The signal I get on certain close-miced drums is too hot for my mixer, which
doesn't have any built-in attenuation. Do I need to buy an attenuator from a
supplier or can I just build my own with resistors? If I need 10-20dB of
attenuation, what values should I use, and what about impedance issues? Is
there any significant difference in sound quality between using different
components (resistors, wires, solder etc), or just a difference in accuracy
of level reduction?

Thanks,

--
tj hertz

tjhertz at gmail dot com

"sycochkn" wrote in message
ink.net...
So what is the impedence of the microphone? What does it like to see. 300
ohm 50k ohm? Single ended balanced?


Would you use a different attenuator for every mic in your collection? The
mics in question are an SM57 and D112 or D6.

I understand basic electronics but I'm not really up to speed on impedences.

Thanks

--
tj hertz

"Scott Dorsey"

But there was a recent thread here
in which Phil insisted everyone was wrong about attenuator resistor
values,



** You are a stinking liar, Scott Dorsey.




.......... Phil

TJ Hertz wrote:

"sycochkn" wrote in message
link.net...
So what is the impedence of the microphone? What does it like to see. 300
ohm 50k ohm? Single ended balanced?


Would you use a different attenuator for every mic in your collection?

Probably not.

The
mics in question are an SM57 and D112 or D6.

A single design will serve these quite nicely.

I understand basic electronics but I'm not really up to speed on impedences.

For the purposes of a mic attenuator you can simply think of the impedances
as resistances. I found a page that describes a design approach at
http://www.isce.org.uk/engnotes/ENote3.pdf

I agree with Scott, though. You should buy the Shure A15AS. It's pre-
engineered, can be set to three different levels of attenuation, can
usually be purchased for less than it will cost you for the equivalent
parts, and will probably be more rugged and reliable than homebrew.

--
================================================== ======================
Michael Kesti | "And like, one and one don't make
| two, one and one make one."
| - The Who, Bargain

"TJ Hertz" wrote in message
k...
"sycochkn" wrote in message
ink.net...
So what is the impedence of the microphone? What does it like to see.
300
ohm 50k ohm? Single ended balanced?


Would you use a different attenuator for every mic in your collection? The
mics in question are an SM57 and D112 or D6.

I understand basic electronics but I'm not really up to speed on
impedences.

Thanks

--
tj hertz



most modern mixer inputs are what you would consider to be high impedance,
therefore a high Z pad would seem to be the way to go to me

but then most of the mixers i run across lately don't need the pad as the
input can be trimed enough to keep from clipping.

which unit are you working with?

In article . net writes:

So what is the impedence of the microphone? What does it like to see. 300
ohm 50k ohm? Single ended balanced?

Depends on the microphone, but for a general purpose "won't harm
anything" pad, make it between 1.5K and 2.5K ohms. It's only critical
if you know exactly which mic it will be used with, and then only
makes a noticable difference with some mics.


--
I'm really Mike Rivers )
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me he double-m-eleven-double-zero at yahoo

"dan lavry"


Also, I would not go through all the trouble with 5% resistors. Go fro
some good resistors with nice tolerance; say 0.1%, for the pair of the
series resistors. The single load resistor tolerance is not important,
but the series resistors tolerance matters a hack of a lot for common
mode rejection!!! Remember that 1% mismatch is only -40dB, and even
0.1% mismatch is -60dB rejection.


** Because the network has no connection to ground, resistor tolerance does
not impact CMRR as badly as it otherwise would. I did a few tests as
follows.


With values of 560, 470 and 100 ohms ( ie an error of 20% in the series
resistors) the CMRR was 50 dB.

With values of 493, 470 and 100 ohms ( ie an error of 5% in the series
resistors) the CMRR was 54 dB.

With values of 470, 470 and 100 ohms ( all 1% types chosen at random ) the
CMRR was 79 dB.

With the inputs shorted together the CMRR was 80 dB.




.............. Phil

In article writes:

** Because the network has no connection to ground, resistor tolerance does
not impact CMRR as badly as it otherwise would. I did a few tests as
follows.


With values of 560, 470 and 100 ohms ( ie an error of 20% in the series
resistors) the CMRR was 50 dB.

With values of 493, 470 and 100 ohms ( ie an error of 5% in the series
resistors) the CMRR was 54 dB.

