How do you star ground computer with firewire audio?

[Speedskater]

While it's not a PC or an interface, this is what Audio Precision does.

OK, my image didn't load.
So it's the last one in this paper.

AP High Performance Audio Analyzer & Audio Test Instruments : Service & Support

[Richard Crowley]

Is your computer a desktop or portable? Does it use an internal power supply and an IEC power cord? Or does it use a "line-lump" "brick" charger?
How is your "firewire audio interface" powered?
How are your speakers powered?

When asking questions like this, it is no time to be secretive about revealing actual make and model numbers, etc. Shielding/grounding/hum prevention/ground loops is not particularly suitable for generic discussion.

Portable computers (notebook, laptop, et.al.) are particularly notorious for this kind of noise problem. Many people use isolation transformers in the analog audio path to break the ground loop problem.

[OpusOfTrolls]

It is a desktop with switching power, plugged into APC UPS via C13 power cable. The Roland firewire interface has option for firewire bus power on 6-pin or 2 pin PSU, and I am using the 2 pin. Grounding point would be the back of the UPS, but the interface is not attached to ground via chassis at this time. The monitors, KRK rokit 8, are powered by the UPS on the battery bypass circuits, so they do not drain the battery.

Major question is how can I star ground this equipment? Do I need to use Firewire 4-pin to interface?

[tINY]

You might try powering the interface from the 6-Wire. If the converter company did a good job, that might be the cleanest (inductor coupling the power, for instance...).

Star grounding is a good technique, but it's not always needed. If the converter/interface can be the center of the universe and the speaker grounds reference it, the only ground path would be through the speakers' power supplies. If these are double-insulated types without a ground connection (transformer isolated). You should be in good shape.

A lot of ifs there....




-tINY

[OpusOfTrolls]

When you say double-insulated types, what are you referring to?

I am most interested in getting rid of the noise from the computer. I want to make sure my audio equipment does not see any bullshit from the computer. I have been thinking to use optical connection, but I will need a new interface before I can do that. Any amp I hook up from the interface via TS sounds dogshit. I just want to get that minimized.

The monitors have grounds from power cable (obviously), and I plug them into a power strip then into the UPS. I also plug the interface into this power strip, but it's only 2 pin. There is a earth screw on the UPS itself. Should I connect the interface to it?

[Speedskater]

from:
Appliance classes - Wikipedia, the free encyclopedia

Class II
A Class II or double insulated electrical appliance is one which has been designed in such a way that it does not require a safety connection to electrical earth (ground).

The basic requirement is that no single failure can result in dangerous voltage becoming exposed so that it might cause an electric shock and that this is achieved without relying on an earthed metal casing. This is usually achieved at least in part by having two layers of insulating material surrounding live parts or by using reinforced insulation.

In Europe, a double insulated appliance must be labelled Class II, double insulated, or bear the double insulation symbol (a square inside another square).

[tINY]

I'd be looking into other speakers. Your interface doesn't have a ground connection on it's power connection and your speakers do.

Read this and see if it makes sense: Sound System Interconnection

If you like your speakers and nothing else helps, these will work:

LT 22 Line Transformer

Ebtech Ebtech HE Series Audio Hum Eliminators Audio Hum & Noise Eliminators at Markertek.com

JENSEN TRANSFORMERS, INC. - ISO-MAX




-tINY

[OpusOfTrolls]

So for now the grounding is okay? I should just use the line transformer and leave everything as it is?

What about the interface? It works great and I don't get shocked usually. There was one time the case did get hot when I fed +4 outputs into -10 inputs (TRS to XLR) to rerecord. Is that normal? Would a ground wire have prevented this?

Also, since the FW is grounded through the computer, what effect could that have? What blocks the FW ground from the rest of the circuitry?

[tINY]

The FW has to have a common ground with the computer - otherwise the high-speed signaling wouldn't work. But, the DC-DC converter that takes the power off the interface can isolate the analog power ground from the digital ground.



-tINY

[enginefire]

Could also just be your computer. I had an old laptop that bled noise thru the speakers any time the hard drive was accessed. Wasn't loud but it was definitely there. Really wrecked a lot of recordings because it was just loud enough to hear after you started hearing it and then it was like the only thing I could hear. I can't explain it, but it was synced rhythmically to the hard drive spinning.

sent from my m-3500 using the gearslutz app

[dcollins]

Originally Posted by tINY

[FONT="comic sans ms"][COLOR="indigo"]
The FW has to have a common ground with the computer - otherwise the high-speed signaling wouldn't work.

Doesn't FW use LVDS? No common ground should be needed.

Hst, these computer grounding issues can be trick to solve, and usually resort to some trial-and-error type experiments.

For whatever reason PC's seem to be more prone to trouble than Mac's when it comes to grounding noise.


DC

[tINY]

The problem with the high speed stuff is that it doesn't couple differentially on the PCB - it's dual-single ended with some differential coupling. Converting this to true differential at the jack is difficult, so the ground is connected at both ends of the cable....

Theoretically, you could capacitor couple the ground at one end. But, if you use too big a cap, you end up conducting signals in the audio band (which may as well be DC for the interfaces we are talking about). Of course, too small a cap, the power-line harmonics could cause trouble with your link...




