Best cheap USB-C headphone dongles?

[jkim]

I have a setup where I connect a 3.5mm from my laptop's dock to my desktop's line-in, so I can hear both audio at once.

However, this inevitably introduces ground loops, so I had to use a USB isolator, then connect the DAC to the USB isolator, then connect the DAC to the desktop through 3.5mm. The generic ADuM3160 isolator only supports USB Full Speed, which is UAC1.

When connected to the isolator, the DA06 is able to detect that the connection is USB Full Speed, and downgrade it to 96kHz/24bit UAC1. The ALC5686 and the Hi-Max seems to be able to detect the connection downgrade, but they provide a 48kHz/32bit output instead of the 96kHz/24bit of the DA06, and crackles when audio is played. Crackling happens even when the audio output is set to 48kHz/16bit, so it is probably an implementation issue on the DAC. No crackling on DA06.

I have also tried extremely cheap dongles such as a unbranded 48kHz/16bit "KM_B2 Digital Audio" dongle and a Bluetrum AB132A 48kHz/16bit dongle. They work fine with the isolator, but they are UAC1 from the start anyways.

Its more than crackling, but you can hear what I mean by the file below. It's 5 seconds of silence, followed by about 30 seconds of Aimyon's Marigold in a .wav file, for the Hi-MAX.

Not on topic, but for your purpose, can't you use audio streaming software between PCs, like the Voicemeeter?

 

[markanini]

It also does appear to have a stereo ADC, which I found quite unique.

I haven't found a dongle with a stereo ADC, only mono ADC. Thanks for sharing this.

 

[CedarX]

I haven't found a dongle with a stereo ADC, only mono ADC. Thanks for sharing this.

Though with a 4-poles 3.5mn, only one ADC is used, or the two ADC are connected in parallel.

 

[Kurumi]

Not on topic, but for your purpose, can't you use audio streaming software between PCs, like the Voicemeeter?

I did not know that was a thing. Searched all over the place for transmitting audio over network and all I found was some abandonware or only worked on Mac OS. Thanks!

 

[mc.god]

I haven't found a dongle with a stereo ADC, only mono ADC. Thanks for sharing this.

For completeness about this topic and for what it matters, CX31993 ADC supports more formats than CB1200, at least in the way they are exposed in the two dongles:

Grave DA06 CX31993
Capture:
Status: Stop
Interface 2
Altset 1
Format: S16_LE
Channels: 1
Endpoint: 0x81 (1 IN) (SYNC)
Rates: 8000, 16000, 32000, 44100, 48000, 96000
Data packet interval: 125 us
Bits: 16
Channel map: FL
Interface 2
Altset 2
Format: S24_3LE
Channels: 1
Endpoint: 0x81 (1 IN) (SYNC)
Rates: 8000, 16000, 32000, 44100, 48000, 96000
Data packet interval: 125 us
Bits: 24
Channel map: FL


Hi-max CB1200
Capture:
Status: Stop
Interface 1
Altset 1
Format: S16_LE
Channels: 1
Endpoint: 0x81 (1 IN) (SYNC)
Rates: 44100, 48000
Data packet interval: 125 us
Bits: 16
Channel map: FL
Interface 1
Altset 2
Format: S24_3LE
Channels: 1
Endpoint: 0x81 (1 IN) (SYNC)
Rates: 44100, 48000
Data packet interval: 125 us
Bits: 24
Channel map: FL


On both you can see that Channel map is only on FL, so mono input as noted by @CedarX

 

[HissingFree]

Jcally JM7 Max has finally reached the "dirt cheap" category, I can buy it for less than 6€ as a new user:

https://es.aliexpress.com/item/1005009622342774.html

Still no good test of this dongle, probably won't improve on the mediocre things of regular CX31993 such as sinad or weird jitter performance, unless there are improvements to the usb bridge or other enhancements. But 120mW at 32 ohms and decent noise floor make it very attractive at this price point compared to the disappointing alternatives below CS43131. The only thing I would like to know is if this dongle crackles in Windows especially, I would like more power than Jcally JM12, without the crackling that it has on regular and JA11 firmware and without the hissing and high noise floor that it has on tinhifi firmware. If it doesn't crackle, this should be the GOAT sub 10$ dongle

 

[HissingFree]

I bought the JCALLY JM7 Max during an AliExpress sale.

