Hi Friends,
I'm using this simple line mixer with an analog 8-trk, taking a sort
of a "straight wire" approach. The mixer sounds worlds better after I
swapped out the 4741 op amps for BB OPA4134s. But now I'm looking at
the signal coupling caps. There are five 22uf NPs in the signal path
from input to output: 1 pre-input fader, 1 post-input fader, 1 between
input opamp and summing amp, 1 post-summing amp, and 1 post-output amp.
That seems to me three too many. I'm thinking of upping the input and
output caps to around 100uf and jumpering out the three in between.
Would this lead to problems? Would the improvement in sound be worth
the effort? Thanks for any help.

Regards,
Joe Kramer

Joe Kramer wrote:
I'm using this simple line mixer with an analog 8-trk, taking a sort
of a "straight wire" approach. The mixer sounds worlds better after I
swapped out the 4741 op amps for BB OPA4134s. But now I'm looking at
the signal coupling caps. There are five 22uf NPs in the signal path
from input to output: 1 pre-input fader, 1 post-input fader, 1 between
input opamp and summing amp, 1 post-summing amp, and 1 post-output amp.
That seems to me three too many. I'm thinking of upping the input and
output caps to around 100uf and jumpering out the three in between.
Would this lead to problems? Would the improvement in sound be worth
the effort? Thanks for any help.

Do you need any input caps at all?

Can you replace the output caps with, say, 20 uF film capacitors?

Can you eliminate any of the internal caps? Measure the DC offset across
each one; if it's small enough and you're going into another gain stage
(and not a fader), try replacing it with a jumper.
--scott

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

Dear Scott,
Thanks for your reply.

Scott Dorsey wrote:

Do you need any input caps at all?

I guess I don't, assuming anything else I use the mixer with is AC
coupled--I can't imagine what wouldn't be these days. The inputs are
normalled to the 8trk and brought out to the front via patchbay should I
choose to FX individual channels or use it to mix whatever else.

Can you replace the output caps with, say, 20 uF film capacitors?

I would if I could, but sadly, no room. The SM82 is eight stereo channels
in, FX send/receive, two out, all in a tightly packed one-space rack. But I
might be able to bypass the electrolytics with .1 uF polypropylenes.

Can you eliminate any of the internal caps? Measure the DC offset across
each one; if it's small enough and you're going into another gain stage
(and not a fader), try replacing it with a jumper.

--scott

With the FET op amps and the fact that it's a bipolar power supply,
shouldn't the DC offset be negligable? The signal does go through "balance"
slider pots, basically panpots, in two places where the caps are. Will I get
scratchiness there if I take out the caps? I guess I should just try doing a
channel and see what happens.

Thanks again,
Joe Kramer

Joe Kramer wrote:

With the FET op amps and the fact that it's a bipolar power supply,
shouldn't the DC offset be negligable?

Dunno, what do the numbers on the data sheets say? Some op-amps are a lot
better than others.

If you are using FET op-amps, you can get by with much smaller input
blocking caps in places where you need them, since the input Z is so
high.

The signal does go through "balance"
slider pots, basically panpots, in two places where the caps are. Will I get
scratchiness there if I take out the caps? I guess I should just try doing a
channel and see what happens.

You should probably put caps on the input of the sliders, even if the offset
is pretty low, since it doesn't take much to clobber pots. But I bet a lot
of others can be eliminated.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

If you make sure there's no offset on the output amps, you can get rid
of the output caps (C41, 43, 45, and 48). Any offset coming into the
output level" amp is removed by C40 and C44, so as long as that amp and
the output amps aren't adding any significant offset, you should be
fine. I wonder if you couldn't get even better DC performance on these
output stages by changing a couple resistances here and there to
balance the input bias currents, but we're talking about less than 1mV
in the case of a good FET amplifier anyway.

The inputs are less straightforward because there's an unbuffered fader
right at the input. I would probably keep the caps around the fader
(C23, C25, C28, C30, for example) but then try to get rid of C27, C31,
C33, and C42. Here's another spot where you might want to look at the
bias currents and the voltage they drop across, for example, R17 and
R18.

Oh, but now that I look at it, you probably need to keep C27 ans C31 in
there unless you can make sure there's really no offset going to hit
those balance pots.

I learned the hard way that you want to keep those caps around the
potentiometers. Pots really don't like DC on them, and op amps don't
like having their DC change suddenly.

This thing is only a couple of years old, but maybe if you have a look
you'll find that you can replace these bipolar caps with new parts that
will be more advanced, or at least fit more capacitance into the space
alotted. I don't think you really need 25V parts in most of these
spots, either. I bet you can get 100uF Panasonic SU caps in there.
That may seem like overkill, but if you have too many HP filters in a
row like this, you want to give them a really low corner frequency.

While you're at it, replace all the power supply electrolytics with the
biggest Panasonic NHGs (pre-regulator) or FCs (post-regulator) you can
fit in there.

