Fiio K13 R2R - Owners Thread

[windhoek]

All analog outputs in the back behave the same way; when you choose Pre out, the volume IC attenuates the signal. When you choose Line out mode, volume fixes to max.

Thank you, that helps with regard to how I might use a K13 should I go ahead and buy one: in my case, two variable unbalanced outputs would be better than one fixed and one variable output.

 

[SoySalsaPicante]

Thank you, that helps with regard to how I might use a K13 should I go ahead and buy one: in my case, two variable unbalanced outputs would be better than one fixed and one variable output.

I used it in a similar manner, with the balanced out going to monobloc power amps and the single-ended out going to a subwoofer (this is not the place to discuss bass management). The subwoofer and speaker levels, once matched initially, remained level when adjusting volume at the K13 R2R.

 

[gerard11]

You should ask yourself how it it possible to sound different if they measure the same?

I don’t have other measurement device but my iPhone with RTA software called RTA Audio and I set to 512 or 1024 points resolution and set the speed to Low and display and line.

Then I play 2:24 duration sine sweep on Youtube under the following link

I only take the peak (blue line) from the measurement thus I can always get same result. Despite very subtle difference in the graph (few small DBs), there is always difference in measurement when I swapped opamp, speaker cables, put on and put off power conditioner, swap DAC.

Below are samples of measurement when I try to compare my audio system when grounding (earth) is not connected due to adapter plug (img_8094) and when it is connected (img_8095).

 

[antcollinet]

I don’t have other measurement device but my iPhone with RTA software called RTA Audio and I set to 512 or 1024 points resolution and set the speed to Low and display and line.

Then I play 2:24 duration sine sweep on Youtube under the following link

I only take the peak (blue line) from the measurement thus I can always get same result. Despite very subtle difference in the graph (few small DBs), there is always difference in measurement when I swapped opamp, speaker cables, put on and put off power conditioner, swap DAC.

Below are samples of measurement when I try to compare my audio system when grounding (earth) is not connected due to adapter plug (img_8094) and when it is connected (img_8095).

If you are using a microphone to record speakers, then you will get run to run varitions as big as that. Your own position in the room (alone) will alter the sound field massively more than any changes from op amps, cables, power conditioner DAC etc etc.

You are not hearing changes from those things.

 

[gerard11]

If you are using a microphone to record speakers, then you will get run to run varitions as big as that. Your own position in the room (alone) will alter the sound field massively more than any changes from op amps, cables, power conditioner DAC etc etc.

You are not hearing changes from those things.

I used a tripod and place the iphone on my listening position (home audio setup) with the phone height at my ear level. Then I moved away before the measurement started.

As I managed to get same result when nothing was modified or swapped from the system, I considered the measurement being consistent and reliable to what it does (measuring peak level output from a sine sweep that run from 20hz - 20,000hz). Definitely this does not measure anything that relate to timing, multi instruments, distortion, or other measurement dimensions. But this is the simplest measurement for a consumer user to start measuring the tonal (frequency response) output from their audio gears.

I have swapped NE5532, LF353, AD823, OPA2134, OPA2604, LME4562, Burson v5i, burson v5vivid, burson v6 classic, burson v6vivid and burson v7 classic and use this measurement approach for the last 8 years.

If you overlay the image and make one slightly above and transparent then its easier to see what that swapped component does to the overall tonal (freq. Response) of the system.

 

[gerard11]

My intention was simply to hear how well the equalizer performed—that was all.

I use FIIO K13 and when trying to use PEQ, what I observed
1. The volume were lower when PEQ activated in FIIO app. If it is bypass EQ not so much. But when selecting custom user it went down quite a lot.
2. When trying to increase the gain to bring the level back to bypass or EQ off, I started to notice clipping in the music.

Similar observation also happened when choosing PRE with volume 100 compared to LO (bypass volume control).

So if you have high power amp (200W) the use of the PEQ with lower volume may still be OK if you are listening at moderate to low volume. But if you have low wattage amp, that reduced DB from activating the EQ alone will affect your dynamic range head room if you need to increase the volume from the amp to compensate the lower volume.

