Hi All - Standard disclaimer that I'm a software engineer who is into embedded applications as a hobby. I've had zero formal training in electronics, and even less in audio electronics. Everything I've learned has been thru dabbling with microprocessors and some various sensors over the last few years.
I happened to pickup a through-hole MAX541 DAC. This is a 16 bit DAC. The datasheet shows the following typical circuit used:
I'm confused about the grounds. I of course am familiar with the typical "3 line" ground symbol, but not the upside-down triangle, which I understand is called a "digital or common" ground.
I don't understand how the typical ground and the digital/common ground differ. I of course only have one ground going from my microprocessor to the breadboard. Am I supposed to use some sort of circuit to create the digital/common ground?
Any assistance would be very much appreciated. Thank you.
I happened to pickup a through-hole MAX541 DAC. This is a 16 bit DAC. The datasheet shows the following typical circuit used:
I'm confused about the grounds. I of course am familiar with the typical "3 line" ground symbol, but not the upside-down triangle, which I understand is called a "digital or common" ground.
I don't understand how the typical ground and the digital/common ground differ. I of course only have one ground going from my microprocessor to the breadboard. Am I supposed to use some sort of circuit to create the digital/common ground?
Any assistance would be very much appreciated. Thank you.
The two grounds are tied together at the AGND pin so notionally they're the same potential. But the distinct symbols are used to indicate which of the two physically separate (yet joined) groundplanes to connect decoupling caps to. The REF is decoupled to the AGND and the VDD decoupled to the DGND. This discourages currents from running through the 'star' junction of the two planes.
Digital grounds have a lot of noise on them. A vertical edge has fourier components running into the high megahertz. High frequency RF waveforms transmits through holes better than lower frequencies. Analog grounds can't stand the RF noise, and typically will run to preamps, mixers, or power amps. Ring of an RCA jack or pin 1 of XLR are analog ground. Cheap 3 pin analog interfaces have AGND on pin 2. Note AGND is in between the +and- supplies used to run the op amp following the D/A converter. There may be no AGND on a digital board other than the ring of the RCA jack out, unless you make one with a voltage regulator, zener diode, and filter caps.
I keep my digital devices as the PC in a different room, on a different AC circuit, than the analog equipment. I don't run the PC or DTV to the analog stereo. The PC & DTV have their own power amp, with lower quality speakers that won't reproduce high frequencies as well.
Also the 3 line symbol for grounds is not typically used for AGND. That symbol is usually restricted to earth, which is the 3rd pin of an AC socket, which is tied to a gas or water pipe at the service entrance of a building. Earth is tied to case ground of power amps with metal cases, or the metal case of PC's. In such power amps that came with 3 pin power cords, great care is taken to isolate case ground from analog ground, without causing danger of shock to persons touching the speaker return or the ring of an RCA plug.
I keep my digital devices as the PC in a different room, on a different AC circuit, than the analog equipment. I don't run the PC or DTV to the analog stereo. The PC & DTV have their own power amp, with lower quality speakers that won't reproduce high frequencies as well.
Also the 3 line symbol for grounds is not typically used for AGND. That symbol is usually restricted to earth, which is the 3rd pin of an AC socket, which is tied to a gas or water pipe at the service entrance of a building. Earth is tied to case ground of power amps with metal cases, or the metal case of PC's. In such power amps that came with 3 pin power cords, great care is taken to isolate case ground from analog ground, without causing danger of shock to persons touching the speaker return or the ring of an RCA plug.
Last edited:
Analog ground and digital ground are clearly marked on the main IC there, which points out the symbols. As said above, all the switching noises from the digital circuits are best left separate from the analog, ie signal, circuits. They all need to reference the same point somewhere, but we don't want analog signal path ground returns sharing copper with digital circuits.
I find grounding schemes fascinating. In terms of drawings, there is also the symbol I call the garden rake, which usually means chassis. I know of one tube amplifier that has at least six ground groups within it. There are various conventions for the drawing. When you separate grounds, you could just use the three parallel lines one for all but note AGDN or DGND next to each one, but by using arbitrarily different symbols it is easier to keep them apart visually.
