[sdiy] Where to buy ceramic caps

David G Dixon dixon at mail.ubc.ca
Sat Dec 30 08:47:08 CET 2017


> > Actually, my simulation suggests that even 5% tolerance caps would 
> > give suitably precise phase displacement.
> 
> Repeat after me:
> do not place absolute faith in "simulations"
> do not place absolute faith in "simulations"

I've already built one PDN with trimmers instead of precise fixed resistors,
and it works great -- exactly like the simulation.  The simulation gave
exactly what my calculations gave: 90-degree phase displacement between 15
Hz and 15000 Hz with a maximum of +/-0.2 degree error.  The only thing I did
with the latest simulation was to specify 5% capacitors and turn on the "Use
Tolerances" feature of Multisim.  The feature randomizes the capacitance
values (so you have to run the simulations over and over to get a sense of
the performance).  With the tolerances on, the phase error was only about 1
degree maximum, and the Lissajous figures still looked nice and circular.
Since I used Lissajous figures to tune my working PDN in the first place, I
believe that the simulation is telling me that the PDN built with 5% caps
and 1% fixed resistors (selected assuming perfectly accurate caps) will do
what I need.

> I have read of many "simulations" suggesting precision parts 
> in the design phase that resulted in a circuit that never worked.

I always build actual circuits, but I find that my simulations are usually
perfect.  Several million dollars worth of my module designs have been sold
to eurorack users around the world, so I think I know how to design a
circuit that works.

> > I'm trying to achieve capacitors for which I can guarantee 
> the value, 
> > for a 90-degree phase-displacement network, so I can avoid 
> putting 12 
> > multiturn trimmers on my PCB.
> 
> There are far easier designs out there for quadrature phase 
> displacement networks that don't rely on precision components.
> 
> You don't specify if your network is passive or active.

Well, my design is about as easy as it gets: Two parallel sets of six active
all-pass stages in series, each stage tuned to a certain frequency as
predicted by the awesome QuadNet program.  And it works, very very well.
However, the RC values have to be fairly precise to get really superior
performance.

> > I like ceramic caps because they are small
> 
> If you chose ceramic dielectric purely for their size then 
> you're doing it wrong.  ESPECIALLY if the network is passive.

It ain't.  It's active.  We have had this discussion about good ceramic caps
(such as C0G ceramics) many times before, and the consensus has always been
that they are every bit as good as, if not better than, film caps for VCOs
and VCFs.  (Ian Fritz has made this argument, persuasively.)  However, since
you are so convinced that I'm doing it wrong, perhaps you like to share
exactly why...?

> > and I've already laid out the board for them.
> 
> You laid out a circuit board with total faith in a simulation 
> without actually proving the design?  BRILLIANT!

I make my own boards at home, so it's really no big deal if they don't work.
However, they always do.

> > I've found some of the caps I need on various sites, sometimes even 
> > for less than a dollar per.  I f*ck*n' HATE buying 
> capacitors.  It's 
> > as if the makers don't actually want anyone to have them.  
> They are 10 
> > times more expensive than they should be, when you can find them at 
> > all.  They're impossible to read, even with specs on.  It's 
> bullshit, 
> > actually, and I don't know why the electronics world puts 
> up with it.
> 
> Professional audio designers over the last forty years have 
> played with $$$ opamps and opted for alternate circuits that 
> achieve the same goals with cheaper opamps.  Like you, these 
> same designers didn't want to pay 10 times more for opamps.  
> So they sought out other designs that would accomplish the 
> same goal, or rolled their own.

Well, you're singing to the choir here, because I have always used the
trusty TL07X for everything (except S&H, for which I use LF444).  If I need
reasonably low input voltage offset, I use TL074BCN.

> Complaining about OEM pricing isn't going to solve any design problem.

I don't have a design problem.  I just need accurate RC networks, and I
don't want to have to tune them.  Is that so wrong?  Why the hell do they
bother to make close-tolerance caps anyway, if not to get precise RC values?
I can't see any other reason.  So, I don't understand why it is a "design
problem" to want accurate RC values.




More information about the Synth-diy mailing list