i made a few posts under the political threads, and in an effort to
redeem myself, i'll bring up this audio question that i've been
mulling over:

it's not too complicated, i don't think. i just want to build a few
simple passuve RC filters -- a pot in series with a capacitor
(although i might put a switch for multiple cap values).

i'd rather not buffer the circuits, but leave them as simple tone
controls -- i don't think any of the gear here has transformers in the
ins/outs, so i think i'm dealing mostly with opamps, which i
understand would be more or less immune to a changing load such as a
RC filter, right?

anyone recommend a particular type/make of capacitor for this?

thanks,
chris deckard, recovering political opinionator.

mr c deckard wrote:
i made a few posts under the political threads, and in an effort to
redeem myself, i'll bring up this audio question that i've been
mulling over:

it's not too complicated, i don't think. i just want to build a few
simple passuve RC filters -- a pot in series with a capacitor
(although i might put a switch for multiple cap values).

i'd rather not buffer the circuits, but leave them as simple tone
controls -- i don't think any of the gear here has transformers in the
ins/outs, so i think i'm dealing mostly with opamps, which i
understand would be more or less immune to a changing load such as a
RC filter, right?

I did an article in Recording magazine recently on a non-passive filter
network using RLC filters, that sounds pretty good.

If you want any low-pass stuff and you want it to be passive, you will
need inductors.

There is a really good discussion of this stuff in the Audio Cyclopedia.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

"mr c deckard" wrote in message
om

i made a few posts under the political threads, and in an effort to
redeem myself, i'll bring up this audio question that i've been
mulling over:

it's not too complicated, i don't think. i just want to build a few
simple passuve RC filters -- a pot in series with a capacitor
(although i might put a switch for multiple cap values).

i'd rather not buffer the circuits, but leave them as simple tone
controls -- i don't think any of the gear here has transformers in the
ins/outs, so i think i'm dealing mostly with opamps, which i
understand would be more or less immune to a changing load such as a
RC filter, right?

RC filters include both the source impedance and the load impedance among
their parameters. That's why buffers are so frequently used with them.

mr c deckard wrote:

i made a few posts under the political threads, and in an effort to
redeem myself, i'll bring up this audio question that i've been
mulling over:

it's not too complicated, i don't think. i just want to build a few
simple passuve RC filters -- a pot in series with a capacitor
(although i might put a switch for multiple cap values).

i'd rather not buffer the circuits, but leave them as simple tone
controls -- i don't think any of the gear here has transformers in the
ins/outs, so i think i'm dealing mostly with opamps, which i
understand would be more or less immune to a changing load such as a
RC filter, right?

anyone recommend a particular type/make of capacitor for this?

SNORT !

Christ ! I just nearly ****ed myself laughing !

As it isn't April 1 - I assume you're just inexperienced in these matters.

A course in practical electronics may be of assistance.


Graham

RC filters include both the source impedance and the load impedance among
their parameters. That's why buffers are so frequently used with them.

perhaps i was thinking in the wrong direction. i suppose the low
impedance of the opamp output will change the time constant of the Rc
network . . .

i guess i can still do this, but i won't be able to calibrate them as
well as i would've liked, since the -3dB frequency (ie, the cutoff
frequency (in a LPF, eg
)) will change depending on which piece of gear is patched in front of
it.

although, 90% of the time, it will be in a channel insert, so, given
the in and out impedances of the console insert, i can calibrate the
front panel to that case.

but, in any case, it's not too critical, i really want a simple LPF
and HPF without having to switch in 4 gain stages with the onboard eq.

thanks,
chrisdeckard
saintlouismo

I did an article in Recording magazine recently on a non-passive filter
network using RLC filters, that sounds pretty good.

i still wish you and mike could one day release a collection of your
articles.




If you want any low-pass stuff and you want it to be passive, you will
need inductors.

i'm ok with that, and had considered it. but what's wrong with just
putting a .01 uF cap to ground paralleled with a pot, like a tone
control on a guitar? since these are opamps, i don't have to worry
too much about loading a 10ohm (theoretical) output impedance, and a
(theoretical) 10kohm input imp., right?



There is a really good discussion of this stuff in the Audio Cyclopedia.

which i have, so i'll check it out.


--scott



cheers,
chris deckard
saint louis mo

mr c deckard wrote:

perhaps i was thinking in the wrong direction. i suppose
the low impedance of the opamp output will change the time
constant of the Rc network . . .

