Op-amp mystery

[Liz Tuddenham]

It's always the stupid things that cause the most trouble:

I was working on a fairly straightforward experimental audio circuit
using TL074 op-amps laid out on Veroboard, which is prone to inter-track
shorts unless you are very careful with your soldering. The gain of a
virtual-earth stage was too high, so I altered some resistor values to
cut it down.

As this was a stereo system, I experimented with one channel first and
then switched off and replaced the equivalent resistors in the second
channel (one of the other op-amps in the same chip).

The input signal was a 6 Kc/s sinwave at about -10 dBu. After the
modification, the volume control, which was in a feedback loop, didnt
seem to make much difference. Then I realised that applying the signal
to the left channel gave a distorted version of it in the right channel
and vice-versa. Obviously I had shorted two tracks somewhere - but
where?

Things got even worse when I realised that the input signal was
appearing on the virtual earth input of each op-amp, this definitely
can't happen! I decided it was time to switch off the power supply and
go and ponder the problem over a cup of tea.

That was the point where I discovered I hadn't switched the power supply
on in the first place.

[Don Y]

Post by Liz Tuddenham
I was working on a fairly straightforward experimental audio circuit
using TL074 op-amps laid out on Veroboard, which is prone to inter-track
shorts unless you are very careful with your soldering. The gain of a
virtual-earth stage was too high, so I altered some resistor values to
cut it down.
As this was a stereo system, I experimented with one channel first and
then switched off and replaced the equivalent resistors in the second
channel (one of the other op-amps in the same chip).
The input signal was a 6 Kc/s sinwave at about -10 dBu. After the
modification, the volume control, which was in a feedback loop, didnt
seem to make much difference. Then I realised that applying the signal
to the left channel gave a distorted version of it in the right channel
and vice-versa. Obviously I had shorted two tracks somewhere - but
where?
Things got even worse when I realised that the input signal was
appearing on the virtual earth input of each op-amp, this definitely
can't happen! I decided it was time to switch off the power supply and
go and ponder the problem over a cup of tea.
That was the point where I discovered I hadn't switched the power supply
on in the first place.

Too funny!

Reminds me of a remote display/control device I made for one
of our products in the 70's. All designed with CMOS and other
low power technologies as it had to sit at the end of a long
cable (e.g., for use on a flying bridge).

I got the prototype board assembled and debugged properly.
Then, discovered the power switch (an alternate acting pushbutton)
had been OFF the whole time! The circuit was able to run off
power derived from the data signals.

Hmmm... now, how do I make it LOOK like it's "off" vs. "on"?

[john larkin]

Post by Liz Tuddenham
I was working on a fairly straightforward experimental audio circuit
using TL074 op-amps laid out on Veroboard, which is prone to inter-track
shorts unless you are very careful with your soldering. The gain of a
virtual-earth stage was too high, so I altered some resistor values to
cut it down.

Show us a pic of your breadboard!

Post by Liz Tuddenham
As this was a stereo system, I experimented with one channel first and
then switched off and replaced the equivalent resistors in the second
channel (one of the other op-amps in the same chip).
The input signal was a 6 Kc/s sinwave at about -10 dBu. After the
modification, the volume control, which was in a feedback loop, didnt
seem to make much difference. Then I realised that applying the signal
to the left channel gave a distorted version of it in the right channel
and vice-versa. Obviously I had shorted two tracks somewhere - but
where?
Things got even worse when I realised that the input signal was
appearing on the virtual earth input of each op-amp, this definitely
can't happen! I decided it was time to switch off the power supply and
go and ponder the problem over a cup of tea.
That was the point where I discovered I hadn't switched the power supply
on in the first place.

Sometimes people note that a CMOS circuit works properly with no power
supply, because some ESD diodes are rectifying a logic input.

I never use those dreadful Veroboard or plastic block plugin things.

Try this:

Loading Image...

Loading Image...

Get artistic!

[Jan Panteltje]

On a sunny day (Tue, 10 Sep 2024 07:47:46 -0700) it happened john larkin

Post by john larkin

Post by Liz Tuddenham
I was working on a fairly straightforward experimental audio circuit
using TL074 op-amps laid out on Veroboard, which is prone to inter-track
shorts unless you are very careful with your soldering. The gain of a
virtual-earth stage was too high, so I altered some resistor values to
cut it down.

