Understanding the EHX Y-Triggered Filter

[rankot]

10k.

Thanks a lot! Then, what about R14/R15? 10k too?

[Scruffie]

Pretty sure R14 is 10k, R15 is 1k.

Also, where did R105 & 106 come from?

And how come you've left out Q4 from the original schematic/found on the back of some of the boards.

[rankot]

R105&106 are there to enable parallel usage of both OTAs in LM13700.

Q4 shown on those schematics was not very logical to me, so I've found a photo of a PCB with that transistor added, and it is actually shown on my schematic to the left of MOD switch, together with two resistors marked with * (R13 and R38). R37 is also seen on some of the schematics' but I'm almost sure it's unnecessary.



If you carefully look at that old schematic (showing Q4), there's a wire shorting bot emitter and collector of Q2 to ground, which is not logical. I traced this photo and to me it seems that Q4 shall be connected like show on my last schematic. But please check, four eyes see much better than two

[Scruffie]

You can still use the 13700 in parallel without them.

Sorry I was looking at an older version of your schematic.

The wire connecting Q2 to Q4 and thus Q2's collector to ground is actually a mistake on my original drawing, it's not on the vintage EHX drawing ('twas nearly 10 years ago I did it!).

So Q4 is working as a buffer before the up/down pot, EHX wouldn't have added it unless it was a worthwhile addition (I assume it improves the action on the pot). I think in the photo you just can't see a trace cut under that extra 10k resistor.

So R38 wouldn't be there, R13 would connect to B+, the emitter connects to R13 & the CW side of the pot, base connects to R19/C10 junction and collector goes to ground.

[rankot]

Like this?

Please notice LINK which I placed between Q4 base and Q2 collector, I couldn't find that trace on photos, but it is present in your old schematic. I presume it shouldn't be there, but I left it just in case.

[Scruffie]

Get rid of link and then yes.

[Mark Hammer]

I hate to be picky, but I'm quickly losing track of what is referring to whatIs it possible to identify the different transistors on the board by number?  I.E., which one is Q2, Q4, etc.  I'm hoping to be able to salvage this build without having to etch en entirely different board.  Changing R4 from 1k to 10k only succeeded in producing an "wiggly" oscillation, so I changed it back. 

Changing R14 to 10k improved things a bit, but it's still twitchy and dark.

[rankot]

Sorry Mark, we were referencing the 'original' schematic, but I also tried to have the same part references on mine. Here's mine schematic with all the changes Scruffie noted, and with transistors numbered too.

[rankot]

I hate to be picky, but I'm quickly losing track of what is referring to whatIs it possible to identify the different transistors on the board by number?  I.E., which one is Q2, Q4, etc.  I'm hoping to be able to salvage this build without having to etch en entirely different board.  Changing R4 from 1k to 10k only succeeded in producing an "wiggly" oscillation, so I changed it back. 

Changing R14 to 10k improved things a bit, but it's still twitchy and dark.

Try to remove C23 (10u) if you have installed it, I didn't see it on 'original' schematic, but found it while tracing. Maybe just my error? I've checked all the other values and they're the same as on 'original', except already mentioned resistors (10/1k), and R15 and R19. Here is the updated schematic, with all references the same as in 'original', and all values which looked different to me on the PCB photos I had are shown in ME/ORIGINAL form, for better tracking.

[Scruffie]

I hate to be picky, but I'm quickly losing track of what is referring to whatIs it possible to identify the different transistors on the board by number?  I.E., which one is Q2, Q4, etc.  I'm hoping to be able to salvage this build without having to etch en entirely different board.  Changing R4 from 1k to 10k only succeeded in producing an "wiggly" oscillation, so I changed it back. 

Changing R14 to 10k improved things a bit, but it's still twitchy and dark.

You're not the only one! 

That's not surprising if you just changed R4, you'd need to change R4,5 & 6.

[Mark Hammer]

So, I removed C23, and the resulting sound is once again clear as a bell, and perfect volume balance to the bypassed signal...except there is now no sweep. Moving the up/down pot works as filter-tuning, but yields nothing that responds to picking.  Still a bit to go.  I'll need to comb through this thread for details I've overlooked.

[rankot]

My PCB has finally arrived, so I will populate it right now!

[rankot]

First report: IT'S ALIVE! 

