Is this how EQ is supposed to work?
Jun 30, 2016 at 3:41 PM Thread Starter Post #1 of 16

reginalb

1000+ Head-Fier
Joined
Aug 11, 2011
Posts
1,061
Likes
295
Alright, so I'm pretty new to messing with equalizers. I don't really trust myself to do it well. But I figure, correcting a bad FR in a DAP to flat is a good start. My assumption was if I set a value at, say 100Hz to +1db, and 150Hz to +.5 dB, it should be at about -.75dB at 125Hz. If it's at +1dB at both 100Hz and at 150Hz, it's reasonable to expect it to also be at +1dB at 125 Hz. 
 
I flattened my HM-602 a bit with no load, and got this:
 

 
This was a pretty quick and rough adjustment, but it worked about how I expected. So then I tried to EQ my AK300 based on my Merlin load, so that it would be a bit closer to neutral when I use it with the Merlins. The EQ in the AK300 isn't quite as flexible as the Rockbox EQ, but you can set the levels at lots of steps (20 of them). So I loaded up the FR graph from it loaded with the Merlins, set adjustments at each point based on how far over or under the FR was, and got...
 

 
So, everywhere it was originally at +.5dB for example, in the low end of the curve, I adjust it, and at those adjustment points it's close to +0, then it jumps back up, then dips at the next adjustment point, etc. I am a troubleshooter, so I thought, well, maybe I just don't get how loads work, so I measured with my profile set, but with no load, and get:
 

 
 
It's just shifted, except where it crosses, which is weird. But I probably didn't make adjustments there, or something. But mainly it's just the same weird step EQ but shifted. 
 
This is not how I expected an equalizer to work, so what gives? Was I incorrect in my expectation? Or is this actually weird behavior?
 
Jun 30, 2016 at 3:54 PM Post #3 of 16
  What are the center frequencies and bandwidths?

 
centers are 30Hz, 45, 60, 90, 120, 180, 250, 380, 500, 750, 1,000, 1.5k, 2k, 3k, 4k, 6k, 8k, 12k, 14k, 18k
 
I have no idea the bandwidths. I'll consult the manual, though I guess if I can change that somehow, that would resolve the issue.
 
EDIT: That's it, there is a q-factor adjustment when you open the PEQ. Thanks!
 
Jun 30, 2016 at 4:00 PM Post #4 of 16
   
centers are 30Hz, 45, 60, 90, 120, 180, 250, 380, 500, 750, 1,000, 1.5k, 2k, 3k, 4k, 6k, 8k, 12k, 14k, 18k
 
I have no idea the bandwidths. I'll consult the manual, though I guess if I can change that somehow, that would resolve the issue.

 
Yeah seems like the BWs are a bit small for what you're trying to do. What exact settings did you use in Rockbox?
 
Jun 30, 2016 at 5:56 PM Post #5 of 16
   
Yeah seems like the BWs are a bit small for what you're trying to do. What exact settings did you use in Rockbox?

 
Regarding Rockbox, I had set it all to Q values of 1, and changed some of the centers. A couple of the Q values are lower. I changed them on accident. I didn't even know what they did until today, haha.
 
The AK300 allows you to change Q-Values, but doing so does nothing, I think it might be broken. 
 
Jun 30, 2016 at 6:06 PM Post #6 of 16
   
Regarding Rockbox, I had set it all to Q values of 1, and changed some of the centers. A couple of the Q values are lower. I changed them on accident. I didn't even know what they did until today, haha.
 
The AK300 allows you to change Q-Values, but doing so does nothing, I think it might be broken. 

 
Would seem an odd feature to allow to be broken :frowning2:  Here's a handy page for Q → octave conversion, just in case:
http://www.sengpielaudio.com/calculator-bandwidth.htm
 
Jul 7, 2016 at 4:32 AM Post #7 of 16
 

Regarding Rockbox, I had set it all to Q values of 1, and changed some of the centers. A couple of the Q values are lower. I changed them on accident. I didn't even know what they did until today, haha.

