FiiO has released the USBDAC JA11, with an initial price of only $9 (69RMB) and support for PEQ DSD128 PCM384

[jose_user]

You should be able to install JA11 firmware using a PC I think.

I cannot install any firmware, because I bricked the dongle. Neither the TInHifi tool, nor Fiio tool can install any firmware. I think the firmware that I installed was corrupted, who knows.

Thank you for your support.

 

[jose_user]

It turns out that UAC 1.0 was the cause!
That's why I experienced this recently after I switched from UAC 2.0 to UAC 1.0 to fix a ticking sound issue on Linux.
As far as I can remember, this never happened before.
And after I switched back to UAC 2.0, this problem disappeared.
Tested on Windows, Linux, and Android.

Actually, I wanted to order the BGVP C01 with the CB1200 chip and Walkplay support, but ended up ordering the Fiio Melody instead.

I'm not 100% sure, but I think my dongle was also using UAC 1.0 (when it worked...). So that makes sense.

 

[jose_user]

Did some more tests of the JM12 with different firmware options.

The erratic, unsteady responses in FFTs are there regardless of firmware. I believe this is a clocking issue. This unsteady response is shown in jitter tests, too. Shown below is a an FFT of a J-test signal when it is normal and steady:
View attachment 463048

The above is fine performance, but the output signal is not steady. At times, the FFT also becomes:
View attachment 463051

This instability is there regardless of the FW. One thing notable is that the multitone response observed with the JA11 by @trungdtmc always occurs with the JM12 on JA11 FW:
View attachment 463054

The JM12 with its stock or TihHifi FW also occasionally shows a similar response:
View attachment 463056

Recorded the C Major test signal. First up is with the JM12 stock FW:
View attachment 463061
Not as clear distortion as with CS431xx, but definitely there's distortion along with elevated noise whenever the signal peaks. In particular, there's greater distortion while the signal level increases and decreases.

Same with the JA11 FW:
View attachment 463059

BUT with the TinHifi FW, we see different behavior:
View attachment 463058
There is still a distortion but substantially less of it when the signal level increases and decreases. More interesting is its noise level which is definitely higher than with the other two FW options, which is clearly seen when there's a lapse between signals.

It turns out that some kind of dynamic range enhancement (DRE) is involved in some FW options:
View attachment 463064
I forgot to make the same measurements of the stock FW before flashing, but I'm sure it must have DRE in action as well. Anyway, it appears that DRE is turned off with the TinHifi FW, but turned on with the other two FW options.

This means that the TinHifi FW shows the device's "real" noise performance without DRE. And unsurprisingly it also exhibits less distortion. The dynamic ranges were also measured:

​	Left Ch. DR​	Right Ch. DR​
JM12 w/ TinHifi FW​	102.2 dB​	101.6 dB​
JM12 w/ JA11 FW​	114.3 dB​	112.9 dB​

The DR of TinHifi FW is not that great, but not disastrous, either. You may hear slight hiss with a very sensitive IEM, but will be fine with others. I barely hear hiss using my 7Hz x Crinacle Zero:2.

By the way, I played with the "Noisegate" parameters in the KTmicro tool:
View attachment 463082
Unfortunately, change of any parameters there has NO effects on measurements at all. Other features like the volume and EQ work, but not this 'Noisegate' feature which is supposed to control the DRE effect.

In any case, I for one will take the elevated noise of the TinHifi FW in exchange of another FW's higher distortion.

However, all in all I wouldn't recommend this device to anyone looking for a cheap PEQ-enabled 1V headphone dongle. Even at less than $10. Why? Because there's an alternative! It is the Hi-Max CB1200AU.

Below is a sneak peak of some measurements of the Hi-Max CB1200AU. First is the recording of the C Major signal played by the Hi-Max:
View attachment 463060
About the same noise level as the JM12 on TinHifi FW, but there's no distortion at all.

