Some notes on Fiio KA17

[CuriousAli]

Due to audiophile uncontrollable reasons, KA17 and Neutron HIFI V1 DAC are on my 'not-needed-but-wanted' list.
Any thoughts on KA17 vs Neutron , strictly for phone usage with Tidal/ streaming ?

I haven't used Neutron HIFI V1 DAC, but given your needs I think you should compare their power usage since you want to use them with your phone.
And I saw the Neutron one can do EQ per channel which is nice if you need it.
They differ in maximum sampling rate (384 vs 768khz) and DSD support (256 vs 512), but I personally don't care about that.

 

[d1m0n]

Hello everyone,

I was comparing the PEQ response on my KA17 with other PEQ implementations and stumbled upon a problem.

I noticed that the Fiios PEQ frequency response for a peak filter differs from the others (Apple AUNbandEQ, Mac SoundSource, Qudelix 5K, and Neutron DAC V1). All other PEQ implementations produce the same measurable change when the same filter is applied. However, to achieve approx. the same response a different Q value must be used on the KA17. The Q deviation also increases with the gain value, as far as I can tell.

I don't have any professional measuring equipment. Just a clone coupler in a flatplate and REW for basic measuring. I think it's perfectly sufficient for this purpose. Perhaps one of you can double-check and confirm this?

I discovered the same problem with the BTR17 in USB connection mode. PEQ via Bluetooth worked fine. A similar problem was also observed with the KA15 shelf filter: https://www.audiosciencereview.com/...e-dac-headphone-amp-review.62928/post-2365957

I reported this to FiiO Support some time ago for the KA17. So far, I've only got an answer that my issue was forwarded to R&D. Unfortunately, the problem still exists with the current released firmware version 2.25.

I hope Fiio might give higher priority to fix it if more people notice and report it or someone with better contact to them from the forum report the issue.

BR

 

[jkim]

Hello everyone,

I was comparing the PEQ response on my KA17 with other PEQ implementations and stumbled upon a problem.

I noticed that the Fiios PEQ frequency response for a peak filter differs from the others (Apple AUNbandEQ, Mac SoundSource, Qudelix 5K, and Neutron DAC V1). All other PEQ implementations produce the same measurable change when the same filter is applied. However, to achieve approx. the same response a different Q value must be used on the KA17. The Q deviation also increases with the gain value, as far as I can tell.

I don't have any professional measuring equipment. Just a clone coupler in a flatplate and REW for basic measuring. I think it's perfectly sufficient for this purpose. Perhaps one of you can double-check and confirm this?

View attachment 472866

View attachment 472868

I discovered the same problem with the BTR17 in USB connection mode. PEQ via Bluetooth worked fine. A similar problem was also observed with the KA15 shelf filter: https://www.audiosciencereview.com/...e-dac-headphone-amp-review.62928/post-2365957

I reported this to FiiO Support some time ago for the KA17. So far, I've only got an answer that my issue was forwarded to R&D. Unfortunately, the problem still exists with the current released firmware version 2.25.

I hope Fiio might give higher priority to fix it if more people notice and report it or someone with better contact to them from the forum report the issue.

BR

The Q factor being mistranslated even for peak filters is not a small problem. I thought this issue is only with the BTR15. It is clear that FiiO engineers simply ported similar PEQ code from the BTR15 to some other devices released later.

Let's ask FiiO to provide solutions to affected devices. FiiO has been responsive to the DRE distortion issue. Hopefully they will do the same.

 

[jkim]

Measured the BTR15 (both USB and BT connections) and KA15 for their peak and low-shelf filters.

Peak filters first:

Issues to be solved:

• BTR15 in USB connection: The Q factor input NOT correctly translated. It seems that Q = 1.0 is misinterpreted as Q = 0.707.
• BTR17 (USB mode) and KA17: Same issue was observed.
• BTR15 in Bluetooth connection: Although the response is overall correct, there is a peaking around 20 kHz (see my next post for comparison with Qudelix 5K).

Low-shelf filter responses:

Note. The latest firmware updates (V4.4 for Bluetooth and V125 for XMOS) of the BTR15 were installed before testing.

Issues to be solved:

• BTR15 in USB connection: Completely wrong response.
• KA15: The Q factor input NOT correctly translated. It seems that Q = 0.71 is misinterpreted as a "shelf slope" S = 0.71 (or Q = 0.5).
• BTR17 (USB mode) and KA17: Same issue of Q factor mistranslation was observed.
• BTR15 in Bluetooth connection: Although the response is overall correct, there is a peaking around 20 kHz (see my next post for comparison with Qudelix 5K).

I will contact FiiO and ask them to provide solutions.

 

[jkim]

I further tested the BTR15's PEQ performance in Bluetooth connection by comparing it with the Qudelix 5K:

Note. The Bluetooth (LDAC) playback of test signals was from the USB Audio Player Pro (UAPP) installed on Google Pixel 8.

There is certainly a problem with the BTR15's LDAC performance, not achieving the sample rate of 96 kHz.

Measured the dynamic range of both devices' LDAC modes:

It seems that the BTR15's Bluetooth LDAC mode has some kind of clocking/jitter issue, making its FFT unstable---the skirt around the test tone is fluctuating over rolling averages.

