Caps on 741 supply voltages

[[email protected]]

Hi there, simple question that must be in H&H, but I can't find it. I
have a PCB with about 18 LM741 op-amps on it. I supply +/-15V to V+ and
V- via traces on the board (~1mm width -- pretty big). Now, I've heard
that you should always have a cap. to ground right at V+ and V- because
the traces on the board carrying power have inductance, and you want to
be able to supply current to the op-amp fast. Is this right? I pretty
much swing the output of the op-amps from 0 to -10V at no faster than
~1kHz. If so, what is a good value? I assume uF-ish is fine.

Thanks!

Jesse Wodin

 

[Tim Wescott]

Hi there, simple question that must be in H&H, but I can't find it. I
have a PCB with about 18 LM741 op-amps on it. I supply +/-15V to V+ and
V- via traces on the board (~1mm width -- pretty big). Now, I've heard
that you should always have a cap. to ground right at V+ and V- because
the traces on the board carrying power have inductance, and you want to
be able to supply current to the op-amp fast. Is this right? I pretty
much swing the output of the op-amps from 0 to -10V at no faster than
~1kHz. If so, what is a good value? I assume uF-ish is fine.

Thanks!

Jesse Wodin

You want to prevent parasitic oscillations in the chips. I always drop
a 100nF cap from each supply to ground at each chip, and I usually try
to make sure there's a 10uF electrolytic somewhere on the board.

 

[[email protected]]

Excellent, thanks! What's the 10uF electrolytic for though? I've
always been taught to stay away from electrolytics, though I guess
that's for RF purposes only.

jesse

 

[Tim Wescott]

Excellent, thanks! What's the 10uF electrolytic for though? I've
always been taught to stay away from electrolytics, though I guess
that's for RF purposes only.

jesse

The little caps decouple each chip at high frequencies, the big caps
decouple the whole board at lower frequencies.

 

[[email protected]]

The little caps decouple each chip at high frequencies, the big caps
decouple the whole board at lower frequencies.

I've got a nasty suspicion that the big electrolytics have enough
equivaent series resistance to damp any high-Q resonances you might get
between the wiring inductances and the 100nF ceramic disks.

 

[Joseph2k]

Excellent, thanks! What's the 10uF electrolytic for though? I've
always been taught to stay away from electrolytics, though I guess
that's for RF purposes only.

jesse

The 100nF caps handle 200Hz and up pretty well, the 10uF is for lower
frequencies.

 

[Rich Grise]

Hi there, simple question that must be in H&H, but I can't find it. I have
a PCB with about 18 LM741 op-amps on it. I supply +/-15V to V+ and V- via
traces on the board (~1mm width -- pretty big). Now, I've heard that you
should always have a cap. to ground right at V+ and V- because the traces
on the board carrying power have inductance, and you want to be able to
supply current to the op-amp fast. Is this right? I pretty much swing the
output of the op-amps from 0 to -10V at no faster than ~1kHz. If so, what
is a good value? I assume uF-ish is fine.

When I was growing up, [ ;-) ], these were called bypass capacitors. I'm
anal about them - in this case, I'd put probably about a 1~10 uF aluminum
from each V to ground - you _do_ have a good low-resistance ground plane,
right?

There may be people who disagree with me, but for audio and maybe even
video/low band HF, and always, always, with digital, you can't
overcapacitate.

(something in me wants to put a smiley there, but the answer isn't
supposed to be flip - just its form. ;-) )

Cheers!
Rich

 

[Winfield Hill]

[email protected] wrote...

I've got a nasty suspicion that the big electrolytics have enough
equivaent series resistance to damp any high-Q resonances you might
get between the wiring inductances and the 100nF ceramic disks.

Might, indeed! Your answer is the correct one, Bill, because a sea
of 0.1 caps certainly makes a nice resonant circuit, somewhere in
the 5 to 20MHz region. A single crummy electrolytic is just what
the doctor ordered to kill that resonance. Better be crummy tho,
no low-esr tantalums need apply. Ahem, crummy, but not too crummy!

 

[mook Johnson]

We ARE talking about 741s here. Adequate for mid-fi audio but not high
performance by any stretch.

Hi there, simple question that must be in H&H, but I can't find it. I
have
a PCB with about 18 LM741 op-amps on it. I supply +/-15V to V+ and V- via
traces on the board (~1mm width -- pretty big). Now, I've heard that you
should always have a cap. to ground right at V+ and V- because the traces
on the board carrying power have inductance, and you want to be able to
supply current to the op-amp fast. Is this right? I pretty much swing the
output of the op-amps from 0 to -10V at no faster than ~1kHz. If so, what
is a good value? I assume uF-ish is fine.

When I was growing up, [ ;-) ], these were called bypass capacitors. I'm
anal about them - in this case, I'd put probably about a 1~10 uF aluminum
from each V to ground - you _do_ have a good low-resistance ground plane,
right?

There may be people who disagree with me, but for audio and maybe even
video/low band HF, and always, always, with digital, you can't
overcapacitate.