With values of 470, 470 and 100 ohms ( all 1% types chosen at random ) the
CMRR was 79 dB.

With the inputs shorted together the CMRR was 80 dB.

I'm surprised that someone as dedicated to specifications and number
as you would find the difference between 50 dB CMRR (with 20%
imbalance) or 54 dB (with 5% imbalance) and 79 dB (with 1% imbalance)
to be a small impact. What is your criteria if not a loss of 25 dB or
more of CMRR?

I agree that there isn't a big difference between 79 dB and 80 dB. I
also will note that CMRR isn't important until you need it. (there are
lots of nice and quiet unbalanced connections) But with the empirical
evidence that you've offered, given the low cost of resistors with 1%
tolerance, there's no reason not to use them.



--
I'm really Mike Rivers )
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me he double-m-eleven-double-zero at yahoo

"Mike Rivers"
Dan Lavry
Phil Allison

Also, I would not go through all the trouble with 5% resistors. Go fro
some good resistors with nice tolerance; say 0.1%, for the pair of the
series resistors. The single load resistor tolerance is not important,
but the series resistors tolerance matters a hack of a lot for common
mode rejection!!! Remember that 1% mismatch is only -40dB, and even
0.1% mismatch is -60dB rejection.

** Because the network has no connection to ground, resistor tolerance
does
not impact CMRR as badly as it otherwise would. I did a few tests as
follows.


With values of 560, 470 and 100 ohms ( ie an error of 20% in the
series
resistors) the CMRR was 50 dB.

With values of 493, 470 and 100 ohms ( ie an error of 5% in the series
resistors) the CMRR was 54 dB.

With values of 470, 470 and 100 ohms ( all 1% types chosen at random )
the
CMRR was 79 dB.

With the inputs shorted together the CMRR was 80 dB.


I'm surprised that someone as dedicated to specifications and number
as you would find the difference between 50 dB CMRR (with 20%
imbalance) or 54 dB (with 5% imbalance) and 79 dB (with 1% imbalance)
to be a small impact. What is your criteria if not a loss of 25 dB or
more of CMRR?


** The total ****wit, incontinent, smelly parrot has not read my words in
context - as per bloody usual

I am suggesting that cheap, readily available *** 1 % *** resistors are
adequate - that is what the numbers say.

No need to go for exotic 0.1 % ones as Dan Lavry assumed.





.............. Phil

In article writes:

** The total ****wit, incontinent, smelly parrot has not read my words in
context - as per bloody usual

I read exactly what you wrote.

I am suggesting that cheap, readily available *** 1 % *** resistors are
adequate - that is what the numbers say.

That's what they say to me, but you never suggested it until this
message. You said that

** Because the network has no connection to ground, resistor tolerance
does
not impact CMRR as badly as it otherwise would. I did a few tests as
follows.

"Because the network has no connection to ground" is a red herring. I
read "does not impact CMRR as badly as it otherwise would" and then
you proceeded to show the difference between 20%, 5%, and 1% values.
Had you simply written that best case CMRR is 80 dB and 1% resistors
reduce it to 79 dB, your meaning would be clear.

The master of obfuscation for the sake of argument continues his modus
operandi.

No need to go for exotic 0.1 % ones as Dan Lavry assumed.

Depends on how fussy you are. 80 dB CMRR isn't all that great, so if
that's all you have to work with, then 1% resistors are fine. If you
have an input capable of better CMRR, you may choose not to degrade it
when you can avoid it by simply purchasing better matched resistors.



--
I'm really Mike Rivers )
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me he double-m-eleven-double-zero at yahoo

"Phil Allison" wrote in message ** Because the I'm surprised that someone as dedicated to specifications and number
as you would find the difference between 50 dB CMRR (with 20%
imbalance) or 54 dB (with 5% imbalance) and 79 dB (with 1% imbalance)
to be a small impact. What is your criteria if not a loss of 25 dB or
more of CMRR?


** The total ****wit, incontinent, smelly parrot has not read my words in
context - as per bloody usual

I am suggesting that cheap, readily available *** 1 % *** resistors are
adequate - that is what the numbers say.

No need to go for exotic 0.1 % ones as Dan Lavry assumed.
............. Phil


I did not assume a thing. It is not about assumptions, and certainly
about foul language. Well, I did not want to assume that there is no
ground path, it was not a given. You are stating it is so, but it is
someone elses gear.