Show us a pic of your breadboard!

Post by Liz Tuddenham
As this was a stereo system, I experimented with one channel first and
then switched off and replaced the equivalent resistors in the second
channel (one of the other op-amps in the same chip).
The input signal was a 6 Kc/s sinwave at about -10 dBu. After the
modification, the volume control, which was in a feedback loop, didnt
seem to make much difference. Then I realised that applying the signal
to the left channel gave a distorted version of it in the right channel
and vice-versa. Obviously I had shorted two tracks somewhere - but
where?
Things got even worse when I realised that the input signal was
appearing on the virtual earth input of each op-amp, this definitely
can't happen! I decided it was time to switch off the power supply and
go and ponder the problem over a cup of tea.
That was the point where I discovered I hadn't switched the power supply
on in the first place.

Sometimes people note that a CMOS circuit works properly with no power
supply, because some ESD diodes are rectifying a logic input.
I never use those dreadful Veroboard or plastic block plugin things.
https://www.dropbox.com/scl/fi/f3redjl7umja5v8tsi2p5/Z382_1.JPG?rlkey=epclqs4lkx2cmdakne390bbae&raw=1
https://www.dropbox.com/scl/fi/ydvcds95zvzjq56bzeimr/Z412_Proto.JPG?rlkey=hyejukxbbnk3573engf0if4zt&raw=1
Get artistic!

Veroboard is nice, all .1 inch specing, easy soldering, easy mounting connectors, no drilling required:
Loading Image...
Loading Image...
that one is about 1.5 GHz out..
etc etc...

Loading Image...

[Lasse Langwadt]

Post by john larkin

Post by Liz Tuddenham
I was working on a fairly straightforward experimental audio circuit
using TL074 op-amps laid out on Veroboard, which is prone to inter-track
shorts unless you are very careful with your soldering. The gain of a
virtual-earth stage was too high, so I altered some resistor values to
cut it down.

Show us a pic of your breadboard!

Post by Liz Tuddenham
As this was a stereo system, I experimented with one channel first and
then switched off and replaced the equivalent resistors in the second
channel (one of the other op-amps in the same chip).
The input signal was a 6 Kc/s sinwave at about -10 dBu. After the
modification, the volume control, which was in a feedback loop, didnt
seem to make much difference. Then I realised that applying the signal
to the left channel gave a distorted version of it in the right channel
and vice-versa. Obviously I had shorted two tracks somewhere - but
where?
Things got even worse when I realised that the input signal was
appearing on the virtual earth input of each op-amp, this definitely
can't happen! I decided it was time to switch off the power supply and
go and ponder the problem over a cup of tea.
That was the point where I discovered I hadn't switched the power supply
on in the first place.

Sometimes people note that a CMOS circuit works properly with no power
supply, because some ESD diodes are rectifying a logic input.

until all the inputs are low, then it only works until the bypass cap is
discharged ;)

seem to remember seeing a boost converter made from just a cmos gate and
an inductor, using the output as input and the supply pins as output via
the ESD diodes, but I can't find it again, but I found this

https://www.edn.com/efficient-voltage-doubler-is-made-from-generic-cmos-inverters/

[Don Y]

Post by Lasse Langwadt
until all the inputs are low, then it only works until the bypass cap is
discharged ;)

Never a problem when your Vcc looks like a 2 ohm load and you've only got
a handful of TTL I/Os to/from your "source"!

Problem is when you're given an interface that was *designed* to
connect one board to another INSIDE an instrument and then told to
extend it a couple of feet OUTSIDE the instrument (so the device
can be hidden to reduce clutter on the bridge).

THEN, told to make that *30* ft and denied a power supply at
the remote end of the cable!

"We won't be able to use PGDs because we can't deliver that
much power FROM THE INSTRUMENT. And, all of the control logic
draws way too much current for the functionality it provides.
And, drop-in CMOS counterparts aren't available. And, there's
way too much C in the cable for those TTL signals. And you've
already committed to the aluminum castings for the case. And..."

