Qudelix-5K Bluetooth DAC & Headphone Amp

[amirm]

So, for the test with Windows PC, you may need to select 44/48/88/96KHz Full Fs mode via the app.

I am using the standard Windows class driver in exclusive mode so there should not be any need to change the sample rate.

 

[AlexScan]

I currently don't have a balanced 33 ohm load so can't do that. I think you can safely interpolate from the 50 ohm load that it can produce a ton more power in balanced.

Ok, but you think Normal and/or Performance mode can improve SINAD ?

 

[Baphomet]

Ok, but you think Normal and/or Performance mode can improve SINAD ?

Connected to USB, it automatically switches to Performance mode. Normal will likely improve the numbers, but the output will be only 1Vrms (although this will make an interesting comparison to other "dongles").

There are a lot of settings to play with and compare (Standard vs Performance, Filters, setting correct developer options for Bluetooth).

 

[bobbooo]

There's no such an application in the market.
ES9218p is a DAC/AMP, which menas D/A converter with integrated HPAMP.
ES9218p receives the digital audio through I2S interface.

Hi, and thanks for what seems to be a great product you've created I understand the ES9218P is purely a DAC/amp, but where do the Kalimba DSP cores embedded in the QCC5124 that process the PEQ come into the chain? I thought it might be something like ES9218P>QCC5124(ADC)>Kalimba>QCC5124(DAC), which I thought could explain why the SINAD of the Qudelix 5K is lower than other devices using the ES9218P (e.g. the LG G7), and seems to be what you would expect from adding the THD of the ES9218P (-114dB) with the QCC5124's ADC (-91dB) and DAC (-90.5dB), which would equal -87.7dB, around what was measured in this review test. Unless the Kalimba cores can be used in isolation of the QCC5124's DAC and something else is causing the 5K's lower than expected SINAD?

Oh and about the PEQ, as I understand, each individual filter gain is currently limited to +/-12dB in the app. Several EQ profiles created by Oratory here, and AutoEQ here, require higher (or lower) gain values for some filters than this +/-12dB limit e.g. when a notch filter with large negative gain is needed within the bandwidth of a shelf filter with positive gain. Would it be possible for you to remove this filter gain limit in a firmware/app update (or at least increase the limit, to say +/-15dB)? I am considering purchasing the Qudelix 5K myself, but I don't think I will if this limit remains, as I have headphones/IEMs that I use with Oratory's settings which require filter gain over this limit, which would be a real shame as otherwise the 5K seems to be a fantastic little device.

 

[PeteL]

Hi, and thanks for what seems to be a great product you've created I understand the ES9218P is purely a DAC/amp, but where do the Kalimba DSP cores embedded in the QCC5124 that process the PEQ come into the chain? I thought it might be something like ES9218P>QCC5124(ADC)>Kalimba>QCC5124(DAC), which I thought could explain why the SINAD of the Qudelix 5K is lower than other devices using the ES9218P (e.g. the LG G7), and seems to be what you would expect from adding the THD of the ES9218P (-114dB) with the QCC5124's ADC (-91dB) and DAC (-90.5dB), which would equal -87.7dB, around what was measured in this review test. Unless the Kalimba cores can be used in isolation of the QCC5124's DAC and something else is causing the 5K's lower than expected SINAD?

Oh and about the PEQ, as I understand, each individual filter gain is currently limited to +/-12dB in the app. Several EQ profiles created by Oratory here, and AutoEQ here, require higher (or lower) gain values for some filters than this +/-12dB limit e.g. when a notch filter with large negative gain is needed within the bandwidth of a shelf filter with positive gain. Would it be possible for you to remove this filter gain limit in a firmware/app update (or at least increase the limit, to say +/-15dB)? I am considering purchasing the Qudelix 5K myself, but I don't think I will if this limit remains, as I have headphones/IEMs that I use with Oratory's settings which require filter gain over this limit, which would be a real shame as otherwise the 5K seems to be a fantastic little device.