And its dynamic range is also about the same as the JM12 on TinHifi FW:

​	Left Ch. DR​	Right Ch. DR​
Hi-Max CB1200AU​	102.5 dB​	104.0 dB​
JM12 w/ TinHifi FW​	102.2 dB​	101.6 dB​
JM12 w/ JA11 FW​	114.3 dB​	112.9 dB​

And it supports 8-band PEQ that works correctly. What more would you ask for at $6?

I may provide a more extended review of the Hi-Max if I get a chance later...

Hello Jkim.

First to all, thanks for such detailed research.

Due to the fact that I bricked my JCally JM12 installing the TinHifi firmware and I cannot fix it, I bought a Hi-Max CB1200AU following your recommendation. I haven't received it yet. I know it's very cheap, but I'm a bit concerned with the durability of this product. Could you tell us something about it, please? Do you think it's a "homogeneous" product, or is it a lottery to find a good one?

Thank you.

 

[jkim]

Hello Jkim.

First to all, thanks for such detailed research.

Due to the fact that I bricked my JCally JM12 installing the TinHifi firmware and I cannot fix it, I bought a Hi-Max CB1200AU following your recommendation. I haven't received it yet. I know it's very cheap, but I'm a bit concerned with the durability of this product. Could you tell us something about it, please? Do you think it's a "homogeneous" product, or is it a lottery to find a good one?

Thank you.

I wouldn't worry about its durability unless I treat it badly. I have tested several Hi-Max dongles. They are consistent. For all of them, the left channel is slightly noisier than the right ch in measurements. But I cannot hear the difference in normal use.

 

[audiosnek]

It seems so.
The https://www.audiocheck.net/audiotests_polaritycheck.phps have no effect on my headphones (Takstar HD2000).
But they do affect my 4 iems.

The JA11 firmware has a problem skipping about 0.5 seconds at the beginning of the audio (occurs on Android and PC).
Steps to Reproduce:
1. Open the music player app and play a new song.
2. Or pause playback, scroll to the beginning, wait about 5 seconds, then play again.
3. Or you can also pause playback, wait about 5 seconds, then resume playback.

This also occurs when streaming movies/videos or music videos.
This is the most annoying issue for me right now on the JA11 firmware.

Hi, just registered to reply you this and share a couple of quirks:

With my JA11 on Windows 10 running driverless, the time it takes to wake up from standby to a sound event in UAC 1.0 is noticeably higher than in UAC 2.0 mode. The CP speaker test can often completely miss the first channel for how long it takes to wake up, and also noticeably skip the very beginning of a song when directly launching a player + song under regular usage (p.e, opening music file or playlist from explorer, so to immediately play in AIMP).

Switching to UAC 2.0 reduces this delay a bit, making it less obvious under usage (while CP test remains far from OK) - pretty comparable to my euro apple dongle's delay under Windows (driverless, UAC 1.0). I guess you reached the same conclusion in the post that followed.

But then, since installing the previous v5.68.0 driver it improved further, so the issue became unnoticeable during regular usage. In the CP channel test it also seems to skip a bit less from silence at the first channel, then fully play the second more consistently.
Note that the driver is only meant for UAC 2.0 and requires Microphone input to be disabled on the app, so TRRS devices playback properly; the JA11 under driver will also be listed as "Speakers (JadeAudio JA11 (2.0))" instead of "Headphones (...)".

Now on the latest v5.74.2 driver (updated for the new JIEZI), there's a tab with an option to keep it "Always On":


It remains applied on the driver after a reboot even if this CP is manually prevented run in tray at log in. When set as "Always On", my JA11 doesn't audibly skip in the CP channel test, behaving like the SB Z. It's running on the latest v2.2 FW, in case it matters.

