampFwdDropV + 5vdc-cpu) < (ExtClampFwdDropV +5vdc-analog)
Then current from input short to battery will flow primarily through the internal clamp diode.

Hope I got this right .

Thanks,
Huff

[Fred]

OK, I understand now. I can't comment yet. 430am = too tired to read manuals. Back to the code.

[Dan]

Huff,

Interesting point there.....
I too need to consult the CPU datasheet before I comment on this. Give me 24 - 48 hours if possible.

Cheers,

Dan

[DaWaN]

Thanks! Do you have any thoughts/feedback on my dual Zener plus current limit idea?

I must check tolerances on Zener diodes and Current / voltage diagrams to see how much a 5V1 Zener leaks at 5V

If the 5V1 zener leaks and dissipates more than 50mW it will mean extra heat in the box, which is something I do not like

My idea: use a 5V6 hard zener and a TL-431 connected to one of the microcontroller pins or a LED
If the TL-431 is triggered at about 5.1V you can iniate a CEL so one knows the analog inputs are wrong and one can take action to fix it

I do not know if the microcontroller has protection diodes from the analog voltage to the digital voltage. We have to check the absolute maximum values from the datasheet to check what extra protection might be needed.
Most likely a single external schottkey will do I think.

[Fred]

Nice idea! :-)

[Dan]

I don't think the TL-431 is a good choice. I will explain more later tonight once I get home. I am now working on this as a priority as it is the last remaining item before a first revision RavAGE PCB can be drafted and hopefully manufactured.

More to come very soon!

EDIT: What I meant by "I don't think the TL-431 is a good choice." is that I don't think it alone would work sufficiently.

I also think a simple 5V1+resistor combo added to the power supply schematic and the input protection circuits left as-is might do the trick (Fred's idea), but I agree that it is ugly and there is probably a more tactful, technical, resourceful and overall better approach. Will further justify in my next post in this thread. *research/testing continues*

[Dan]

Now I need to firstly test Fred's idea. I think it will work, but it would/will be ugly.

Secondly, I need to test the circuit that is attached if the above "Fred" test doesn't prove itself worthy.

Thirdly, the TL-431 probably won't work exactly the way we may want it to because it is still a zener and will probably conduct with our regulator working normally, it is only capable of handling 100ma (i don't think this matters fully as the 5V6 zener upstream would handle the brunt of it) and being a zener it would be just as ugly as sticking a 5V1 in there and calling it a day IMO.

I think we could get the precision we need by doing maybe something like what I have drawn up in the attached PDF or maybe I have just had too much beer. But it is a good, stable, precise method of setting a trip/clamp for over-voltage scenarios. Coupled with the 5V6 zener upstream, I think heat dissipation will be quite minimal and it would work.

But that depends on how we want to handle the over-voltage "fault" condition. Do we want to try to maintain the integrity of 5V switched whilst the fault is happening and try to still function safely with the fault in place OR do we want to safely handle the fault and cause a brief shutdown of the 5V switched rail which would inherently alert the user that something is not quite right? (the IC I have added in the attached PDF has a pin that could be used to tell our CPU that the fault is present also).

Need thoughts on all the above ASAP please people!