The AK300 allows you to change Q-Values, but doing so does nothing, I think it might be broken. 


Would seem an odd feature to allow to be broken :frowning2:  Here's a handy page for Q → octave conversion, just in case:
http://www.sengpielaudio.com/calculator-bandwidth.htm


But given the customer base, reginalb might well be the first AK user on the planet to recognize the broken Q parameter :p :xf_eek: :frowning2:
 
HiBy Stay updated on HiBy at their facebook, website or email (icons below). Stay updated on HiBy at their sponsor profile on Head-Fi.
 
https://www.facebook.com/hibycom https://store.hiby.com/ service@hiby.com
Jul 10, 2016 at 5:42 AM Post #8 of 16
Peak filter when Q = 1.0 (cyan) and BW/oct = 1.0 (red).
 

 
When you adjust gain in tiny range as like +/-1dB then the difference isn't that much to worry about.
 
IMO, using PEQ instead of GEQ would be better choice.
 
Jul 11, 2016 at 5:34 PM Post #9 of 16
But given the customer base, reginalb might well be the first AK user on the planet to recognize the broken Q parameter
tongue.gif
redface.gif
frown.gif

 
Should have bought an X7, but I thought it would be fun to try to hack some apps on to this thing myself. That's the sort of thing I like to do. A well sorted DAP isn't as much fun to tinker with 
wink.gif

 
  Peak filter when Q = 1.0 (cyan) and BW/oct = 1.0 (red).
 

 
When you adjust gain in tiny range as like +/-1dB then the difference isn't that much to worry about.
 
IMO, using PEQ instead of GEQ would be better choice.

 
Well...allegedly I am using a PEQ...allegedly...
 
Jul 12, 2016 at 2:48 AM Post #11 of 16
  Well...allegedly I am using a PEQ...allegedly...

 
Definition for PEQ can be found from here: https://en.wikipedia.org/wiki/Equalization_%28audio%29
 
IMO, Astell & Kern's fixed band PEQ is just a GEQ with variable Q and also with that narrow frequency division it's useless in hands of user who is unaware of what happens under the hood.
 
Jul 12, 2016 at 2:53 AM Post #12 of 16
The thing is, the allegedly variable Q doesn't do anything either right now.
 
HiBy Stay updated on HiBy at their facebook, website or email (icons below). Stay updated on HiBy at their sponsor profile on Head-Fi.
 
https://www.facebook.com/hibycom https://store.hiby.com/ service@hiby.com
Jul 12, 2016 at 8:23 AM Post #14 of 16
   
Is that confirmed?
 
Dunno how OP runs the test but there's a note on AK300 specs:

 
I have not used it as a DAC. Not above believing that I've done something wrong, but it seems like a fairly simple task. I see what you mean that you're supposed to be able to shift the centers in a PEQ. Anyhow, it seems that, running a test signal, changing the Q values, and running a second, confirming it didn't change the results at all is a pretty solid indicator that somethings not quite right...
 
Jul 12, 2016 at 8:48 AM Post #15 of 16
   
I have not used it as a DAC. Not above believing that I've done something wrong, but it seems like a fairly simple task. I see what you mean that you're supposed to be able to shift the centers in a PEQ. Anyhow, it seems that, running a test signal, changing the Q values, and running a second, confirming it didn't change the results at all is a pretty solid indicator that somethings not quite right...

 
OK, I mentioned this because of no mention how you have got those plots.
 
Yes, it's easy to check if there's something wrong in filter bandwidth setting ... just use one frequency band (1kHz or close) and higher gain values than what you have mentioned earlier ... like maximum -/+ dB values then plot the response for few Q values (0.1, 0.5, 1, 2, 4). If response curve in plot remains same for all of the Q values then it's broken (dunno the implementation they used).
 
BTW, does the equalizer work for the headphones connected directly into AK300?
 

Users who are viewing this thread

Back
Top