I've built it w/o C23 (10u), and I also didn't install Q4 which should drive Up/Down pot, I've just connected pot directly to joint of R19, C10 and R13.

R4/R5/R5 = 10k
R10 = 12k
R19 = 1k8
R105/R106 = 510Ω

R21/R37 not installed.

It is working, but I've noticed that there is some kind of 'reverberated' filter - it makes a quack on strum, but then there's another one (or two) quacks with less intensity after that. It doesn't happen always, I must check what's producing that - but it seems to be normal, and sounds almost the same as in this demo:



Also, Up/Down pot works OK, but not when turned totaly CW - then there's almost no sound - tone is heard only at the beginning of the note (during pluck), then nothing. So probably Q4 is necessary, but I will try that in the morning.

And one more thing - to my ears, that Range switch doesn't do much. I will also install Filter switch and Sensitivity pot (as proposed by Mark) in the morning, and report my findings.

[rankot]

Not to mention that I'm extremely happy with this, because my first PCB for this filter didn't work at all!

 

[rankot]

I've been ltspicing this this morning, and here are the conclusions:

- If R2/4/5/6 are 1k, then C23 should be there. If they're 10k, then no C23.
- R21 is making a difference in envelope shaping, so it can be switchable - I'll test it.
- The most difference is made with C12 value, and I will test it within 1u-10u range.
- R19 should be 1k8.
- R10 value is not so important (12/18k).
- R37 doesn't seem to be necessary.
- Q4 can make a lots of difference, so it can be also a nice idea to put it on a switch during testing (and maybe after).

So I will test this on a real pedal and report.

Cheers!

[rankot]

OK, here are the first findings:
- R21 doesn't make too much audible difference, regardless simulation, but it seem that makes things better with Up/Down pot at CW. I also tried to add C23 (10u) and it made things better a little bit, so I left it there too - it didn't make things worse, if nothing at all.
- I've found a point where U/D pot starts making trouble and measured it - it seems that it needs some 3k6 added resistance at CW side to work well. Maybe I should try R19 with 4k7? That's next step.

[rankot]

Increasing R19 to 4k7 didn't make any improvement. So I've tried to add Q4 and R20 between C10/R19/R/13 junction and U/D CW. It seems to be a little bit better and filter works a little bit different, but there is still a dead zone when U/D is rotated to CW. I will add that resistance between pin 3 of CW and Q4 emitter. I will also try to make a switch so Q4 is switchable, to easily make comparisons.

Tested Sensitivity pot as proposed by Mark, but it didn't bring audible difference to me, so I've removed it.

Next is C12 value, hope it holds some surprises!

[rankot]

I tried to replace R10 from 12k to 18k, but it didn't bring any improvement to Up/Down action. After that, I tried a Filter switch, as proposed by Mark, and it brought some changes, although I like it the most when switched to 3n3. However, I'll leave it as on my schematic.

Switching that switch produces a thump, so any advice how to avoid that is welcome!

Next, I started experimenting with C12 value. Putting 1u there made sweeping shorter and also brought bigger "dead" zone at Up/Down pot CW side (and perhaps some at CCW), so it seems that this capacitor is the main problem. At 1u, dead zone is 22k big, compared to 3k6 with 10u.

[rankot]

And finally - I decided to use 2u2 as a base value for C12, paralleled with another 2u2 and 10u mounted on On-Off-On switch. It gives a nice variation of sounds. There is also a 4k3 resistor between Up/Down pot's CW and rest of the circuit, to skip the dead zone. Seems to fit all above values for C12, but it can be up to 10k, try what suits you.

There is also a switch called Drive, to enable/disable Q4. I left it there for testing purposes, because I didn't have enough time to test impact of Q4.

As I have promised, here is the final version of a schematic, together with PCB files. Please note that there are two SMD (or you can try with small MLCC capacitors as I did) 100n capacitors for noise filtering above both ICs. Also, I used patterns for MLCC capacitors wherever I could (those not in a signal chain), but if you want to use polarized for some reason, check polarity according to the original schematic.

Merry Christmas and a Happy New Year!

https://docdro.id/N6ioW9N

[rankot]

I've been tweaking this in ltspice for a while, and it seems that switching Q4 could be interesting - if it is out of the circuit, width of pulse is shorter. Also, R19 seems to be in reverse relation to pulse amplitude, so I will try to put 1k resistor in series with 5k pot there, to is if there is any real sonic impact.