Cheers

PS: A bigger issue I face with my JA11 and no FW ever corrected it, is that the volume cap always must be left at max 60/60 in the app, so I can't lower it for sensible IEMs. Else, it'll very noticeably distort with a light trailing fuzz when listening to singled out instruments against silent background (as in the rumble test), both in Windows and Android. Easy to notice on this song's intro, right from 0-17s, before more instruments fill in.
But for some reason, keeping it at 60/60 but enabling the EQ flat and set to -12db (-24db effective) remains usable, rather than the fuzzy mess caused by an equivalent lower value than 60. I posted this finding on another forum at Fiio's sponsored thread, but got no comment from the rep, while a user confirmed it on his 2 units.

Finally and to note, on stock FW (v1.6?) and up to FW v2.0, when set to UAC 2.0 it produced a ticking/clicking noise while playing (solely) on my Zen3+X470 system (regardless of USB ports), while another Zen3+B450 system was free from it. FW v2.1 finally fixed this, and v2.2 remains OK on that.

 

[pers]

Hi, just registered to reply you this and share a couple of quirks:

With my JA11 on Windows 10 running driverless, the time it takes to wake up from standby to a sound event in UAC 1.0 is noticeably higher than in UAC 2.0 mode. The CP speaker test can often completely miss the first channel for how long it takes to wake up, and also noticeably skip the very beginning of a song when directly launching a player + song under regular usage (p.e, opening music file or playlist from explorer, so to immediately play in AIMP).

Switching to UAC 2.0 reduces this delay a bit, making it less obvious under usage (while CP test remains far from OK) - pretty comparable to my euro apple dongle's delay under Windows (driverless, UAC 1.0). I guess you reached the same conclusion in the post that followed.

But then, since installing the previous v5.68.0 driver it improved further, so the issue became unnoticeable during regular usage. In the CP channel test it also seems to skip a bit less from silence at the first channel, then fully play the second more consistently.
Note that the driver is only meant for UAC 2.0 and requires Microphone input to be disabled on the app, so TRRS devices playback properly; the JA11 under driver will also be listed as "Speakers (JadeAudio JA11 (2.0))" instead of "Headphones (...)".

Now on the latest v5.74.2 driver (updated for the new JIEZI), there's a tab with an option to keep it "Always On":
View attachment 475391

It remains applied on the driver after a reboot even if this CP is manually prevented run in tray at log in. When set as "Always On", my JA11 doesn't audibly skip in the CP channel test, behaving like the SB Z. It's running on the latest v2.2 FW, in case it matters.

Cheers

PS: A bigger issue I face with my JA11 and no FW ever corrected it, is that the volume cap always must be left at max 60/60 in the app, so I can't lower it for sensible IEMs. Else, it'll very noticeably distort with a light trailing fuzz when listening to singled out instruments against silent background (as in the rumble test), both in Windows and Android. Easy to notice on this song's intro, right from 0-17s, before more instruments fill in.
But for some reason, keeping it at 60/60 but enabling the EQ flat and set to -12db (-24db effective) remains usable, rather than the fuzzy mess caused by an equivalent lower value than 60. I posted this finding on another forum at Fiio's sponsored thread, but got no comment from the rep, while a user confirmed it on his 2 units.

Finally and to note, on stock FW (v1.6?) and up to FW v2.0, when set to UAC 2.0 it produced a ticking/clicking noise while playing (solely) on my Zen3+X470 system (regardless of USB ports), while another Zen3+B450 system was free from it. FW v2.1 finally fixed this, and v2.2 remains OK on that.

Thank you for the information.
My experience is only with 2.2, so I've never tried firmware before that.
Yes, I use it without drivers from FiiO (this is not my Windows device, so I can't install anything without permission).
I primarily use it on Linux and Android, so issues related to Windows and its fixes don't really affect my usage.
It's a shame because UAC 1.0 doesn't have the ticking/clicking sound issue on Linux like UAC 2.0 does, but it has the fading/delay/skip issue at the beginning of the audio on Linux/Android/Windows (UAC 1.0 + driverless), which UAC 2.0 doesn't have.
At least for me...

 

[popthere]

It remains applied on the driver after a reboot even if this CP is manually prevented run in tray at log in. When set as "Always On", my JA11 doesn't audibly skip in the CP channel test, behaving like the SB Z. It's running on the latest v2.2 FW, in case it matters.

