(Unofficial) Review of JCALLY JM20 MAX: USB-C Headphone Dongle with High Unbalanced Output Power

[pioterd]

anticipate

Anticipate yes. Be sure, no single dac + single opamp shall give similar results if implemented comparably well. I had KA11 and will soon have UA mini and I can write back what I think about it. I mostly use it as a line level sources (setting 1.5V - 2.5V max output depending on headphones) for bigger amplifier (ifi ZenCan) and for me, I do not need anything more.

Dual dac/ dual opamp, like Alpha XI1 or KA15 are REALLY nice and strong devices, sounding great with nice voltage and current even for low impedances.

 

[MC_RME]

Crosstalk is dictated from the quality of the ground connection from TRS female plug to internals of the USB-A connector. That value will change with cables plugged in, both by the basic design of the cable and the contact resistance of the TRS male jack. So whatever me or anyone else measures has no meaning for your personal setup.

I played around with several IEM cables and connectors and can give a more useful answer. For the Jcally JM20 Max I measured crosstalk of around -73 dB, no matter what frequency. This is most probably caused by the ground connection via female TRS connector and woven cabling with soldering onto the DAC PCB inside the USB-C connector. That value as such is very good for a TRS output.

More interesting is the output impedance. I measured 0.544 Ohm on left output channel, 0.531 Ohm on right output channel. This is with the most perfect connection to the dongle's female TRS output, and without any external cabling. For comparison, the original Apple dongle measures 3 Ohms 0.257 Ohms per channel. Good again!

The cable resistance can be quite critical. Years ago there were IEMs shipping with a single, shared ground wire for left/right that had several Ohms resistance and caused significant, even audible crosstalk. After checking the cables currently on my desk it seems this failure is no longer an issue. All (!) my cables have separated cabling for left and right channel which ends in the connector. Means ground is combined at the ideal point, at the very end of the cable. Crosstalk by shared ground cable resistance is completely avoided this way.

You should be able to verify your cable being wired correctly just by looking at it. With the cheap Zero:2 cable the left and right cable goes in parallel into the connector. With modern woven designs I think they all do the same and are soldered only in the connector, not in the middle at the split point. Examples are KBEAR ST12 or TRN-T2 Pro.

The resistance itself can be different. KBEAR ST12 measures 0.440 Ohm per wire, TRN-T2 Pro only 0.139 Ohms. In real-world you won't notice any difference, neither in sound nor in volume. Remember the old rule - output impedance x 8. Even combining output impedance with cable resistance the factor is typically much higher than 8.

Unless the IEM is unusually low in impedance. Example: the latest Blue Zero:2 and its 5 Ohms impedance, rising a lot below 1 kHz (at DC 66 Ohms). Using the Apple dongle with its 3 Ohms output impedance will give a slight bass boost, similar to the included 5 Ohms connector that comes with the Blue Zero:2 (see @amirm 's review). So here you have an example why two DACs that measure better than you can hear will sound different with the Blue Zero:2. Edit: Would have been a nice showcase, but no.

 

[multiformous]

Some have measured the Apple USB-C dongle as having between 0.3-0.5 Ohms of output impedance, and the review on site here states 0.9 Ohms.

 

[markanini]

Maybe @MC_RME 's dongle is different? Either the product changed, or a case of where counterfeits are sold as legitimate-it happens: https://www.forbes.com/sites/amitch...of-genuine-iphone-chargers-on-amazon-are-fake

 

[MC_RME]

I'll check tomorrow if I find a second Apple dongle and will measure again.

 

[jkim]

More interesting is the output impedance. I measured 0.544 Ohm on left output channel, 0.531 Ohm on right output channel.

My sample's output impedance is 0.57 to 0.58 ohm. I measured both channels using 100 Hz and 1 kHz tones. Used just a simple method with a true rms meter.

 

[jkim]

Could you please provide results also for maximum line level output (2.5V)?

1 kHz sine tone SINAD results at its maximum unloaded output have been added to the review. In fact, I made these measurements originally but did not include them for simplicity.

 

[MC_RME]

Indeed the Apple dongle had fooled me. I used the external signal insertion method to measure the output impedance. There is 3 Ohms, but it is a standby state of the dongle. If you play something it wakes up and has 0.257 Ohms. Exactly 6 seconds after playback it drops back to 3 Ohms. I fixed the text in my post #42.

