For what it's worth, I normally do ground all connectors straight to the chassis and make the whole thing double insulated to avoid ground loops via mains earth (also because most of the mains outlets here have no earth anyway). With star grounding you make lots of coupling loops for external RF interference to couple in. For extremely hum sensitive inputs, such as phono amplifier inputs, I connect the return of the amplifier straight to the connector.
To answer penmarker's question, the in/out terminals are insulated from the chassis - confirmed with ohmmeter.
To my surprise I haven't taken any pictures of the finished amp yet, but I have some I took during the build. I'm not home yet so this will have to do for now:
https://i.imgur.com/SILDeze.jpg
https://i.imgur.com/s5ivToc.jpg
https://i.imgur.com/1zPvQvn.jpg
https://i.imgur.com/it3arUR.jpg
To my surprise I haven't taken any pictures of the finished amp yet, but I have some I took during the build. I'm not home yet so this will have to do for now:
https://i.imgur.com/SILDeze.jpg
https://i.imgur.com/s5ivToc.jpg
https://i.imgur.com/1zPvQvn.jpg
https://i.imgur.com/it3arUR.jpg
Last edited:
each circuit requires a source and receiver component. The circuit MUST use two wires to transfer current from the source to the receiver and return the current from the receiver to the source.
That is the purpose you mention where you can use twisted pairs or close coupled pairs. The Flow and Return currents pass along traces/wires that have low loop area.
If you split those low loop area pairs to run one of the pair to a remote connection you will INCREASE the LOOP AREA. That makes interference worse.
The FIRST PRIORITY is to maintain each close coupled pair in EVERY circuit.
The Star you refer to is a voltage reference. Do not split existing Flow and Return pairs to take one half of the source to receiver current on a big loop area route.
Mark Whitney, all my equipment is RCA only. I don't have any XLR cables and a pseudo-differential cable simply does exactly the same thing that the cable I used internally in the amp does anyway with its 2 conductors + shield. I have the negative wire and the shield soldered together onto the shield pin of the RCA connector mounted on the back of the chassis.
I'm a beginner at Hifi design, but I've worked on other sorts of analogue electronics.
I must admit, I'm more at home with the kind of PCb that has a ground plane, and probably planes for power rails too.
For instance I have sorted out circuits that were failing EMC when taken to a type approval lab.
I would say there are ways of doing things with ground planes and there are ways of doing things with star earths. Doing either right can require a bit of subtlety.
For instance I have designed things where it's best to make cuts in a ground plane to prevent currents flowing under a sensitive point.
I have designed things where a ground plane needs to avoid creating circulating currents.
I have fought with circuits which are supposedly star ground, but then a system uses two of them and you've got a nice (?) earth loop.
90% comes down to 'where are the currents flowing?'
Some of the rest is electrostatic coupling.
Sometimes it's not that the circuit is exposed to too much interference, the problem is the circuit is too sensitive. E.g.something has broken and the common mode rejection is gone.
Unfortunately it's hard to diagnose without schematics, layout, actually getting your hands on it.
I must admit, I'm more at home with the kind of PCb that has a ground plane, and probably planes for power rails too.
For instance I have sorted out circuits that were failing EMC when taken to a type approval lab.
I would say there are ways of doing things with ground planes and there are ways of doing things with star earths. Doing either right can require a bit of subtlety.
For instance I have designed things where it's best to make cuts in a ground plane to prevent currents flowing under a sensitive point.
I have designed things where a ground plane needs to avoid creating circulating currents.
I have fought with circuits which are supposedly star ground, but then a system uses two of them and you've got a nice (?) earth loop.
90% comes down to 'where are the currents flowing?'
Some of the rest is electrostatic coupling.
Sometimes it's not that the circuit is exposed to too much interference, the problem is the circuit is too sensitive. E.g.something has broken and the common mode rejection is gone.
Unfortunately it's hard to diagnose without schematics, layout, actually getting your hands on it.
You may want to try connecting both conductors together and leaving the shield connected to ground. I have also successfully used two conductors that are twisted tightly together with a drill.
Good Luck!
I have had a number of hum issues thru time, from various sources with various simple solutions.
It helps to do a backward trace from a "no hum" situation" to where one, and only one, connection produces hum. That will narrow it down a bit.
You've already stated that with only the speakers connected to the amp there is silence. So that's a start.
You've next connected a source. Perhaps you should just connect ONE interconnect with no source on the end, then go to TWO interconnects, then ONE interconnect to a source, and so on. Once you get hum with just the one connection you can stop and try to figure out why.
Yes I know you think you are already there, but I only read that you connect everything up and there is hum. You need to do one thing at a go.
More often than not loud ground hum is produced by something at the interconnect/rca interface on the back of the amp or elsewhere, at least that has been my experience.
You also state that you are using shielded twisted pair for the connection of rca to the amp inputs with screen only connect at rca. Perhaps you could disconnect the shield from the rca ground so that you just have a single wire connection at "hot" and "signal ground".
Whatever you do you need to try something different from what you have.
Hope this helps and does not add to the confustion.
What you described in your first post is basically a "ground plane" (differing for the etched circuit type in that your current loops will be larger).
And a ground plain would probably work pretty well if everything were in that chassis, on that ground plane. But of course everythig isn't.