(QCC5124 BT and USB receivers)>QCC512 Kalimba DSP>QCC512 IIS digital out>ES9218P

 

[bobbooo]

(QCC5124 BT and USB receivers)>QCC512 Kalimba DSP>QCC512 IIS digital out>ES9218P

Thanks, so the Kalimba cores can be used in isolation then. If correct (and it seems to be considering @wslee has liked your post), what do you think is the cause of the ~25dB higher THD of the Qudelix 5K compared to other devices using the ES9218P like the LG G7 or the Dense Audio Adapt Reference? I would say maybe it's the Kalimba DSP processing but I believe EQ/DSP was disabled for the measurements taken in this review. That just leaves the QCC5124 USB receiver and digital out, but I don't see how that could add 25dB of THD when there's no D/A conversion and it's essentially just passing a digital signal to the ES9218P, no?

 

[PeteL]

Thanks, so the Kalimba cores can be used in isolation then. If correct (and it seems to be considering @wslee has liked your post), what do you think is the cause of the ~25dB higher THD of the Qudelix 5K compared to other devices using the ES9218P like the LG G7 or the Dense Audio Adapt Reference? I would say maybe it's the Kalimba DSP processing but I believe EQ/DSP was disabled for the measurements taken in this review. That just leaves the QCC5124 USB receiver and digital out, but I don't see how that could add 25dB of THD when there's no D/A conversion and it's essentially just passing a digital signal to the ES9218P, no?

The "Dense Audio Adapt" If I read correctly, is a 209$ Dac, with no dsp, no battery, no Antenna. You certainly expect it to have better numbers, where from, impossible to know really, maybe with a teardown. The G7 is a much less powerful device, and has no USB input, so such comparisons are difficult, because we are not measuring the same thing. It's a misconception to think that anything digital is perfect and equal. I couldn't tell you where theses limitations are from, it can be many things., but a SOC like QCC5124 sure does a lot of stuff in one single chip. It is not just "passing a digital signal to the ES9218P"

 

[bobbooo]

The "Dense Audio Adapt" If I read correctly, is a 209$ Dac, with no dsp, no battery, no Antenna. You certainly expect it to have better numbers, where from, impossible to know really, maybe with a teardown. The G7 is a much less powerful device, and has no USB input, so such comparisons are difficult, because we are not measuring the same thing. It's a misconception to think that anything digital is perfect and equal. I couldn't tell you where theses limitations are from, it can be many things., but a SOC like QCC5124 sure does a lot of stuff in one single chip. It is not just "passing a digital signal to the ES9218P"

The Dense Audio Adapt Reference is a portable DAC/amp, and it's $129 (minus a $20 new member discount making it $109, exactly the same as the Qudelix 5K). The LG G7 has the same max 2Vrms output in high impedance mode as the Dense Audio and the Qudelix's unbalanced output (the THD measurements of which we're comparing to here). I never said "anything digital is perfect and equal", I was just wondering how exactly a relatively large 25dB difference in THD could come about in a purely digital signal path (with DSP disabled) leading to the same DAC/amp chip at the same output level. It looks like the developer of Qudelix, @wslee is the only one who might know. Regardless, the 5K's THD as is will add an inaudibly tiny amount of distortion to any headphone in practice, so these questions are largely academic.

 

[bobbooo]

Need to review the not so new fiio btr5 as a direct competitor.

Fiio BTR5, pleasseee...

The FiiO BTR5 has been measured by Soomal here, showing pretty much the same performance as the Qudelix 5K, but the latter is smaller and lighter, with a longer battery life and a parametric equalizer (that can also be used over LDAC), so I don't see a good reason for getting the BTR5 over the 5K really.

 

[faheem]

The FiiO BTR5 has been measured by Soomal here, showing pretty much the same performance as the Qudelix 5K, but the latter is smaller and lighter, with a longer battery life and a parametric equalizer (that can also be used over LDAC), so I don't see a good reason for getting the BTR5 over the 5K really.

Aptx LL, which the 5k lacks.

 

[fgfgfg]

@amirm could you measure AAC codec as well?

 

[bobbooo]

Aptx LL, which the 5k lacks.