View attachment 474902

I connected it to my Windows 11 PC and played a test audio in the sound settings, but it played with a fade-in.
There's a glitch where playing music from silence causes it to fade in. It's awful.

Right out of the box, I was disappointed when playing the test audio. “Not again,” I thought.

The JCALLY JM30 has a similar issue.
It's not a fade-in, but when playing an audio file, the first 0.5 seconds are missing.
Information about this is being shared in a dedicated thread.

I own two JCALLY products. Are they all like this?
I've wasted money twice now.

I missed your post. Does it crackle when doing the sound test in sound settings (ignoring this fade-in issue)? by that I mean if the sound test sounds clean or if it sounds a bit distorted, compared to JM30? also could you test this rumble test to see if it also crackles there? rumble in phase or out of phase, does it sound distorted?

Online Stereo Polarity (Phase) Sound Test

Checks your loudpeakers wiring.

www.audiocheck.net

 

[CedarX]

Side note on JCally: their website https://jcally.com is now up & running. It includes what appears to be Audio Precision measurements for some characteristics.
It is Chinese-only, but I have had good success with website translation features.
JM7max and JM98max are their newest dongles.

 

[mc.god]

Jcally JM7 Max has finally reached the "dirt cheap" category, I can buy it for less than 6€ as a new user:

https://es.aliexpress.com/item/1005009622342774.html

Still no good test of this dongle, probably won't improve on the mediocre things of regular CX31993 such as sinad or weird jitter performance, unless there are improvements to the usb bridge or other enhancements.

Poor jitter in CX31993 is due to its SYNC usb transfer, and being it a full featured chip with integrated adc and bridge for super cheap, i doubt anyone would pair it with an external bridge supporting ASYNC mode.

 

[wiggum]

Most recorded music is max 90-96dB(16 bits). Maybe you should create som test files for you own pupose to determine where the threshold is for you, on your best DAC.

Most recorded music has a maximum dynamic range of 12 dB.

Some classical music tracks are 30 dB, but they are rare.

 

[Peterinvan]

My FiiO KA3 dongle works well for me. My old iPhone 6 has sufficient power to drive it.

 

[staticV3]

Most recorded music has a maximum dynamic range of 12 dB.

Crest factor ≠ Dynamic range

 

[wiggum]

Crest factor ≠ Dynamic range

Crest factor is also below 30 dB in most recorded music AFAIK.

 

[staticV3]

Crest factor is also below 30 dB in most recorded music AFAIK.

My point was that 12dB and 30dB refer to crest factor in music, not the dynamic range.

Dynamic range is easily above 60dB, even in highly compressed pop.

All it takes is a fade to/from silence and just like that, you have the full 96dB of dynamic range (or 144dB for 24bit).

 

[wiggum]

My point was that 12dB and 30dB refer to crest factor in music, not the dynamic range.

I was referring specifically to dynamic range of music, not crest factor. There are have been some published papers that evaluated music over the decades.

Dynamic range is easily above 60dB, even in highly compressed pop.

All it takes is a fade to/from silence and just like that, you have the full 96dB of dynamic range (or 144dB for 24bit).

DR doesn't take into account the silence, only the ratio of the softest and loudest signals present in an entire musical score.

https://help.roonlabs.com/portal/en/kb/articles/dynamic-range#Where_can_I_see_Dynamic_Range_in_Roon

 

[staticV3]

DR doesn't take into account the silence, only the ratio of the softest and loudest signals present in an entire musical score.

And a fade to/from silence generates a signal approaching -96dBFS for 16bit.