In my six-ground amp I see also your two symbols, and then the same symbols but with circles around them. So triangle versus triangle/circled. I have seen some with circles around one symbol and then a hexagon box around it elsewhere, giving us plain or two surrounded versions for one symbol type.
In my tube amp, they break down ground groups into power supply, preamp, middle amp. etc. They want the preamp not to react to other sections to prevent crosstalk or instability.
I find grounding schemes fascinating. In terms of drawings, there is also the symbol I call the garden rake, which usually means chassis. I know of one tube amplifier that has at least six ground groups within it. There are various conventions for the drawing. When you separate grounds, you could just use the three parallel lines one for all but note AGDN or DGND next to each one, but by using arbitrarily different symbols it is easier to keep them apart visually.
In my six-ground amp I see also your two symbols, and then the same symbols but with circles around them. So triangle versus triangle/circled. I have seen some with circles around one symbol and then a hexagon box around it elsewhere, giving us plain or two surrounded versions for one symbol type.
In my tube amp, they break down ground groups into power supply, preamp, middle amp. etc. They want the preamp not to react to other sections to prevent crosstalk or instability.
Hi Guys - I can't tell you how much I appreciate the informative replies even if some of it is a bit beyond me.
If I'm understanding correctly, it seems the two grounds, even if connected at one point, need to be kept as separate as possible to help with avoiding RF noise.
My plan was to prototype this DAC on a breadboard and possibly have a small board manufactured. If I do so have a small board made, it sounds like I need to have separate layers (planes) for each of the two grounds.
At least the basic principle and the schematic makes more sense now. Thanks again!
If I'm understanding correctly, it seems the two grounds, even if connected at one point, need to be kept as separate as possible to help with avoiding RF noise.
My plan was to prototype this DAC on a breadboard and possibly have a small board manufactured. If I do so have a small board made, it sounds like I need to have separate layers (planes) for each of the two grounds.
At least the basic principle and the schematic makes more sense now. Thanks again!
DOn't know that you need layers, just careful routing should do. But now we are leaving my turf, ground plane layers may be desirable for your project.
ADCs are probably the most difficult part to optimize a layout for. If you use a breadboard you'll not easily be able to create planes. Planes are used to create the lowest possible impedance (both resistance and inductance) for grounds. With PCBs being so affordable nowadays I'd recommend not prototyping a 16bit ADC on a matrix board.
Got it, and thanks for the heads up. Interestingly enough, I came across an eval kit for the MAX541, datasheet here. And indeed they only use one layer.
Got it. Just to be clear, matrix board and bread boards are different, right? This is a bread board and this is matrix board.
I also know matrix board to be called "perf board", if I recall correctly.
Matrix board is used for actually soldering in through hole components to create a prototype where breadboards are maybe the most basic type of prototyping? (Guessing here on the terminology)
Of course no soldering is done with breadboards.
If my above terminology assumptions are correct, point taken. You're saying that making a prototype using matrix board will just result in a noisy prototype and you should jump right to having an actual printed circuit board manufactured.
Good advice, thanks. Please correct me if any of my above assumptions are incorrect.
That eval board from maxim sounds perfect.
Yes, running noisy high freq digital electrons in the same wire or right next to slow <22 khz analog electrons can cause coupling between the two and induce buzzing & howling noises in the analog parts that drive sound to your ear. Even switcher power supplies like modern wall supplies can cause noises that transformer supplies will not.
Those push wire in board products produced so many bad connections in my college physics lab, I've never bought one. I have done bare wire circuits on bare holey board with a soldering iron, which is okay as long as you don't run into any RF problems as I did on my LM1875 failure. Building a d/a board with wire or vector board sounds like a recipe for RF interference problems.
Personally I buy my d/a converters in $2 CD players from Salvation Army resale shop, or in PC's I get for $50 from friend's businesses. Then change rubber switch pads, corroded battery clips, or expired electrolytic caps (lots of those in PCs). Digital sources are 99.9% MP3 anyway, which sounds almost as bad as cassette tapes to me. But MP3 is what everybody is doing now, since most people have been deaf to high frequencies since childhood fireworks.