You mention the noise comes up when you connect to another source. Bingo. That source isn't on your ground plane and connects to it via (a least) two paths. The noise you describe almost certainly arises from mains noise current that is flowing in the shield ("ground") side of your cable as it find part of its way back around through the power safety grounds. Basically, the current that flows on the shields develops a voltage at the ends that depends on the noise current level and the series impedance of the shields. (forget resistance, it's impedance that matters). The problem is probably that your circuit has no way to inhibit or reject noise from shield current in relation to desired current.
If you are familiar with simple voltage/impedance/current math, here's a sample analysis of this kind of problem and a way to configure the input circuit to greatly mitigate the issue. (Of course, the best fix is to go balanced line, but that's more of an extensive mod than is practical in a lot of equipment). -- Ground loop noise, why it is and what you might do about it - diyAudio
There is a probably better treatment of this in the first issue of Linear Audio, but if you are handy with a soldering iron you might be able to work with the blog article I linked. All it takes is adding a resistor and moving all the signal (not power!) ground connections of an input circuit to the top of the resistor.
And a ground plain would probably work pretty well if everything were in that chassis, on that ground plane. But of course everythig isn't.
You mention the noise comes up when you connect to another source. Bingo. That source isn't on your ground plane and connects to it via (a least) two paths. The noise you describe almost certainly arises from mains noise current that is flowing in the shield ("ground") side of your cable as it find part of its way back around through the power safety grounds. Basically, the current that flows on the shields develops a voltage at the ends that depends on the noise current level and the series impedance of the shields. (forget resistance, it's impedance that matters). The problem is probably that your circuit has no way to inhibit or reject noise from shield current in relation to desired current.
If you are familiar with simple voltage/impedance/current math, here's a sample analysis of this kind of problem and a way to configure the input circuit to greatly mitigate the issue. (Of course, the best fix is to go balanced line, but that's more of an extensive mod than is practical in a lot of equipment). -- Ground loop noise, why it is and what you might do about it - diyAudio
There is a probably better treatment of this in the first issue of Linear Audio, but if you are handy with a soldering iron you might be able to work with the blog article I linked. All it takes is adding a resistor and moving all the signal (not power!) ground connections of an input circuit to the top of the resistor.
Last edited:
None of the standard solutions for grounding are going to work here.
Modulus-86 Rev. 2.0: Composite amplifier achieving 0.000067 % THD.
http://www.thatcorp.com/datashts/THAT_1200-Series_Datasheet.pdf
Modulus-86 Rev. 2.0: Composite amplifier achieving 0.000067 % THD.
http://www.thatcorp.com/datashts/THAT_1200-Series_Datasheet.pdf
Would you attach the images rather than uploading them to imgur, please? When I try to view your images, I see attached image saying that imgur is over capacity. You can attach pictures by clicking "Go Advanced" and clicking on the paper clip "attach" icon.
It was brought to my attention (thanks Mark!) that you're building one of my amps, so I'm a bit surprised you haven't contacted me directly. I'm always available to help.
Anyway. The Modulus-86 shouldn't make any rude noises. I run a 4-channel Modulus-86 amp powered by a Power-86 and an Antek AS-3222 power transformer. If I really jam my ear into the tweeter I can just barely make out a very, very faint hiss which may very well be from the source (MiniDSP 4x10HD). I built a similar 4xMOD86 + PWR86 amp for a friend where I built in the MiniDSP 2x4. I used pseudo-balanced connections from the 2x4 to the amp modules. That amp runs without issues as well.
I'm not familiar with the universal DIY Audio supply (aside from being aware of its existence) so I can't provide specific data for that supply. However, I do show in the documentation for the Modulus-86 (pg. 20 in the doc for Rev. 2.2) how to connect the amp to a generic unregulated power supply.
As for your original question about star grounding: The point of a star ground is to keep sensitive ground nodes (those in series with the signal path, so signal ground) away from high-current grounds (power ground). If done properly, the star ground can in theory meet this goal, but it also maximizes the impedance of all ground connections. This means that any error current in the ground nets will create the maximum error voltage across these connections. I've seen these types of grounding schemes degrade the THD of a generic LM3886 amp by over 100x.
The best grounding scheme that I've found is to minimize the ground impedance by using a ground plane. Create one plane for the power ground and connect the high current grounds (decoupling, speaker return, power supply) here. Create a separate plane for the signal ground and keep all the low-current sensitive nodes tied to this ground. Connect the two planes at the speaker output ground terminal. This minimizes the error voltage induced into the signal path.
You can see more here:
LM3886 chip amp grounding.
LM3886 PCB vs Point-to-Point (with data)
Tom
Attachments
Last edited:
There was a chap on ebay selling amplifiers a few years ago, with no metal parts in the construction. The case was wood and plastic and the circuit layout was completely random with no parallel tracks.
His theories justifying this strange design sounded plausible, but it certainly looked like a rat's nest inside. Can't say I was staggered by the sound either.
His theories justifying this strange design sounded plausible, but it certainly looked like a rat's nest inside. Can't say I was staggered by the sound either.
All my prototypes and early listening are done without any chassis.
I do include RF filtering and I don't hear interference clicks and buzzes.
But I have hard wired my house rather than relying on wireless coms.
Except my wireless telephone and my mobile phone.
I don't use WiFi and have no other transmitters.
I do occassionaly see interference from low energy lighting.
I do include RF filtering and I don't hear interference clicks and buzzes.
But I have hard wired my house rather than relying on wireless coms.
Except my wireless telephone and my mobile phone.
I don't use WiFi and have no other transmitters.
I do occassionaly see interference from low energy lighting.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.