But the point is simple: One is going to invest say 1 hour of their
time in building a simple resistor network. It sounded like someone
that needs to buy the resistors anyway. Only factories labs and alike
have a large selection of parts on hand. So you call Digikey or
Mouser, and find that you can do it with 5% resistors that will cost
you a couple of cents, or 1% resistors that will still only cost a
couple of cents, or you can go for 0.1% and pay a whole couple of
dollars for a pair, keep that RO at 1% or whatever.

Now you add shipping costs, and the materials are NEAR NEGLEGABLE! It
is not like I am sugesting someone to spend $1000 on matetrials. I
stand by my sugestion to go for .1%, even if it only buys you a couple
of dB. It will probably do better than that.

So instead of swearing a lot, listen to what I said - and I am telling
you that computing is a must, not just measuring, unless you want to
specify exactly the value of the resistors used in the circuit you
tested, which is a real waste of time...

My advice: just go for .1% at 25ppm/C or similar and pay a couple of
bucks.
You could in the future use that attenuator with gear that does not
have great isolation. I am not yet sure the mixer we are talking about
does.

BR
Dan Lavry

"dan lavry" wrote in message
m...

My advice: just go for .1% at 25ppm/C or similar and pay a couple of
bucks.

Or...

Digi-Key sells its 1% resistors in packets of five. Within a batch, my
experience is that the spread is less than 1%; if you buy two packets,
you're likely to get at least two pairs that match within 0.1%. If you have
a digital multimeter, the job of matching's easy. (And if you don't, you
probably ought to have one anyway. Radio Shack sells 'em at very reasonable
prices.)

Peace,
Paul

In article writes:

Digi-Key sells its 1% resistors in packets of five. Within a batch, my
experience is that the spread is less than 1%; if you buy two packets,
you're likely to get at least two pairs that match within 0.1%. If you have
a digital multimeter, the job of matching's easy.

How do you think they make 0.1% resistors? g

What you aren't likely to find are two that are within 0.1% of the
marked value, because those get branded as 0.1% and sold for more
money because there are fewer of them than those within 1% of the
marked value. But for this application, the absolute value isn't
critical.



--
I'm really Mike Rivers )
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me he double-m-eleven-double-zero at yahoo

"Mike Rivers"
Paul Stramler

Digi-Key sells its 1% resistors in packets of five. Within a batch, my
experience is that the spread is less than 1%; if you buy two packets,
you're likely to get at least two pairs that match within 0.1%. If you
have
a digital multimeter, the job of matching's easy.

How do you think they make 0.1% resistors? g



** Nothing a demented, incontinent, squawking Parrot would ever know -
that is for sure.



What you aren't likely to find are two that are within 0.1% of the
marked value, because those get branded as 0.1% and sold for more
money


** Squawk ... squawk .... squawk ..... squawk ..........

The parrot has got a bad dose of avian diarrhoea !!!



because there are fewer of them than those within 1% of the
marked value. But for this application, the absolute value isn't
critical.



** Oh my god - now the featherless abomination is parroting my words
!!




............. Phil

In article writes:

But for this application, the absolute value isn't
critical.

** Oh my god - now the featherless abomination is parroting my words

A very lucid and creative interpretation, much more than a parrot
could do. At least people won't misinterpret what I wrote because I
said what's important up front, not hid it in a diversion. But then
you make reading this newsgroup such a challange.


--
I'm really Mike Rivers )
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me he double-m-eleven-double-zero at yahoo

"Phil Allison" wrote in message ...
"Mike Rivers"
Paul Stramler

** Nothing a demented, incontinent, squawking Parrot would ever know -
that is for sure.

** Squawk ... squawk .... squawk ..... squawk ..........

The parrot has got a bad dose of avian diarrhoea !!!

** Oh my god - now the featherless abomination is parroting my words
............ Phil

Phil, you win, I am out of here. I do not like your style. You
probably think it is cool or cute. It may be so for some, but the
internet has all sorts of people, from various cultures and walk of
life. I, for one, at nearly 60 years old, find your comments do not
fit my style. I'll say that much: all that hoopla with parrot and
diarrhhoa is very un inviting to have a dialog. It is a distruction,
making people less willing to have a dialog. Of course you are right,
I must be an old fart and a nurd for saying so.