Yes, running noisy high freq digital electrons in the same wire or right next to slow <22 khz analog electrons can cause coupling between the two and induce buzzing & howling noises in the analog parts that drive sound to your ear. Even switcher power supplies like modern wall supplies can cause noises that transformer supplies will not.
Those push wire in board products produced so many bad connections in my college physics lab, I've never bought one. I have done bare wire circuits on bare holey board with a soldering iron, which is okay as long as you don't run into any RF problems as I did on my LM1875 failure. Building a d/a board with wire or vector board sounds like a recipe for RF interference problems.
Personally I buy my d/a converters in $2 CD players from Salvation Army resale shop, or in PC's I get for $50 from friend's businesses. Then change rubber switch pads, corroded battery clips, or expired electrolytic caps (lots of those in PCs). Digital sources are 99.9% MP3 anyway, which sounds almost as bad as cassette tapes to me. But MP3 is what everybody is doing now, since most people have been deaf to high frequencies since childhood fireworks.
Last edited:
There are many ground symbols and no real standards for their use. The main point here - as others have pointed out as well - is to keep the quiet analog ground separate from the bouncy digital ground.
The catch is that the two grounds will need to meet at some point. You want to be careful with the routing - especially of the digital signals - to prevent the noisy digital return ("ground") currents from corrupting the quiet analog ground.
M.I. Montrose, Printed Circuit Board Design Techniques for EMC Compliance would be a good place to start.
Tom
The catch is that the two grounds will need to meet at some point. You want to be careful with the routing - especially of the digital signals - to prevent the noisy digital return ("ground") currents from corrupting the quiet analog ground.
M.I. Montrose, Printed Circuit Board Design Techniques for EMC Compliance would be a good place to start.
Tom
Hi guys, I just wanted to post back with some results. I've got the MAX541 DAC setup in a breadboard being fed 16 bit audio at roughly a 44.1 KHz sample rate from a Texas Instruments TM4C1294 development board.
The audio the the dev board sends to the DAC is very short. It's about a second of professional piano recording (typical CD quality 16 bit, 44.1 KHz) I converted to a data array that the program running on the TM4C1294 repeatedly loops through, sending the audio to the DAC.
I'm sure it sounds no where near as good as a boutique professionally built audiophile DAC, but for what it is (an relatively inexpensive chip on a breadboard), I'm surprised how good it sounds. Here's an example as a wave file:
Google Drive: Sign-in
That recording was done by hooking up the audio output pin of the DAC, and a ground line, to a 3.5mm jack and running it into the line-in input of my PC and recording it using Audacity.
The signal was quite low. I used Audacity to normalize it. I suppose I need to feed the DAC's output into an amp chip and then take the output of that. I suspected this when buying the DAC and other parts, so I bought a couple different MAX amps as well.
If anyone's interested in the code or the schematic, they can be found here:
Misc/TivaMAX541 at main * tmdarwen/Misc * GitHub
Many thanks for the help with understanding the different ground symbols.
The audio the the dev board sends to the DAC is very short. It's about a second of professional piano recording (typical CD quality 16 bit, 44.1 KHz) I converted to a data array that the program running on the TM4C1294 repeatedly loops through, sending the audio to the DAC.
I'm sure it sounds no where near as good as a boutique professionally built audiophile DAC, but for what it is (an relatively inexpensive chip on a breadboard), I'm surprised how good it sounds. Here's an example as a wave file:
Google Drive: Sign-in
That recording was done by hooking up the audio output pin of the DAC, and a ground line, to a 3.5mm jack and running it into the line-in input of my PC and recording it using Audacity.
The signal was quite low. I used Audacity to normalize it. I suppose I need to feed the DAC's output into an amp chip and then take the output of that. I suspected this when buying the DAC and other parts, so I bought a couple different MAX amps as well.
If anyone's interested in the code or the schematic, they can be found here:
Misc/TivaMAX541 at main * tmdarwen/Misc * GitHub
Many thanks for the help with understanding the different ground symbols.
This gets more complicated when you have two ADC chips. Single point connection is not possible