Schematic Check - Tyler Frequency Splitter Clone

forgotmyeeclass · Post by **forgotmyeeclass** » 12 Mar 2021, 18:56

Howdy folks,

I am trying to put together a box that will fulfil the same purpose as this
frequency splitter: https://www.kma-machines.com/m_tyler.html

I based this on these circuits: https://sound-au.com/project155.htm and http://www.runoffgroove.com/splitter-blend.html. However I haven't done any circuit design or analysis forever, so I was hoping someone would be willing to take a look and advise on anything I got wrong or could've implemented better:

Thanks!

stolen · Post by **stolen** » 13 Mar 2021, 07:54

Hi!

Firstly, Q1 seems to be flipped - base and collector should be swapped! Also, you used the BJT symbol for a JFET there, but that's nitpicking

.

Around IC1B there's a bigger issue: There's a virtual ground connection missing, and the potential is undefined! C9 blocks DC, and there lacks another DC reference, so the bias of that stage is left to the whims of the input bias current of IC1B, which would probably make it drift into the rails.

If you had a buffered VR, you could simply use it as a ground connection between R12 and the potentiometer half, just like Elliot did in their drawing - since 680R is quite stiff coupling, your existent VR might get pulled around quite a lot in the highest setting though. One buffer-less solution would be to establish a secondary VR with a say 100k resistor from VR to a huge electrolyte (22-100uF) to give the sufficient AC stiffness and define DC.

The double buffers IC1A/IC2A seem redundant, but due to the high worst-case load of ~340R they might not be the worst idea. If you really need all that sweep range, sure, otherwise you could cut the parts count there. Not sure how useful 20Hz are, maybe 50Hz are sufficient? You could use that extra OPA to replace Q1.

The resistor values around IC3A are a bit on the high side, with R19, R20 = 22k and R18 = 47k you could decrease noise quite a bit.

Lastly, the output volume pot is wired backwards, swap CW and wiper. R21 is not necessary but would make a good job between input and ground, so that if you install true bypass you don't get DC pops from a floating input cap. The same goes for the input caps of the returns.

Hope this helps!

forgotmyeeclass · Post by **forgotmyeeclass** » 13 Mar 2021, 15:52

That is tremendously helpful, thank you!
Regarding this part I'm a bit unclear:

stolen wrote: ↑13 Mar 2021, 07:54
If you had a buffered VR, you could simply use it as a ground connection between R12 and the potentiometer half, just like Elliot did in their drawing - since 680R is quite stiff coupling, your existent VR might get pulled around quite a lot in the highest setting though. One buffer-less solution would be to establish a secondary VR with a say 100k resistor from VR to a huge electrolyte (22-100uF) to give the sufficient AC stiffness and define DC.

Would I set this up like this?

stolen · Post by **stolen** » 13 Mar 2021, 16:01

Hi, precisely like that! Sorry for not making a drawing. We'd go for 47k-100k instead of 100R though, bias might be a bit unstable or even latch up otherwise depending on the input source since you'd introduce quite some positive feedback at very low frequencies.

forgotmyeeclass · Post by **forgotmyeeclass** » 13 Mar 2021, 16:28

Thanks so much, stolen. That was meant to be a 100k resistor, not 100R

I appreciate the help!

stolen · Post by **stolen** » 13 Mar 2021, 17:29

Sure! Oh, also one more thing, not sure if you're aware: At the moment your circuit operates at unity gain with a passive crossfade, meaning that each of your return signals would be reduced to half volume with the blend in middle position and volume at full. With distortion pedals this is fine, but for modulation etc. you might want to add a little bit of gain near the output so you can compensate if necessary. Is this relevant to you?

stolen · Post by **stolen** » 13 Mar 2021, 18:51

Uhm,,,,

sorrysorrysorry we must've been sleeping earlier our thing above is unstable! Here's the fixed up section, this should work happily:

: splitter.png (108.16 KiB) Viewed 1786 times

Hope this didn't cause any problems!

forgotmyeeclass · Post by **forgotmyeeclass** » 13 Mar 2021, 22:16

stolen wrote: ↑13 Mar 2021, 17:29 Sure! Oh, also one more thing, not sure if you're aware: At the moment your circuit operates at unity gain with a passive crossfade, meaning that each of your return signals would be reduced to half volume with the blend in middle position and volume at full. With distortion pedals this is fine, but for modulation etc. you might want to add a little bit of gain near the output so you can compensate if necessary. Is this relevant to you?

I was aware that it's a passive volume, the main use of this was to have compression in the lo-pass loop and distortion in the hi-pass loop. I might want to use modulation also though.
Thanks for the updated drawing! No problems at all, I'm trying to get the schematic sorted before ordering any parts.

stolen · Post by **stolen** » 15 Mar 2021, 11:07

Okay! We made another mistake, that really wasn't our day, C12 is a little bit on the low side to carry the load of the proposed decreased R18-20 values, maybe multiply them by 2 or increase C12 to 470n (which would be quite a large film cap though).

If you wanna plan ahead and don't mind adding another OPA, we drew up a simple modified merger stage that includes an output booster. With Vol and Blend in the middle position gain for each channel should be unity in the given configuration.
http://stolencircuits.net/fsb/merger.jpg

Ben N · Post by **Ben N** » 15 Mar 2021, 12:45

Did you see this thread: viewtopic.php?f=7&t=30823?

forgotmyeeclass · Post by **forgotmyeeclass** » 15 Mar 2021, 16:07

stolen wrote: ↑15 Mar 2021, 11:07 Okay! We made another mistake, that really wasn't our day, C12 is a little bit on the low side to carry the load of the proposed decreased R18-20 values, maybe multiply them by 2 or increase C12 to 470n (which would be quite a large film cap though).

If you wanna plan ahead and don't mind adding another OPA, we drew up a simple modified merger stage that includes an output booster. With Vol and Blend in the middle position gain for each channel should be unity in the given configuration.
http://stolencircuits.net/fsb/merger.jpg

Thanks again

Ben N wrote: ↑15 Mar 2021, 12:45 Did you see this thread: viewtopic.php?f=7&t=30823?

I did see that thread, thanks. I was looking to build something that did basically the same things but was relatively a bit simpler (although maybe as it turns out, it isn't).