As we all know, USB is digital. I always believed that it either worked or it didn't but couldn't just add noise to the sound. Apparently I was wrong. My wife had a MOTU M4 for her PC and uses headphones. Today she told me that her sound had a loud hiss. I checked it and yes the hiss was there and was volume dependant. I changed to a different MOTU M4 I have on my desk and there was no change. I loaded the latest driver for the MOTU and rebooted. The hiss was still there with the signal. Finally I went back to her own MOTU M4 because it wasn't the problem and then pulled and moved the USB cable at the PC. After doing that it worked perfectly. Same signal but with zero hiss. Can anyone tell me how a bad usb connection added hiss along with the normal signal?
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
USB cable adds very audible noise
- Thread starter JeffGB
- Start date
DVDdoug
Major Contributor
- Joined
- May 27, 2021
- Messages
- 3,023
- Likes
- 3,976
With USB powered devices, sometimes you can get noise into the analog electronics through the USB power. But, that's more common with microphone inputs (with the noise amplified by the preamp) and it's usually more of a "digital whine" from a switching power supply or other "digital noise", rather than hiss.
radix
Major Contributor
- Joined
- Aug 1, 2021
- Messages
- 1,398
- Likes
- 1,335
There are different USB ports on a computer. They could be white, gray, blue, green, or orange (in a type-A socket). Were both connectors the same color? I wonder if you had a low-speed or under-powered USB port the first time.
Sometimes usb ports go bad and have flaky connections, but I would have expected that to be more digital noise than hiss.
Sometimes usb ports go bad and have flaky connections, but I would have expected that to be more digital noise than hiss.
- Thread Starter
- #4
No. I didn't even change the port or the cable. Just wiggled it a bit and it was suddenly fine. The hiss was what surprised me. I would have expected it to cut out or just become a loud noise but the hiss with the existing signal was weird.
Yea, common-mode computer noise are usually wideband (hissy). Tho sometimes you can hear your mouse or HDD through it.
First we have to know what the "good" USB connection is supposed to sound like. Because sometimes the "good" connection is the one that is hissy and we resort to cutting ground connections to remove the hiss.
Common-mode noise happens when the ground (or, the "common", or reference) voltage at the source is different from the ground voltage at the load. The signal voltage, without going into details explaining why, can be assumed to be identical on both ends. So if ground voltage is also the same on both ends, then "signal minus ground" at source is the same as "signal minus ground" at the load and everything is happy.
Now, the computer is a very noisy thing, and this can be easily demonstrated by getting two sound cards inside the same computer to measure themselves. Individually they may measure -120dB SNR, but when measuring each other they can drop to -100dB or less, which make them on-par with onboard audio. That's what having lots of ground current flowing in ground planes can do to SNR. Do remember tho, -100dB @ 2V = just 0.2mV, so this is not a concern for electronics operation; audio just happens to be sensitive.
So when it comes to generating the signal for your headphone, what ground do you think it is referencing to? Not a rhetorical question, because I also don't know how exactly the insides of the DAC/amp/preamp is wired. But let's say it is referencing the ground of the original analog signal. But at the same time, your DAC ground (I'm guessing) is also connected to your computer USB. So what ground is your MOTU at since it is being connected to both? Answer is we don't know. Hence the noise.
First we have to know what the "good" USB connection is supposed to sound like. Because sometimes the "good" connection is the one that is hissy and we resort to cutting ground connections to remove the hiss.
Common-mode noise happens when the ground (or, the "common", or reference) voltage at the source is different from the ground voltage at the load. The signal voltage, without going into details explaining why, can be assumed to be identical on both ends. So if ground voltage is also the same on both ends, then "signal minus ground" at source is the same as "signal minus ground" at the load and everything is happy.
Now, the computer is a very noisy thing, and this can be easily demonstrated by getting two sound cards inside the same computer to measure themselves. Individually they may measure -120dB SNR, but when measuring each other they can drop to -100dB or less, which make them on-par with onboard audio. That's what having lots of ground current flowing in ground planes can do to SNR. Do remember tho, -100dB @ 2V = just 0.2mV, so this is not a concern for electronics operation; audio just happens to be sensitive.