I don't have one of these, I'm just looking at the schematic from
Rane's website. I can't tell what the pin spacing is on the caps that
are in there, so I don't know exactly how much you'll be able to do.

ulysses

In article , Joe Kramer
wrote:

Hi Friends,
I'm using this simple line mixer with an analog 8-trk, taking a sort
of a "straight wire" approach. The mixer sounds worlds better after I
swapped out the 4741 op amps for BB OPA4134s. But now I'm looking at
the signal coupling caps. There are five 22uf NPs in the signal path
from input to output: 1 pre-input fader, 1 post-input fader, 1 between
input opamp and summing amp, 1 post-summing amp, and 1 post-output amp.
That seems to me three too many. I'm thinking of upping the input and
output caps to around 100uf and jumpering out the three in between.
Would this lead to problems? Would the improvement in sound be worth
the effort? Thanks for any help.

Regards,
Joe Kramer

Dear J U M,
Thanks for your reply. (more below . . .)

Justin Ulysses Morse wrote:

If you make sure there's no offset on the output amps, you can get rid
of the output caps (C41, 43, 45, and 48). Any offset coming into the
output level" amp is removed by C40 and C44, so as long as that amp and
the output amps aren't adding any significant offset, you should be
fine. I wonder if you couldn't get even better DC performance on these
output stages by changing a couple resistances here and there to
balance the input bias currents, but we're talking about less than 1mV
in the case of a good FET amplifier anyway.

The BB OPA4134s indicate .5mV input offset, but I suppose I would have to
measure output offset for anything askew in the power supply.

The inputs are less straightforward because there's an unbuffered fader
right at the input. I would probably keep the caps around the fader
(C23, C25, C28, C30, for example) but then try to get rid of C27, C31,
C33, and C42. Here's another spot where you might want to look at the
bias currents and the voltage they drop across, for example, R17 and
R18.

The part numbers you're citing here are actually for the FX Receive inputs,
but the normal inputs have essentially the same configuration. You're saying
get rid of the shunt caps in the feedback loop, right? I still don't quite
get why there has to be caps on both side of the input fader though.
Shouldn't one after the fader take care of any problems?

Oh, but now that I look at it, you probably need to keep C27 ans C31 in
there unless you can make sure there's really no offset going to hit
those balance pots.

I learned the hard way that you want to keep those caps around the
potentiometers. Pots really don't like DC on them, and op amps don't
like having their DC change suddenly.

What happens to the pots? I have built several simple mixers with unbuffered
pots at the input, followed by caps at the op amp input. Nothing has ever
gone bad. What should I beware of?

This thing is only a couple of years old, but maybe if you have a look
you'll find that you can replace these bipolar caps with new parts that
will be more advanced, or at least fit more capacitance into the space
alotted. I don't think you really need 25V parts in most of these
spots, either. I bet you can get 100uF Panasonic SU caps in there.
That may seem like overkill, but if you have too many HP filters in a
row like this, you want to give them a really low corner frequency.

While you're at it, replace all the power supply electrolytics with the
biggest Panasonic NHGs (pre-regulator) or FCs (post-regulator) you can
fit in there.

I'll definitely go with the Panasonics. I usually just use FCs all around,
and I know these will fit size-wise, but are the SUs better for signal path?
Thanks again for all you help.

Regards,
Joe Kramer

Joe Kramer wrote:

The BB OPA4134s indicate .5mV input offset, but I suppose I would have to
measure output offset for anything askew in the power supply.

Another source of offset, which is greatly diminished in FET amps, but
still important to consider when dicking around with an existing
design, is the input bias current. There's a small amount of current
flowing through the inverting and non-inverting inputs of the
amplifier. In the case of a FET device, it's vanishingly small. But
this current will cause a voltage drop across the resistance seen
between that pin and ground. Now, suppose your non-inverting input is
grounded while your inverting input has a large resistance to ground.
The resulting voltage at each of these inputs would therefore try not
to be equal. This is effectively a DC input signal, and it will be
amplifief by the gain of the stage, and by any subsequent DC-coupled
stages.

The part numbers you're citing here are actually for the FX Receive inputs,
but the normal inputs have essentially the same configuration.

Right. I just used those as an example because they're all the same,
according to the schematic.

You're saying get rid of the shunt caps in the feedback loop, right? I still don't quite
get why there has to be caps on both side of the input fader though.
Shouldn't one after the fader take care of any problems?

Okay, let's look at C25 and C31 together. Imagine for a moment that
you get rid of C25. Now the DC resistance to ground seen by the
non-inverting input of the amplifier will be that 100k resistor (R17)
in parallel with the input pot, R16. So it will vary between 0 ohms
and 50k ohms, depending on the setting of the pot. Meanwhile, the
inverting input sees an open DC circuit to ground, because of C31. So
you've got a fader-dependent input offset, and you've got a varying
offset on the pot. Both are bad news. By keeping C25 there, you can
prevent both problems. You've still got 100k to ground on the
non-inverting input and nothing on the inverting input though, so you
could improve things (IF you use an amp with good DC performance) by
shorting C31, and making the DC resistance to ground of each input
equal. This reduces the DC offset at the output to whatever error is
inherent in the op amp, multiplied by the gain. Since this is a unity
gain stage, it will remain very very small.