Volume aside, the PEQ in FIIO K13 is quite rich, with 10 slot and you can choose the frequency (can be customized), select filter type (peak, high shelf, low shelf, etc), Q and Gain. It can save the EQ setting on the device so you can select the EQ using the remote.

 

[gerard11]

I use FIIO K13 and when trying to use PEQ, what I observed
1. The volume were lower when PEQ activated in FIIO app. If it is bypass EQ not so much. But when selecting custom user it went down quite a lot.
2. When trying to increase the gain to bring the level back to bypass or EQ off, I started to notice clipping in the music.

Similar observation also happened when choosing PRE with volume 100 compared to LO (bypass volume control).

So if you have high power amp (200W) the use of the PEQ with lower volume may still be OK if you are listening at moderate to low volume. But if you have low wattage amp, that reduced DB from activating the EQ alone will affect your dynamic range head room if you need to increase the volume from the amp to compensate the lower volume.

Volume aside, the PEQ in FIIO K13 is quite rich, with 10 slot and you can choose the frequency (can be customized), select filter type (peak, high shelf, low shelf, etc), Q and Gain. It can save the EQ setting on the device so you can select the EQ using the remote.

You can calculate your amp power, speaker sensitivity and listening distance to obtain the max DB you can get without distortion / clipping from this link by Geoff the Grey:

Understanding Speaker Sensitivity - Geoff the Grey Geek

A simple explanation on how to understand and use Speaker Sensitivity specifications. Includes calculators to determine how loud your speakers will be.

geoffthegreygeek.com

Music can have dynamic range of 20-25 DB while movies like Horror movies can have dynamic range up to 45DB.

So with low wattage amp, that reduced DB from activating the PEQ may cause the amp to play at higher wattage just to play at the same volume level and thus leave less room for the dynamic range.

 

[RandomEar]

So with low wattage amp, that reduced DB from activating the PEQ may cause the amp to play at higher wattage just to play at the same volume level and thus leave less room for the dynamic range.

That's generally not correct. Ignoring extreme EQs changing the balance between bass and treble, same SPL naturally means same amplifier power. Where else would the power from the amp go, if not into the sound?

You may be operating at a higher gain with a significant negative pre-amp gain in your EQ setup, but higher gain ≠ higher power.

 

[RandomEar]

I used a tripod and place the iphone on my listening position (home audio setup) with the phone height at my ear level. Then I moved away before the measurement started.

As I managed to get same result when nothing was modified or swapped from the system, I considered the measurement being consistent and reliable to what it does (measuring peak level output from a sine sweep that run from 20hz - 20,000hz). Definitely this does not measure anything that relate to timing, multi instruments, distortion, or other measurement dimensions. But this is the simplest measurement for a consumer user to start measuring the tonal (frequency response) output from their audio gears.

I have swapped NE5532, LF353, AD823, OPA2134, OPA2604, LME4562, Burson v5i, burson v5vivid, burson v6 classic, burson v6vivid and burson v7 classic and use this measurement approach for the last 8 years.

If you overlay the image and make one slightly above and transparent then its easier to see what that swapped component does to the overall tonal (freq. Response) of the system.

Measuring op amps with the speaker, room and microphone in the chain of measurement is just introducing more sources of error. Any measurement involving mics is rather unreliable outside of soundproof chambers. This is the wrong way to measure what you're trying to measure.

Changes in the frequency response in the audible range would be trivial to detect using a reliable method. And I have not seen any indication from tests here on ASR that op amp swapping influences the FR like you describe. One very clear counter-example:

Sparkos SS3602 Opamp Rolling In Fosi P4

This is a detailed measurements of effect of swapping the stock NE5532P op-amp in Fosi P4 preamplifier with Sparkos SS3602 "discrete" opamp. The SS3602 costs US $79.80 and the NE5532P, US $0.47. Swapping opamps in P4 is more challenging due to back connectors/switches that need to be...

www.audiosciencereview.com

 

[gerard11]

That's generally not correct. Ignoring extreme EQs changing the balance between bass and treble, same SPL naturally means same amplifier power. Where else would the power from the amp go, if not into the sound?

You may be operating at a higher gain with a significant negative pre-amp gain in your EQ setup, but higher gain ≠ higher power.