So when it comes to generating the signal for your headphone, what ground do you think it is referencing to? Not a rhetorical question, because I also don't know how exactly the insides of the DAC/amp/preamp is wired. But let's say it is referencing the ground of the original analog signal. But at the same time, your DAC ground (I'm guessing) is also connected to your computer USB. So what ground is your MOTU at since it is being connected to both? Answer is we don't know. Hence the noise.
Yep, crappy incompetent design from whoever manufacturer the DAC in a way which didn't galvanically isolate the USB input from the rest of the audio circuitry. While it seems 99% of DACs out there transformer isolate their coax inputs, still far too many of them ignore / screw up the USB side. Lack of isolation combined with poor internal signal / power routing in the DAC can lead to noise getting to the audio circuits.
KSTR
Major Contributor
@garbz,
Galvanic isolation of USB-capable DACs is neither trivial nor cheap. It would approximately double the price of entry-level DACs with only RCA outputs. There is a reason why designers only seldom are allowed to implement isolation, it drives cost without any significant merit for the majority of use cases.
Galvanic isolation of USB-capable DACs is neither trivial nor cheap. It would approximately double the price of entry-level DACs with only RCA outputs. There is a reason why designers only seldom are allowed to implement isolation, it drives cost without any significant merit for the majority of use cases.
Killingbeans
Major Contributor
Ground loop. The cancer of audio reproduction.
It is when you only isolate the I2S lines and the actually DAC part and leaf the whole control part at the USB ground
KSTR
Major Contributor
No, you need all control/data pins to the DAC chip be isolated and that's about 10 total, . The isolators for the signals, notably the clock signals (including MCLK which is usually is above 20MHz) must be of extreme high quality wrt noise/jitter. And you need galvanically isolated supplies at multiple voltages (+-15V, 5V, 3.3V). Sums up very quickly to significant BOM cost.
Sure but it is normally not <10 and most of them are low speed. so a simple optocoupler is sufficient.
The jitter sensitivity depends on the topology of the DAC. if it is Asynchronous re clocking anyways it shuld be no problem?
something like tihs has ~1ns of jitter.
how would this translate analog output performance desegregation?
You need them anyways. if you look at it the otter way around you only isolated supply for digital part and and for this the 5V from the USB can be used.
I couldn't disagree more. An ISO724x digital isolator with more than enough bandwidth to completely galvanically isolate two I2S devices (such as a USB receiver and an audio interface / DAC) is cheaper than a pulse transformer used to isolate coax lines.
If you're sending 10 control signals you're doing it wrong. Isolate I2S, and isolate I2C, use a port expander to get the additional signals if you can't speak I2C to the devices directly.
Or better still don't isolate any control signals. You only need to isolate USB. There's zero reason to send control signals to USB from within the equipment. There's zero reason to setup a powersupply for it as well. Have the USB receiver powered by the PC and controlled by the PC. Done. Now you only have 3 signals to isolate (I2S) and in practice since they always come in sets of 4 on the isolation chips you can use one of them to carry the reset signal from your DAC's control circuit.
You're making assumptions about a design that is causing you to engineer yourself into a corner. Don't do that. Don't shoehorn a solution together with an isolator. Design with isolation in mind from the ground up and many of your problems disappear.
Don't beg the question. There are many designs that implement proper isolation. You complaining about cost in an industry well known for insane markups over BOM, and saying there's no merit in a thread very much discussing a problem which this would solve.
fastfreddy666
Member
- Joined
- Dec 16, 2021
- Messages
- 56
- Likes
- 100
Maybe the cable itself is broken. I know this from experience. One of my mini USB3 portable HDD's only works with one specific cable. The same thing happened with some of my HDMI cables. But the standard is still developing. The newest one being HDMI 2.1 so it could be that I had some older cable which look the same but are incompatible at certain resolutions and/or framerates.