Are there any problems when setting it to "on when needed"?

 

[audiosnek]

Setting "On when needed" (which is default) will return previous driver behavior, so expect a very small wake up delay. See my comment regarding v5.68.0, it'll then apply to v5.74.2.

 

[o0o]

Can someone check if the original JA11 has the polarity inverted? Thanks in advance!

Some quick notes based on my unit. 1. Looks like the polarity is inverted, here's the impulse response

Can't believe I missed this....

Just to confirm, the inverted polarity was found on your genuine JA11, right, @danadam ? I'd assume it's still inverted with the latest fw v2.2. Gotta report that to FiiO engineers.

 

[danadam]

Just to confirm, the inverted polarity was found on your genuine JA11, right, @danadam ?

Yes.

I'd assume it's still inverted with the latest fw v2.2.

That I don't know. I've never updated the firmware on it.

 

[o0o]

One more observation with JA11's fw - abnormal clock drift.

For a 2 min 40 sec track the drift is +135 ms

Here's an additional calculation for ppm value made by grok:

But it's not fixed, it can be a bit higher or a bit lower, but still too high, regardless of the fw verion. In UAC 1.0 mode it's definitely better and in the acceptable range.

With JM12 original fw and TinHiFi fw the clock is almost perfectly in sync with Focusrite Scarlett that was used for the test and only surpassed by 100% matching clock of CX31993.

That I don't know. I've never updated the firmware on it.

Why not? FiiO fixed and improved a few things there.

 

[jkim]

Can't believe I missed this....

Just to confirm, the inverted polarity was found on your genuine JA11, right, @danadam ? I'd assume it's still inverted with the latest fw v2.2. Gotta report that to FiiO engineers.

Why hung up about the JA11/JM12? It has not only the worst noise floor with DRE disabled, but also exhibits peculiar, unstable responses in FFTs, as shown in this thread. No reason to consider it over other cheap dongles...

 

[danadam]

Why not? FiiO fixed and improved a few things there.

Back then I had a cursory look into how to do the upgrade and, IIRC, the whole process didn't fill me with confidence, i.e., download an app from weird looking url, reports with problems, up to bricking the dongle, etc. It may be better now but I don't really use the dongle, so I don't feel like doing it

 

[o0o]

Why hung up about the JA11/JM12? It has not only the worst noise floor with DRE disabled, but also exhibits peculiar, unstable responses in FFTs, as shown in this thread. No reason to consider it over other cheap dongles...

In a $5 range it's a perfectly viable option with a built-in eq.

For example, JM12 (KT02H20) with TinHiFi fw even with its limitations and drawbacks offers more power and resolution comparing to Hi-Max (CB1200AU), which is more on the level of AB13X, subjectively speaking. Durabilty of JM12 seems better too, particularly the cable part. So yeah, I'd say it definitely worth consideration.

If we ignore the built-in eq part and focus on sound quality alone, CX31993 would be my choice in $5 range. KT02H20 and ALC5686 are ok too. CB1200AU just doesn't cut it for me.

Back then I had a cursory look into how to do the upgrade and, IIRC, the whole process didn't fill me with confidence, i.e., download an app from weird looking url, reports with problems, up to bricking the dongle, etc. It may be better now but I don't really use the dongle, so I don't feel like doing it

It's safe, fast and easy on windows. But I get your point.

 

[jkim]

In a $5 range it's a perfectly viable option with a built-in eq.

For example, JM12 (KT02H20) with TinHiFi fw even with its limitations and drawbacks offers more power and resolution comparing to Hi-Max (CB1200AU), which is more on the level of AB13X, subjectively speaking. Durabilty of JM12 seems better too, particularly the cable part. So yeah, I'd say it definitely worth consideration.

If we ignore the built-in eq part and focus on sound quality alone, CX31993 would be my choice in $5 range. KT02H20 and ALC5686 are ok too. CB1200AU just doesn't cut it for me.

