PCB Schematic and layout review.

[LexLabs]

Greetings,  I'm new here, Glad to see so many people.

I designed this board to control AC motor using a TRIAC with a microcontroller. I've attached the Schematic and layout files. Would love to add more details if required.

Would love to have it reviewed and any sort of feedback would be much appreciated

[nuclearcat]

1)Can't see pull-ups for I2C.
2)gnd pour on low voltage part of pcb might be good idea.
3)Flywheel diode for relay is common practice, but worst one, causes relay to "lag". I can't recall exact article, i read in "art of electronics", its about relay "release time" on deenergize. Zener + diode might be way better. Another route is interesting "transistors" like NUD3105LT1G, they might do all this in one package (pick correct one for ur voltage!).
4)might add more when i get sober

[LexLabs]

1)Can't see pull-ups for I2C.
2)gnd pour on low voltage part of pcb might be good idea.
3)Flywheel diode for relay is common practice, but worst one, causes relay to "lag". I can't recall exact article, i read in "art of electronics", its about relay "release time" on deenergize. Zener + diode might be way better. Another route is interesting "transistors" like NUD3105LT1G, they might do all this in one package (pick correct one for ur voltage!).
4)might add more when i get sober

Thank you very much for commenting.  I Gotta add the Pull-ups for I2C.  The relay thing is interesting will definitely look up the same.  I was more concerned about the AC part and Noise suppression.

[nuclearcat]

Snubber circuit is separate magic, i guess heavily depends on your load. I hope some Triac expert will give him opinion here, as i'm curious about this too.

[LexLabs]

Snubber circuit is separate magic, i guess heavily depends on your load. I hope some Triac expert will give him opinion here, as i'm curious about this too.

Oh yeah!, Totally forgot about that. will definitely look into how to implement it. Thanks again

[S. Petrukhin]

Snubber circuit is separate magic, i guess heavily depends on your load. I hope some Triac expert will give him opinion here, as i'm curious about this too.

Oh yeah!, Totally forgot about that. will definitely look into how to implement it. Thanks again

A very simple zero detector will give you a significant phase shift. I do not know exactly PC817, but it will need at least 0.5mA. With a current-limiting resistor of 60k, even without taking into account the voltage drop on the 2 bridge diodes and on the opto-pair LED, the current of 0.5mA will be reached only at 30V, and this is far from zero.

[S. Petrukhin]

Another route is interesting "transistors" like NUD3105LT1G, they might do all this in one package (pick correct one for ur voltage!).

Personally, I do not like the built - in protective diode-it will merge the self-induction EMF into the supply circuit. All the same, a protective diode is needed at the relay terminals to kill the self-induction EMF on the spot.

The same thing, but with a lower level will be when using a pair of diode-zener diode. I do not think that the delay of switching off the relay is any noticeable and important, except in the harsh theory. 

[LexLabs]

Another route is interesting "transistors" like NUD3105LT1G, they might do all this in one package (pick correct one for ur voltage!).

Personally, I do not like the built - in protective diode-it will merge the self-induction EMF into the supply circuit. All the same, a protective diode is needed at the relay terminals to kill the self-induction EMF on the spot.

The same thing, but with a lower level will be when using a pair of diode-zener diode. I do not think that the delay of switching off the relay is any noticeable and important, except in the harsh theory. 

Thanks for the comment, So is like the diode-zener pair the best deal when trying to kill the EMF?

[sicco]

Is a 4 diode bridge rectifier on the zero-crossings detection the right choice here? Will that not get you twice the AC mains frequency instead of once? Plus as already mentioned above I think, the bridge circuit will have more phase uncertainties. Better I think would be one anti-parallel diode on the optocoupler LED, with the same tow resistors.
But maybe the 4 diodes bridge works for you - as the triac should trigger twice per AC mans cycle anyway. All depends on what code you've got in the uC.

[S. Petrukhin]

Another route is interesting "transistors" like NUD3105LT1G, they might do all this in one package (pick correct one for ur voltage!).

Personally, I do not like the built - in protective diode-it will merge the self-induction EMF into the supply circuit. All the same, a protective diode is needed at the relay terminals to kill the self-induction EMF on the spot.

The same thing, but with a lower level will be when using a pair of diode-zener diode. I do not think that the delay of switching off the relay is any noticeable and important, except in the harsh theory. 

Thanks for the comment, So is like the diode-zener pair the best deal when trying to kill the EMF?

