Low leakage cap for synth CV

[NoisyIndividual]

Hi,

I have a Roland SH09 synthesiser that doesn't hold the CV voltage very well once a key is released.
The CV from the keyboard charges a 0.33uF cap, then when the key is released, the voltage is just held by this cap.
You then don't want this voltage to drop and have the held note drop in pitch.

The original is a large 200v type, way beyond the few volts actually required, so size is very much unimportant here, there's a lot of room.

So I'm just wondering what type cap construction is best to use/avoid here, so it will hold its voltage as long and as reliably as possible.

 

[Khron]

Polypropylene film, by the looks of things.

https://en.wikipedia.org/wiki/Film_capacitor#Polypropylene_(PP)_film_capacitors
"In addition, PP film capacitors have the lowest dielectric absorption, which makes them suitable for applications such as VCO timing capacitors, sample-and-hold applications, and audio circuits. They are available for these precision applications in very narrow capacitance tolerances."

 

[Newmarket]

Hi,

I have a Roland SH09 synthesiser that doesn't hold the CV voltage very well once a key is released.
The CV from the keyboard charges a 0.33uF cap, then when the key is released, the voltage is just held by this cap.
You then don't want this voltage to drop and have the held note drop in pitch.

The original is a large 200v type, way beyond the few volts actually required, so size is very much unimportant here, there's a lot of room.

So I'm just wondering what type cap construction is best to use/avoid here, so it will hold its voltage as long and as reliably as possible.

What sort of rate of voltage / pitch drop are you experiencing ?
Wondering if there might be an unintentional current path in parallel with the cap.

 

[NoisyIndividual]

It pretty much start dropping within a second.

I need to eliminate the Cap first, then look for other things as necessary.

 

[JohnRoberts]

+1 polypropylene is probably your most practical choice. Polystyrene is a great dielectric but I've never seen one that large. I've seen as large as 0.1uF polystyrene caps and they were huge (the size of my thumb).

Most caps will list leakage on their data sheet. DA (dielectric absorption) is a slightly different characteristic but Low DA is very useful for accuracy and stability of rapidly charged capacitors in sample and hold circuits, or the like.

JR

 

[AnalogPackrat]

Polystyrene or polycarbonate types were used in many analog synths for their lower loss and temp stability.

You might also check the PCB for possible leakage paths due to moisture damage or prior repairs near the hold cap.

 

[gyraf]

+1 for leakage somewhere else

 

[NoisyIndividual]

I was really hoping it isn't going to be leakage elsewhere, I had considered that.

The PCB is very clean, and I just cleaned it again. The op-amp that buffers the voltage is one of those metal can spiders that I really don't want to touch if I don't have to, then there is a FET that connects the cap to the keyboard CV via the Gate signal that could be leaking, but again I don't want to touch a decades old FET if I don't have to.
So I will replace the cap, (which is possibly a good idea anyway if I want the synth to last a few more decades) and if it isn't the cap, then I will have to look further afield at these other items.
Which I feel I am going to have to, unfortunately.

https://manuals.fdiskc.com/flat/Roland SH09 Service Manual.pdfPage 4.

 

[zamproject]

Hi

Recently help someone (another forum, syth dedicated) about a SH2 that don't hold well the KB CV any more, like yours.
Lost about 1semitone, 12mV in 30sec, even after cap replacement.
I then made the math and 5GΩ is enough leak for that lost.
I suspect a light PCB deterioration over time, or impregnated moisture.
Suggested him to try floating this part of the circuit, simple and SH09 is the same circuit there, a fet the cap and the opamp.

So pull out the leg of the fet (Q2) as pin 3 of IC2, make a floating junction with resistor leg or whatever wire you have, and joint C2 on it too.

After that the SH2 drop 1 semitone in 15min !!!

 

[zamproject]

The PCB is very clean, and I just cleaned it again

Paradoxically that may be counterproductive...especially for this part of the circuit that need higher possible R/Z isolation.
IIRC pcb from this Roland era is brown, so no FR4 but single sided phenolic/paper...
Putting liquid on it with solvent may be absorbed, event a little may influence parasitic Ω

As I say a 30 sec drop from 5.000 to 4.917V in 330n is about 5GΩ, it's high but way enough.
The same drop in 15min is equal to 100GΩ parallel discharge resistor, that's the benefit of floating this part of the circuit !!!

The op-amp that buffers the voltage is one of those metal can spiders that I really don't want to touch if I don't have to, then there is a FET that connects the cap to the keyboard CV via the Gate signal that could be leaking, but again I don't want to touch a decades old FET if I don't have to.

I understand your fear, but it's probably a move you can't avoid. If I'm right about my analysis here you'll be more than happy with the result.

Cheers
Zam

 

[ccaudle]

So I will replace the cap, (which is possibly a good idea anyway if I want the synth to last a few more decades)

What type of capacitor? Electrolytic capacitors have degradation mechanisms, but I have never heard any discussion of film cap degradation. As far as I am aware most people do not replace film caps pre-emptively, there just isn't a common enough failure mechanism to make it worth the effort.

 

[NoisyIndividual]

Well, I'm not about to spend a huge amount of time looking for a leak without first proving the leak isn't the cap.
I like working with known knowns, then you have something you can test with.

 

[NoisyIndividual]

So pull out the leg of the fet (Q2) as pin 3 of IC2, make a floating junction with resistor leg or whatever wire you have, and joint C2 on it too.

After that the SH2 drop 1 semitone in 15min !!!

OK, so I went ahead and removed the FET. This has reduced the drop drastically, but it is still an unacceptable semitone in 30 seconds.
So I guess I'll look for another FET, replace the cap to see if that does anything, and continue from there.

So, multiple leaks it would seem.
And I bet one of them is the PCB.

 

[Khron]

You don't know until you know...

 

[zamproject]

Ho !
I did not get that the drop is faster than 30sec ?!? so yes probably failing junction in the fet too ?!?
How did you test without the FET ?

 

[AnalogPackrat]

What is the op-amp model #?

 

[NoisyIndividual]

Ho !
I did not get that the drop is faster than 30sec ?!? so yes probably failing junction in the fet too ?!?
How did you test without the FET ?

Soldered a couple of wires to the PCB drain and source pads and briefly shorted while holding a key.

What would you say was a reasonable semitone drop time to aim for, given this synth is now an old lady?

 

[zamproject]

What is the op-amp model #?

LF13741 which seem a 741 with jfet input ?
5x10*11Ω input according to datasheet I find

What would you say was a reasonable semitone drop time to aim for, given this synth is now an old lady?

I have no idea what was the spec in the 80' strait out from the factory, but the 15min for one semitone drift the other guy report after my suggestion to float this node is quite impressive and (to me) a must do upgrade/mod.
There is chance that with phenolic/comp-paper pcb even band new SH don't perform that well, not talking time degradation which is probably unavoidable...

 

[NoisyIndividual]

OK, so that could mean the FET was entirely the problem then, and the circuit was always a bit crap from the factory.

 

[zamproject]

It's difficult to take a time machine and test an SH strait from factory but I won't say crap !
To summary, probably the FET is faulty if without him you get aprox 30 sec for a semitone detune ?
But 30 sec is still short (depending of music you want to perform), so if you have already everything opened and pcb out you'll have to float that node