Driver to drive?

On Thursday, June 12, 2014 1:04:02 PM UTC-4, Jim Thompson wrote:
> On Sun, 08 Jun 2014 14:22:20 -0600, m II <C@in.the.hat> wrote:
<snip>
<snip>

Here's some outside the box thinking for 21st century: get a real multimeter.
The Extech EX330 measures ( and displays ) duty cycle over 0.1 to 99.9% range with resolution of 0.1% and accuracy of +/- 1.2% reading and 0.5Hz - 150KHz fequency. And this meter is cheap, even Lowes sells it, although not at the best price.

 

[Bill Sloman]

On 23/06/2014 00:41, Robert Martin wrote:

I would like to build a 50Hz sharp notch filter for an instrumentation amp.

It seems that all low pass and high pass filters create varying degrees
of phase and time distortion. Since I want to avoid this effect, does it
also apply to notch filters?

Given the above preference, can anyone point me to a notch filter design
that would be most suitable?

My understanding is that sharp notch filters introduce nasty
frequency-dependent phase shifts, but you can smooth these out by adding a
complementary all-pass filter that has no effect on the amplitude of the
signal but can produce more-or-less arbitrary frequency dependent
phase-shifts (the order of the all-pass filter has to be at least as high
as the order of the notch).

If I got stuck into my copy of Williams and Tyalor's "Electronic Filter
Design Handbook" I could probably give you chapter and verse if my brain
were actually working. I got off my flight from the Netherlands at 7:30pm
last night so my brain is actually jet-lagged into a state where it's
output
is not to be trusted - I won't dare drive a car until tomorrow.


As far as notch filters go, I like the bridged differentiator in figure
5.23
of "The Art of Elelctronics". It's remarkably easy to tune, and it is a 2nd
order filter, so an all-pass filter to get the over-all phase response
close
to phase-linear shouldn't be too difficult.

Analog Devices seems to list it as the Boctor Notch, figure 8.78 on page
8.105, but I'm jet-lagged ...

http://www.analog.com/library/analogdialogue/archives/43-09/edch%208%20filter.pdf

--
Bill Sloman, Sydney

 

I would like to build a 50Hz sharp notch filter for an instrumentation amp.



It seems that all low pass and high pass filters create varying degrees

of phase and time distortion. Since I want to avoid this effect, does it

also apply to notch filters?

any filter with sharp transitions in the freq domain will have ugly artifacts in the time domain. You can equalize the phase response or use FIR type filters with linear phase. This will linearize the phase at the expense of overall time delay or latency. BUT...even if you do this, there will still be nasty ringing and other artifacts in the time domain.

If is a fundamental fact of nature that you cannot sharply filter in one domain without screwing up the other domain.

If for example, you create a sharp 50 Hz notch to block mains hum for example, you WILL introduce 50 Hz ringing into the step response. No way around it.

Mark

 

[miso]

A filter rings if the impulse response goes negative. This is obvious if you
remember that filtering in the frequency domain is convolution in the time
domain.

If you go FIR, this is even easier to visualize. The tap coefficients must
all be positive for no ringing to occur. The step response of a FIR filter
is the sequential sum of the taps. A negative tap means you went down
instead of up.

There is this misconception that linear phase filters don't ring. If the
impulse response goes negative, the filter will ring. High order Bessel
filters ring.

 

On Mon, 23 Jun 2014 08:41:23 +0200, Bill Sloman
<cutlersloman@tpg.com.au> wrote:

On 23/06/2014 00:41, Robert Martin wrote:

I would like to build a 50Hz sharp notch filter for an instrumentation amp.

These days, the mains interference is not just clean fundamental hum,
but contains a lot of harmonics due to electronics (rectifier,
capacitor) loads. In countries, in which three phase distribution is
common, the third harmonics will add up (while the fundamental is
canceled), causing a 150 Hz (for 50 Hz) fundamental current with
harmonics in the neutral line.

For this reason, some form of comb filtering may ne needed.

For required bandwidths below the mains frequency (but want to avoid
sharp low pass filtering), the sample rate is set to be exactly the
mains period (preferably frequency locked to the actual mains
frequency e.g. with aggregates), so that a full hum cycle fits within
a sample integration period and hence cancel out. Also the harmonics
are canceled out, forming a comb filter.

Of course, this is not very useful, if the required passband is higher
than the mains frequency.

 

[Jan Panteltje]

On a sunny day (Mon, 23 Jun 2014 13:34:19 -0700 (PDT)) it happened
makolber@yahoo.com wrote in
<a622ac70-833b-47ba-ad41-ad1322c66375@googlegroups.com>:

I would like to build a 50Hz sharp notch filter for an instrumentation =
amp.