Until the zener diode opens (it is needed to allow the relay to release self-induction and turns off quickly), the self-induction will merge into the power circuit with a diode built into the transistor - this does not make sense because it will give interference and it will not affect the speed of release - the self-induction will be shunted anyway. This was an option when a diode and a zener diode were connected in parallel to the coil.

But, if you turn on the diode in series with the coil, and the zener diode in parallel with the coil, the situation will change, but slightly. The rate of increase of self-induction is very high, it is much ahead of the beginning of the release of the armature, the diode will open long before the armature begins to move back. This is some strange theory with puffed-up cheeks for some special occasions. In 99.99%, you should not count on instant disconnection of the relay or acceleration of disconnection. And in 99.99% of relay usage, this is not necessary. 

[Shadowfire]

To put it another way:
The current in the relay coil is what is holding the contact on the activated position.
When you turn off the drive voltage, any energy stored in the coil (1/2*L*I*I) continues to cause current to flow through the snubber (and the coil).
This energy isn't immediately destroyed, but you do need to get rid of it to cause the current in the coil to fall (to below the contact holding voltage).
If you are using a typical silicon snubber diode with a forward voltage of around 0.8V at the operating current of the coil, your power dissipated in the diode is I*0.8V
When you insert a series zener diode, the power dissipated is I*(0.8+Bvzener), which causes the current to plummet faster (and thus turn off faster).
The tradeoff here, is that now you are dissipating more heat in the zener diode, and also you are presenting a more negative voltage to whatever is driving the relay coil.
There's nothing wrong with using just a regular diode to suppress the back EMF, as long as the relay turnoff time is adequate for your application.

[nuclearcat]

Another route is interesting "transistors" like NUD3105LT1G, they might do all this in one package (pick correct one for ur voltage!).

Personally, I do not like the built - in protective diode-it will merge the self-induction EMF into the supply circuit. All the same, a protective diode is needed at the relay terminals to kill the self-induction EMF on the spot.

The same thing, but with a lower level will be when using a pair of diode-zener diode. I do not think that the delay of switching off the relay is any noticeable and important, except in the harsh theory. 

Thanks for the comment, So is like the diode-zener pair the best deal when trying to kill the EMF?

Until the zener diode opens (it is needed to allow the relay to release self-induction and turns off quickly), the self-induction will merge into the power circuit with a diode built into the transistor - this does not make sense because it will give interference and it will not affect the speed of release - the self-induction will be shunted anyway. This was an option when a diode and a zener diode were connected in parallel to the coil.

But, if you turn on the diode in series with the coil, and the zener diode in parallel with the coil, the situation will change, but slightly. The rate of increase of self-induction is very high, it is much ahead of the beginning of the release of the armature, the diode will open long before the armature begins to move back. This is some strange theory with puffed-up cheeks for some special occasions. In 99.99%, you should not count on instant disconnection of the relay or acceleration of disconnection. And in 99.99% of relay usage, this is not necessary. 

Thats very interesting and broad subject to discuss. I tried to search about longevity of typical relays with different flywheel/kickback supression circuits, and not able to find any.
Appeal to circuit with zener and diode i saw in "Art of electronics" and here
https://www.te.com/commerce/DocumentDelivery/DDEController?Action=srchrtrv&DocNm=13C3311_AppNote&DocType=CS&DocLang=EN
I have to admit you are right, it is easier to put diode, otherwise it is required to simulate specific relay to see how much it affects VCC on kickback with zener supression, and to handle that properly.

[LexLabs]

Is a 4 diode bridge rectifier on the zero-crossings detection the right choice here? Will that not get you twice the AC mains frequency instead of once? Plus as already mentioned above I think, the bridge circuit will have more phase uncertainties. Better I think would be one anti-parallel diode on the optocoupler LED, with the same tow resistors.
But maybe the 4 diodes bridge works for you - as the triac should trigger twice per AC mans cycle anyway. All depends on what code you've got in the uC.

Hi, Thanks for your reply.  Yes, it is true that the bridge rectifier used for Zero Crossings detection would output twice per wave.  But, you've answered that yourself as well, the application demands two signals per wave (Starting point and halfway point), The uC is programmed as such.  I still have to study the phase implications the bridge rectifier has on the circuit.  The anti-parallel diode on the optocoupler seems like a cool idea, will definitely use when required.  Thanks for the pointers

[LexLabs]

Another route is interesting "transistors" like NUD3105LT1G, they might do all this in one package (pick correct one for ur voltage!).

