Resetting a slave from master mcu

[Demon]

See post #22:

https://www.picbasic.co.uk/forum/sho...323#post156323

A SN7407 Buffer works great!

--------------------------------------------------------------------------------

I have a reset button on the master PIC, and I'd like to reset all the slaves internally. This is what I have so far:




I found formulas to calculate the base resistor, like this one:

https://breadboardcircuits.com/how-transistors-work/


But they're dependent on preliminary calculations done for the load current, and I have no clue how that is calculated with a PIC pin. I see several current loads in the datasheet for the 16F1933, but no clue which is the appropriate one to use in the formulas.

[Ioannis]

The main Load Current for the PNP transistor will come from the 10K, R5 resistor, that goes to +Vdd and not from the PIC /MCLR pin. Do not worry about that.

Ioannis

[Demon]

I had gotten that circuit from "somewhere", but I can't find any more. So, I'm looking at this guy:

https://www.electro-tech-online.com/...d-reset.17256/

He's using a NPN transistor, and I notice he's missing a pull-down on the base of the NPN.

I added labels to make it clear what attaches where, and I end up with this:




Do you see a problem if I have the programmer on the slave PIC at the same time as the pin from master PIC is connected?

I can always unconnect everything when I program, but I know "one day", I'm gonna rush and program with everything connected.

[Ioannis]

The base of the transistor is grounded through R6. The other end of R6 is Vss, meaning ground.

To Reset the system you have to feed high to the transistor in this circuit.

I am a bit worry about the collector that is connected to a 100 ohm and this to the programmer. In programming, there might be 13 or more volts. At this time you have to be sure that the transistor will be in OFF state, that is [/MCLR Slave from master PIC] be low. If you can be sure of this then I see no problem.

Ioannis

[Demon]

Yeah, this seems so sketchy to me. I'm thinking it would be safer to be able to connect only one at a time; either the programmer, or the master PIC.

[Demon]

I ended up with this design, using tri-state pin on master to reset all the slaves when required. This will not be used for programming the slaves, only resets.




It's simple, uses a single general I/O pin on master, and no extra components on slaves.

I include that 100R res on slaves "just in case" I need them one day with my Pik-Kit.


EDIT: For those that want to ICSP as well, look at the SN7407, and thank Mr E for the info (master on Input A, slave on Output Y)

[Ioannis]

What drives the MCLR pin of the slaves? A PIC output?

I am asking, because the MCLR pin needs a level of 0,2*Vdd to see LOW (for the 16F177x series, but I guess it is the same for others).

It depends on output characteristics if you are within limits of the MCLR. So be sure that what you think as low is actually sensed as low on the MCLR pins.

Ioannis

[amgen]

you could do in software..... send out to one or all listeners a command to goto a restart routine like.... which "provides a way to execute a MCLR in software"
,
,
,
restarting:
@ RESET

[Demon]

you could do in software..... send out to one or all listeners a command to goto a restart routine like.... which "provides a way to execute a MCLR in software"
,
,
,
restarting:
@ RESET

But will that work if the slave is caught in a loop or some such?

I'm aiming for a complete reset with no decision from the slave.

[Demon]

What drives the MCLR pin of the slaves? A PIC output? ...

A tri-state general I/O pin on the master.

Output low = reset
Input = do nothing

...

I am asking, because the MCLR pin needs a level of 0,2*Vdd to see LOW (for the 16F177x series, but I guess it is the same for others).

It depends on output characteristics if you are within limits of the MCLR. So be sure that what you think as low is actually sensed as low on the MCLR pins.

Ioannis

I can't say. This is all theory for now, but is supported by posts on google and a forum elder, Mr E.

So I blame him if my house burns down.

[amgen]

But will that work if the slave is caught in a loop or some such?

I'm aiming for a complete reset with no decision from the slave.

could do.........
1)
'
restarting:
pause 1000 ............... wait a second or 2 to reset all
@ RESET
'
or 2)................
if all slaves are on same 5 or 3 volt supply, then use a small NC (normally closed) relay to turn off power for few seconds to restart all slaves

[Demon]

I figured out an easier way to MCLR a PIC from another; a simple diode.




I was fixated on that darn pull-up to VDD:

- pull-down MCLR on Slave.
- 1N4001 diode from GP pin on Master.
- 100R towards ICSP.

And that's it. I just tested it, and there's never more than 5V on Master-side of diode.

I can't believe it was so simple.


I have one last test to do; check how much current is running from Master.

EDIT: I see a quick pulse of 16-17mA when I'm programming via ICSP. I'm not sure how I can manage that.

[amgen]

MCLR's have internal weak pullups in chip.....

[Demon]

What diode are you using? Did you put it on backwards?

Like Ioannis said, ICSP runs a whole bunch of volts through that line.

[Demon]

I got my better multimeter and got:

- 15mA surge
- 0.43mA continuous

The surge is just that, an instantaneous burst. Considering I/O pins on the 16F18877 are rated for absolute maximum 50mA, and most PICs are rated for 2-25mA, I don't think a burst of 15 or a few more mA is something to be worried about.

