To ground plane or not to ground plane

Hello All -

I’m designing my first PCB (2 layer) based off of this schematic. http://home.comcast.net/~houta69/M1-R.bmp

My background is in software and my hardware skills leave a lot to be desired. One thing that I’m very confussed about is the proper grounding to use on the board. I’ve seen where the ground rail is provided with a big fat trace running around the perimeter of the board, others recommend providing a ground plane. I’ve seen recomendations for seperate ground planes that tie back together at the supply if RF or mixed analog and digital circuits are used. I’ve also read arguments for using a single ground plane for the entire board regardless of the circuitry involved.

I was hoping that someone might shed a little additional light on this topic for me. Are there general rules of thumb for which techniques should be used in a given situation or is every situation completely unique?

This is also my first CAD schematic diagram, so any critiquing of that is welcomed too.

Thanks much

I’d just use fat supply tracks, if it’s a prototype for your own use. If it is something that is going into production, you might need to use copper pour and take other measures to avoid EMC problems. The term ‘ground-plane’ is usually reserved for multi-layer boards.

Spend some time getting the components placed optimally. Route critical nets like the supply and ground, and oscillator, first.

Leon

Why do you have your LM1117 taking as its input voltage the output of the 7805? Such a setup can lead to badness like weird oscillations. With your board - I don’t see any need for that setup.

As to the topic of this thread - I’d go with a solid ground plane on one layer of the PCB if you’re having this board professionally made. There really is no reason not to - it’s a simple enough board that the layout should be easy to keep all on one layer.

If you were doing RF stuff on the board itself maybe you’d have to worry more about separate ground planes - but all the RF is handled within the RF module you’re using, so it’s not a big deal.

Thanks for taking the time to respond.

Why do you have your LM1117 taking as its input voltage the output of the 7805? Such a setup can lead to badness like weird oscillations. With your board - I don’t see any need for that setup.

The RF receiver runs at 5V while the MCU runs at 3.3V so I need to have two supply levels on the board. That’s also why I’m using the voltage divider on the RF data line that feeds into the MCU. As I mentioned previously, my harware skillset is still pretty poor. How should I supply both 5V and 3.3V to the board? I thought that supplying the 3.3V LDO (it’s actually an AME1117) from the LM7805 was the way to do that. I don’t need to have two separate supplies feeding the regulators do I?

Thanks again.

Using two regulators in series like that can be useful, as it reduces the dissipation of the second one. Provided you use the recommended capacitors you won’t have problems.

Leon

Appreciate all the info Leon.

houta69:
Thanks for taking the time to respond.

Why do you have your LM1117 taking as its input voltage the output of the 7805? Such a setup can lead to badness like weird oscillations. With your board - I don’t see any need for that setup.

The RF receiver runs at 5V while the MCU runs at 3.3V so I need to have two supply levels on the board. That’s also why I’m using the voltage divider on the RF data line that feeds into the MCU. As I mentioned previously, my harware skillset is still pretty poor. How should I supply both 5V and 3.3V to the board? I thought that supplying the 3.3V LDO (it’s actually an AME1117) from the LM7805 was the way to do that. I don’t need to have two separate supplies feeding the regulators do I?

Thanks again.

I understand why you want two separate supplies. My point is that I see no reason to run the 3.3V supply off of the 5V supply. It is very poor design practice and is something that should only be done when absolutely needed. As Leon says it can be done to limit power dissipation in a single part, but it also can cause nasty noise and other shenanigans that you don't want to deal with.

To be clear: the change I’m saying you need to make is have both the LM1117 and the 7805 use the raw 12V supply as their input voltage.

Why should it cause noise etc? I often use a regulated bench power supply or a wall-mounted supply which contains a regulator to power a PCB with a regulator on it, as do most people, and that doesn’t cause any problems.

Leon

Assuming you use the proper smoothing capacitors recommended by the datasheet, I can’t imagine there being any problem with multiple regulators in series like that.

I’ve used regulators in series many times without any problem, even in an RF (cellular) application. Although some regulators might conflict, regulators such as the 78xx or LD1805Vxx don’t seem to have problems. Leon also brings up a good point about benchtop power supplies using regulators. People put regulators in series all the time then and there’s no problem.

As far as ground planes, I usually have one on the bottom side of my PCBs (but I don’t do much high frequency stuff). To get technical, it can add a small amount of capacitance to nearby traces. That can be a good or bad thing (good - decoupling, bad - parasitic capacitance). If you’re doing RF stuff, most designs have restrictions on where you should put the ground plane, so double check the reference designs first (if they exist).

Also, if you’re not going to have a soldermask on your board, keep in mind that it’s real easy to bridge pads to the ground plane when you’re soldering if your trace clearance is too small.

You want separate caps from v+ to gnd and v- to gnd on the sp3232.

/mike

Andrew02E:
I’ve used regulators in series many times without any problem, even in an RF (cellular) application. Although some regulators might conflict, regulators such as the 78xx or LD1805Vxx don’t seem to have problems. Leon also brings up a good point about benchtop power supplies using regulators. People put regulators in series all the time then and there’s no problem.