Personally, I do not like the built - in protective diode-it will merge the self-induction EMF into the supply circuit. All the same, a protective diode is needed at the relay terminals to kill the self-induction EMF on the spot.

The same thing, but with a lower level will be when using a pair of diode-zener diode. I do not think that the delay of switching off the relay is any noticeable and important, except in the harsh theory. 

Thanks for the comment, So is like the diode-zener pair the best deal when trying to kill the EMF?

Until the zener diode opens (it is needed to allow the relay to release self-induction and turns off quickly), the self-induction will merge into the power circuit with a diode built into the transistor - this does not make sense because it will give interference and it will not affect the speed of release - the self-induction will be shunted anyway. This was an option when a diode and a zener diode were connected in parallel to the coil.

But, if you turn on the diode in series with the coil, and the zener diode in parallel with the coil, the situation will change, but slightly. The rate of increase of self-induction is very high, it is much ahead of the beginning of the release of the armature, the diode will open long before the armature begins to move back. This is some strange theory with puffed-up cheeks for some special occasions. In 99.99%, you should not count on instant disconnection of the relay or acceleration of disconnection. And in 99.99% of relay usage, this is not necessary. 

Oh yeah, I was thinking the same when going through the theory.  I guess it is very peculiar and is probably not needed for regular usage except for when the turn off time is critical for the application.  Thank you for taking the time to explain. _/\_

[LexLabs]

To put it another way:
The current in the relay coil is what is holding the contact on the activated position.
When you turn off the drive voltage, any energy stored in the coil (1/2*L*I*I) continues to cause current to flow through the snubber (and the coil).
This energy isn't immediately destroyed, but you do need to get rid of it to cause the current in the coil to fall (to below the contact holding voltage).
If you are using a typical silicon snubber diode with a forward voltage of around 0.8V at the operating current of the coil, your power dissipated in the diode is I*0.8V
When you insert a series zener diode, the power dissipated is I*(0.8+Bvzener), which causes the current to plummet faster (and thus turn off faster).
The tradeoff here, is that now you are dissipating more heat in the zener diode, and also you are presenting a more negative voltage to whatever is driving the relay coil.
There's nothing wrong with using just a regular diode to suppress the back EMF, as long as the relay turnoff time is adequate for your application.

Thank you for the reply, Thanks again for accurately pointing out the tradeoffs between each of these component usage.  It makes more sense to me now.

[LexLabs]

Another route is interesting "transistors" like NUD3105LT1G, they might do all this in one package (pick correct one for ur voltage!).

Personally, I do not like the built - in protective diode-it will merge the self-induction EMF into the supply circuit. All the same, a protective diode is needed at the relay terminals to kill the self-induction EMF on the spot.

The same thing, but with a lower level will be when using a pair of diode-zener diode. I do not think that the delay of switching off the relay is any noticeable and important, except in the harsh theory. 

Thanks for the comment, So is like the diode-zener pair the best deal when trying to kill the EMF?

Until the zener diode opens (it is needed to allow the relay to release self-induction and turns off quickly), the self-induction will merge into the power circuit with a diode built into the transistor - this does not make sense because it will give interference and it will not affect the speed of release - the self-induction will be shunted anyway. This was an option when a diode and a zener diode were connected in parallel to the coil.

But, if you turn on the diode in series with the coil, and the zener diode in parallel with the coil, the situation will change, but slightly. The rate of increase of self-induction is very high, it is much ahead of the beginning of the release of the armature, the diode will open long before the armature begins to move back. This is some strange theory with puffed-up cheeks for some special occasions. In 99.99%, you should not count on instant disconnection of the relay or acceleration of disconnection. And in 99.99% of relay usage, this is not necessary. 

Thats very interesting and broad subject to discuss. I tried to search about longevity of typical relays with different flywheel/kickback supression circuits, and not able to find any.
Appeal to circuit with zener and diode i saw in "Art of electronics" and here
https://www.te.com/commerce/DocumentDelivery/DDEController?Action=srchrtrv&DocNm=13C3311_AppNote&DocType=CS&DocLang=EN
I have to admit you are right, it is easier to put diode, otherwise it is required to simulate specific relay to see how much it affects VCC on kickback with zener supression, and to handle that properly.

Hi nuclearcat, Thanks for the reply.  I went through 'The art of electronics' after you've pointed out and found the same concern and I come to find that 'RC snubbers' are equally effective, and I also guess another user 'Shadowfire' pointed out the tradeoffs between the usage of both the components. I now have better understanding of the topic.  Thank you very much. _/\_