[Demon]

could do.........
1)
'
restarting:
pause 1000 ............... wait a second or 2 to reset all
@ RESET
'
or 2)................
if all slaves are on same 5 or 3 volt supply, then use a small NC (normally closed) relay to turn off power for few seconds to restart all slaves

What is the @ RESET command? (I'm using PBP3.1)

I prefer to avoid relays and mechanical devices. Contacts generally consume more current than ICs and discretes, and contacts wear out.

I've got an order for some SN7407N from Digikey; gonna check out Mr E's suggestion.

[amgen]

diodes can be like signal diodes 1N4148. Not backwards in order to pull MCLR low to reset and that is considering you do LOW VOLTAGE PROGRAMMING setting, if not doing that then would use the 2n3906 or any NPN to pull down all the slaves through the diodes. The @ RESET is an assembly instruction using the @ for PBP to do machine code directly ( I think you know all that stuff)...... so if you want a hard reset after ICSP then yes to use MCLR but if chips are running, you could give instruction to GOTO reset.

[Demon]

... diodes can be like signal diodes 1N4148. Not backwards in order to pull MCLR low to reset and that is considering you do LOW VOLTAGE PROGRAMMING setting...

Nah, I'm doing regular ICSP programming for now, but I'm considering using ZIF socket adapter for programming, and then installing the PICs on IC sockets.

The nice thing with IC sockets, I can't damage the PIC while soldering. A dab of hot glue at both ends should stop the PIC from falling out.

... The @ RESET is an assembly instruction using the @ for PBP to do machine code directly ( I think you know all that stuff). ...

I graduated back in '83, was 2nd in my assembler classes. But I've never used it in my professional career as mainframe programmer.

... if not doing that then would use the 2n3906 or any NPN to pull down all the slaves through the diodes.

Can you explain this like I'm 5? I looked at NPN and PNP transistors but I can't figure out how to do that.

[amgen]

crap, I figured you were a kid ..... I always put the pics in sockets, saves a lot of F-ups. For transistor, logic level MOSFET is good to use. The logic level means it will be full on with 5 volts on gate, no resistors required.... N-channel like BS170 ... with this, the reset is a high pulse on master...(dah)

[richard]

I'm doing regular ICSP programming for now

any pic programmer worth its salt can program chips that support LVP in "regular ICSP" mode without that nasty hv pulse being needed at all

[Demon]

any pic programmer worth its salt can program chips that support LVP in "regular ICSP" mode without that nasty hv pulse being needed at all

Low-voltage programming is not used by melabs programmers and should always be disabled in the configuration settings.

https://melabs.com/support/icsp.htm


I'm gonna take my PicKit4 out of storage and give it another shot.

[Demon]

crap, I figured you were a kid .....

I'm a kid at heart, and I get out of bed at the same speed as them.

..... I always put the pics in sockets, saves a lot of F-ups....

Yeah, I'd hate to notice a mistake at the end of assembling my boards.

... For transistor, logic level MOSFET is good to use. The logic level means it will be full on with 5 volts on gate, no resistors required.... N-channel like BS170 ... with this, the reset is a high pulse on master...(dah)

I finally got a few SN7407 thanks to Mr E's suggestion a while back. This runs awesome, and I have room for 6 slaves, that should be enough for my boards.




I don't have to worry about the surges up towards the Master during programming the Slave.

(Which is good, cause my MeLabs U2 programmer doesn't do LVP, and I can't get my PicKit4 to run properly).

[richard]

I don't have to worry about the surges up towards the Master during programming the Slave.

No, but the 120 ohm pullup will be a pin killer if the vdd is 5v

[Demon]

No, but the 120 ohm pullup will be a pin killer if the vdd is 5v

Yeah, I was blindly following SN7407 datasheet. I don't really care about timing, so I upped it to 1K pull-up.

[richard]

Yeah, I was blindly following SN7407 datasheet. I don't really care about timing, so I upped it to 1K pull-up.

given that I-inh max for a SN7407 gate is 40uA and a pic output pin could drive more than 20 of them and still produce a clean TTL signal.
i cannot see that signal timing would even be a consideration. just what need does the pullup fulfill ?

[Demon]

given that I-inh max for a SN7407 gate is 40uA and a pic output pin could drive more than 20 of them and still produce a clean TTL signal.
i cannot see that signal timing would even be a consideration. just what need does the pullup fulfill ?

Except from preventing the input pin on the SN7407 from floating, I don't see one for my application.

I don't really care about "precision timing"; I just trigger the MCLR reset on the Slave with a 10usec pause. I still have a pull-up downstream on the Slave PIC.

SlaveMCLR = 0
PauseUS 10
SlaveMCLR = 1

EDIT: I figured I'd read the instructions AFTER posting that reply.

11.1 Layout Guidelines: ... All unused inputs of digital logic devices must be connected to a high or low bias to prevent them from floating.

Promptly runs over to breadboard to pull down the other 5 unused inputs.