Generally speaking you will have to have series resistors on the outputs
to get "convenient" capacitor values.

but, in any case, it's not too critical, i really want a simple
LPF

Why LPF, this one is _complicated_(x), it needs two components, a series
resistor and a capacitor to ground.

and HPF

A "suitably" dimensioned non-electrolytic capacitor in an insert jack
plug can do. I made a handful once upon a time and used two of them
once.

(x) _complicated_ == more difficult to fit in a jack plug ....

chrisdeckard
saintlouismo


Kind regards

Peter Larsen

--
*******************************************
* My site is at: http://www.muyiovatki.dk *
*******************************************

mr c deckard wrote:

I did an article in Recording magazine recently on a non-passive
filter network using RLC filters, that sounds pretty good.

i still wish you and mike could one day release a collection of your
articles.

If you want any low-pass stuff and you want it to be passive,
you will need inductors.

No. Series resistor and cap to ground will in my understanding do if
first order is OK.

i'm ok with that, and had considered it. but what's wrong with just
putting a .01 uF cap to ground paralleled with a pot, like a tone
control on a guitar? since these are opamps, i don't have to worry
too much about loading a 10ohm (theoretical) output impedance, and a
(theoretical) 10kohm input imp., right?

Opamp output stages have series resistors, usually no less than 50 Ohm,
preferably 100, to protect the opamp from capacitive load.

There is a really good discussion of this stuff in the Audio Cyclopedia.

--scott

chris deckard


Kind regards

Peter Larsen

--
*******************************************
* My site is at: http://www.muyiovatki.dk *
*******************************************

"mr c deckard" wrote in message
om

RC filters include both the source impedance and the load impedance
among their parameters. That's why buffers are so frequently used
with them.

perhaps i was thinking in the wrong direction. i suppose the low
impedance of the opamp output will change the time constant of the Rc
network . . .

Absolutely.

i guess i can still do this, but i won't be able to calibrate them as
well as i would've liked, since the -3dB frequency (ie, the cutoff
frequency (in a LPF, eg )) will change depending on which piece of gear
is patched in front of
it.

Absolutely.

although, 90% of the time, it will be in a channel insert, so, given
the in and out impedances of the console insert, i can calibrate the
front panel to that case.

If the impedances are stable, known and included in the filter's design,
then you have predictable results.

but, in any case, it's not too critical, i really want a simple LPF
and HPF without having to switch in 4 gain stages with the onboard eq.

Why do you fear the onboard eq?

mr c deckard wrote:
RC filters include both the source impedance and the load impedance among
their parameters. That's why buffers are so frequently used with them.

perhaps i was thinking in the wrong direction. i suppose the low
impedance of the opamp output will change the time constant of the Rc
network . . .

i guess i can still do this, but i won't be able to calibrate them as
well as i would've liked, since the -3dB frequency (ie, the cutoff
frequency (in a LPF, eg
)) will change depending on which piece of gear is patched in front of
it.

This is another reason why everything should be run 600 ohm constant
impedance. It makes passive EQ modules much easier to deal with.

although, 90% of the time, it will be in a channel insert, so, given
the in and out impedances of the console insert, i can calibrate the
front panel to that case.

but, in any case, it's not too critical, i really want a simple LPF
and HPF without having to switch in 4 gain stages with the onboard eq.

So, build my box. Only a single gain stage, unbalanced inputs and outputs,
real inductor sound.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

In article ,
mr c deckard wrote:
I did an article in Recording magazine recently on a non-passive filter
network using RLC filters, that sounds pretty good.

i still wish you and mike could one day release a collection of your
articles.

So do I.

DIY Active Equalizer
Recording, Vol. 17, No. 3 (Dec 2003)


If you want any low-pass stuff and you want it to be passive, you will
need inductors.

i'm ok with that, and had considered it. but what's wrong with just
putting a .01 uF cap to ground paralleled with a pot, like a tone
control on a guitar? since these are opamps, i don't have to worry
too much about loading a 10ohm (theoretical) output impedance, and a
(theoretical) 10kohm input imp., right?

It's not very controllable for one thing. Using only series networks
means you're going to need a lot of make-up gain. The problem is that
finding inductors of any quality today is hard.

