Do you really tie Analog GND and Digital GND at pin chip level in DAC/Receivers

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
This is the general suggestion of TI, ADI and Maxim for low digital current chip, like DAC and Digital receivers with separate AGND and DGND.
(Not suggested for high digital current devices like DSP)

Don't want to debate the articles, but I would like to know if you really do in this way (join GNDs at chip pin level) with DAC and receivers (like CS84xx) even if you have separate Analog/Digital Power supply regulators (or even separate rectifiers and transformers).

http://www.hit.bme.hu/~papay/edu/Acrobat/GndADCs.pdf

https://www.analog.com/media/en/training-seminars/tutorials/MT-031.pdf?doc=cn288.pdf

http://www.ti.com/lit/an/slyt499/slyt499.pdf
 
Last edited:
Don't want to debate the articles, but I would like to know if you really do in this way (join GNDs at chip pin level) with DAC and receivers (like CS84xx) even if you have separate Analog/Digital Power supply regulators (or even separate rectifiers and transformers).

Yes. Both grounds are needed by, say, a dac chip, and they shouldn't be at different voltages at the chip so they need to be tied together there. Also, it often does have some beneficial effect on sound quality to use multiple regulators and or power supplies.
 
I am also pondering this for my latest dac pcb.

Why isn't it that the noisier digital ground (from ground bounce) will pollute the clean analogue ground?

I would have thought it best to keep them separate up to the 0v pcb connection to the psu, or even connect the two grounds direct to the psu separately.

Analogue Devices AD1862 dac datasheet supports this by referring to analogue ground common and digital ground common points and they connect at one point only:-

GROUNDING RECOMMENDATIONS The AD1862 has two ground pins, designated analog ground (AGND) and digital ground (DGND). The analog ground pin is the “high-quality” ground reference for the device. The analog ground pin should be connected to the analog common point in the system. The reference bypass capacitor, the noninverting terminal of the current-to-voltage conversion op amp, and any output loads should be connected to this point. The digital ground pin returns ground current from the digital logic portions of the AD1862 circuitry. This pin should be connected to the digital common point in the system. As illustrated in Figure 7, AGND and DGND should be connected together at one point in the system.
 
quoting Douglas Self in his book Small signal audio design:
Separate analog and digital ground planes should be used. Most conversion ICs have their analog and digital interfaces at opposite ends or opposite sides of the package, facilitating the use of separate ground planes. It is usually best to run the analog ground plane under the IC to minimize the coupling of digital noise. The two ground planes must of course be connected together at some point, and this should be implemented by a single junction close to the IC.

Also, about Analog Devices, yes, have a look at MT031 tutorial (Grounding Data Converters and Solving the Mystery of "AGND" and "DGND" )
 
Last edited:
I am also pondering this for my latest dac pcb. <snip>

At some point, inside the DAC chip, the digital part must interface with the analog side of the chip. If, for example, you have separate ground planes and join them physically away from the DAC chip, then at said interface point, there will be noise between the grounds which means that noise will be in the signal.

The grounds must be joined at the DAC.
 
That makes sense. So what about the other components connected to analogue ground? Should these all be connected to the dac analogue ground pin by traces and then the analogue and digital ground pins be joined at the dac. Or, can everything simply be connected to the ground planes at the component and dac ground pins? (ie there are no ground traces)
 
Regarding ground planes, I have only been using one, but the ground return connection going back to the digital power supply is at the side of the digital end of the dac board. The analog ground return is down at the other end where the output stage is. If you look at the dac board and try to visualize where ground currents have to be flowing, mostly they are pretty much out of each other's ways. That means digital ground currents aren't modulating the ground voltage in the analog circuitry and vice versa (at least to a reasonable approximation). Of course, the digital and analog grounds are only tied together at the ground plane.
 
Last edited:
"Separate analog and digital ground planes should be used... The two ground planes must of course be connected together by a single junction close to the IC. "

So, this seems to be supported by the majority of DIY hobbysts.

Professional manufacturers seems to be in partial disagreement, as it looks that the DAC chip should be connected exclusively to the Analog ground plane. :confused:

See fig.5 in MT-031

And what about digital receivers like CS8412 (I'm still using it.. :nod:). Single junction close to the IC too (?) , and to which ground plane Analog o Digital?
 