Sure, the JM12 has an advantage for its output. It is in fact not a 1V dongle. Its max output under no load reaches 1.25 Vrms.

But how can you accept the following response from it?

And this and this?


And while these responses occur randomly, they do not occur rarely. Every time I put it on my testbench (I did quite a few times), I always observe these unstable signals from it without an exception. Same regardless of signal levels and regardless of FW.

How can one recommend this DAC to others? I cannot.

And its PEQ app on the KTMicro tool is not reliable, either. Sometimes works, other times not. Sometimes the center frequency doubles. But it's just a minor issue compared to the above.

By the way, here's a take we can have about the Hi-Max vs. CX31993 based on measurements data:

Best cheap USB-C headphone dongles?

JCALLY JM7 Max measurements by huihifi.com

www.audiosciencereview.com

 

[markanini]

I wonder if there's an audio sample that would make it easy to audibly detect the FFT distortions on KT02H20. I've tried standard test signals like sine waves, white noise and music and nothing sound abnormal.

BTW can someone with a CX31993 dongle confirm that the audible crackling on the Dune soundtrack disappears on the lower volumes?

 

[jkim]

Analysis of Maximum Loaded Output of JA11/JM12 versus Hi-Max CB1200AU

I realized that no one has tested the JM12's maximum output capability under different loads. Also, an ASR member (@o0o in Post #534) said "JM12 offers more power and resolution comparing to Hi-Max (CB1200AU)". It piqued my interest as I have both devices at hand.

Note that these dongles need some tweaking to unleash their maximum output:

• JM12 w/ TinHifi FW (which I tested): "DAC L&R" level on the KTmicro tool's Volume tab must be set to 3.0 dB.
• Hi-Max CB1200AU w/ latest FW: "DAC Playback Volume" on the WalkPlay app's EQ Effect tab must be set to +3.0 dB.

First, THD+N versus output voltage under 300 ohm load:

Note. Small steps were used near clipping so that each device's maximum output level can be precisely measured.
Right channel is shown with both channels being driven.
The two devices were tested in exactly the same condition.

The JM12 produces higher output under 300 ohm load. In my view, however, 300 ohm load is not a critical testing condition since it is not likely for people with hi-impedance headphones to use these dongles.

Next under 32 ohm load:

Under 32 ohm, their max outputs are virtually identical. But the THD level of the Hi-Max---noise is actually lower for the Hi-Max---starts rising before 0.3 V. One may think this indicates the Hi-Max's greater stress under low impedance. Actually it does not. It is simply the Hi-Max's characteristic of harmonic products, which is not necessarily a bad property. Read further to see why.

Under 12 Ohm load:

Note. The weird looking curve b/w 280 - 620 mV for the JM12 is due to its random, unstable noise elevation. To remove it, this stepped test should've been done many times only to spot one stable run, which is absurd to do.

Under this very low impedance, the Hi-Max finally surpasses the JM12 albeit by a small margin. Note that the Hi-Max's measured output voltage reflects the effect of its 2.5 ohm output impedance---so, a voltage drop due to its output impedance is not a concern since clipping is the main limiting factor here.

Now, I will show how nice the Hi-Max's distortion behavior is. According to the above tests, its THD+N is higher than the JM12's at higher output levels. It is entirely due to THD. But THD is just a single summary value. Here is an FFT spectrum of the Hi-Max @ 0.55 Vrms into 12 Ohm:

As you can see, the higher THD is mostly due to low-order (2nd-order, in particular) harmonics. The spectrum is very clean for higher-order harmonics.

Here's an FFT of the JM12:

Although the THD value is lower than the Hi-Max's, its tall-order harmonics are nasty. In fact, both devices would be fine audibly. But if I have to pick one response of the above two, I will choose the Hi-Max's any day. Noise levels at higher frequencies are worse, too, compared to the Hi-Max.

