Hello guys, I have reading some documentation in internet and some people recommend to use two different ground planes (AGND + DGND) but other people recommend to use just one good GND plane with the digital part and the analog part very well placed and differenced. In my last designs I have been using only one ground plane and I placed the analog part and the digital part in the right way.And I got good results (I follow the next TI paper: http://www.ti.com/lit/an/scaa082/scaa082.pdf), However, I would like to know which are your experience? Thanks too much.
I use always separate planes and connect them only on a single point (net-tie). Since planes are also resistors and inductors, like every wire. So digital current should not flow during the analog area.
Gast
#4093597
It depends on so much things. Which digital frequency range you use? How about disturbances on your ground plane because of connections with other devices? (Connection with PC etc.) Best things you can employ: - proper layout is paramount! => keep trace lengths as short as possible => keep sensible / high frequency current flows in the smallest possible loop area => use series resistance in your digital output lines to reduce slew rate of your signals => Place high speed and low level analog as far away as possible => shield digital signals with surrounding ground lines => decouple DC with ferrite beads and special low-impedance ceramic capacitors For low level / high accuracy analog measurements (ADC > 14 Bits) it might be best practice to keep analog & digital ground (and power supply!)seperated and use optocouplers in between the domains, but this introduces time delays of 20ns to µs-seconds.
Hi! We do use only single ground planes over a complete layer in our high speed designs (> 1 GHz) at work. And it works well. Having split or interrupted planes always gives you impedance changes at the gap and therefore the signal quality goes down. And as you wrote: One single ground plane worked well in your last design, so why change a running system? For ripple-pickups in audio circuits, it is a different story. Cheers, Carsten
Carsten Wille schrieb: > For ripple-pickups in audio circuits, it is a different story. I assumed, exact this was the question. Many Notebooks, MP3-Players and so one have such problems with audible noise because wrong decoupling of the analog GND/VCC and digital signals to the analog circuits.
Peter Dannegger schrieb: > Carsten Wille schrieb: >> For ripple-pickups in audio circuits, it is a different story. > > I assumed, exact this was the question. > Many Notebooks, MP3-Players and so one have such problems with audible > noise because wrong decoupling of the analog GND/VCC and digital signals > to the analog circuits. According to the quoted datasheet, it is all about high speed design, not audio.
Gast
#4094357
mmmm I see that there are different opnions respecting to this issue. I would like to know if there is some kind of gold rule regarding to the ground planes which is valid for all kind of designs: video, audio, etc. For example, in theory IPHONE(Apple) is a good product and in therory it is an example to follow respecting to the PCB design..., Anybody knows how is the ground plane here?
Enrique schrieb: > mmmm I see that there are different opnions respecting to this issue. I > would like to know if there is some kind of gold rule regarding to the > ground planes which is valid for all kind of designs: video, audio, etc. > For example, in theory IPHONE(Apple) is a good product and in therory it > is an example to follow respecting to the PCB design..., Anybody knows > how is the ground plane here? Good question... The iPhone 4 uses a 10 layer PCB http://www.eetimes.com/document.asp?doc_id=1259324 and the tear down shows that the audio codec is more or less beneath the CPU https://de.ifixit.com/Teardown/iPhone+4+Teardown/3130 For the iPhone 6 https://de.ifixit.com/Teardown/iPhone+6+Teardown/29213
Gast
#4094760
First look at the current flows.
Antwort schreiben
Bitte melde dich an, um einen Beitrag zu schreiben.