Digitakt Filter Cheat sheet

[spill]

frequencies.pdf (22.5 KB)

Hey Guys
I’ve made a cheat sheet of Digitakt filter values to actual frequencies
hope you’ll find it useful

P>S> IMO this is can be used with other Elektron machines as well

[Lauli]

Oh what a blessing you have given

When mixing situation required it, I’ve always always had a bit of trouble being precise in the mids with the filter!

[Phurniture]

Very nice. I love it when people do stuff like this. One suggestion - although it’s obvious once you look at it for a moment, adding labels for “Elektron Filter Value” and “Real-life frequency” (or better but similar labels) might be useful.

[heretical_audio]

Truly dope, thank you sir. I’ve always just ‘used my ears’, but this should be interesting to see what freq I usually gravitate towards. EQ1 FTW.

[LoomieDarko]

Great! Thanks a lot

[tagaz]

Having not found a summary of the precise filter values and having forgotten my math lessons, I measured everything.

If it can help anyone.

[otzcz]

Hi, good work Is it valid for “EQ” on page 2 also?

[spill]

Yep, as far as I’m concerned.

[pokk]

Amazing! Super useful, thanks

[juaniko]

Hi!
Yesterday I made a few measures with the Open Sound Meter software to check what types of filters uses the Digitakt MKI, and also in witch Digitakt values are synchronized with each frequency. After measuring, we can observe some conclusions:

• I realized the Lo pass filter and Hi pass filter have diferent slopes (contrary to what the manual specifies, that is both filters are 2-pole; what means they have -12dB of attenuation by octave).

While the Lo pass filter has 2-pole slope (filter LR-12), the Hi pass filter has another one (BUT-6). Lo Pass filter (2-pole / LR-12) has an attenuation of -6dB in the cut-off frequency (1kHZ / 80’77 on Digitakt) as we can see in the following graph:

Red trace: measure on the Digitakt (with +1dB offset for better comparasion)
Blue trace: LR-12 filter added to compare (both have the same slope, and attenuation on cut-off freq.)

Hi Pass filter (apparently is a BUT-6 filter) with an attenuation of -3dB in the cut-off frequency (1kHz, “91” on Digitakt):

Green trace: Hi pass filter measure on Digitakt.
Brown trace: a BUT-6 filter added to compare.
Note that the slopes aren’t exactly the same but is the filter whose slope most closely resembles Digitakt’s Hi pass filter.

• Lo pass filter and Hi pass filter are not matched in frequency. Hi pass filter (BUT-6) with cut-off frequency in 500Hz (“80’77” on Digitakt). In this graph we can prove that if we put the same value “80’77” in the Hi pass filter (the same we put on Lo pass ) the cut-off frequency isn’t 1kHz but 500Hz. Because of the difference in the type filters, the value to frequency ratio is not the same in the Lo pass filter and Hi pass filter.
  
  
  
  Red trace: Lo pass filter (LR-12) cut-off freq at 1kHz (80’77 on Digitakt)
  Purple trace: Hi pass filter (BUT-6), cut-off freq at 500Hz (80’77 on Digitakt)
  Green trace: Hi pass filter (BUT-6), cut-off freq at 1kHz (91 on Digitakt)

If the slopes and values were the same the crossover freq would be 1kHz, we can observe this doesn’t happen, so Lo pass filter and Hi pass filter doesn’t have same slope.

• EQ1/EQ2/EQ3/EQ4/EQ5 have differents slopes and gains, but they are matched in frequency with Lo pass filter (1kHz / 80’77 on Digitakt):
  
  
  
  EQ1: black / EQ2: blue / EQ3: green / EQ4: purple / EQ5: orange

I realized the Lo Pass filter and EQ1, EQ2, EQ3, EQ4 and EQ5 the frequency is matched at the same value (aprox 1kHz is on value “80’77”). But the Hi Pass filter doesn’t share the same value, as you can check in the measures (aprox 1kHz is on value “91” on Digitakt).

• The cut-off frequency is not match between Lo pass filter and Hi pass filter, the resonance is matched, so if you aply maximun gain at resonance parameter at same frequency value in both filters (e.g 80’77 at freq and 127 at resonance) the frequency with max amplitude is the same in both filters (low pass and hi pass). So cut-off frequency is not matched in both filters because of the diference between filters slopes but the resonance is matched in both filters. We can observe this in the next graph:

Hi pass filter comparasion “with resonance vs without resonance”:

Blue trace: Hi pass filter with cut-off frequency at 500Hz without resonance.
Dark blue trace: Hi pass filter with cut-off frequency at 500Hz, but with maxium resonance matching 1000Hz
Red trace: Hi pass filter with cut-off frequency at 1000Hz without resonance.
Dark red trace: Hi pass filter with cut-off frequency at 2000Hz, with maxium resonance at 2000Hz aprox.

We can observe when the filter is set and 1kHz and we add max resonance the peak desviates to 2kHz. When the filter is set at 500Hz and we add max resonance the peak desviates to 1kHz.

Lo pass vs Hi pass “with and without resonance” comparasion. Here we can observe when there isn’t resonance the filters doesn´t match, but when adding max resonance in both filters the frequency is matched.