 

[jkim]

I wonder if there's an audio sample that would make it easy to audibly detect the FFT distortions on KT02H20. I've tried standard test signals like sine waves, white noise and music and nothing sound abnormal.

I would not bother to do a listening test. Two reasons why a listening test is tricky:

1. The phenomenon occurs randomly. It's difficult to spot by hearing rather than seeing FFTs.
2. It is not a type of distortion/noise that is easy to notice like crackling sound. In terms of SNR with a single tone, it lowers SNR roughly by 10 dB. Sure, some golden ears should be able to notice it if spotting it at a right time with a right signal (e.g., at certain frequencies).

 

[o0o]

But how can you accept the following response from it?

From pure tech/measurement perspective it doesn't look good, very true. From listener perspective though, it sounds pretty fine. Good dynamics and resolution, tight fast lows, open mids and decent highs for $5.

...offers more power and resolution comparing to Hi-Max (CB1200AU)...

Well, better resolution might not be necessarily the case here, gotta correct myself. Did another and this time the last comparison between the two (volume matched +/- 0.1dB) and on certain tracks with fast and busy arrangements CB1200AU managed to deliver a bit more details and separation in high frequencies. Also it is "cleaner" sounding than KT02H20, which has some saturation/harmonics added across the entire frequency range. Probably the audible side of these measured anomalies you generously reported, @jkim . Many people may actually like this added flavor.

It is worth mentioning that the 3 firmwares that we're actively discussing in this thread have 3 different tunings. If we take the JA11 FW as a flat baseline (after correcting the polarity inversion) the stock JM12 FW will have slightly more "bass and treble" (or slightly recessed midrange) in comparison and TinHiFi will be the one with slightly boosted midrange.

I realized that no one has tested the JM12's maximum output capability under different loads. Also, an ASR member (@o0o in Post #534) said "JM12 offers more power and resolution comparing to Hi-Max (CB1200AU)". It piqued my interest as I have both devices at hand.

First of all thank you again for your work with all these measurements and analysis!
One thing that makes the power-difference simple for me is a planar IEM (16Ω in my case). When maxed-out CB1200AU (+6dB) is roughly 2dB lower comparing to KT02H20 (+3dB), when both set at 0dB KT02H20 is obviously more powerful too.

Am I liking CB1200AU more now? No, not really. It sounds kinda "cheap and lifeless" to me. I still have a sensation of it being not resolving enough and occasionally muddy and smeared sounding. Is it its tuning or some real-time dsp-processing - I don't know, probably just my personal perception. Frankly, I'm not using my JM12 either, because I don't like its sound signature and I have much better alternatives too.

For all interested, imo both Hi-Max and JM12 (with JA11 fw) are fine if you need an on-device eq strictly under $5.
- JM12 with JA11 fw supports FiiO Control app (no registration/login, incl. web version) and FiiO drivers (incl. ASIO), supports up to 384kHz including muliples of 44.1 and 48, accepts DSD64 and DSD128 (with volume control unlike stock JM12). This FW has inverted polarity.
- Hi-Max supports Walk Play app (requires registration/login, incl. web version), accepts up to 96kHz without downsampling and up to 384kHz with downsampling, (only 44.1 and multiples of 48 are supported).

 

[jkim]

One thing that makes the power-difference simple for me is a planar IEM (16Ω in my case). When maxed-out CB1200AU (+6dB) is roughly 2dB lower comparing to KT02H20 (+3dB), when both set at 0dB KT02H20 is obviously more powerful too.

Although I cannot comment on your subjective impressions, I can on the observation you made above based on my measurements. At 16 ohm, Hi-Max should be slightly more powerful than KT02H20. The two devices have different gain levels when set to the same preamp gain value. For example, if you set both devices to 0 dB, Hi-Max will be gain/voltage-limited whereas KT02H20 is current-limited. The Hi-Max can go a bit louder if you increase its preamp gain. Firmware can be another factor. There was a bug in the Hi-Max's previous firmware concerning preamp gain adjustment. If you do your test properly again, you will have a different result.