Custom micro-controller reset circuit

Thread Starter

przemek64

Joined Dec 11, 2019
2
Hi All,

I have to design a circuit which will detect if there's TTL communication from a micro-controller.
When there's no communication for 4-5 minutes it should send a 1s reset signal (positive).

I'm trying to do that with discrete elements but I'm wondering if there's already a device (IC/circuit) which does it.
I have it working now but I used 3x 555-timer, 5x transistor, 12x resistor and 6x capacitor.
I have a feeling that it can be done much easier and I'm just reinventing the wheel here.

Some considerations:
It can send only one signal even if reset doesn't work (that is: TTL communication never comes back).
I don't mind if the circuit sends the reset signal at power on.
The circuit will work on 3.3V

Cheers for any thoughts.
 

ronsimpson

Joined Oct 7, 2019
2,989
LTC6995
data sheet
1576088354447.png
This little IC has a timer that can be set from mS to 10 hours.
Look at the examples in the data sheet. There are many options.
I think, when there is communications each bit of data will strobe the reset pin and hold the timer in reset mode. With no reset the timer is set to count for 5 minutes. I think there is a way to get a approximant 1 second pulse.

I know there are other ICs that do this job.
 

schmitt trigger

Joined Jul 12, 2010
872
Maxim Semi and others offer watchdog timer ICs which will do precisely what you are requiring.

Many of them also include brown out detectors, which help if the supply voltage drops below safe operating limits.
 

danadak

Joined Mar 10, 2018
4,057
Low end PSOC 4 family part would do this. Note many resources for other tasks
not used, available for other stuff (right hand window) -

1576200583641.png


One chip solution, and a cap to bypass its power pins all thats needed.

Note solution produces no glitches on power up.




Regards, Dana.
 
Last edited:

danadak

Joined Mar 10, 2018
4,057
Simple - ?

Since he mentions 555's, my interpretation of "discrete elements" is no programmable devices.
Smells like a job for a CD4060. 256 seconds is 4.27 minutes, so 1 bit-time in the last 8 bits meets both the delay and output pulse width requirements.

ak
Discretes, I missed that TS comment.

But over T & V and component accuracy question is how much tolerance in his 5 min and 1 sec
time events can he tolerate ? CD4060 has huge timing variation using RC based unless xtal used.
Lets see how TS responds.


Regards, Dana.
 

danadak

Joined Mar 10, 2018
4,057
Looks like 4.5 s +/-11%

prezmek: what is the ambient temperature range where the circuit will be operating?

ak
Thats with known accurate RC elements, just the chip, fixed known supply and fixed T, right ?

Otherwise its probably >> 50%.....


Regards, Dana.
 

Thread Starter

przemek64

Joined Dec 11, 2019
2
Hi there,

Many thanks for all replies.

A watchdog... looks like I just didn't know what I was building... :)

I couldn't source LTC6995 without ~20 USD DHL/UPS charge to Ireland.
It's a bit crazy here. If Farnell, RS-Online or TME doesn't have it, you need to add nuts shipping charge to even cheapest components or wait weeks for a delivery from China.

Maxim chips don't have delays long enough so I ended up with TPL5010 which does exactly what I want.

In case someone needs something similar in the future:

1576584926218.png

I think it would work without that MOSFET but I had an impression TPL5010 wasn't triggered every time so I left it there.

Dana's solution sounds interesting. I've never used such things before.
What chip would you use?
I would give it a try when life gets less busy (if ever).

Cheers,
Przemek
 

danadak

Joined Mar 10, 2018
4,057
Top