When I enable PEQ even without any adjustment yet, the volume already went lower. So just trying to adjust the gain in the PEQ to compensate the volume back without changing the amp produce the clipping effect.

So the clipping might happened at the DAC itself rather than at the

And with the lower volume from activating the PEQ, I need to increase the volume in the amp to compensate. The sound produced are more harsh than without the PEQ (even with no adjustment). Not sure whether my amp already reached it peak distortion limit or the PEQ component itself add something.

 

[gerard11]

Measuring op amps with the speaker, room and microphone in the chain of measurement is just introducing more sources of error. Any measurement involving mics is rather unreliable outside of soundproof chambers. This is the wrong way to measure what you're trying to measure.

Changes in the frequency response in the audible range would be trivial to detect using a reliable method. And I have not seen any indication from tests here on ASR that op amp swapping influences the FR like you describe. One very clear counter-example:

Sparkos SS3602 Opamp Rolling In Fosi P4

This is a detailed measurements of effect of swapping the stock NE5532P op-amp in Fosi P4 preamplifier with Sparkos SS3602 "discrete" opamp. The SS3602 costs US $79.80 and the NE5532P, US $0.47. Swapping opamps in P4 is more challenging due to back connectors/switches that need to be...

www.audiosciencereview.com

I used the 2:24 sine sweep as it is slowly moving from 20hz all the way up to 20,000hz. And I use the peak measure in the RTA Audio without any smoothing.

So I can see how that sine sweep move up and down like an ECG chart showing peak and valleys across the audible frequency range during that 2 minutes and 24 seconds measuring process.

As that iPhone sits at my listening position, I consider it is hearing what I am hearing when I sat there.

Anyway I always wonder how frequency response chart always show smooth curve? did they do some smoothing or post processing. But my experience with RTA audio (peak mode) and sine sweep (2:24) has always with such jagged lines.

Did all sort of things with that RTA app and sine sweep from rolling op amps, rolling amps, rolling speaker driver, adjusting speaker placement, rolling DAC, interconnect cables, speaker cables, power cables and using/not using a power voltage stabilizer. And the chart reflects the perceived listening effect (in my humble opinion).

Also I used equalizer APO to perform EQ from PC and did the same approach
A. Measure with RTA audio + sine sweep before the adjustment
B. Perform the adjustment based on the measurement
C. Measure with RTA audio + sine sweep after the adjustment.
The measurement reflects the changes I made at Equalizer APO.

its what consumer user can do best with an iphone and a youtube, anyway

 

[RandomEar]

When I enable PEQ even without any adjustment yet, the volume already went lower. So just trying to adjust the gain in the PEQ to compensate the volume back without changing the amp produce the clipping effect.

Yes, if you increase the digital pre-amp gain in the EQ you use above 0 dB, it might clip.

So the clipping might happened at the DAC itself rather than at the

Rather then at the amp? Yes, you can digitally clip the signal if you use a positive pre-amp gain. That's why one should generally not do that. There are exceptions, where you know that you digital signal is far from 0 dB and you need to push it to achieve a reasonable listening level. With most commercially available music, that's not the case - if anything, it is usually mastered too hot.

And with the lower volume from activating the PEQ, I need to increase the volume in the amp to compensate. The sound produced are more harsh than without the PEQ (even with no adjustment). Not sure whether my amp already reached it peak distortion limit or the PEQ component itself add something.

Assuming you end up with the same RMS voltage at the speaker outputs and just needed to turn up the volume dial, the amp's volume has no influence on "harshness" unless there is something broken in your unit. You should also never reach the clipping or distortion limits in the amplifier if you simply lower the input volume (-> lower input voltage) and increase the amps gain to compensate (-> same output voltage). That is, unless your amps internal pre-amp is badly designed and starts to clip at full gain.

Here an example for the amplifier output voltage calculation in both cases:

• Low gain example:
  Input 3.16 V_RMS, amp set to 10 dB gain -> Output 10 V_RMS
• High gain example:
  Input 1 V_RMS, amp set to 20 dB gain -> Output 10 V_RMS

The amp's output power and the rersulting SPL would be the same in both examples. If the amplifier does not clip at 10 V_RMS, it will also not clip if used with high gain at 10V_RMS - with the mentioned caveat of amplifiers with an active pre-amp and a design error in that pre-amp.

