Understanding low frequency delay added by DSP

[Angst92]

I'm looking for help understanding measurements taken from the Fiio K15. From my limited understanding, the below spectrograms show that the K15 adds a time delay to the low frequencies when it has its EQ DSP enabled(shown in "bypass" mode, where it is enabled but set to make 0 changes to the signal).
Is this normal, and what is the expected audible impact? My brain says there's a slight difference when A/B testing, but I can't pin down what it is or decide if its placebo or not.

 

[Keith_W]

Welcome to ASR!

Can you please post the MDAT. You will need to zip it first.

 

[Angst92]

Morning, I've attached the mdat, It contains 4 measurements.

1. EQ disabled
2. EQ enabled but in bypass mode
3. EQ enabled but using a custom profile that does nothing but add a slight negative pre-gain
4. EQ enabled making significant changes.

A slight correction from the first post, both the disabled and bypass seem to behave very similarly; it's only when changes are made(measurements 3 and 4) that the delay looks to be added to the low frequencies.

 

[AnalogSteph]

A slight correction from the first post, both the disabled and bypass seem to behave very similarly; it's only when changes are made(measurements 3 and 4) that the delay looks to be added to the low frequencies.

#3 is a bit of a head-scratcher then. Can you verify that the sub-bass response is actually flat?

 

[Sokel]

#3 is a bit of a head-scratcher then. Can you verify that the sub-bass response is actually flat?

You can tell by the curves, the only non-flat one is #4:

 

[AnalogSteph]

I would want to verify that with a direct (electrical) loopback measurement. Looks like there might be a fraction of a dB worth of difference when approaching 20 Hz. I wouldn't rule out that there's some sort of subsonic filter going. Something has to be responsible for this delay, and an IIR highpass of sorts seems the most logical explanation. Nobody puts an allpass or a Hilbert transform into their signal path willy-nilly.

 

[Sokel]

I would want to verify that with a direct (electrical) loopback measurement. Looks like there might be a fraction of a dB worth of difference when approaching 20 Hz. I wouldn't rule out that there's some sort of subsonic filter going. Something has to be responsible for this delay, and an IIR highpass of sorts seems the most logical explanation. Nobody puts an allpass or a Hilbert transform into their signal path willy-nilly.

Did a quick one with some filters to check (as an external electrical loopback through foobar) :

 

[Angst92]

I would want to verify that with a direct (electrical) loopback measurement. Looks like there might be a fraction of a dB worth of difference when approaching 20 Hz. I wouldn't rule out that there's some sort of subsonic filter going. Something has to be responsible for this delay, and an IIR highpass of sorts seems the most logical explanation. Nobody puts an allpass or a Hilbert transform into their signal path willy-nilly.

I cant test anything further till I am back next week however frrom what you are saying and the reading that I have done I understand that phase shift of low frequencies would be expected if a high pass filter was used as part of the EQ to attenuate low frequencies but if i understand it right that should not be expected when the eq is enabled but not altering any frequencies or i guess altering the entire band with the negative pre-gain?
My other question is, is this delay seen here excessive and would it have an audible impact?

 

[nagster]

Level, phase and group delay of the FIIO K15. I compared the EQ settings below.
EQ OFF
EQ ON, Bypass
EQ ON, Global gain -0.5 dB, EQ Flat
EQ ON, Global gain -10 dB, EQ Significant changes