Yellow trace: Lo pass filter at 1000Hz, no resonance
Dark yellow trace: Lo pass filter at 1000Hz, with max resonance
Blue trace: Hi pass filter at 500Hz, no resonance
Dark blue trace: Hi pass filter at 500Hz, with max resonance.

When adding resonance in both filters the freq with max resonance is matched. When resonance is set to “0” the cut-off freq aren’t matched.

• Band pass filters (second layer in Digitakt’s filter menu), they are BUT-6 filters, cut-off frequency set at 1kHz (value “75” in Digitakt):
  
  
  
  Green trace: Low pass filter measured in Digitakt, (BUT-6)
  Blue trace: Hi pass filter measured in Digitakt (BUT-6)

Both filters make perfect cross-over at 1kHz at same values.

I made a table with some usefull values:
You can choose the parameter you want to match; freq cut-off or frequency at max resonance (the values only changes in HPF due to the displacement previously commented, in LPF the values remains the same):

LPF			HPF
FREQ(Hz) MATCH	VALUE DT FREQ	VALUE DT RES	FREQ(Hz) MATCH	VALUE DT FREQ	VALUE DT RES
20	22,41	22,41	20	35,28	22,41
40	31,88	31,88	40	45,5	31,88
60	38,15	38,15	60	51,68	38,15
80	42,5	42,5	80	56,31	42,5
100	46,24	46,24	100	59,4	46,24
200	56,52	56,52	200	69,52	56,52
300	62,78	62,78	300	75,33	62,78
400	67,25	67,25	400	79,72	67,25
500	70,6	70,6	500	82,65	70,6
600	73,37	73,37	600	85,22	73,37
700	75,8	75,8	700	87,35	75,8
800	77,87	77,87	800	89,21	77,87
900	79,67	79,67	900	90,5	79,67
1000	81,13	81,13	1000	91,85	81,13
2000	91,8	91,8	2000	100,3	91,8
3000	97,97	97,97	3000	104,41	97,97
4000	102,41	102,41	4000	107,08	102,41
5000	105,71	105,71	5000	108,89	105,71
6000	108,48	108,48	6000	110,26	108,48
7000	110,97	110,97	7000	111,2	110,97
8000	112,77	112,77	8000	112,06	112,77
9000	114,55	114,55	9000	112,58	114,55
10000	116,29	116,29	10000	113,02	116,29
12000	119,24	119,24	12000	113,81	119,24
14000	121,65	121,65	14000	114,26	121,65
16000	123,4	123,4	16000	114,54	123,4
18000	125,22	125,22	18000	114,73	125,22
20000	127	127	20000	114,82	127

I hope this information is helpful to you.
Cheers!

[basic_helix]

Reverse engineering skills here are getting more and more sophisticated with every dedicated answer, respectively

[tubefund]

Your hired!
Jokes aside, do others notice samples sound a tidy more open in the high end when the filter is turned off compared to, say filter on with LP fully open? As if there’s always a slight high rolloff on max setting vs filter off? Looking at your chart it runs to 20k, that implies I imagine it but i would bet i could hear the difference in a blind test.

[Bitsnbobs]

it’s beautiful thank you

[juaniko]

Well, this makes sense because when the LP filter is turn on but with max cut-off frequency (in this case value 127 in Digitakt) we have an attenuation caused by the slope of the filter.

We can see it in this graph:

Red trace: Digitakt LP filter off, the frequency response is totally flat
Blue trace: Digitakt LP filter on, with max cut-off freq (value 127 on Digitakt)

We can observe this attenuation if we compare with the no filter frequency response.

[B_LD]

Hi that’s a nice amount of information you have been collecting there! I’ve been thinking about doing some similar tests/measurements with other boxes. But I’m not familiar with Open Sound Meter, and the documentation is sparse. Would you mind sharing some insight on how you set this up physically and in the software itself?

[juaniko]

Yes, You can contact me via DM, and I can guide you since the proccedure when learning is long.

The method in essence is the same in all measurements; You compare a known signal (reference signal) vs this same signal (measured signal) after going thru the gear or after processing, the difference is represented as Phase, Magnitude and Impulse response (and others ones too). The only thing that changes the proccedure is the gear to mesure (available inputs, outputs, if it’s analog or digital, if it’s an acoustical measurement you will need a measurement microphone, etc).
In this video you can learn the basics of measurements:

Don’t doubt to ask me.

In the case of Digitakt I record a pink noise(this is my reference signal) and assigned to 2 tracks on DT, so the track 1 is the know signal (reference) and the track 2 is the measured signal (where i placed the filters, etc), the software compares both and gives you the result. On Digitakt is the way I figured to do the measurements, because we can’t put filter on “input L/R” like in normals mixers.

Cheers!

[B_LD]

Cool, thank you very much! I’ll take that as a starting point, wrap my head around it and see (if I find the time) how far I get!

[autechpan]

Brilliant. Thanks so much.

[thumbslapper]

anyone know why elektron just doesn’t list the real frequency instead of a arbitrary value? i was gobsmacked the first time i saw this and realize those numbers aren’t the actual hz but some intermediate number that needs decoding.