 

[wiggum]

And a fade to/from silence generates a signal approaching -96dBFS for 16bit.

Easy way to settle this.

Run

sox testfile.wav -n stats which gives the crest factor directly while the DR is calculated as difference of abs(RMS Pk dB) and abs(RMS Tr dB)

 

[jkim]

I was referring specifically to dynamic range of music, not crest factor. There are have been some published papers that evaluated music over the decades.

DR doesn't take into account the silence, only the ratio of the softest and loudest signals present in an entire musical score.

https://help.roonlabs.com/portal/en/kb/articles/dynamic-range#Where_can_I_see_Dynamic_Range_in_Roon

And a fade to/from silence generates a signal approaching -96dBFS for 16bit.

Easy way to settle this.

Run

sox testfile.wav -n stats which gives the crest factor directly while the DR is calculated as difference of abs(RMS Pk dB) and abs(RMS Tr dB)

You guys are talking about all different things.

• Audio device's (e.g., DAC's or amp's) Dynamic Range = difference (in dB) between the device's loudest possible undistorted output signal level and its noise floor (the AES17 procedure is one standard method).
• Roon's Dynamic Range of an audio track = EBU R 128 Loudness Range of an audio track = difference between the 10th percentile and the 95th percentile of the distribution of an audio track's signal levels over equally spaced, short time windows (excluding silent windows).
• Dynamic Range of an audio track (commonly considered by recording engineers?) = difference between the peak and trough RMS levels of an audio track (near-silent windows are included?)
• Crest Factor of an audio track = difference between the peak level and average level of an audio track (silent windows are included in the calculation of the average level).

Based on the above definitions, we can say that either the EBU R 128 loudness range (Roon's DR) or the crest factor is NOT suitable (from from being ideal) for considering the noise performance of an audio device. For example, it is not appropriate to say, "because most audio tracks' loudness range, or its crest factor, is just XX to YY dB, any DAC with a dynamic range of ZZ dB should be more than sufficient."

Maybe we can say something more sensible using the third definition of an audio track's DR. But this still does not seem to be measured with a standardized procedure.

 

[wiggum]

You guys are talking about all different things.

• Audio device's (e.g., DAC's or amp's) Dynamic Range = difference (in dB) between the device's loudest possible undistorted output signal level and its noise floor (the AES17 procedure is one standard method).
• Roon's Dynamic Range of an audio track = EBU R 128 Loudness Range of an audio track = difference between the 10th percentile and the 95th percentile of the distribution of an audio track's signal levels over equally spaced, short time windows (excluding silent windows).
• Dynamic Range of an audio track (commonly considered by recording engineers?) = difference between the peak and the trough RMS levels of an audio track (near-silent windows are included?)
• Crest Factor of an audio track = difference between the peak level and average level of an audio track (silent windows are included in the calculation of the average level).

Based on the above definitions, we can say that either the EBU R 128 loudness range (Roon's DR) or the crest factor is NOT suitable (from from being ideal) for considering the noise performance of an audio device. For example, it is not appropriate to say, "because most audio tracks' loudness range, or its crest factor, is just XX to YY dB, any DAC with a dynamic range of ZZ dB should be more than sufficient."

Maybe we can say something more sensible using the third definition of an audio track's DR. But this still does not seem to be measured with a standardized procedure.

I think the required DR of a DAC = headroom + (peak - trough) + 60 dB?

60 dB because when the softest part of the music is playing the THD+N should be below 0.1%(60 dB) which is threshold of audibility of THD.

Any extra DR is nice, but won't be used at all?

 

[jkim]

I think the required DR of a DAC = headroom + (peak - trough) + 60 dB?

60 dB because when the softest part of the music is playing the THD+N should be below 0.1%(60 dB) which is threshold of audibility of THD.

Any extra DR is nice, but won't be used at all?

Are there some reliable statistics about (peak - trough) measured with consistent methods/software for music recordings?