It seems that all low pass and high pass filters create varying degrees

of phase and time distortion. Since I want to avoid this effect, does i=
t

also apply to notch filters?



any filter with sharp transitions in the freq domain will have ugly artifac=
ts in the time domain. You can equalize the phase response or use FIR typ=
e filters with linear phase. This will linearize the phase at the expense =
of overall time delay or latency. BUT...even if you do this, there will st=
ill be nasty ringing and other artifacts in the time domain.

If is a fundamental fact of nature that you cannot sharply filter in one do=
main without screwing up the other domain.

If for example, you create a sharp 50 Hz notch to block mains hum for examp=
le, you WILL introduce 50 Hz ringing into the step response. No way around =
it.

Mark

But you could add 50 Hz in anti phase.
You can make that 50 Hz by using a small notch filter on the signal.
The real signal does not pass the noth filter.
Shift phase all you want on the 50 Hz you make, change amplitude too,
make harmonics needed.

Just an idea.

 

On Tue, 24 Jun 2014 05:18:51 GMT, Jan Panteltje <panteltje@yahoo.com>
wrote:

On a sunny day (Mon, 23 Jun 2014 13:34:19 -0700 (PDT)) it happened
makolber@yahoo.com wrote in
a622ac70-833b-47ba-ad41-ad1322c66375@googlegroups.com>:


I would like to build a 50Hz sharp notch filter for an instrumentation amp.

It seems that all low pass and high pass filters create varying degrees
of phase and time distortion. Since I want to avoid this effect, does it
also apply to notch filters?

any filter with sharp transitions in the freq domain will have ugly artifac=
ts in the time domain. You can equalize the phase response or use FIR typ=
e filters with linear phase. This will linearize the phase at the expense =
of overall time delay or latency. BUT...even if you do this, there will st=
ill be nasty ringing and other artifacts in the time domain.

If is a fundamental fact of nature that you cannot sharply filter in one do=
main without screwing up the other domain.

If for example, you create a sharp 50 Hz notch to block mains hum for examp=
le, you WILL introduce 50 Hz ringing into the step response. No way around =
it.

Mark

But you could add 50 Hz in anti phase.
You can make that 50 Hz by using a small notch filter on the signal.
The real signal does not pass the noth filter.
Shift phase all you want on the 50 Hz you make, change amplitude too,
make harmonics needed.

Just an idea.

I guess that you should also get at least the 150 Hz harmonics "off
the air" in addition to the 50 Hz fundamental and not try to generate
the harmonics from the fundamental.

While in a single phase environment, using a very narrow band PLL to
generate the antiphase fundamental should work, things can get quite
ugly in a three phase environment. If the input signal is moved from
close to one mains line phase to an other mains phase, the resultant
hum signal will experience a 120 degree phase shift. A too narrow PLL
might not be able to track this movement.

 

[Robert Martin]

On 23/06/14 08:41, Robert Martin wrote:

Thanks everyone for setting me straight on this. I will check out the
circuit suggestions mentioned.

Robert Martin

 

[Jan Panteltje]

On a sunny day (Tue, 24 Jun 2014 09:32:43 +0300) it happened
upsidedown@downunder.com wrote in
<fb6iq99ib2t3fmk1n7oqg5sa5o4k0rs95f@4ax.com>:

On Tue, 24 Jun 2014 05:18:51 GMT, Jan Panteltje <panteltje@yahoo.com
wrote:

On a sunny day (Mon, 23 Jun 2014 13:34:19 -0700 (PDT)) it happened
makolber@yahoo.com wrote in
a622ac70-833b-47ba-ad41-ad1322c66375@googlegroups.com>:


I would like to build a 50Hz sharp notch filter for an instrumentation amp.

It seems that all low pass and high pass filters create varying degrees
of phase and time distortion. Since I want to avoid this effect, does it
also apply to notch filters?

any filter with sharp transitions in the freq domain will have ugly artifac=
ts in the time domain. You can equalize the phase response or use FIR typ=
e filters with linear phase. This will linearize the phase at the expense =
of overall time delay or latency. BUT...even if you do this, there will st=
ill be nasty ringing and other artifacts in the time domain.

If is a fundamental fact of nature that you cannot sharply filter in one do=
main without screwing up the other domain.

If for example, you create a sharp 50 Hz notch to block mains hum for examp=
le, you WILL introduce 50 Hz ringing into the step response. No way around =
it.