 

[RandomEar]

I used the 2:24 sine sweep as it is slowly moving from 20hz all the way up to 20,000hz. And I use the peak measure in the RTA Audio without any smoothing.

So I can see how that sine sweep move up and down like an ECG chart showing peak and valleys across the audible frequency range during that 2 minutes and 24 seconds measuring process.

As that iPhone sits at my listening position, I consider it is hearing what I am hearing when I sat there.

Anyway I always wonder how frequency response chart always show smooth curve? did they do some smoothing or post processing. But my experience with RTA audio (peak mode) and sine sweep (2:24) has always with such jagged lines.

Yes, frequency response measurements are typically slightly smoothed, maybe 1/48 or 1/24 octave or using psychoacoustic smoothing. The small peaks you see are problematic, because especially in higher frequencies (>7 kHz) they are extremely dependent on positioning. A couple of millimeters of movement on the microhpone will usually have a measureable impact. The frequency resolution does also depend on your phones' capture frequency, the FFT length and the sweep length and type.

In short: What you see in those plots is only roughly representative of what you hear. The measurement is not reliable with regards to small details in the FR plot.

If you want reliable FR data, measure the voltage at the speaker terminals of the amplifier using an ADC designed for speaker level voltages. Maybe a cheap speaker-to-line level converter + a bog standard microphone input on your PC might work as well as a cheap alternative. That will eliminate dozens of potential errors in measurements using a mic, including subtle sources like a car driving by or the fridge or AC turning on during the measurement.

Did all sort of things with that RTA app and sine sweep from rolling op amps, rolling amps, rolling speaker driver, adjusting speaker placement, rolling DAC, interconnect cables, speaker cables, power cables and using/not using a power voltage stabilizer. And the chart reflects the perceived listening effect (in my humble opinion).

Also I used equalizer APO to perform EQ from PC and did the same approach
A. Measure with RTA audio + sine sweep before the adjustment
B. Perform the adjustment based on the measurement
C. Measure with RTA audio + sine sweep after the adjustment.
The measurement reflects the changes I made at Equalizer APO.

its what consumer user can do best with an iphone and a youtube, anyway

It's the best you can do without extra equipment, but it's not a reliable way to measure what you want to measure. That's just how it is. It is decent for room EQ and broader corrections like dialing in a bass shelf. It is absolutely not suitable to prove supposed FR changes from op amps.

 

[NTTY]

It is absolutely not suitable to prove supposed FR changes from op amps.

Indeed. I tried like 80 combinations or so of OPA swaps in a DAC, between IV and LPF functions, and one thing that never changed was the FR
I obviously got issued to properly insert some OPAs into socket though

 

[gerard11]

Yes, frequency response measurements are typically slightly smoothed, maybe 1/48 or 1/24 octave or using psychoacoustic smoothing. The small peaks you see are problematic, because especially in higher frequencies (>7 kHz) they are extremely dependent on positioning. A couple of millimeters of movement on the microhpone will usually have a measureable impact. The frequency resolution does also depend on your phones' capture frequency, the FFT length and the sweep length and type.

In short: What you see in those plots is only roughly representative of what you hear. The measurement is not reliable with regards to small details in the FR plot.

If you want reliable FR data, measure the voltage at the speaker terminals of the amplifier using an ADC designed for speaker level voltages. Maybe a cheap speaker-to-line level converter + a bog standard microphone input on your PC might work as well as a cheap alternative. That will eliminate dozens of potential errors in measurements using a mic, including subtle sources like a car driving by or the fridge or AC turning on during the measurement.


It's the best you can do without extra equipment, but it's not a reliable way to measure what you want to measure. That's just how it is. It is decent for room EQ and broader corrections like dialing in a bass shelf. It is absolutely not suitable to prove supposed FR changes from op amps.

Thanks for sharing and explaining this.

Out of curious, if I create an Equalizer APO EQ config with 50 peak filter with 0.25DB adjustment gain or cut across the frequency range but at frequency

Yes, frequency response measurements are typically slightly smoothed, maybe 1/48 or 1/24 octave or using psychoacoustic smoothing. The small peaks you see are problematic, because especially in higher frequencies (>7 kHz) they are extremely dependent on positioning. A couple of millimeters of movement on the microhpone will usually have a measureable impact. The frequency resolution does also depend on your phones' capture frequency, the FFT length and the sweep length and type.