Opamp Labs may still make some networks with RLC constants and a single
op-amp module. It's worth asking Bel over there.
--scott


--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Christ ! I just nearly ****ed myself laughing !

As it isn't April 1 - I assume you're just inexperienced in these matters.

A course in practical electronics may be of assistance.


Graham



i'm glad i was able to give a good laugh. actually it's not april,
but i am trying to help the signal to noise a bit on this list, and
thought i'd bring this idea i've had up. of course a capacitor and
resistor are easy enough, but the devil's in the details, no?

i have an old RCA tube manual that has this cool chart of resistances
at different frequencies for different valued capacitors. i also
understand simple time constant equations.

what is a bit beyond me are the nuances of how an unbuffered first
order passive RC(maybe L) filter will interact with the in and output
stages of the (opamp buffered) inserts of my console.

as far as what kind of capacitor . . . i'm just fishing for any
possible options i MUST avoid, or MUST use. otherwise, i'll end up
using whatever the surplus store has in the bin.

i know there's this kind of upturned nose to people who ask "what kind
of microphone" question, but i think it's all valid. i still read to
see what people are using on what, because i don't like working in a
vaccuum, and sometimes saint louis is very much one.

also, i like to know what theoretical perfection is, so that i can get
a sense for how how close i NEED to get, and how far i can deviate.

all the best,
chris deckard
saint louismo

but, in any case, it's not too critical, i really want a simple LPF
and HPF without having to switch in 4 gain stages with the onboard eq.

Why do you fear the onboard eq?

it's a tascam m3700.





i think there are 4 gain stages per eq section.

i'll be honest here, tho. in the very non-critical tests i've done,
i can't repeatedly tell a difference between the eq being out and in
(set flat). i know this means i shouldn't worry about it, so i guess
it's more of a principle/housekeeping kinduva thing.

well, i suppose it would give me some options, too.

cheers,
chris deckard
saint louis mo

"mr c deckard" wrote in message
om...

Why do you fear the onboard eq?

it's a tascam m3700.

cue "Psycho" shower scene music
--


Neil Henderson
Saqqara Records
http://www.saqqararecords.com

mr c deckard wrote:

Christ ! I just nearly ****ed myself laughing !

As it isn't April 1 - I assume you're just inexperienced in these matters.

A course in practical electronics may be of assistance.


Graham

i'm glad i was able to give a good laugh. actually it's not april,
but i am trying to help the signal to noise a bit on this list,

A welcome move !


and
thought i'd bring this idea i've had up. of course a capacitor and
resistor are easy enough, but the devil's in the details, no?

i have an old RCA tube manual that has this cool chart of resistances
at different frequencies for different valued capacitors. i also
understand simple time constant equations.

Well... time constant is easy enough - simply multiply R * C

To get the - 3 dB point of either a low or high pass filter ( depending how you
configured the components ) is easy enough too with f = 1 / ( 2*pi*R*C )


what is a bit beyond me are the nuances of how an unbuffered first
order passive RC(maybe L) filter will interact with the in and output
stages of the (opamp buffered) inserts of my console.

Here we are talking about *loading*.

Passive EQ is generally expected to be loaded with a high impedance since,
otherwise, the load impedance becomes part of the filter equation.

Sadly for this application, desk inserts are likely to have 'mid-impedance'
inputs - typically around 10kOhms and *will* load most passive EQs based around
financially sensible C values !

As a simple 'rule of thumb' - you want the load impedance to be *at least* 10x
the R in your passive EQ. So, if you use your desk's inserts as they are,
anticipate using an R of 1 kOhm in your filters. This is sadly likely to yield
rather large C values that will be expensive.

The desk's output impedance is normally very low and will not likely influence
your passive filter.

If you want to do passive EQ practically, you are really best advised to include
your own high impedance buffer amplifier after the RC(L) section.

as far as what kind of capacitor . . . i'm just fishing for any
possible options i MUST avoid, or MUST use. otherwise, i'll end up
using whatever the surplus store has in the bin.

Errr.. well avoid using electrolytic ( polarised or non-polarised ) and most
ceramic caps and you'll be OK.

In short, use plastic film types for EQ.


i know there's this kind of upturned nose to people who ask "what kind
of microphone" question, but i think it's all valid. i still read to
see what people are using on what, because i don't like working in a
vaccuum, and sometimes saint louis is very much one.

