Vintage Mistress ticking

[yeeshkul]

Yes mate, the layout is a textbook example of how to induce bollocks to your guitar tone.

[Scruffie]

The layout on those boards is horrible.  The tracks that run to the rate pot snake through the whole circuit.  Nice big square waves coupling to all the other tracks.  Doesn't explain why one version ticks less than the other though as the layouts are pretty much the same according the pics on the EM Mystery Page.  I would guess the voltage regulation was the problem and audio probe the supply pins on the chips in the audio section.

The voltage isn't the issue, trust me, i've played with this circuit alot, it is the design, even the brand of 339/4013 used can have an impact on tick and even the range of the LFO.

Of course the layout doesn't help.

Adding additional decoupling capacitors directly on the 4013 and 339 can help a bit, some units have this cap already, others don't.

[yeeshkul]

I ended up making my own PCB with somehow cautious layout and additional filtering.
The basic idea is two separate DC lines for LFO/VCO and power/signal portion. The LFO/VCO noise that is likely spread by DC is filtered twice before it hits the signal DC line.
There are also two separate grounds stargrounded to one reference point.
I will report when i have it tested.

As you may have noticed, i used a common voltage stabilizer powering as well as a simple output amp stage for True Bypass.

[DrAlx]

I built a V2 EM on perfboard using my own layout.  I used a 78L12 regulator to cut down on the parts count and managed to get the layout down to 21 x 25 holes.  It doesn't "tick" but it does have self-heterodyning noise (i.e. bird tweets on the clock line) which I think is unavoidable with that particular VCO.  The VCO on the V6 and Deluxe EM VCO is much better IMO.

The layout image is here:
https://1drv.ms/i/s!AvrH61utWEtEhzJbikswAzgRULD7

The software I used to make the design can be downloaded for free here:
https://sourceforge.net/projects/veroroute/files/?source=navbar

and the design file for the circuit "18V_EM.vrt" for loading into the software is here:
https://1drv.ms/u/s!AvrH61utWEtEhzOYvzndNcyxTvAE

I added some extra caps to the circuit too (e.g. 1n shunting 6k8 in the pre-emphasis section, and 100p shunting pin1 of the LM339 to extend the low end of the sweep). There are several 100nF ceramics (not shown on the diagram) that I put directly onto the solder side of the board in various places to cut down on noise.  I did that between all IC supply pins. I also manually held a 100nF ceramic between ground and the various pins of the LM339 and listened for noise reduction in the self-heterodyning.  Some pins gave a clear benefit.

[Scruffie]

I'm surprised you still have aliasing noise with the 1n cap at the front, I found that was the easier thing to get rid of compared to the tick, are you sure that's not your power supply?

If you're making your own layout you may as well put the SAD1024 in parallel multiplex to cut down on noise too, perhaps that's why you still had the noise. I know it looks like it is in the schematic but it's not.

[DrAlx]

Hi Scruffie.  You are refering to heterodyning between RF on the audio line and the clock signal, and that would get reduced by the 1nF filtering out RF on the audio as you say.
I am not talking about that though.  I am refering to modulation on the clock line itself.
You hear it at various points of the LFO sweep as you audio probe the clock line directly.
Those tweets on the clock line leak through to the BBD output.


It could be my power supply as you say.  I will have a try with baterries.

[Scruffie]

Huh, i've not encountered that one.

Really is a crappy VCO isn't it.

I did try building the VCO with just a dual comparator, using a dual opamp lfo and dropping all that biasing resistor network, IIRC it was a hair more stable, though not by much. It was a low speed comparator though, I don't know if it would be improved with a Dual 311 type.

[stallik]

No idea if this helps but I built a deluxe electric mistress from a thread on this forum. It was designed to use 12v. No ticking. Ive temporarily hooked it up with 9v and it sound fine but ticks gently in the background. I can live with it

Don't know if yours might behave in the same way

[yeeshkul]

DrAlx - do you use a charge pump 9->18V? That one may cause heterodyning.

[DrAlx]

DrAlx - do you use a charge pump 9->18V? That one may cause heterodyning.

Just a wall wart, and I am pretty sure it's linear, not switching. Regarding the modulation on the clock lines, I hear it if I audio probe the clock lines in my EM3207 (a clone using the V6 VCO) but it's much quieter than in the V2 VCO and is not audible at the BBD output.

[DrAlx]

It isn't the power supply. I just tried batteries on both the V2 and the V6 circuits.  Audio probe the clock lines and you can clearly hear heterodyne type noise on the clock lines themselves. You get it even in filter-matrix mode with the pitch of the whistle changing as if you are tuning in on an AM radio. Audio probe the BBD output pins on the V6 circuit (before it reaches the smoothing cap) and you can here the whistles there too (although fainter).