(something in me wants to put a smiley there, but the answer isn't
supposed to be flip - just its form. ;-) )

Cheers!
Rich

 

[Pooh Bear]

Hi there, simple question that must be in H&H, but I can't find it. I have
a PCB with about 18 LM741 op-amps on it. I supply +/-15V to V+ and V- via
traces on the board (~1mm width -- pretty big). Now, I've heard that you
should always have a cap. to ground right at V+ and V- because the traces
on the board carrying power have inductance, and you want to be able to
supply current to the op-amp fast. Is this right? I pretty much swing the
output of the op-amps from 0 to -10V at no faster than ~1kHz. If so, what
is a good value? I assume uF-ish is fine.

When I was growing up, [ ;-) ], these were called bypass capacitors. I'm
anal about them - in this case, I'd put probably about a 1~10 uF aluminum
from each V to ground - you _do_ have a good low-resistance ground plane,
right?

There may be people who disagree with me, but for audio and maybe even
video/low band HF, and always, always, with digital, you can't
overcapacitate.

Yes you can !

Lots of bulk capacitance to ground simply provides a path for any ripple. You
end up with a dirty ground.

Graham

 

[Pooh Bear]

The 100nF caps handle 200Hz and up pretty well, the 10uF is for lower
frequencies.

What do you mean by 'handle 200Hz and up'. Is the power supply no good
for that on its own ?

I think you have a fundamental misunderstanding of their purpose.

Graham

 

[[email protected]]

We ARE talking about 741s here. Adequate for mid-fi audio but not high
performance by any stretch.

The original 741 suffered from "pop-corn" noise, so-called because that
was what it sounded like in audio circuits. Back in 1970, one of my
then colleagues was using LM308s in a high-quality audio frequency
front end purely because the processing necessary to make the
super-beta transistors eliminated the pop-corn noise.

Modern 741's are supposed to be free from this vice, but the last time
I replace a 741 in a circuit, back in 1988, we did seem to get rid of a
lot of pop-corn noise, and the effect was quite noticeable - it was
controlling the induction heaters on a GaAs crsytal puller, and they
went from bang-bang operation to the sort of proportional control
they'd been designed to provide. The operaters were grateful and the
GaAs crystals seemed to have fewer defects. This was the machine that
made 95% of the single-crystal GaAs produced in the West (as it was
then).

 

[Winfield Hill]

[email protected] wrote...
Hey, even 741s can suffer from ringing on their supply bus lines.
They pass this right on to the next stage, where common input-stage
BJT nonlinear RF rectification turns this into distortion.

The original 741 suffered from "pop-corn" noise, so-called because that
was what it sounded like in audio circuits. Back in 1970, one of my
then colleagues was using LM308s in a high-quality audio frequency
front end purely because the processing necessary to make the
super-beta transistors eliminated the pop-corn noise.

Modern 741's are supposed to be free from this vice, but the last time
I replace a 741 in a circuit, back in 1988, we did seem to get rid of
a lot of pop-corn noise, and the effect was quite noticeable - it was
controlling the induction heaters on a GaAs crsytal puller, and they
went from bang-bang operation to the sort of proportional control
they'd been designed to provide. The operaters were grateful and the
GaAs crystals seemed to have fewer defects. This was the machine that
made 95% of the single-crystal GaAs produced in the West (as it was
then).

Bill, what did you replace the 741s with in 1988?

 

[[email protected]]

[email protected] wrote...

Bill, what did you replace the 741s with in 1988?

OP-01. I subsequently did a fair bit to the front end electronics - we
weighed the crystal as it grew with an LVDT on a spring, and the LVDT
monitoring circuit got somewhat fancier under my care - but that was
the sinble most effective improvement.

 

[Pooh Bear]

OP-01. I subsequently did a fair bit to the front end electronics - we
weighed the crystal as it grew with an LVDT on a spring, and the LVDT
monitoring circuit got somewhat fancier under my care - but that was
the sinble most effective improvement.

That's an interesting case.

1/f noise may actually have been more problematic than 'poporn' in your
instance though.

Graham

 

[[email protected]]

That's an interesting case.

1/f noise may actually have been more problematic than 'popcorn' in your
instance though.

The 741 hasn't got any spec for either, and neither does the OP-01.
IIRR the OP-27 and OP-37 were amongst the first that did specify !/f
noise - the LT1001 certainly does (0.3uV ptp typical 0.6uV max for
0.1Hz to 10Hz) and I'd have gone for something with a specification if
1/f noise had been a real problem.

The LVDT was excited at a few kHz, and I amplified the output as much
as I could before demodulation, so the OP-01 was more of a buffer than
an amplifier by the time I'd finished.

I'd have loved to have used a proper AC-bridge on the LVDT output, but
there wasn't enough space for much electronics directly above the LVDT,
and there was a great deal of expensive mechanical engineering in the
way of getting any more space.

 

[Pooh Bear]

No, it is not, that many trace inches away. That is why they are adjacent
to the parts and not clustered elsewhere.

You *do* have a fundamental misunderstanding.

Graham

 

[Joseph2k]

What do you mean by 'handle 200Hz and up'. Is the power supply no good
for that on its own ?

I think you have a fundamental misunderstanding of their purpose.

Graham

No, it is not, that many trace inches away. That is why they are adjacent
to the parts and not clustered elsewhere.