[The idea/feasability of using an MCU for such a function, back
then, was just plain silly!]

That's what happens when a company lets "sales/marketing" drive
their product development, unconditionally. You need "adults"
that can make them understand the consequences of their fantasies!

OTOH, the salesman who boasted he could sell hundreds of these
"if you could extend the cable to 30 ft" was never heard from,
again!

[Valuable lesson learned. There and at all subsequent employers
and clients, I did my own market research USING THEIR SALES RECORDS
to argue against the inevitable "kitchen sink" requirements lists
that invariably came along. "Yeah, I know you, as a salesman,
don't want to lose ANY sale because of some missing capability.
BUT, you have to realize that these capabilities come with costs;
figure out the ESSENTIAL features and then learn how to spin those
to your customers!"]

So, I turned lemons into lemonade: left the hardware run continuously
(eliminating the start-up delay necessary to "acquire signal") and
just turned the power switch into a "display blank" and "controls
inhibit" function.

But, the prototype board -- instead of being ready for production
for the previous design requirements -- had to be completely
redone (we had a small etch tank so we could print 2 layer protos
in-house so this was just an annoyance).

Post by Lasse Langwadt
seem to remember seeing a boost converter made from just a cmos gate and an
inductor, using the output as input and the supply pins as output via the ESD
diodes, but I can't find it again, but I found this
https://www.edn.com/efficient-voltage-doubler-is-made-from-generic-cmos-inverters/

[john larkin]

Post by Lasse Langwadt

Post by john larkin

Post by Liz Tuddenham
I was working on a fairly straightforward experimental audio circuit
using TL074 op-amps laid out on Veroboard, which is prone to inter-track
shorts unless you are very careful with your soldering. The gain of a
virtual-earth stage was too high, so I altered some resistor values to
cut it down.

Show us a pic of your breadboard!

Post by Liz Tuddenham
As this was a stereo system, I experimented with one channel first and
then switched off and replaced the equivalent resistors in the second
channel (one of the other op-amps in the same chip).
The input signal was a 6 Kc/s sinwave at about -10 dBu. After the
modification, the volume control, which was in a feedback loop, didnt
seem to make much difference. Then I realised that applying the signal
to the left channel gave a distorted version of it in the right channel
and vice-versa. Obviously I had shorted two tracks somewhere - but
where?
Things got even worse when I realised that the input signal was
appearing on the virtual earth input of each op-amp, this definitely
can't happen! I decided it was time to switch off the power supply and
go and ponder the problem over a cup of tea.
That was the point where I discovered I hadn't switched the power supply
on in the first place.

Sometimes people note that a CMOS circuit works properly with no power
supply, because some ESD diodes are rectifying a logic input.

until all the inputs are low, then it only works until the bypass cap is
discharged ;)
seem to remember seeing a boost converter made from just a cmos gate and
an inductor, using the output as input and the supply pins as output via
the ESD diodes, but I can't find it again, but I found this
https://www.edn.com/efficient-voltage-doubler-is-made-from-generic-cmos-inverters/

This will drive a blue or whatever LED from a low voltage supply.

Loading Image...

[piglet]

Post by Lasse Langwadt

Post by john larkin

Post by Liz Tuddenham
I was working on a fairly straightforward experimental audio circuit
using TL074 op-amps laid out on Veroboard, which is prone to inter-track
shorts unless you are very careful with your soldering.   The gain of a
virtual-earth stage was too high, so I altered some resistor values to
cut it down.

Show us a pic of your breadboard!

Post by Liz Tuddenham
As this was a stereo system, I experimented with one channel first and
then switched off and replaced the equivalent resistors in the second
channel (one of the other op-amps in the same chip).
The input signal was a 6 Kc/s sinwave at about -10 dBu.  After the
modification, the volume control, which was in a feedback loop, didnt
seem to make much difference.  Then I realised that applying the signal
to the left channel gave a distorted version of it in the right channel
and vice-versa.  Obviously I had shorted two tracks somewhere - but
where?
Things got even worse when I realised that the input signal was
appearing on the virtual earth input of each op-amp, this definitely
can't happen!  I decided it was time to switch off the power supply and
go and ponder the problem over a cup of tea.
That was the point where I discovered I hadn't switched the power supply
on in the first place.

