Sub-osc with CD4013 chip issue

[snk]

Hello, I am trying to build a sub-oscillator module for a Yamaha CS10 analog synthesizer.
I am using the Yusynth schematic, adapted for modern CD4013 chips (the transistor is setup as common emitter).
This is the original schematic :
https://yusynth.net/gear/CS10_en.html


This is the layout I have drawn from it :


The circuit is very simple, and it is already wired to the synth.
My concern is that I don't hear anything coming out from the sub-oscillator circuit so far :
When I am an audio probe, I can hear the original signal coming to the CD4013 pin 3, but i hear nothing coming out from pins 11 and/or 12 (which are supposed to be the -1 and -2octave out) if i'm not mistaken.

I have checked the voltages with a multimeter, and this is what i am getting :
pin 01 : 14,98v
pin 02 : 15,04
pin 03 : 7,93
pin 04 : 0,07
pin 05 : 14,98
pin 06 : 0,05
pin 07 : ground
pin 08 : 0
pin 09 : 15,06
pin 10 : 0
pin 11 : 15,02
pin 12 : 15,05
pin 13 : 0,02
pin 14 : 15,06

Do these values seem correct ?
I don't have a spare CD4013 to swap and check with another chip:(

Thank you in advance for any hint:)

[duck_arse]

what does your datasheet tell you about the transistor orientation in your layout, please? that's for a start.

[snk]

what does your datasheet tell you about the transistor orientation in your layout, please? that's for a start.

That i'm not the smartest guy on earth ?  

[snk]

With the transistor in the right direction (i'm so dumb : I previously made a couple layouts with the 2N3904 in the right orientation, and turned it 180° the other day  ), i'm getting some octave down signal from the 4013 pins, but not from the synth...
I will investigate the wiring further, because something lmight be wrong there...

(Also, one thing which surprises me is that when I check the circuit using an audio probe, sometimes the octave down sounds are clean, and sometimes "wobbly")

[snk]

Ok, I'm suspecting either a dodgy 4013 or a dodgy socket or a badly soldered veroboard (but the solders looks fine) : when "just" testing the circuit, I can't hear anything, but if i push on the top of the chip with my finger, I can hear the sound...

[ElectricDruid]

There was a thread about a similar problem on Modwiggler. Basically, the modern 4013's are a lot faster than the old ones and it screws up some of the old synth sub-osc circuits because they're fast enough to react to capacitively-coupled spikes from other pins. The SH-101 one was the example in that case, I think. This is the thread:

https://modwiggler.com/forum/viewtopic.php?t=221841

The TLDR version is that the fix is to add a 1K resistor between D and Q! and a 100n cap from D to ground.

HTH

[snk]

Wonderful, Electric Druid, thank you so much for the hint!

[snk]

When you mean "D" and "Q", which pins exactly do you mean?
(there are several)



From this picture, it seems to be D=pin9 and Q=pin13?

[snk]

Ok, from this schematic, it's pin 2 to 5 and pin 9 to 12


The schematic shows polarized caps, is it required?
Also, the schematic seems to call for 2.2k resistors, while you suggested 1k : which would be the difference?

[snk]

This is what I am coming with...

[ElectricDruid]

Ok, from this schematic, it's pin 2 to 5 and pin 9 to 12


The schematic shows polarized caps, is it required?
Also, the schematic seems to call for 2.2k resistors, while you suggested 1k : which would be the difference?

Yes, exactly. The modern chips require that you deliberately *slow down* the feedback from the !Q output to the D input (on either flip-flop, hence the choice of pin numbers).

2K2 is going to provide even more smoothing/slowing down than 1K. If you go with too large a value, you'll find the sub-octave effect doesn't work beyond certain frequencies, but as long as it covers all the stuff you care about, a bigger value is probably better.

HTH

[Rob Strand]

My concern is that I don't hear anything coming out from the sub-oscillator circuit so far

From that I suspect the input square-wave doesn't have sufficient swing. For the original application of a mod the square-wave is nice and clean and swings virtually rail to rail.  Not an audio signal.  It's not clear if the intended input swings negative.  The diode protects from negative swings but is there negative swing?   In some circuits with negative swings you might find a series cap on the input.  It all depends on the details.

Have you got the R and S inputs on the first gate grounded?   The schematic has three pin numbers but only one connection.  All those pins need to be grounded.

The schematic shows polarized caps, is it required?
Also, the schematic seems to call for 2.2k resistors, while you suggested 1k : which would be the difference?

This schematic has a Schmitt-trigger front-end which might make the circuit more sensitive  However the aim of the Schmitt-trigger and the 2k2 +100n filters is about making the circuit more robust.

If you aren't hearing anything you aren't even getting the clock on the first gate to trigger.   If you were hearing something and it was glitchy then perhaps the second circuit would help.

If we knew what the square-wave input was it might be possible to make a better suggestion.   A one transistor inverting switch might work.  This idea but maybe different values and put a reverse diode across base and emitter for protection; can also put a resistor across base and emitter to tune sensitivity.