Mark

But you could add 50 Hz in anti phase.
You can make that 50 Hz by using a small notch filter on the signal.
The real signal does not pass the noth filter.
Shift phase all you want on the 50 Hz you make, change amplitude too,
make harmonics needed.

Just an idea.

I guess that you should also get at least the 150 Hz harmonics "off
the air" in addition to the 50 Hz fundamental and not try to generate
the harmonics from the fundamental.

Why is that?
geting harmonics of the filtered 50 Hz would follow amplitude changes.

While in a single phase environment, using a very narrow band PLL to
generate the antiphase fundamental should work, things can get quite
ugly in a three phase environment.

Yes, but if the 50 Hz amplidude changes over time,
then you need a way to track that too.

If the input signal is moved from
close to one mains line phase to an other mains phase, the resultant
hum signal will experience a 120 degree phase shift. A too narrow PLL
might not be able to track this movement.

Yes I see plenty of problems with a PLL,
but it would be great if the hum amplitude is a constant?

 

On Tue, 24 Jun 2014 05:18:51 GMT, Jan Panteltje <panteltje@yahoo.com>
wrote:

If for example, you create a sharp 50 Hz notch to block mains hum for examp=
le, you WILL introduce 50 Hz ringing into the step response. No way around =
it.

Mark

But you could add 50 Hz in anti phase.
You can make that 50 Hz by using a small notch filter on the signal.
The real signal does not pass the noth filter.
Shift phase all you want on the 50 Hz you make, change amplitude too,
make harmonics needed.

One way to inject this antiphase hum would be to use a paired cable
with double shields. Drive the inner shield with the antiphase signal
(obtained from a normal mains transformer) adjust amplitude
accordingly. This might work at least in single phase mains
environment.

A similar arrangement is to reduce the capacitance to ground from the
two signal conductor is driving the inner shield with the cable with
the common mode voltage, which can be easily obtained from a three
op-amp instrumentation amplifier.

Of course, the best way to get away with the hum is to use proper
cabling, sufficient balancing, proper shielding and a very detailed
attenuation how and where these shields are grounded. Of course, this
is not always possible with old installation.

 

yes I want to clarify my remarks...

sharp filtering in the frequency domain does create artifacts in the time domain...but that doesn't mean that a good notch filter won't be effective at removing single frequency hum.

Usually the artifacts are not audible and are much less objectionable compared to the hum, so yes build and use a notch filter, or even a comb filter if the hum has harmonics.

This is a good example of theory vs practice, it does work well in practice.


Mark

 

[Frank Miles]

On Tue, 24 Jun 2014 17:06:44 +1000, Robert Martin wrote:

On 23/06/14 08:41, Robert Martin wrote:

Thanks everyone for setting me straight on this. I will check out the
circuit suggestions mentioned.

Robert Martin

Depending on bandwidth requirements, etc., it may be possible to subtract a line-phase-locked average
from your signal. This completely avoids the distortions of a notch filter. The commercial product
"Hum Bug" does this. There's a fairly simple implementation of this concept that was described in one
of the trade publications, but I've (at least temporarily) lost the link - sorry.

 

[Jan Panteltje]

On a sunny day (Tue, 24 Jun 2014 11:33:26 +0300) it happened
upsidedown@downunder.com wrote in
<7hdiq9l5ccjunc8be1ioemv8rf4bihlfgi@4ax.com>:

On Tue, 24 Jun 2014 05:18:51 GMT, Jan Panteltje <panteltje@yahoo.com
wrote:

If for example, you create a sharp 50 Hz notch to block mains hum for examp=
le, you WILL introduce 50 Hz ringing into the step response. No way around =
it.

Mark

But you could add 50 Hz in anti phase.
You can make that 50 Hz by using a small notch filter on the signal.
The real signal does not pass the noth filter.
Shift phase all you want on the 50 Hz you make, change amplitude too,
make harmonics needed.

One way to inject this antiphase hum would be to use a paired cable
with double shields. Drive the inner shield with the antiphase signal
(obtained from a normal mains transformer) adjust amplitude
accordingly. This might work at least in single phase mains
environment.

A similar arrangement is to reduce the capacitance to ground from the
two signal conductor is driving the inner shield with the cable with
the common mode voltage, which can be easily obtained from a three
op-amp instrumentation amplifier.

Of course, the best way to get away with the hum is to use proper
cabling, sufficient balancing, proper shielding and a very detailed
attenuation how and where these shields are grounded. Of course, this
is not always possible with old installation.

yes,
my experience does include 'rescuing' some audio and video _recordings_.
I was thinking about combining your PLL (to get a very pure sine wave)
with amplitude control derived by comparing against teh filtered 50Hz
Bets of both systems.
Make harmonics from the PLL output like you suggested.