Only really useful for hardcore mobile gamers really, which is a pretty small niche. Unlike the BTR5, the 5K supports the newest aptX Adaptive codec, which offers the best balance between quality and latency for gaming and videos. The Qudelix also supports LDAC with a 100ms latency mode as opposed to the BTR5's 200ms.

 

[faheem]

Only really useful for hardcore mobile gamers really, which is a pretty small niche. Unlike the BTR5, the 5K supports the newest aptX Adaptive codec, which offers the best balance between quality and latency for gaming and videos. The Qudelix also supports LDAC with a 100ms latency mode as opposed to the BTR5's 200ms.

As well as pc gamers and tv/movie watching. 100ms is still too high for gaming or movie/tv. the 5k is great, but the Btr5 has it's place and, has a substantial benefit with it's ll ability over the adaptive stuff, of which there are currently no transmitters.

 

[Haint]

I'm not so sure if I'm allowed to write a post here.

But, you need to check the 5K USB FS setting first.
Unlike other USB DAC, 5K USB DAC offers an option to change the device supported sample rate.
Please go to the app INPUT --> USB, and check the device supported sample rate.

In the beginning, 5K delivers its supported sample rates to the USB HOST, and the USB HOST selects one of them.

Thanks!

@wslee is it possible to get a firmware update that disables the USB DAC's "OS Hardware Volume" or to lock it to 100%? JDS Labs just released such an update for their Atom to address issues with Playstation 5. Playstation consoles do not expose a DAC volume control in the OS and as such your product seems to be limited to ~50% volume (when compared to a PC) which does not allow sufficient headroom to drive most headphones. In other words the Playstation OS drops the max volume ceiling on your product ~30dB which effectively renders it unusable for that purpose.

 

[amirm]

@amirm could you measure AAC codec as well?

I can't. My Macbook Pro refuses to work and I don't have a working iPhone right now.

 

[bobbooo]

As well as pc gamers and tv/movie watching. 100ms is still too high for gaming or movie/tv. the 5k is great, but the Btr5 has it's place and, has a substantial benefit with it's ll ability over the adaptive stuff, of which there are currently no transmitters.

Most TVs have adjustable A/V sync options these days, as does PC video player software (and even mobile apps e.g. VLC), so low latency isn't really needed for video. Adaptive is effectively Qualcomm's replacement for LL, incorporating low latency technology itself - they've retired LL due to poor market adoption, probably because the cross-section of people who want ultra-low latency (e.g. hardcore gamers) and wireless at the cost of audio quality is pretty small. If a user really cares about either the highest sound quality or lowest latency, they'll go wired (the latter goes for gaming mice and keyboards too). Oh and there are plenty of smartphones that support aptX Adaptive transmitting..

 

[mSpot]

they've retired LL due to poor market adoption, probably because the cross-section of people who want ultra-low latency (e.g. hardcore gamers) and wireless at the cost of audio quality is pretty small.

A big factor was that aptX LL requires a separate antenna. Manufacturers don't want to add an extra antenna only for LL.

https://en.wikipedia.org/wiki/AptX#aptX_LL
AptX Low Latency requires a dedicated, wireless antenna, so it did not achieve much adoption in smartphones and was retired by Qualcomm in favor of aptX Adaptive​

 

[bobbooo]

@Cahudson42 or any other Qudelix 5K owners, would you mind checking what the max possible positive and negative gain is for each filter in the parametric equalizer? (After updating to the latest firmware and app version.)

 

[Cahudson42]

@bobbooo
With firmware v1.5, it appears the 12 db max has been removed. I was able to enter 24 in all the filter types. Each time it 'took' and the graph reflected it.

Unfortunately, updating causes a minor problem. All the PEQ custom names were deleted and replaced by default Custom 1 etc. But the actual profiles were retained.

 

[Baphomet]

Unfortunately, updating causes a minor problem. All the PEQ custom names were deleted and replaced by default Custom 1 etc. But the actual profiles were retained.

This happens every time I re-connect to 5k. Restarting the app solves this and all custom profile names are visible again.