Supply voltage not important at this point.  Point O would connect to the clock input.

With diode:

[snk]

Thank you all for the help, that's very valuable.

If we knew what the square-wave input was it might be possible to make a better suggestion.   

I am going to give a bit more information about the synth and the mod :

- The synth is a vintage analog synth made by Yamaha in the late 70s. You can find the service manual here : https://www.synthxl.com/yamaha-cs-10/

It's lovely, but the single oscillator design doesn't help getting deep basses...

- The Sub-Osc mod was designed by Yves Usson (Yusynth), who later designed the MiniBrute. The good point is that it's a talented and well-respected synth designer... The bad point is that he did the mod on his very own synth... 30 years ago : it's a one-shot,  not a commercial product, and back in the days he may have used the older version of the CD4013, which may explain why his schematic doesn't work anymore "right out of the box" and requires some tweaking.

2K2 is going to provide even more smoothing/slowing down than 1K. If you go with too large a value, you'll find the sub-octave effect doesn't work beyond certain frequencies, but as long as it covers all the stuff you care about, a bigger value is probably better.

Thank you. I will start with 1k.
I can try with Mylar 100nF caps (instead of polarized electrolytics)?

If you aren't hearing anything you aren't even getting the clock on the first gate to trigger.   If you were hearing something and it was glitchy then perhaps the second circuit would help.

When I am applying a small pressure on the chip with my finger, I can hear the -1 and -2 suboctaves coming out of the chip, so i am suspecting either a bad socket or a faulty CD4013. I have ordered some more 4013, and I am willing to rebuild the circuit if required (but first I'd like to know what works and what doesn't and if the schematic is fine before building another wrong layout).

[snk]

Ok. I build another veroboard, with the additional mods (1k resistor + 100nF electro cap). I used another CD4013 (which I borrowed from another circuit).
The result is strictly the same 
No sub-octave sound, unless if I touch the chip and push it slightly, i often get a buzzy squarewave (but not a clean one).
The build is not the same, the chip is not the same, the socket is not the same... the issue is the same 

[snk]

Have you got the R and S inputs on the first gate grounded? The schematic has three pin numbers but only one connection.  All those pins need to be grounded.

Sorry, Rob, I'm not following you  : what do yo mean by "R and S"?

[duck_arse]

you have the datasheet in front of you, yes? observe the drawing of the flip flops, two square boxes with pins/legs sticking out of them. ALL the pins of flipflop one will have "1" on them, and all the pins for flipflop the second, with nary a scent of imagination, will have "2" in their names.

each FF will have a D data input, S set input, R reset input, Q output, !Q bar|not Q - an inverted Q output, and a CLK clock input. look how the circuit presents the connections for a divider - an output fed back to an input, of the same half.


ALL unused INPUTS of a cmos chip, ALL CMOS CHIPS, must be taken to a logic level. this rule is not negotiable, cannot be overruled by some web guy telling you otherwise, is not negated by a wrong drawn circuit diagram. a level can be high or low, supply voltage or ground voltage. to get your flipflops to work properly, and it might be your finger effect proving this, you need to connect all the S pins and all the R pins to ground, because that's what the datasheet tells you.


obvs supply and ground is common to both FF's. you probably should be using poly/mylar/plastic/ceramic caps for the slowdown timings. stay with the parts you have at the moment, you will get them working right eventually.

[snk]

Thank you so much, Duck Head (and sorry Rob for being a bit thick).
I was looking at the colored pinout posted above, which doesn't feature "R"...
With this one it's better

you need to connect all the S pins and all the R pins to ground,

Ok, so I see that pins 8 & 10 (flipflop #2's Set & Reset) are already grounded, but not pins 4 & 6 (flipflop #1's Set & Reset).
I will put them to ground and let you know how it goes.

[snk]

you need to connect all the S pins and all the R pins to ground, because that's what the datasheet tells you.

It works!
Thank you so, so much Duck Head, Rob Strand, Electric Druid!   
[insert hippie, flowers, hurrah emojis]

[snk]

I would have one last question : the audio out from the mod circuit is supposed to be wired to the square wave mixer pot.
It's nice and tidy because you don't have to mess the original pcb (or remove it from the front panel).
But the issue is that if I mute the square wave, the sub-octave also gets muted. I wish I could for instance have the sawtooth wave + the sub-octave mixed together, without having to have also the regular square wave...
I will browse the synth schematic to find another possible place to put the output from the sub-osc circuit, but i'm open to suggestions (because my knowledge is quite limited and I wouldn't want to damage this venerable synth).


(the plan B would be to wire the sub-osc out to a jack socket, and then put a jack cable into that socket, going then to the audio in jack... but it is not very elegant, as I would very much prefer a solution where everything is inside the chassis, and permanently wired).

[snk]

Hello
Ok, I found a good place (imho), I am putting here the screenshots from the schematic and the circuit for future reference.
Again, thank you so much all for your help and good advices : I am so glad to have this mod working, and I couldn't have managed to make it work without you.