In short: What you see in those plots is only roughly representative of what you hear. The measurement is not reliable with regards to small details in the FR plot.

If you want reliable FR data, measure the voltage at the speaker terminals of the amplifier using an ADC designed for speaker level voltages. Maybe a cheap speaker-to-line level converter + a bog standard microphone input on your PC might work as well as a cheap alternative. That will eliminate dozens of potential errors in measurements using a mic, including subtle sources like a car driving by or the fridge or AC turning on during the measurement.


It's the best you can do without extra equipment, but it's not a reliable way to measure what you want to measure. That's just how it is. It is decent for room EQ and broader corrections like dialing in a bass shelf. It is absolutely not suitable to prove supposed FR changes from op amps.

Thank you for sharing @RandomEar

Just out of curiosity, If I create an equalizer APO config file with 50 peak filters that each either add 0.25db gain or 0.25db cut and a large Q factor and spread it across the audible frequency range but avoid the major key note frequencies. Would the standardized measurement using the 1/48 or 1/24 octave psychoacoustic smoothing can capture the difference when that EQ is on or off. Cause definitely the music would sound different due to this EQ in the listener's ear.

But if measured without smoothing those small peak and valley of 0.25DB should appear everywhere.

 

[RandomEar]

Just out of curiosity, If I create an equalizer APO config file with 50 peak filters that each either add 0.25db gain or 0.25db cut and a large Q factor and spread it across the audible frequency range but avoid the major key note frequencies. Would the standardized measurement using the 1/48 or 1/24 octave psychoacoustic smoothing can capture the difference when that EQ is on or off. Cause definitely the music would sound different due to this EQ in the listener's ear.

But if measured without smoothing those small peak and valley of 0.25DB should appear everywhere.

You would need to test that. I dont think it would be invisible, but it might only show in certain frequency ranges depending on which smoothing method is applied. An advantage is that measurements of the FR at the speaker binding posts of the amp don't need any smoothing.

Those 50 filters would also be very narrow and 0.25 dB isn't much. So the assumption that those would be audible would need to be tested, too.

 

[gerard11]

You would need to test that. I dont think it would be invisible, but it might only show in certain frequency ranges depending on which smoothing method is applied. An advantage is that measurements of the FR at the speaker binding posts of the amp don't need any smoothing.

Those 50 filters would also be very narrow and 0.25 dB isn't much. So the assumption that those would be audible would need to be tested, too.

Valid points @RandomEar

For this experiment, I think we can design this way:
1. Create 8 Equalizer APO EQ config file. Where the gain/cut varies from 0.01DB (base reference) followed by 0.25DB, 0.5DB, 1.0DB, 1.5DB, 2.0DB, 2.5DB and 3.0DB
2. As there are 31 freq band in normal graphic EQ, I will create 60 peak filter to have 2 peak filter in each range within each band with one of them add DB gain and the other add DB cut. I will use large Q to make the peak and the valley very narrow. For each of the config I will add a pre-amp gain of -3DB to make the base volume equal for each EQ config.

We can then test A-B-A-B via listening and FR measurement test to see:
1. At which DB level you start hearing difference between base reference (0.01DB) and the other test config file from listening test.
2. (If you have measurement device) At what DB level you start seeing visible difference between base reference (0.01DB) and the other test config file from measurement test.

It will be interesting to discover whether your ear is more sensitive compared to the measurement or vice versa. Which of course would vary between tester due to age, ear health, listening habit, work environment, etc.

Let me find time to create this config file.

 

[RandomEar]

@gerard11 It would be better to create test files using such filters, because using an EAPO config would require sighted testing which will give biased results. Test files allow you to use tools such as the foobar2000 ABX plugin to do unbiased blind tests. You could still play and measure the files using whatever method you would like, so there's no disadvantage there.

Let me say that I don't have the equipment to measure at the speaker terminals. I own a UMIK for room correction and that's about it. I also think that this is getting off-topic in a DAC thread, so it may be time to start a fresh one for the test