[yeeshkul]

I have finally built the clone on my own PCB design as pictured above.
I am feeding it with 7812 stabilizer. I omitted the voltage divider behind it, so there is a strong and stable voltage source.
The unwanted flanging drone sound is really quiet, almost not there, which is great, so as for this noise the new PCB helped.
The ticking is still present however, and very annoying.

Here are my observations:
1. I was listening by audio-probe on the output of 4013 (pin 12,13). The clicking is there.
2. The clicking is present only when the Range knob is down. It starts on like 3/4 of the path, and it is the most prominent when the Range knob is all the way down. The click appears at the deep end of the flanging cycle.
3. When i listen to the actual pedal output, the flanging effect is suppressed or completely gone when the clicking comes out.
Is it possible that the ever changing DC levels cause the VCO dropping out of the oscillation?
4. The flanging range is shorter than on my original EM, which has 20s cycle. My has only 15s cycle. I guess i can help it by additional capacity in the between pin 2 and 4 of LM339.

[DrAlx]

Yes, if I remember correctly the VCO can only handle a certain range of CV levels.  I recall that lowering the range pot in that circuit not only lowers the maximum CV level in the sweep but also lowers the minimum CV level in the sweep. I suspect there is a CV level below which the VCO stops and you are reaching it.

Try filter matrix with Max color and see if there is a CV level that makes the bell-like filter matrix sound disappear.

[StephenGiles]

I have a cloned Electric Mistress I bought around 1977 - it must have been because my son was with me when I bought the board, so before November 1977, and it didn't click at all! If I can find it I'll fire it up over the weekend and see if that is still the case.

[DrAlx]

I dug out some CV voltage measurements I took from an 18VEM clone I built a few years ago:

Rate Pot is at minimum (i.e slow sweep).

Range Pot at Max:  CV:  0.340V to 8.700V   (giving clock rates from 222 kHz to 30.76 kHz respectively)
Range Pot at Min:   CV:  0.068V to 0.527V  (giving clock rates from 256 kHz to 212 kHz respectively).

So the minimum CV level dropped as the Range pot decreased.
But I did not get the VCO stopping.

Having said that there is no trimmer on the current source charging the clock cap in that circuit ,
so there will be variations in how CV maps to clock rate between different builds.

[yeeshkul]

SOLVED
Guys i am so sorry, i put the transistor on the VCO input upside down. So the current was insufficient. I for some weird reason thought it was a buffer, but it really is a common base current driver.
Well, now there is no ticking at all, and the circuit is very quiet, so i'd say ... WIN! 

[yeeshkul]

I am gonna post the actual board if anyone is interested.
By the way, one of my original units has the transistor also upside down - that was the noisy unit i was gonna repair first, and then decided to build my own clone, making the same mistake on my own board, hahahaha.
So if anyone owns a ticking unit, you know where to look now

[yeeshkul]

Here is my clone

[DrAlx]

Sorry to bump an old thread Yeeshkul.
Just looking at your layout, I can see two "grounds" (light blue and dark blue).
The dark blue ground with all the text etched into it is clearly connected to the central ground pin of the voltage regulator.
My problem is that I can't see how the light blue section gets a DC connection to the central pin of the regulator.
There is no obvious wire connecting the two (and if there were you wouldn't actually have two grounds but one)
and I can't spot any low resistance connection either.

To have two decoupled sections "A" and "B" I would expect to see the sections having their own decoupling caps AND resistors going back to Reg_Out and Reg_Gnd.  I mean something like this.

Code Select Expand


Vcc_A <-----+-----10R-----Reg_Out-----10R-----+-----> Vcc_B
            |                |                |
          220uF            220uF            220uF
            |                |                |
Gnd_A <-----+-----10R-----Reg_Gnd-----10R-----+-----> Gnd_B



or at the very least this...

Code Select Expand


Vcc_A <--------------Reg_Out-----10R-----+-----> Vcc_B
                        |                |
                      220uF            220uF
                        |                |
Gnd_A <--------------Reg_Gnd-------------+-----> Gnd_B



The resistors ideally should be in both the "supply" and "ground" arms, but at least in one of the arms.  If the resistance is zero then very little filtering is going on to decouple the sections and you basically have a common supply.

From your layout it seems you have a few 220uF caps, but many of them are in parallel to each other, just distributed about the board.  So how is the light blue section connected to the dark blue?  Is the diagram you posted correct?

[DrAlx]

I see it now in the photo.  A big fat wire going to the light blue section. Seems the light blue and dark blue sections are directly connected after all.
That means that although you have a star ground, you don't in fact have separate decoupled supplies for the audio and LFO/VCO sections.  You really have a single supply node and a single ground node, with a lot of capacitance between those two nodes.
It just happens to be distributed over the board.