As far as ground planes, I usually have one on the bottom side of my PCBs (but I don’t do much high frequency stuff). To get technical, it can add a small amount of capacitance to nearby traces. That can be a good or bad thing (good - decoupling, bad - parasitic capacitance). If you’re doing RF stuff, most designs have restrictions on where you should put the ground plane, so double check the reference designs first (if they exist).

Also, if you’re not going to have a soldermask on your board, keep in mind that it’s real easy to bridge pads to the ground plane when you’re soldering if your trace clearance is too small.

The point is... sometimes as you say they *do* conflict. There is no need for them to be in series on this board - it offers no advantages to him as power dissipation does not look like it'll be an issue.

One doesn’t have a choice with devices such as the Telit GE-863 GPRS/GPS module. It runs off 3.7V which can be supplied from a regulator, and it has a built-in regulator supplying 2.8V for auxiliary circuitry.

Leon

Thanks to all for the feedback. This was more than I had anticipated.

leon_heller:
One doesn’t have a choice with devices such as the Telit GE-863 GPRS/GPS module. It runs off 3.7V which can be supplied from a regulator, and it has a built-in regulator supplying 2.8V for auxiliary circuitry.

My MCU does the same thing. It takes VDD from the LM1117 and runs it through an on-chip regulator to drop it down to 2.5V from which it runs the core. I do not say this to dismiss NleahciM in any way, but it does make me feel a little less embarrassed about my ignorance when I see that there are points of contention even amongst the experts.

houta69:
Thanks to all for the feedback. This was more than I had anticipated.

leon_heller:
One doesn’t have a choice with devices such as the Telit GE-863 GPRS/GPS module. It runs off 3.7V which can be supplied from a regulator, and it has a built-in regulator supplying 2.8V for auxiliary circuitry.

My MCU does the same thing. It takes VDD from the LM1117 and runs it through an on-chip regulator to drop it down to 2.5V from which it runs the core. I do not say this to dismiss NleahciM in any way, but it does make me feel a little less embarrassed about my ignorance when I see that there are points of contention even amongst the experts.

I think the problem is that you're assuming everybody knows what they're talking about.

I think the problem is that you’re assuming everybody knows what they’re talking about.

This is true to some extent. I know that I don’t chime in on something unless I felt confident in what I am saying (bad information is worse than no information), so I assume that most others behave accordingly (at least in technical forums such as this).

I don’t have any insight into the members on this site, but I am fimilar with Leon from the microchip forums and I’ve never seen him steer anyone wrong. Again, this doesn’t mean that I discount what you are telling me (I’m “assuming” that you’re knowledgeable), this just tells me that I need to do more homework and now I am a little better prepared to do that.

Thanks

I have the exact situation … one of my components likes 2.8V, the other likes 3.8V. There is really no way around it. the board works fine.

there will be times that ‘yes’ the regulators will conflict…

but doing precautions and using caps around them usually work, all the time…

a lot of IC will be doing an internal regulation too (to somewhere around 2.++ volts)…

even industrial and commercial PSU’s do use regulators in series with pre-regulators and other regulators…

it should just be noted that the load of the regulator is also a regulator in which the control should be able to compensate…

on the ground plane, try reading this…

http://www.hottconsultants.com/pdf_file … signal.pdf

http://www.hottconsultants.com/pdf_file … _plane.pdf

using planes or using fat traces won’t be of any good effect if layout is not taken into consideration…when doing layout always try to look at the current return path of each signal…it should be as close as possible to the signal path so loop areas are minimized…the proper use of ground plane instantly makes the loop area almost zero…

then always remember to put a dividing line where the RF, Analog, Digital, and Power sections of your are…you may use a single ground plane for all of them but you can never put them exactly where the other ones are…

one more thing, don’t use 90 degree corners on RF and high speed signals…a circular corner is ideal but 45 degree corners are sufficient…

good luck on your first CAD… :wink:

rdpzycho:
there will be times that ‘yes’ the regulators will conflict…

but doing precautions and using caps around them usually work, all the time…

a lot of IC will be doing an internal regulation too (to somewhere around 2.++ volts)…

even industrial and commercial PSU’s do use regulators in series with pre-regulators and other regulators…

it should just be noted that the load of the regulator is also a regulator in which the control should be able to compensate…

I think everybody is missing the point of what I was saying: there is no advantage to him on this board to put the two regulators inline with each other. There are, however, potential disadvantages. We are not talking about every single possible situation. We are talking about this one dude’s PCB.

NleahciM:
I think everybody is missing the point of what I was saying: there is no advantage to him on this board to put the two regulators inline with each other. There are, however, potential disadvantages. We are not talking about every single possible situation. We are talking about this one dude’s PCB.

He needs 5VDC for the RF device and 3.3VDC for the PIC.

Instead of saying “there is no advantage to him doing xyz”, and ending your paragraph, why not suggest the possible alternative so everybody knows what you are talking about?

The alternative would be to put both regulators in parallel, as both his selected regulators can handle 12VDC.

6 of 1, half-dozen of the other. You could do it either way.