Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

mixed-signal PCB design questions

Status
Not open for further replies.

reedlaw

Newbie level 1
Joined
Jun 4, 2010
Messages
1
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Activity points
1,320
I've designed a board with analog and digital signals where I've tried to isolate the analog section. I've read about the benefits of keeping analog and digital ground on the same plane. One question I have is should the analog ground be tied directly to the power supply ground, or have to go back through the ferrite and power filters before reaching the power supply ground?

Here's my circuit:
**broken link removed**

As you can see the bright red section is analog ground. The power supply jack is in the bottom right. At this point, the analog ground is not connected directly to the power jack. I'm not very clear about what ferrite does to help keep the signal free from interference.

You can read more about it and download the Eagle files **broken link removed**.
 

I would connect the analog & digital grounds together at the power jack. You could do it with a link so you can split them and experiment later if you get problems.

Keith.
 

isolated analog and digital ground plane regions.
 

Ferrite bead act as just a low pass filter.It filters all the high frequency components (spikes)in either plane and doesnot allow to enter into other plane.

So it is advisable to use ferrite bead to connect the two ground planes


Thanks
Rajan.K
 

Hi Reedlow,
I believe, that your placement isnt optimal :-(...
The better solution is as the example PCB from pcbwing demonstrates it too: both GNDs are at the PWR-source/connection contacted(if you will apply to for connecting your ANGND tover a "link" to the prw-connector; then it will function as an antenna!-not so direct optimal..).
Of course is a ferite bead, as Rajan proposed too, to the connecting both GNDs surly a good idea, but (in my opinion) are the GNDs to be guided/placed into the immediate distance of the pwr-connector...
K.
 

I've read about the benefits of keeping analog and digital ground on the same plane. One question I have is should the analog ground be tied directly to the power supply ground

hi to all
I always try to keep analog grounds and digital ground apart by using two different power supply for each of them, by reading this topic comment it posed me a question that what are the benefits of joing analog ground and digital ground?
 

They need to be connected somewhere. You usually need to interface the analogue to the digital so they need a common ground for the digital signals. The skill is deciding where to connect the two and whether to make it a short circuit or include some resistance/inductance.

Keith.
 

keith1200rs said:
They need to be connected somewhere.

the analog grounds and digital ground are both connected to the corresponding power grounds . dose this method cause some problem?
 

As long as the corresponding power grounds are connected, that's fine. It would not be a good idea if the two power supplies were floating.

Keith.
 

I don't know your IC. some time you must read datasheet of IC. many IC will split ground under IC.
 

keith1200rs said:
As long as the corresponding power grounds are connected, that's fine.

.
I use two different transformers for each power supply and therefore i don't have power grounds connected to each other.


It would not be a good idea if the two power supplies were floating.
but still i don't understand why it's not a good idea to have two power supplies not connected. i have some idea about it and that is that separated grounds may cause coupling noise to each other due to the difference in the electromagnetic potential, is that the reason?
 

You must connect the grounds somewhere because the two circuits I presume are connected - digital and analogue. If you have an analogue to digital converter then the data lines to the digital logic will require signals relative to ground. The analogue input will be relative to ground (or it could be differential, but will still have an analogue ground that must be fixed). The ADC will not allow differences in voltage between the grounds except in special cases.

Keith.
 

Hi,
No one of usual ICs allows more as a diode voltage between both GNDs.
Its AN ABSLUTE MUST TO CONNECT both (or more) GNDS to others at ONE POINT_practically on ADC or DAC!!
Reason is technology_I think Keith can explain it more in detail why is your aspect/idea nt possible...
If you have higher voltages as mentioned U-diode, the parasitic elements will be open, the "tub/wan" will not functin and so on,
= you will kill your IC...
K.
 

keith1200rs said:
You must connect the grounds somewhere because the two circuits I presume are connected - digital and analogue. If you have an analogue to digital converter then the data lines to the digital logic will require signals relative to ground. The analogue input will be relative to ground (or it could be differential, but will still have an analogue ground that must be fixed). The ADC will not allow differences in voltage between the grounds except in special cases.

Keith.

my device has two part , RF circuits and digital circuits, the connection between these two circuits is a DAC chip (AD5363) which has itself two DGND and AGNDs. to isolate these two parts , i decided to use two power supply and i didn't connect the DGND and AGND together. because i wanted to avoid noises from digital part interfere with RF circuits. is that method cause problem?

Added after 1 hours:

karesz said:
Hi,
No one of usual ICs allows more as a diode voltage between both GNDs.
Its AN ABSLUTE MUST TO CONNECT both (or more) GNDS to others at ONE POINT_practically on ADC or DAC!!
Reason is technology_I think Keith can explain it more in detail why is your aspect/idea nt possible...
If you have higher voltages as mentioned U-diode, the parasitic elements will be open, the "tub/wan" will not functin and so on,
= you will kill your IC...
K.
thanks for the answer, i understand that. but is it good to totally separate two grounds? I want to have two AGND(RF Circuits) and DGND separate grounds, can i use Opto Isolator for isolating digital control pin of dac from RF part? in that case i suppose that i have 3 grounds DGND1(digital circuits consist of micro controller, motor drivers and..), DGND2 for DAC digital parts after OptoIsolator, and AGND that is the ground of RF and Analog circuits, DGND2 and AGND are tied together,
is that idea cause any problem?
 

If you opto isolate then it is a different situation, but you have a chip with Agnd & Dgnd so it is not isolated. Look at your DAC datasheet for layout guidance but you MUST connect the Agnd and Dgnd, usually under the DAC. If you don't you are relying on the DAC to do it with its substrate. That is not a good idea!

Keith
 

keith1200rs said:
If you opto isolate then it is a different situation

yes, my question is all about this,that is ," is it better to isolate DGND And AGND or separating these grounds and then tie them together with a ferrit?", is separate grounds useful or not?
 

You cannot optoisolate in your case because you have a chip with both analogue and digital grounds. You MUST connect them.

Read the DAC datasheet to find out the best place to connect them.

Keith
 

The siad place is practically in all cases at/below the IC=regularly are both GNDs not so far to the other(maybe the next pin), they needs a wide "short", a bridge_you dont like it, but you must it believe nba!...
K.
 

Hello all,

I have been working on a design for some time now that is used to amplify low level analog signals from CP (cathodic protection systems) and interface these signals with a PLC.

After some research in noise reduction between the Analog and Digital sections of the circuit, I settled on separating the AGND and DGND with a ferrite bead. Then, as pointed out http://www.hottconsultants.com/pdf_files/june2001pcd_mixedsignal.pdf I realized that I have several signals that have to cross the divide between the two worlds.

This leaves me questioning my design and reluctant to release it for manufacturing. It seems that I would be better off with a single ground plane and possibly separate power supplies. Although both sections are +12V, so again I am reluctant to add a second +12V regulator to the system as this would be costly both in dollar amount and physical layout area.

I would appreciate some input on my attached schematic.

Thanks
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top