Here is a good place for ideas.

Just out of curiosity, why does passive EQ interest you so much ?


also, i like to know what theoretical perfection is, so that i can get
a sense for how how close i NEED to get, and how far i can deviate.

The search for the Holy Grail is a long one !

Most ppl settle for 'pretty damn good' - and these days that means potentially
very high performance indeed.


Graham

Peter Larsen wrote:

snip

Opamp output stages have series resistors, usually no less than 50 Ohm,
preferably 100, to protect the opamp from capacitive load.

True, and they are needed ( especially with some op-amp designs ).

Usually they are placed *inside* the feedback loop - so actually the source Z
is really very low.


Graham

mr c deckard wrote:

but, in any case, it's not too critical, i really want a simple LPF
and HPF without having to switch in 4 gain stages with the onboard eq.

Why do you fear the onboard eq?

it's a tascam m3700.

Hmmm... haven't met one but Tascam aren't exactly known for their consoles.

i think there are 4 gain stages per eq section.

4 op-amps ?

That suggets a 'state-variable' EQ implementation, which is one of the best
around !


i'll be honest here, tho. in the very non-critical tests i've done,
i can't repeatedly tell a difference between the eq being out and in
(set flat). i know this means i shouldn't worry about it,

Yup !

so i guess
it's more of a principle/housekeeping kinduva thing.

well, i suppose it would give me some options, too.

Nothing wrong with experimentation.


Graham

In article ,
Pooh Bear wrote:
Peter Larsen wrote:

snip

Opamp output stages have series resistors, usually no less than 50 Ohm,
preferably 100, to protect the opamp from capacitive load.

True, and they are needed ( especially with some op-amp designs ).

Usually they are placed *inside* the feedback loop - so actually the source Z
is really very low.

If you put them inside the feedback loop, your EQ is no longer passive. But
it's a lot more flexible.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

"Pooh Bear" wrote in message

Peter Larsen wrote:

snip

Opamp output stages have series resistors, usually no less than 50
Ohm, preferably 100, to protect the opamp from capacitive load.

True, and they are needed ( especially with some op-amp designs ).

The need for a building-out resistor (see below) is common enough that op
amps that DON'T require one are exceptional enough that this would be a
feature.

Usually they are placed *inside* the feedback loop - so actually the
source Z is really very low.

A output resistor inside the feedback loop is at best a current limiter. It
primarily relates to durability and clipping recovery. Generally it doesn't
need to be added because its function is already part of the op amp's
design.

A output resistor following the feedback loop take-off point is a
building-out or isolation resistor. It relates to stability, but also to
durability and usuability. I can't remember the last time I saw a line-level
output that lacked one.

i think there are 4 gain stages per eq section.

4 op-amps ?

That suggets a 'state-variable' EQ implementation, which is one of the best
around !

oh, if i'd be so lucky, but no, it's actually 3 gain stages, 4 if i
use the HPF, which i almost always do, on dual opamps -- njm4580D's.

the interesting thing, is that if i switch out the hpf and eq section,
the insert out is tied directly to the tape in, so i suppose the
output impedance then is 10k (assuming that's what the tape machine's
-10dbV out is).

when they're switched in, either one has a 100k resistor strapped to
the output to ground. the hpf runs as unity in inverting mode, the
last gain stage in the eq has a 10k bypassed by a 22pF cap. this
doesn't make the output impedance 100k for each stage -- it's
bypassed by the low internal impedance on the opamp, correct?

anyway, the insert return hits a non-inverting unity buffer, with a
470k strapped to ground (which i suppose i swamped by the internal
resistance), then a 10uF buffer cap before it hits the + in.



Just out of curiosity, why does passive EQ interest you so much ?

it's simple. i like that more out of asthetic than economy.

it's kinda like minimalism, but i suppose if i really took that to its
logical conclusion, i'd have a mono deck with one mic . . .

cheers,
chris deckard
saint louismo

mr c deckard wrote:

i think there are 4 gain stages per eq section.

4 op-amps ?

That suggets a 'state-variable' EQ implementation, which is one of the best
around !

oh, if i'd be so lucky, but no, it's actually 3 gain stages, 4 if i
use the HPF, which i almost always do, on dual opamps -- njm4580D's.

the interesting thing, is that if i switch out the hpf and eq section,
the insert out is tied directly to the tape in, so i suppose the
output impedance then is 10k (assuming that's what the tape machine's
-10dbV out is).