The Qudelix 5K's Bluetooth performance (and its PEQ accuracy) is exemplary, explaining why it has been regarded as the best of its kind for quite a while.

 

[CuriousAli]

Ah, that's a shame, I got it for its PEQ feature mainly.
I want to report it to Fiio too, should I say this?

The Q factor input NOT correctly translated. It seems that Q = 1.0 is misinterpreted as Q = 0.707.

 

[jkim]

Ah, that's a shame, I got it for its PEQ feature mainly.
I want to report it to Fiio too, should I say this?

Yes, and refer to the measurements of @d1m0n and mine in this ASR thread. I actually wrote an email to FiiO Support. Let's see how (and whether) they respond.

 

[d1m0n]

Due to audiophile uncontrollable reasons, KA17 and Neutron HIFI V1 DAC are on my 'not-needed-but-wanted' list.
Any thoughts on KA17 vs Neutron , strictly for phone usage with Tidal/ streaming ?

Litle late but here are my thoughts,

I have both devices (Spotify user).

Solely for phone usage I’d personally take Neutrons V1 due to weight and especially battery drain (even worse if you activate Desktop mode). Of course you could use the estick of them or other power bank as additional power. But this is nothing one want hang on the phone as a dongle.
Fiios App + Web App is a nightmare too imho.

Other than that the KA17 has physical buttons and more juice via 4.4mm + 3 user eq presets (yet suspect to me? PEQ implementation) you can switch w/o app on the go. May be an advantage of course if this is something you need or want.

 

[CuriousAli]

Yes, and refer to the measurements of @d1m0n and mine in this ASR thread. I actually wrote an email to FiiO Support. Let's see how (and whether) they respond.

Thank you, so far their only reply has been they have forwarded the info to their engineers, I will post an update if they reply again.

 

[jkim]

Thank you, so far their only reply has been they have forwarded the info their engineers, I will post an update if they reply again.

I have only received the same response and not heard anything since then. Will make another inquiry.

 

[feconicu]

@jkim @CuriousAli Any updates from them? Is this device issue or is it a problem with their app (which makes it every other device using it)

 

[jkim]

@jkim @CuriousAli Any updates from them? Is this device issue or is it a problem with their app (which makes it every other device using it)

FiiO does not respond. My advice is that one should not buy a FiiO product for its DSP/PEQ feature unless its functionality is thoroughly tested by a third party. Their hardware may be designed properly. But not their software. If you are not going to utilize PEQ, then it may be fine.

 

[CuriousAli]

Yep, no further update since September for me either.

 

[d1m0n]

No response to me either. By the way I tried the new QX13 of them. PEQ response was deviating compared to other PEQ too for me. I don’t even bother to report it to them and returned it a day later.

 

[reallystate]

Snowsky Melody Built-in Equalizer Measurements appeared and unfortunately, the problem described on this page also partially appears in this device

 

[jkim]

Incorrect Q for shelving filters is not a big issue. The automatic -12 dB preamp cut in all FiiO devices is a bad design. I understand they might might want to reduce a return rate due to user error in PEQ setup. I also find similar auto preamp behavior in all TTGK devices (e.g., TRN Black Pearl). Again, a bad design element.

All this makes me truly appreciate the thoughtful design of the Qudelix 5K. It is the only portable DAC with onboard PEQ on the market that has this headroom behavior achieved by DSP volume control.

 

[pettervo]

The automatic -12 dB preamp cut in all FiiO devices is a bad design.

I am new to this, but does this mean that there is no need to adjust the master gain in the PEQ to avoid clipping on the KA17, as long as the maximum volume increase by the PEQ is less than 12 db?

BTW: I also sent a request to FIIO regarding the q-factor issued raised in this thread, linking to this post.

 

[reallystate]

I am new to this, but does this mean that there is no need to adjust the master gain in the PEQ to avoid clipping on the KA17, as long as the maximum volume increase by the PEQ is less than 12 db?

BTW: I also sent a request to FIIO regarding the q-factor issued raised in this thread, linking to this post.

That's true, but it's worth remembering that multiple EQ bands can be summed. Even if each individual band has less than 12 dB of gain, multiple bands affecting the same frequency range can produce a combined gain exceeding 12 dB.
If the combined gain does not exceed 11 dB, then it is worth manually increasing the preamp.

 

[jkim]

The biggest problem is that FiiO does not state this design choice explicitly. We want "predictable, intuitive behavior" from the device to use it properly. What if they quietly change the preamp to -15 dB or -9 dB later in a firmware release?

 

[OHtaru]

That's true, but it's worth remembering that multiple EQ bands can be summed. Even if each individual band has less than 12 dB of gain, multiple bands affecting the same frequency range can produce a combined gain exceeding 12 dB.
If the combined gain does not exceed 11 dB, then it is worth manually increasing the preamp.

It is a misunderstanding that pre-amp gain reduction equals the highest point on the EQ curve. For IIR EQ, phase rotation can cause larger peaks. Even enabling a 1000Hz -6dB EQ can cause the peak to increase (originally 0dBFS)