 

[JSmith]

ASR review is up;

JCALLY JM20 MAX Headphone Dongle Review

This is a review and detailed measurements of the JCALLY JM20MAX DAC and portable headphone amplifier. It was sent to me by Crinacle. Cost is anywhere from US $22 to $65 :eek:. The dongle has a bit of style to it with that logo and such. Otherwise it is well, a dongle. :) Key thing is how it...

www.audiosciencereview.com

JSmith

 

[multiformous]

I bought one the other day based on the strength of your review, before Amir had posted his results. Didn't even know he was testing it. Thanks for taking the time to do these measurements.

I had a Tanchjim Space for a while but never really liked it.. it would disconnect from my phone/PC at the slightest touch and I always found the independent volume adjustment to be annoying. I bricked it doing a firmware update and threw it away. Good riddance... A JM20 Max should be more than I'll ever need.

 

[welwynnick]

Looking at its frequency / phase response, it seems that the default filter is the minimum-phase type:​

Response at 96kHz sampling frequency:
As expected the phase does not shift much in the audible band when the bandwidth is increased. If you're concerned about the phase shift, use a wide bandwidth setting.
SMPTE IMD vs. output voltage under 300 Ohm load:
THD+N vs. frequency is excellent, too:

jkim many thanks for those measurements - in particular - frequency/phase, IMD/level and THDN/frequency/load.

I've been getting more and more sceptical about the value of measurements at 1V / 1W / 1% / 1dB / 1kΩ / 1kHz / 1µV etc.

If you just rely on best-case scenario tests, you convince yourself that all electronics are the same.

For lots of different reasons that are all coming together, I think what we need from audio equipment is:

• Linear frequency response
• Linear amplitude response
• Linear phase response

All those things over the actual operating envelope:

• Whole frequency range
• Whole amplitude range
• Whole load range

It's not difficult to show good performance for a certain parameter at a certain point in the middle of the envelope.
Everything degrades at the edges and the corners of the envelope, and we hear those things, too.
Therefore I'm really happy when people like you perform the more insightful tests as well.

Thanks, Nick

 

[dtaylo1066]

Just bought one on AliExpress for $27 with taxes. Unreal. I will use this on my Samsung Galaxy Tablet S8+. So is volume handled by the Samsung via its volume control or do I have to install a player?

I will use with Tidal.

 

[iGmole]

Just bought one on AliExpress for $27 with taxes. Unreal. I will use this on my Samsung Galaxy Tablet S8+. So is volume handled by the Samsung via its volume control or do I have to install a player?

I will use with Tidal.

It works fine without apps. It'll obey the volume buttons like any other headphone or audio device you might plug in the tablet. But UAPP has some tricks up it's sleeve like matching the dongle sample rate and bit depth to the music format (so as to not have to do conversion inside Android) and hardware volume control and it works with Tidal, so you might like it.

EDIT: remember to turn the volume to zero and carefully increase it when you have your headphoney things on. Guess how I heard the loudest sound of my life?

 

[dtaylo1066]

It works fine without apps. It'll obey the volume buttons like any other headphone or audio device you might plug in the tablet. But UAPP has some tricks up it's sleeve like matching the dongle sample rate and bit depth to the music format (so as to not have to do conversion inside Android) and hardware volume control and it works with Tidal, so you might like it.

EDIT: remember to turn the volume to zero and carefully increase it when you have your headphoney things on. Guess how I heard the loudest sound of my life?

Thank you. That is what I suspected and was going to download UAPP, too, to bypass the Android 48kh thing.

 

[TNT]

Does this cause some unwanted audible effects?

I this kind of design is utterly insulting and I will never buy anything that behaves like this. @amirm should implement a standard test for this kind of characteristic. I strongly suspect that this logic, while always running, will affect music in certain situations - I wonder how....

//

 

[welwynnick]

I this kind of design is utterly insulting and I will never buy anything that behaves like this. @amirm should implement a standard test for this kind of characteristic. I strongly suspect that this logic, while always running, will affect music in certain situations - I wonder how....

It's possible to understand why you might want to avoid a device that does this, but it's not possible to understand why it should be insulting.

This is surely a natural consequence of ASR, and the like, continually advocating objective measurements and specifications.

The year is 2025, and there are far better things to throw out your rattle over.