Sometimes people note that a CMOS circuit works properly with no power
supply, because some ESD diodes are rectifying a logic input.

until all the inputs are low, then it only works until the bypass cap is
discharged ;)
seem to remember seeing a boost converter made from just a cmos gate and
an inductor, using the output as input and the supply pins as output via
the ESD diodes, but I can't find it again, but I found this
https://www.edn.com/efficient-voltage-doubler-is-made-from-generic-cmos-inverters/

It is possible you were remembering my posting this in April 2016:

<https://www.dropbox.com/scl/fi/8n4fz5o0r79rkmbn05j1a/micropowerBoost.pdf?rlkey=v1w9vm0his1pzd3lwd93dsm3v&st=5x8irpk6&raw=1>

All power flows through the inductor. The ESD diodes are only involved
during startup, once oscillator running the output stage n-fet is the
boost switch and the p-fet is the synchronous rectifier. A version built
using 74HC132 was in a successful product which needed micropower 5V
from 2-3V input.

piglet

[Jan Panteltje]

On a sunny day (Thu, 12 Sep 2024 08:27:36 +0100) it happened piglet

Post by piglet

Post by Lasse Langwadt

Post by john larkin

Post by Liz Tuddenham
I was working on a fairly straightforward experimental audio circuit
using TL074 op-amps laid out on Veroboard, which is prone to inter-track
shorts unless you are very careful with your soldering.   The gain of a
virtual-earth stage was too high, so I altered some resistor values to
cut it down.

Show us a pic of your breadboard!

Post by Liz Tuddenham
As this was a stereo system, I experimented with one channel first and
then switched off and replaced the equivalent resistors in the second
channel (one of the other op-amps in the same chip).
The input signal was a 6 Kc/s sinwave at about -10 dBu.  After the
modification, the volume control, which was in a feedback loop, didnt
seem to make much difference.  Then I realised that applying the signal
to the left channel gave a distorted version of it in the right channel
and vice-versa.  Obviously I had shorted two tracks somewhere - but
where?
Things got even worse when I realised that the input signal was
appearing on the virtual earth input of each op-amp, this definitely
can't happen!  I decided it was time to switch off the power supply and
go and ponder the problem over a cup of tea.
That was the point where I discovered I hadn't switched the power supply
on in the first place.

Sometimes people note that a CMOS circuit works properly with no power
supply, because some ESD diodes are rectifying a logic input.

until all the inputs are low, then it only works until the bypass cap is
discharged ;)
seem to remember seeing a boost converter made from just a cmos gate and
an inductor, using the output as input and the supply pins as output via
the ESD diodes, but I can't find it again, but I found this
https://www.edn.com/efficient-voltage-doubler-is-made-from-generic-cmos-inverters/

<https://www.dropbox.com/scl/fi/8n4fz5o0r79rkmbn05j1a/micropowerBoost.pdf?rlkey=v1w9vm0his1pzd3lwd93dsm3v&st=5x8irpk6&raw=1>
All power flows through the inductor. The ESD diodes are only involved
during startup, once oscillator running the output stage n-fet is the
boost switch and the p-fet is the synchronous rectifier. A version built
using 74HC132 was in a successful product which needed micropower 5V
from 2-3V input.
piglet

Interesting!

Classic voltage doubler is the Villard circuit:
https://josepheoff.github.io/posts/diode-capacitors-volts-pt1

And if you need more low power voltage:
Loading Image...
Loading Image...
Loading Image...
has been working for decades now...
also all battery powered
Loading Image...
Loading Image...

[Liz Tuddenham]

Post by john larkin

Post by Liz Tuddenham
I was working on a fairly straightforward experimental audio circuit
using TL074 op-amps laid out on Veroboard, which is prone to inter-track
shorts unless you are very careful with your soldering. The gain of a
virtual-earth stage was too high, so I altered some resistor values to
cut it down.

Show us a pic of your breadboard!