Video fixing is also fun, from painting in each frame to writing algos
to process some artifacts.

 

[Jan Panteltje]

On a sunny day (Tue, 24 Jun 2014 08:34:32 -0700 (PDT)) it happened
makolber@yahoo.com wrote in
<dc95bc54-a7a5-4516-a4a7-cc7875215e1d@googlegroups.com>:

yes I want to clarify my remarks...

sharp filtering in the frequency domain does create artifacts in the time domain...but that doesn't mean that a good notch
filter won't be effective at removing single frequency hum.

Usually the artifacts are not audible and are much less objectionable compared to the hum, so yes build and use a notch filter,
or even a comb filter if the hum has harmonics.

This is a good example of theory vs practice, it does work well in practice.


Mark

There are many types of notch filters.
A comb filer for example,
one that also filters out 2*f, 3*f, 4*f .. n*f, _will_ cause audible audio artefacts.
Here is one:
http://panteltje.com/panteltje/newsflex/download.html#humfilter

 

[Kevin Aylward]

wrote in message
news:a622ac70-833b-47ba-ad41-ad1322c66375@googlegroups.com...

I would like to build a 50Hz sharp notch filter for an instrumentation
amp.



It seems that all low pass and high pass filters create varying degrees

of phase and time distortion. Since I want to avoid this effect, does
it

also apply to notch filters?



any filter with sharp transitions in the freq domain will have ugly
artifacts in the time domain. You can equalize the phase response or use
FIR type filters with >linear phase. This will linearize the phase at the
expense of overall time delay or latency. BUT...even if you do this, there
will still be nasty ringing and other >artifacts in the time domain.

If is a fundamental fact of nature that you cannot sharply filter in one
domain without screwing up the other domain.

I see it as that only signals in the time domain are physical real in
nature. The frequency domain being simply an abstract mathematical construct
defined as the Fourier transform of time signals. It is then a mathematical
fact that sigma_t x sigma_f >=1/2



Kevin Aylward B.Sc.
www.kevinaylward.co.uk
www.anasoft.co.uk - SuperSpice

 

On Tue, 24 Jun 2014 08:34:32 -0700 (PDT), makolber@yahoo.com wrote:

yes I want to clarify my remarks...

sharp filtering in the frequency domain does create artifacts in the time domain...but that doesn't mean that a good notch filter won't be effective at removing single frequency hum.

Usually the artifacts are not audible and are much less objectionable compared to the hum, so yes build and use a notch filter, or even a comb filter if the hum has harmonics.

This is a good example of theory vs practice, it does work well in practice.


Mark

Is this question about (industrial sensor) instrumentation amplifier
or instrument amplifier, such as guitar amplifier ?

For audio applications, the ear is not very sensitive to sounds below
100 Hz at least at moderate sound levels, say below 80 dB, thus quite
a lot hum is tolerated. However, since the ear sensitivity is much
better at higher frequencies, thus mains interference even with quite
low level harmonics can be much more objectionable than a pure 50 Hz
hum with the same level.

If this is for a guitar amplifier, much can be done to eliminate the
hum in the first place. At least previously, electric guitars were
connected with unbalanced circuits to the amplifier causing al kinds
of hum problems. The pick-ups are balanced, but the other end is
connected to the cable shield as well as all metal parts (such as
strings). At least use stereo plugs to carry the balanced signal to
the amplifier and just use the cable shield for connecting the metal
parts on the guitar.

 

[DannyD.]

Oren wrote, on Wed, 25 Jun 2014 07:24:12 -0700:

I was hoping you would mention the % of concentrated
Clorox un-scented bleach :-\

Hi Oren,

Huckleberries are, um, I mean bleach is a strange beast indeed, which,
I admit, is confusing because, while chlorinating liquid is usually
sold by *trade percentage*, in contrast, a regular bleach is usually
sold by *weight percentage* of product (i.e., of sodium hypochlorite).

So 8.25% bleach (sold by weight % of sodium hypochlorite) is only
7.86% available chlorine (by weight), and, assuming a 1.10 g/ml density,
this is 9.08% trade percentage (i.e., % Available Chlorine by volume).

Similarly, a typical 6% bleach is 6% sodium hypochlorite by weight, which
is only 5.7% available chlorine.