 

[Eddie]

My first test was ramp-up issue on a Windows PC (as before with DS2) - none. I also did not hear any short noise on playback start (as reported for the non Max version). I used the already installed Bravo-HD ASIO drivers, but now I also see 'WASAPI headphone (CS43131+SGM826 HIFI Audio)' and that one works as well.

Thanks for this information. It's crucial for me.

In summary: Can you confirm the following behavior for all supported PCM sample rates on Windows 10/11 using WASAPI exclusive mode in the event callback style?
• no pop/click/noise on start
• no pop/click/noise on stop
• no fade-in/ramp-up on start
• no fade-out/ramp-down on stop
• on start: samples at the beginning are NOT skipped (this seems odd, but other popular devices with CS43131 actually do this)

Also:
• no pop/click/noise when changing the PCM sample rate

Thanks.

 

[ZolaIII]

I found an interesting phenomenon about this device. I believe this occurs with any DAC built on CS43131 (or CS43198).

As discussed by @staticV3 and @amirm in the review of SMSL DL100 (see the thread from this post), CS43131 adopts a technique that increases dynamic range (DNR). Basically, it reduces analog gain when it meets with low-level signals and compensates for the decreased signal by applying digital gain. This way noise is reduced but the signal level is preserved. But information on precisely how it is implemented is not documented. Based on the data I observed for the JM20 Max---the only CS43131-based device that I own---, it seems that the way this DNR enhancement function is turned on is different from the way it is turned off.

Shown below are the rms noise level measurements from two separate 1 kHz sinusoid sweeps. The first sweep was from -120 dB to 0 dB, which is how this test is usually done, and the second swept from 0 dB to -120 dB (i.e., loud to soft levels):
View attachment 443074

When the signal starts from a low level (-120 dB in this case), the DNR enhancement is turned on. Once it's on, the amount of noise reduction appears to be in its full state when the signal is lower than -40 dB, but above that level, noise reduction adapts to the signal, which seems to occur between -40 to -12 dB. In contrast, when the signal starts from a loud level, the function is turned off and responds to signal amplitude just like a brick wall. Noise reduction is turned on to its full state as soon as the signal reaches -51 dB.

More generally in music/audio playback, my guess is that whenever the signal goes below -50 dB in a certain time window, the DNR enhancement is turned on to its full state. But once it is turned on, going back to the off state does not occur by applying a single threshold, but rather signal-dependent, fine adjustment of analog/digital gain structure is made.

It has also been found that the DAC chip's digital compensation for analog gain reduction is not 100% accurate, which results in less than perfect linearity. See below for the measurements of linearity calculated from the above two sweeps:
View attachment 443077
View attachment 443078

Does this cause some unwanted audible effects? I doubt it. Just an interesting observation.

No you didn't, Wolf whose obsessed with chasing that tale down and in the end did but you won't like L&P pricing. For me simple - 2 dB to CS amp stage solves the problem on less than optimal implementations (it's really about fast optimal rail switching for H op mode in CS43131 case). For 25$ and +6 dB of output unbalanced this will remain to be taught to beat for a long time and it keeps it up very good under sensitive IEM's (12~16 Ohms load under 50 mV). To some extent even older CS DAC IP's had a same problem (like one's in Creative SB Z range digital clipping). Don't know why I ended up surrounded with CS DAC's especially on the go. I really regret Ivan never hooked up as we would by now had 120 dB (balanced, 117~118 unbalanced) pocket one's at much better prices (than L&P most definitely).

 

[Eddie]

For the Jcally JM20 Max I measured crosstalk of around -73 dB, no matter what frequency.

The specification is an incredible 140dB, although there is no information on load resistance, frequency, or direction.
Even in the best case scenario (e.g., no load, around 1kHz, optimal direction) I doubt this value.
With a maximum output of 2.5V RMS, 140dB lower would result in 0.25µV RMS. That's below the noise of this device, which should be around 1µV.
Can a sine be measured if it actually disappears into noise?
Even for devices with balanced outputs, I've never read a value of 140dB for crosstalk.
With unbalanced outputs, 120dB might be possible.
But 140dB? Isn't that probably a typo?

 

[TNT]

not possible to understand why it should be insulting.

Because it's performance trickery i.e. not real figures. Not that it matters SQ wise (probably..) but I hate to be deliberately fooled...

//