-tINY

[dcollins]

Originally Posted by tINY

[FONT="comic sans ms"][COLOR="indigo"]
The problem with the high speed stuff is that it doesn't couple differentially on the PCB - it's dual-single ended with some differential coupling. Converting this to true differential at the jack is difficult, so the ground is connected at both ends of the cable....

Interfaces like USB and FW are always differential in the cable, so I'm not sure what you mean. They do not require a ground to operate. If you have a powered interface then, yes, you need the common but not for the data.

Istr there is even a ground-lift adapter for FW to address the ground-loop issue.

Now, whether it's a good idea to float the shield for EMI reasons, is another story.


DC

[LeeYoo]

Originally Posted by OpusOfTrolls

...makes unbearable noise when going unbalanced into monitors.

Is this a joke?
Only cheap consumer/prosumer gear is unbalanced.
Sounds like a typical groundloop.
This can/should only be fixed with balancing the audio to your speakers.
Forget about fixes in the digital domain or adding ground wires.
Leo..

[tINY]

Originally Posted by dcollins

Interfaces like USB and FW are always differential in the cable, so I'm not sure what you mean. They do not require a ground to operate. If you have a powered interface then, yes, you need the common but not for the data.

Try disconnecting the shield from either end of a FW or USB2/3 and tell me that they are not required to operate. You may get lucky and have it work (some FW ports are transformer coupled), but most will not be reliable, especially with longer runs.

Bottom line, the ground is tied to both jacks, and computer ground noise ends up on the digital side of your converter. How the interface designer accounts for this in the grounding scheme is the difference between good interfaces and bad ones.



-tINY

[OpusOfTrolls]

Originally Posted by LeeYoo

Is this a joke?
Only cheap consumer/prosumer gear is unbalanced.
Sounds like a typical groundloop.
This can/should only be fixed with balancing the audio to your speakers.
Forget about fixes in the digital domain or adding ground wires.
Leo..

Hi ApeShit, I use balanced XLR to my monitors currently. I would like to add an amplifier in my budget for a pair of passive monitors, but affordable options require either unbalanced input or modding of the circuitry. The ground loop is highly typical, thanks for your insight.

[OpusOfTrolls]

Originally Posted by tINY

Try disconnecting the shield from either end of a FW or USB2/3 and tell me that they are not required to operate. You may get lucky and have it work (some FW ports are transformer coupled), but most will not be reliable, especially with longer runs.

Bottom line, the ground is tied to both jacks, and computer ground noise ends up on the digital side of your converter. How the interface designer accounts for this in the grounding scheme is the difference between good interfaces and bad ones.



-tINY

I have attempted to find a galvanic isolator for a firewire cable before with no luck. The circuitry in the interface is pretty minimal on the account of being a mobile interface. But I don't see how else the computer ground could be leaking into the signal chain?

[tINY]

Your best hope is to ground the monitors and the analog side of the interface to the same place. This isn't always easy, but you have a ground screw...

I'd connect that to the safety ground connection on the UPS. I'd also connect the chassis ground of the monitors to this point as well.



-tINY

[OpusOfTrolls]

This site:
FireWire (IEEE1394) bus interface pinout and wiring @ pinouts.ru

shows the 4 pin as not having a ground. Only differential pairs. I am a bit confused now lol.

[dcollins]

Originally Posted by tINY

[FONT="comic sans ms"][COLOR="indigo"]
Try disconnecting the shield from either end of a FW or USB2/3 and tell me that they are not required to operate. You may get lucky and have it work (some FW ports are transformer coupled), but most will not be reliable, especially with longer runs.

I first learned of this technique on (I think) Head-Fi, where people cured ground-loop noise by cutting the GND pin on the USB cable, so clearly it can work.

I'm by no means an expert on LVDS, but I don't see whether the balancing is galvanicly isolated or not would make a difference.

Although it's certainly possible for the ground lift to make matters worse.



DC

[tINY]

Originally Posted by OpusOfTrolls

This site:
FireWire (IEEE1394) bus interface pinout and wiring @ pinouts.ru

shows the 4 pin as not having a ground. Only differential pairs. I am a bit confused now lol.

...but the shield connects to the shell... A 5th pin if you will.



-tINY

[tINY]

Originally Posted by dcollins

I first learned of this technique on (I think) Head-Fi, where people cured ground-loop noise by cutting the GND pin on the USB cable, so clearly it can work.

I'm by no means an expert on LVDS, but I don't see whether the balancing is galvanicly isolated or not would make a difference.

Although it's certainly possible for the ground lift to make matters worse.

That's good to know - that it can work. USB 2.0 drivers may be pretty robust in that regard now.

With higher speeds, it gets worse. The problem is that, in the 1Gb range, serial interfaces have to work as wide-band transmission lines up to 500MHz. PCB's cannot route these differentially, only as complementary single-ended. At higher speeds, converting this to differential mode at the jack becomes difficult (just do quick calculation for 500MHz vs 50 ohms for the tiny capacitances involved...). But, tying the shield of a shielded-twisted-pair is easy.

-At lower speeds and using signaling like PAM, you can get away with a lot (like in audio grounding). FW, USB1.0, and ethernet are in this camp.
- USB2.0 is out in the middle with 240MHz bandwidth needed.
-USB3.0 and thunderbolt (as well as eSATA and DP), have a ground for each "diff" pair because they need it. You won't be able to float the ground on these interfaces if it causes an audio problem.



-tINY

[OpusOfTrolls]

Thanks for the assistance!