The output Z will still be low - op-amps are like that :-).


when they're switched in, either one has a 100k resistor strapped to
the output to ground. the hpf runs as unity in inverting mode, the
last gain stage in the eq has a 10k bypassed by a 22pF cap. this
doesn't make the output impedance 100k for each stage -- it's
bypassed by the low internal impedance on the opamp, correct?

I suppose that's one way of saying it. Op-amp output impedances are always low by
design.

anyway, the insert return hits a non-inverting unity buffer, with a
470k strapped to ground (which i suppose i swamped by the internal
resistance), then a 10uF buffer cap before it hits the + in.

There should be another R to ground directly connected to the op-amp + terminal.
Calculate the parallel combination of that value and the 470k and you get the
input impedance for that stage. Op-amp actual input impedances are *very* high,
so the impedance of the circuit is set by the external components.


Graham

The output Z will still be low - op-amps are like that :-).

heh. i suppose they are, unless they actually put a 10k in series on
the output . . .


anyway, the insert return hits a non-inverting unity buffer, with a
470k strapped to ground (which i suppose i swamped by the internal
resistance), then a 10uF buffer cap before it hits the + in.

There should be another R to ground directly connected to the op-amp + terminal.


no, there's a 47k on the (-) in to ground . . . incidently, i had the
idea of putting a switch to flip the + and - inputs for a polarity
reversal switch, since the board has none, making phase tests
troublesome.


Calculate the parallel combination of that value and the 470k and you get the
input impedance for that stage. Op-amp actual input impedances are *very* high,
so the impedance of the circuit is set by the external components.


cheers,
chris deckard
saint louismoe

The output Z will still be low - op-amps are like that :-).

heh. i suppose they are, unless they actually put a 10k in series on
the output . . .


anyway, the insert return hits a non-inverting unity buffer, with a
470k strapped to ground (which i suppose i swamped by the internal
resistance), then a 10uF buffer cap before it hits the + in.

There should be another R to ground directly connected to the op-amp + terminal.


no, there's a 47k on the (-) in to ground . . . incidently, i had the
idea of putting a switch to flip the + and - inputs for a polarity
reversal switch, since the board has none, making phase tests
troublesome.


Calculate the parallel combination of that value and the 470k and you get the
input impedance for that stage. Op-amp actual input impedances are *very* high,
so the impedance of the circuit is set by the external components.


cheers,
chris deckard
saint louismoe

Arny Krueger wrote:

"Pooh Bear" wrote in message

Peter Larsen wrote:

snip

Opamp output stages have series resistors, usually no less than 50
Ohm, preferably 100, to protect the opamp from capacitive load.

True, and they are needed ( especially with some op-amp designs ).

The need for a building-out resistor (see below) is common enough that op
amps that DON'T require one are exceptional enough that this would be a
feature.

I had in mind the Texas bifet series. When first introduced, their tendency to
oscillate when configured * esp as voltage followers * was not documented IIRC.
It's in the applications data now though.


Usually they are placed *inside* the feedback loop - so actually the
source Z is really very low.

A output resistor inside the feedback loop is at best a current limiter. It
primarily relates to durability and clipping recovery. Generally it doesn't
need to be added because its function is already part of the op amp's
design.

Well, I beg to differ but there is some extra detail. One of my standard
configurations uses 33R in series with the op-amp output with AF feedback taken
'after' it and RF feedback taken from the op-amp output. Works fine.

A output resistor following the feedback loop take-off point is a
building-out or isolation resistor. It relates to stability, but also to
durability and usuability. I can't remember the last time I saw a line-level
output that lacked one.

Neve have widely used an output inductor in parallel with 100 ? Ohms instead
after 5532 stages. Very low output Z at AF.


Graham

Pooh Bear wrote:

I had in mind the Texas bifet series. When first introduced, their tendency to
oscillate when configured * esp as voltage followers * was not documented IIRC.
It's in the applications data now though.

I should say that the Excaliburs are a textbook example of how not to design
a good audio op-amp.

And it's really depressing, since they are pretty much the only BiFET amps
available, now that Motorola has closed down their lines.
--scott


--
"C'est un Nagra. C'est suisse, et tres, tres precis."