Yesterday I tuned the LPF using my trusty old board o’ potentiometers…
Best sounding values came out like this:
which I, after some thinking, recognized as the original values like in the JUNO6! Very nice! Just had to replace the 100pF capacitor with a 240pF capacitor like in the original. Only thing missing is the “mystery resistor” on the + input of the OTA to ground. For now I just left this res out.
A fitting frequency range for the filter should be achieved by a good control current. According to the IR3109 specifications, transconductance should range from 1µƱ – 10mƱ. The expo converter from the previous post delivers a current ranging from 70nA to 1.65mA which translate to a transconductance of 1.34µƱ to 31mƱ. Ok for now.
I made a mockup of the JUNO6 board and the IR3109 ic:
I’m not sure. Did a lot of prototyping and these are the results.
Since I couldn’t figure out a proper way to calculate the cutoff frequency when using different current and resistor combination, I decided to use the default lowpass filter circuit from the LM13700 datasheet.
Hooked my expo converter up to pin 1 (100nA < Iabc < 1.1mA) and started testing. The Vout of this pole is the Vin of the second pole. Unfortunately I only had 1 LM13700 working left, since I blew up all my other’s ;). I experimented a lot with different capacitor and resistor combination and the only ok sounding combination is a 1nF capacitor with 10K big and 220Ohm small resistors. Check the following sound files for the results:
As seen in the previous post, the following formula determines the cutoff frequency of the typical LM13700 lowpass filter design:
The only variable part in this formula is the Iabc part, the incoming current from the expo converter. The cutoff frequency of the JUNO106 filter varies between 5Hz and 50000Hz (from the manual). To achieve this, Iabc low and Iabc high must differ a factor of 10000. I redesigned the expo converter as following:
Resulting in the following:
The current @ 5Hz is now 100nA and the current @50kHz is 1mA which differ a factor 10000. Nice!
Using the FREQ slider on the JUNO106 results in a exponential filter control current from 32nA to 705µA. So, when feeding 32nA into the filter, its cutoff frequency should be 5Hz. Feeding the filter 705µA results in a 50kHz cutoff frequency. My LM13700 lowpass filterdesign is based on the lectures by Aaron Lanterman. This is a typical design:
resulting in the following plot (Iabc = 1mA)
cutoff frequency in this example is:
using this calculation I will determine the required resistor values for the exponential converter range (32nA – 705µA).
I reversed the current direction by using PNP transistors instead of NPN transistors (plus some other minor alterations). I also realized that the output current of the expo conv in the previous post (7mA) was way too high for the LM13700 (max current 1mA) so I adjusted the resistor in the iref connection to 10 megaohms! Max output to the LM13700 is now 705µA.
plot (current source leads to negative figures in LTspice)
So expo out is at max 705µA and at min 32nA. On to the actual filter!