Multiple ground planes can get expensive, and may not always be needed. In some cases one ground plane may be preferred over two. In other words, it depends. However, for our purposes here in most cases you should use at least one ground plane. If possible try to think about were ground currents would be flowing and try to be sensible about layout to help minimize any ground noise issues. Designing PCBs is not always a one-shot process. Sometimes more than one design iteration might be needed if any unexpected issues are found during testing. I would suggest trying with one ground plane first. Test the boards you get and make sure the circuit works as expected. If you have problems you can always post here and we will try to help. Most likely, it will be okay.
 
"Separate analog and digital ground planes should be used... The two ground planes must of course be connected together by a single junction close to the IC.


In practical DIY terms, a single continuously poured ground-plane likely provides an inaudible ground solution. At most, something like what's shown in Fig. 8 of that A.D. document (essentially, a single ground-plane divided in multiple sections which are still connected at one common point (in star-ground fashion) is the most that's practical, or audible, for DIY purposes. Instruments may reveal inconsequentially lower (for home audio) measured noise for a multiple ground-plane solution.

I design and build my own DIY DAC PCBs utilizing common single continuous ground-planes. What I do in addition, is to impedance isolate each chip via ferrite choke and low a inductance ceramic bypass cap on the power supply pins, and the signal inter-connections via inserting a low value resistor in each signal line. This greatly reduces the high-frequency currents circulating on the ground-plane, greatly mitigating any ground-loop interference consequence of utilizing a single common plane. I recall that Guido Tent, of Tent Labs, once published an nice app. note on isolating digital circuits in that manner.
 
Last edited:
I design and build my own DIY DAC PCBs utilizing common single continuous ground-planes. What I do in addition, is to impedance isolate each chip via ferrite choke and low a inductance ceramic bypass cap on the power supply pins, and the signal inter-connections via inserting a low value resistor in each signal line. This greatly reduces the high-frequency currents circulating on the ground-plane, greatly mitigating any ground-loop interference consequence of utilizing a single common plane. I recall that Guido Tent, of Tent Labs, once published an nice app. note on isolating digital circuits in that manner.

I've allowed for inductors in the power traces to both dacs on my dual 1541 and quad ad1862 boards and also 10r resistors in the 4 digital inputs (which are not routed around the board)

My pcb design software generates upper and lower ground planes automatically but dividing these into separate areas is more work so I have, in the past, kept the ground planes, analogue, digital and power supply grounds all separate up to the common point at the off board 0v pin. (ie the ground planes weren't connected to anything except each other)

As you say, I will be surprised if there is an audible difference on such a dac simple design.
 
High frequency currents will typically return on the reference plane and right under their respective trace (lowest inductance path). If they are causing a problem, you need to re-evaluate your layout and floor-planning. Splitting planes, unless you know exactly what you are doing, is a recipe for EMC issues.
 
Sorry forget to say...

Sorry, I forget to say that i'm using exclusively perfboards and not PCB...:worship:

Hence, not possible for me to create ground planes but I can only tie multiple wires and create multiple transformer/bridge ground separated power supplies


So question is:

should I have two ground wires (digital and analog) from Power supply (transformer/bridge ground separated?) to the dac/receiver digital and analog ground pins

or

one wire from power supply to the dac/receiver joined digital and analog pins?
 
Sorry, I forget to say that i'm using exclusively perfboards and not PCB...:worship:

Hence, not possible for me to create ground planes but I can only tie multiple wires and create multiple transformer/bridge ground separated power supplies


So question is:

should I have two ground wires (digital and analog) from Power supply (transformer/bridge ground separated?) to the dac/receiver digital and analog ground pins

or

one wire from power supply to the dac/receiver joined digital and analog pins?

I would use separate ones but connect a low value inductor between them at the pcb. This keeps HF noise from getting from one to the other, whilst maintaining them at the same DC potential.
 
I would use separate ones but connect a low value inductor between them at the pcb. This keeps HF noise from getting from one to the other, whilst maintaining them at the same DC potential.
It also introduces a ground loop, so you have to make sure its loop area is kept to a minimum to avoid stray magnetic field pickup. Not using the very thinnest wire you can find would be another good idea.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.