If it's an EMI problem you could try a shielded USB cable with Ferrite Chokes (on one or both sides). I have a Motu M2 myself. And I use a USB-C to USB-C connector cable. I don't have any problems (yet)
Archimago did some tests with different cables made absolutely zero difference (with Jitter tests and everything)
archimago.blogspot.com
But every situation is different. With Information Technology and Audio Engineering the weirdest things can happen. You solve them by Trial and error. Try a different USB port. Swap the USB cable. Maybe even reinstall the software. Do a malware check. I even re-installed windows 10 one one of my PC's because my DAW Software Ableton wouldn't work with the newest build (missing some DLL. Reinstallation is faster than to do research on the internet). Ground loops can also cause problems. Connect every peripheral devices to the same outlet. This is also how professionals do it. I'm one myself. But I have to admit I also use a A true-RMS multimeter which is pretty accurate even at the microvolt level to diagnose electronic equipment (PCs mostly). I also use the thing to calibrate measuring software but I have not been using that lately. I'm thinking about buying a E1DA Cosmos ADC so i can verify Amirm's tests. Trust, but verify as the Russians say. I really like to see THD+N vs amplitude tests for some audio interface devices. Amirm does something similar with IMD distortion.
In music harmonics (both odd and even) are kind of a big deal, you know....
This is also a good example how these graphs work. Almost all modern audio devices are noise limited. The constant output noise represents a decreasing percentage of the total output as the signal amplitude increases (It's a ratio) and this is why the the line drops to the right until the clipping point. I digress. I am sorry.
If it's an EMI problem you could try a shielded USB cable with Ferrite Chokes (on one or both sides). I have a Motu M2 myself. And I use a USB-C to USB-C connector cable. I don't have any problems (yet)
Archimago did some tests with different cables made absolutely zero difference (with Jitter tests and everything)
MEASUREMENTS: USB Cables for Audio DACs. [2013-06-18 UPDATE]
A blog for audiophiles about more objective topics. Measurements of audio gear. Reasonable, realistic, no snakeoil assessment of sound, and equipment.
archimago.blogspot.com
But every situation is different. With Information Technology and Audio Engineering the weirdest things can happen. You solve them by Trial and error. Try a different USB port. Swap the USB cable. Maybe even reinstall the software. Do a malware check. I even re-installed windows 10 one one of my PC's because my DAW Software Ableton wouldn't work with the newest build (missing some DLL. Reinstallation is faster than to do research on the internet). Ground loops can also cause problems. Connect every peripheral devices to the same outlet. This is also how professionals do it. I'm one myself. But I have to admit I also use a A true-RMS multimeter which is pretty accurate even at the microvolt level to diagnose electronic equipment (PCs mostly). I also use the thing to calibrate measuring software but I have not been using that lately. I'm thinking about buying a E1DA Cosmos ADC so i can verify Amirm's tests. Trust, but verify as the Russians say. I really like to see THD+N vs amplitude tests for some audio interface devices. Amirm does something similar with IMD distortion.
In music harmonics (both odd and even) are kind of a big deal, you know....
This is also a good example how these graphs work. Almost all modern audio devices are noise limited. The constant output noise represents a decreasing percentage of the total output as the signal amplitude increases (It's a ratio) and this is why the the line drops to the right until the clipping point. I digress. I am sorry.
Isolating the relevant signals is indeed trivial. If the device is to be USB powered, an isolated power supply will also be needed. Compared to a simple two-chip dongle DAC, the BOM cost probably will roughly double, though.
KSTR
Major Contributor
All good points, appreciated. There sure are ways to make isolation quite cost-efficient and I agree it should be done at the right place where it's most feasible to implement.
Looking at an example...
... we see that the I2S + MCLK domain is the most straigthforward to isolate. In a simple application that all what's needed. In a real product with more functionality and features it gets a bit heavier but the suggested use of an I²C bridge plus expanders would keep the isolation interface relatively simple.
It's still additional design work, increased PCB real estate and BOM cost. As we have noted, in the low price segment (with RCA outputs only) this extra effort is seldom supported by the management (which have failed to see the USP aspect) even if the engineering dept. would suggest it and be able to realize it poperly... sadly, I know what I'm speaking of here...
Yes of course. You can't isolate things where the original design requirement prohibits the use of isolation in the first place. But I would wager that doesn't apply to 95%+ of DACs on the market which draw their own powersupplies from elsewhere.