I've posted circuit diagrams and a picture at:

http://www.poppyrecords.co.uk/other/Turntables/parallel-tracker.htm

It was the headphone amplifier that caused the mystery. So simple, yet
so confusing when you aren't expecting simple things to go wrong.

[Liz Tuddenham]

Post by Liz Tuddenham

Post by john larkin

Post by Liz Tuddenham
I was working on a fairly straightforward experimental audio circuit
using TL074 op-amps laid out on Veroboard, which is prone to inter-track
shorts unless you are very careful with your soldering. The gain of a
virtual-earth stage was too high, so I altered some resistor values to
cut it down.

Show us a pic of your breadboard!

http://www.poppyrecords.co.uk/other/Turntables/parallel-tracker.htm
It was the headphone amplifier that caused the mystery. So simple, yet
so confusing when you aren't expecting simple things to go wrong.

Thanks, interesting headphones volume setting.

I was originally just going to copy the circuit of the headphone
amplifier in a Marantz CD player (why bother to re-invent the wheel?),
then I spotted that they used a 100-ohm resistor in series with the
headphone socket. This would remove the damping, so that headphones
with any unwanted resonances wouldn't give as good sound quality as they
would have been capable of if they had been fed from a low impedance.

To keep the output impedance low, I took the feedback from the output
terminals.and put the 100-ohm resistors in the collector circuit, where
they would still limit the output current to a safe value that wouldn't
destroy the headphones.

The volume control circuit was derived from a Doug Self design in which
he used a unity-gain buffer to minimise the loading on the pot wiper. I
had a bagfull of 5k twin-gang pots, so I decided that I could omit the
buffer as long as the feedback resistors around the op-amp were high
compared with the pot itself. The pot is linear but the scale is
approximately in dB, which works a lot better than trying to make two
logarithmic pots track accurately together (or buying some exotic and
expensive controls).

Originally I was aiming for similar gain characteristics to the line
driver amplifier (pot half-way up = 0dBm output) but I found that was
far too much gain and the volume control never needed to go beyond a
quarter of a turn. The input signals going into all the output stages
are around -10dBu with a cartridge giving around -45dBu on an 'average'
record; this allows 30dB headroom before the volume control, which is
necessary to avoid clipping the crackles from a damaged 78.

The circuit I finished up with actually generates a loss, rather than a
gain, but that gives a more favourable pot setting for comfortable
listening. Despite not workoing quite as Doug Self intended, it
manages to produce a smooth transition across the middle of the scale,
without 'crowding' near the top ot the bottom.

[piglet]

Post by Liz Tuddenham

Post by john larkin

Post by Liz Tuddenham
I was working on a fairly straightforward experimental audio circuit
using TL074 op-amps laid out on Veroboard, which is prone to inter-track
shorts unless you are very careful with your soldering. The gain of a
virtual-earth stage was too high, so I altered some resistor values to
cut it down.

Show us a pic of your breadboard!

http://www.poppyrecords.co.uk/other/Turntables/parallel-tracker.htm
It was the headphone amplifier that caused the mystery. So simple, yet
so confusing when you aren't expecting simple things to go wrong.

Thanks, interesting headphones volume setting.

[John Robertson]

Post by Liz Tuddenham
I was working on a fairly straightforward experimental audio circuit
using TL074 op-amps laid out on Veroboard, which is prone to inter-track
shorts unless you are very careful with your soldering. The gain of a
virtual-earth stage was too high, so I altered some resistor values to
cut it down.
As this was a stereo system, I experimented with one channel first and
then switched off and replaced the equivalent resistors in the second
channel (one of the other op-amps in the same chip).
The input signal was a 6 Kc/s sinwave at about -10 dBu. After the
modification, the volume control, which was in a feedback loop, didnt
seem to make much difference. Then I realised that applying the signal
to the left channel gave a distorted version of it in the right channel
and vice-versa. Obviously I had shorted two tracks somewhere - but
where?
Things got even worse when I realised that the input signal was
appearing on the virtual earth input of each op-amp, this definitely
can't happen! I decided it was time to switch off the power supply and
go and ponder the problem over a cup of tea.
That was the point where I discovered I hadn't switched the power supply
on in the first place.