What happens to the pots? I have built several simple mixers with unbuffered
pots at the input, followed by caps at the op amp input. Nothing has ever
gone bad. What should I beware of?

If there's even a little DC on the pot, it will be noisy. As I
described above, the resistance of the pot combined with the bias
current in the amplifier will cause a DC voltage. Furthermore, as I
described above, changing the position of the pot will change that
voltage (ohm's law) and thereby shift the DC operating point of the
op-amp, which can (and often does, in my experience) cause noise,
oscillation, and eratic behavior in the op-amp.

If you can be sure to get rid of any offset coming out of the previous
stage, you can omit the cap feeding the input of the fader. But I
think even the best FET-input amps will cause problems when fed by a
DC-coupled potentiometer.

I'll definitely go with the Panasonics. I usually just use FCs all around,
and I know these will fit size-wise, but are the SUs better for signal path?
Thanks again for all you help.

No, I would expect the FCs to be better caps; but they need to have a
DC bias on them that's greater than the signal swing. In the case of
low-offset circuits like these where the signal is around a volt and
the offset is a couple of millivolts, polarized electrolytic caps will
distort the signal. The solution is either design the circuit so the
polarized caps are DC biased, or else use non-polarized caps such as
film capacitors or Panasonic SU's.


ulysses

Dear Ulysses,
Thanks! This discussion has really helped me a lot--I'm printing it out for my files! I'm
going to try doing a channel or two to see how it sounds and then maybe I'll post the results
for anyone interested. I appreciate your help and your knowledge.

Regards,
Joe Kramer

Justin Ulysses Morse wrote:

Joe Kramer wrote:

The BB OPA4134s indicate .5mV input offset, but I suppose I would have to
measure output offset for anything askew in the power supply.

Another source of offset, which is greatly diminished in FET amps, but
still important to consider when dicking around with an existing
design, is the input bias current. There's a small amount of current
flowing through the inverting and non-inverting inputs of the
amplifier. In the case of a FET device, it's vanishingly small. But
this current will cause a voltage drop across the resistance seen
between that pin and ground. Now, suppose your non-inverting input is
grounded while your inverting input has a large resistance to ground.
The resulting voltage at each of these inputs would therefore try not
to be equal. This is effectively a DC input signal, and it will be
amplifief by the gain of the stage, and by any subsequent DC-coupled
stages.

The part numbers you're citing here are actually for the FX Receive inputs,
but the normal inputs have essentially the same configuration.

Right. I just used those as an example because they're all the same,
according to the schematic.

You're saying get rid of the shunt caps in the feedback loop, right? I still don't quite
get why there has to be caps on both side of the input fader though.
Shouldn't one after the fader take care of any problems?

Okay, let's look at C25 and C31 together. Imagine for a moment that
you get rid of C25. Now the DC resistance to ground seen by the
non-inverting input of the amplifier will be that 100k resistor (R17)
in parallel with the input pot, R16. So it will vary between 0 ohms
and 50k ohms, depending on the setting of the pot. Meanwhile, the
inverting input sees an open DC circuit to ground, because of C31. So
you've got a fader-dependent input offset, and you've got a varying
offset on the pot. Both are bad news. By keeping C25 there, you can
prevent both problems. You've still got 100k to ground on the
non-inverting input and nothing on the inverting input though, so you
could improve things (IF you use an amp with good DC performance) by
shorting C31, and making the DC resistance to ground of each input
equal. This reduces the DC offset at the output to whatever error is
inherent in the op amp, multiplied by the gain. Since this is a unity
gain stage, it will remain very very small.

What happens to the pots? I have built several simple mixers with unbuffered
pots at the input, followed by caps at the op amp input. Nothing has ever
gone bad. What should I beware of?

If there's even a little DC on the pot, it will be noisy. As I
described above, the resistance of the pot combined with the bias
current in the amplifier will cause a DC voltage. Furthermore, as I
described above, changing the position of the pot will change that
voltage (ohm's law) and thereby shift the DC operating point of the
op-amp, which can (and often does, in my experience) cause noise,
oscillation, and eratic behavior in the op-amp.

If you can be sure to get rid of any offset coming out of the previous
stage, you can omit the cap feeding the input of the fader. But I
think even the best FET-input amps will cause problems when fed by a
DC-coupled potentiometer.

I'll definitely go with the Panasonics. I usually just use FCs all around,
and I know these will fit size-wise, but are the SUs better for signal path?
Thanks again for all you help.

No, I would expect the FCs to be better caps; but they need to have a
DC bias on them that's greater than the signal swing. In the case of
low-offset circuits like these where the signal is around a volt and
the offset is a couple of millivolts, polarized electrolytic caps will
distort the signal. The solution is either design the circuit so the
polarized caps are DC biased, or else use non-polarized caps such as
film capacitors or Panasonic SU's.

ulysses