Of course. I'm just calling this out as not being an excuse for this type of device in the audio industry. A not insignificant portion of which will say "screw your delta-sigma DAC chip, I'll design my own DAC from the ground up with FPGAs, and R2R ladders, and gambling, and hookers". Or come up with some other bizarre workaround to a (IMO non-existent) problem. Or better still those who use hyper expensive "boutique" parts because something something audio distortion something. This is not an "efficient" industry from a design or BOM point of view. Sure I absolutely agree there are some companies that are highly cost conscious and would cost optimise this isolation out of design, but that doesn't excuse many companies in this particular industry.
KSTR
Major Contributor
The marketing depts. have not yet discovered the unique selling point aspect of an integrated isolation for all types of digital inputs. Hopefully they will in the future.
I for one am fully convinced that full galvanic isolation is always an asset. My RME interfaces are not isolated (even the SPDIF-coaxial input is not using a input xformer, only the AES input does) and therefore I'm using universal USB2.0 isolators (Intona) and also a special supply to block the mains leakage from the stock supply... for application where it matters. Normal audio playback is not such an application as balanced outputs (even the "unbalanced" TRS outs are impedance balanced) fully mitigate ground loop issues.
tvrgeek
Major Contributor
Testing on my bench, I found from my laptop I was getting a tiny bit if hash, so I just ordered a LTM2884 demo board form AMD. It cost more than the cheap Amazon Chinese wonders, but is not full of snake oil as some audio companies sell. Just the LT isolation chip in a board with the proper connectors. Under $9 delivered from Digi-Key. It is an industrial/medical intended market where any noise is a no-no. Will it be audible in my Server? Don't know, but I have use for it on the test bench. It will be curious if I can measure any difference between the LT board and a $12 dongle with a cheaper AMD chip in it. The quality of the on-board DC converter is suspect on the cheap boards.
I noticed several DAC makers are finally addressing this internally. SCHITT for one. There are others. Now, this device is not doing any re-clocking. The DAC needs to deal with that.
Balanced lines do not always solve ground loops. Good old Jensen transformers do. There are plenty of ways to mess up the ground path in one box or another. Right now, I have a PC feeding a DAC/Pre, RCA to the amp. No hum. Nothing. Can't even feel the woofer cone. So proof you don't have to use balanced lines for no ground loop. FWIW, I built the amp so I dealt with it properly internally. I assume Topping knows what they are ding too.
Most of the isolators have an isolated DC to DC converter, so yes, they are fully isolated, Current is lower of course. Look at the specs. Some DACs don't pull much. All three of mine are line powered though.
I noticed several DAC makers are finally addressing this internally. SCHITT for one. There are others. Now, this device is not doing any re-clocking. The DAC needs to deal with that.
Balanced lines do not always solve ground loops. Good old Jensen transformers do. There are plenty of ways to mess up the ground path in one box or another. Right now, I have a PC feeding a DAC/Pre, RCA to the amp. No hum. Nothing. Can't even feel the woofer cone. So proof you don't have to use balanced lines for no ground loop. FWIW, I built the amp so I dealt with it properly internally. I assume Topping knows what they are ding too.
Most of the isolators have an isolated DC to DC converter, so yes, they are fully isolated, Current is lower of course. Look at the specs. Some DACs don't pull much. All three of mine are line powered though.
D
Deleted member 46664
Guest
One thing most technicians know... The number one cause of failure in electronics is bad connections.
That wiggling the plug fixed it tells me that's probably all it was. Any connector can fall victim to a bad connection caused by dust, nicotine, repeated insertions, humidity or any number of other reasons.
The fix is to clean your plugs and sockets regularly... I use rubbing alcohol (Isopropoline). I just get some in a small dish and then dip the plug from the cable into it and swish it around a bit. Then while still wet, I plug it in and unplug it a few times in rapid succession.
90% of the time the problem goes away.
That hiss you were hearing was probably the USB version of a bad ground.
Killingbeans
Major Contributor
Mechanical failures, in general, are a b¤¤ch. Especially because they often cause intermittent errors.
Similar threads