A cuppa Tea solves EVERYTHING!

John :-#)#

[Don Y]

Post by John Robertson
A cuppa Tea solves EVERYTHING!

Sadly, it takes time to prepare (boil water, steep,
cool to drinking temperature).

I'm debating buying an "electric kettle" so I can
buy *two* and always have one that is ready to
stand in for the one that is just emptying...

[Liz Tuddenham]

Post by Don Y

Post by John Robertson
A cuppa Tea solves EVERYTHING!

Sadly, it takes time to prepare (boil water, steep,
cool to drinking temperature).

I find little tasks that can be done whilst waiting for the tea to
cool. The problem is that I sometimes get so engrossed in them that I
don't remember the tea until it is nearly cold.

Post by Don Y
I'm debating buying an "electric kettle" so I can
buy *two* and always have one that is ready to
stand in for the one that is just emptying...

I habitually top up the kettle straight after pouring from it. This is
a habit I acquired when using a wood-fired Kelly Kettle (AKA Storm
Kettle) to make tea for outdoor work parties. If you didn't top it up
immediately, there was a risk of burning it out.

[Don Y]

Post by Liz Tuddenham

Post by Don Y

Post by John Robertson
A cuppa Tea solves EVERYTHING!

Sadly, it takes time to prepare (boil water, steep,
cool to drinking temperature).

I find little tasks that can be done whilst waiting for the tea to
cool. The problem is that I sometimes get so engrossed in them that I
don't remember the tea until it is nearly cold.

The bigger problem is having to wait/watch while the kettle boils.
And, then while the tea steeps. Steeping for too long with Pu'er
will almost stain *glass*!

I boil half a gallon of water and then let the tea steep in a
large thermos. After it's steeped, I seal the thermos and dispense
12 oz at a time over the course of the next 6 hours.

Then, repeat the process for a second batch. (this usually gets me
through to bedtime)

Post by Liz Tuddenham

Post by Don Y
I'm debating buying an "electric kettle" so I can
buy *two* and always have one that is ready to
stand in for the one that is just emptying...

I habitually top up the kettle straight after pouring from it. This is
a habit I acquired when using a wood-fired Kelly Kettle (AKA Storm
Kettle) to make tea for outdoor work parties. If you didn't top it up
immediately, there was a risk of burning it out.

The thermos will keep the tea hot for ~20 hours -- i.e., steaming
hot even as I pour the last cup. (When I work away from home, I
bring the thermos with me as my "hydration source") So, no need
to keep "hot water" available from which to make more.

["Tea" is just a beverage, for me]

I have a second (identical) thermos (bought to address the
inevitability of this one breaking/failing) but having TWO in
use just means having to MAKE two when they are both empty.

An electric kettle seems like it wold cut down on wear-and-tear
on the (electric) stove -- boilovers are a PITA to clean -- as
well as easing the requirement for me to stand over it waiting
for it to boil.

[I would just use the kettle to boil the water and still steep
it in the thermos]

Having a third tap -- hot, cold, tea -- at the sink would be
ideal!

[legg]

Post by Liz Tuddenham
I was working on a fairly straightforward experimental audio circuit
using TL074 op-amps laid out on Veroboard, which is prone to inter-track
shorts unless you are very careful with your soldering. The gain of a
virtual-earth stage was too high, so I altered some resistor values to
cut it down.
As this was a stereo system, I experimented with one channel first and
then switched off and replaced the equivalent resistors in the second
channel (one of the other op-amps in the same chip).
The input signal was a 6 Kc/s sinwave at about -10 dBu. After the
modification, the volume control, which was in a feedback loop, didnt
seem to make much difference. Then I realised that applying the signal
to the left channel gave a distorted version of it in the right channel
and vice-versa. Obviously I had shorted two tracks somewhere - but
where?
Things got even worse when I realised that the input signal was
appearing on the virtual earth input of each op-amp, this definitely
can't happen! I decided it was time to switch off the power supply and
go and ponder the problem over a cup of tea.
That was the point where I discovered I hadn't switched the power supply
on in the first place.

Stupid op amps!

RL