Hence, the two equations:
1. Cost per pound of available chlorine for liquid chlorine:
$cost/gallon x 1gallon/9.7pounds x 100pounds of liquid/10.78pounds available chlorine

2. Cost per pound of available chlorine for liquid bleach:
$cost/gallon x 1gallon/9.0pounds x 100pounds of liquid/5.7 pounds available chlorine

Sometimes things are just complicated.

 

[Kevin Aylward]

wrote in message news:g7ukq9tmomn0aqj297fv6120rlf5uiphc5@4ax.com...

On Tue, 24 Jun 2014 08:34:32 -0700 (PDT), makolber@yahoo.com wrote:

yes I want to clarify my remarks...

sharp filtering in the frequency domain does create artifacts in the time
domain...but that doesn't mean that a good notch filter won't be effective
at removing single frequency hum.

Usually the artifacts are not audible and are much less objectionable
compared to the hum, so yes build and use a notch filter, or even a comb
filter if the hum has harmonics.

This is a good example of theory vs practice, it does work well in
practice.


Mark

For audio applications, the ear is not very sensitive to sounds below
100 Hz at least at moderate sound levels, say below 80 dB, thus quite
a lot hum is tolerated.

Ahmmmmmmmmmmmmmmmmmmmmm

The low note on a guitar is 82 Hz, the low note on a Bass is 41, sure its
not as sensitive, but if you cant hear that in music, you must be deaf.
60/50 hz hum is bloody annoying indeed in a guitar amplifier.

Kevin Aylward B.Sc.
www.kevinaylward.co.uk
www.anasoft.co.uk - SuperSpice

 

[DannyD.]

DannyD. wrote, on Wed, 25 Jun 2014 20:12:23 +0000:

regular bleach is usually
sold by *weight percentage* of product (i.e., of sodium hypochlorite).

From the net ...

So, for example, 12.25% Trade Percentage is % Available Chlorine by volume
which is is 10.78% Available Chlorine by weight assuming 1.16 g/ml density
and is 11.31% sodium hypochlorite by weight.

Weight % Available Chlorine = Trade % / Specific Gravity
= (Weight % NaOCl) * Cl2_g_mole / NaOCl_g_mole = (Weight % NaOCl) * 0.9525

So Weight % NaOCl = 1.05 * Weight % Available Chlorine

The divisor in those equations calculating pounds of available chlorine
is "% Available Chlorine (by weight)".

While chlorinating liquid is usually sold by Trade %,
the percentage for regular bleach is usually weight % of product
(i.e. sodium hypochlorite) so 8.25% bleach (weight % of sodium hypochlorite)
is 7.86% Available Chlorine (by weight) and assuming a 1.10 g/ml density
this is 9.08% Trade (% Available Chlorine by volume).

 

On Wed, 25 Jun 2014 17:38:39 +0100, "Kevin Aylward"
<ExtractkevinRemove@kevinaylward.co.uk> wrote:

wrote in message news:g7ukq9tmomn0aqj297fv6120rlf5uiphc5@4ax.com...

On Tue, 24 Jun 2014 08:34:32 -0700 (PDT), makolber@yahoo.com wrote:

yes I want to clarify my remarks...

sharp filtering in the frequency domain does create artifacts in the time
domain...but that doesn't mean that a good notch filter won't be effective
at removing single frequency hum.

Usually the artifacts are not audible and are much less objectionable
compared to the hum, so yes build and use a notch filter, or even a comb
filter if the hum has harmonics.

This is a good example of theory vs practice, it does work well in
practice.


Mark



For audio applications, the ear is not very sensitive to sounds below
100 Hz at least at moderate sound levels, say below 80 dB, thus quite
a lot hum is tolerated.

Ahmmmmmmmmmmmmmmmmmmmmm

The low note on a guitar is 82 Hz, the low note on a Bass is 41, sure its
not as sensitive, but if you cant hear that in music, you must be deaf.
60/50 hz hum is bloody annoying indeed in a guitar amplifier.

Take a look at Fletcher Munson curves.

At 100 dB SPL the frequency response is linear at 100 phon from 1000
Hz down to 20 Hz.

Assuming the bass guitar system is capable of producing 100 dB SPL at
50 Hz.

A 50 Hz hum only 20 dB below full power (80 dB SPL) is 65 phon, 40 dB
below (60 dB SPL) will produce 10 phon and -50 dB below full power (50
dB SPL) would be at the threshold of hearing ( 0 phon).

The corresponding figures at 1 kHz would be 80 phon, 60 phon and 50
phon respectively.

Thus, you will need 50 dB better SNR at 1 kHz compared to 50 Hz.