Good luck
Dan Lavry

Phil, you win, I am out of here. I do not like your style. You
probably think it is cool or cute. It may be so for some, but the
internet has all sorts of people, from various cultures and walk of
life. I, for one, at nearly 60 years old, find your comments do not
fit my style. I'll say that much: all that hoopla with parrot and
diarrhhoa is very un inviting to have a dialog. It is a distruction,
making people less willing to have a dialog. Of course you are right,
I must be an old fart and a nurd for saying so.

Good luck
Dan Lavry



Well, I'm 64 and I don't care for it either.
Killfile coming up.


--Wayne

-"sounded good to me"-

On or about Thu, 2 Sep 2004 00:20:11 +1000, Phil Allison allegedly wrote:

** Nothing a demented, incontinent, squawking Parrot would ever know -
that is for sure.

** Squawk ... squawk .... squawk ..... squawk ..........

The parrot has got a bad dose of avian diarrhoea !!!

** Oh my god - now the featherless abomination is parroting my words


I think you must've been hanging out too much with Rod Speed, Phil.



Noel Bachelor noelbachelorAT(From:_domain)
Language Recordings Inc (Darwin Australia)

My 0.1% resistors have much tighter specified tempcos than my 1%
resistors.

Are you suggesting that they select the tempcos as well? or that it
always just happens that if a resistor is very close to its design
value, it automatically has a tighter tempco? Or that the makers lie
about the specs? Or have you just managed to find 0.1% resistors with
bad tempcos?

On 1 Sep 2004 09:12:58 -0400, (Mike Rivers) wrote:


In article writes:

Digi-Key sells its 1% resistors in packets of five. Within a batch, my
experience is that the spread is less than 1%; if you buy two packets,
you're likely to get at least two pairs that match within 0.1%. If you have
a digital multimeter, the job of matching's easy.

How do you think they make 0.1% resistors? g

What you aren't likely to find are two that are within 0.1% of the
marked value, because those get branded as 0.1% and sold for more
money because there are fewer of them than those within 1% of the
marked value. But for this application, the absolute value isn't
critical.

Tony (remove the "_" to reply by email)

"Tony"

My 0.1% resistors have much tighter specified tempcos than my 1%
resistors.


** Naturally - since they are made with different and better materials and
a different process.

The aging characteristics are an order of magnitude better as well.



Are you suggesting that they select the tempcos as well? or that it
always just happens that if a resistor is very close to its design
value, it automatically has a tighter tempco? Or that the makers lie
about the specs? Or have you just managed to find 0.1% resistors with
bad tempcos?


** How you are enjoying sticking the boot into that Septic featherless,
squawking parrot ??





.............. Phil

In article writes:

Are you suggesting that they select the tempcos as well? or that it
always just happens that if a resistor is very close to its design
value, it automatically has a tighter tempco? Or that the makers lie
about the specs? Or have you just managed to find 0.1% resistors with
bad tempcos?

I have never been concerned with "tempcos" (I assume that's Australian
for temperature coefficient) for a device that will always be used at
room temperature.

Isn't the temperature coefficient primarily a function of the material
the resitor is made of? Unless (for example) there are no 0.1% metal
film resistors because the temperature coefficient can't support that
tolerance.

Educate me here, don't just challange my statement. And explain, don't
just point me to a web site. That way I'll be confident that you know
what you're saying.






--
I'm really Mike Rivers )
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me he double-m-eleven-double-zero at yahoo

Mike Rivers wrote:

In article writes:

Digi-Key sells its 1% resistors in packets of five. Within a batch, my
experience is that the spread is less than 1%; if you buy two packets,
you're likely to get at least two pairs that match within 0.1%. If you have
a digital multimeter, the job of matching's easy.

How do you think they make 0.1% resistors? g

What you aren't likely to find are two that are within 0.1% of the
marked value, because those get branded as 0.1% and sold for more
money because there are fewer of them than those within 1% of the
marked value. But for this application, the absolute value isn't
critical.

Actually my experience is different. Commercial 1% resistors measure *far* better than 1% in practice.

I suspect that 0.1% resistors are generally manufactured on a special line. They are a specilaised high
cost item. In any event, the demand for them is inadequate to exhaust those 1% types that would be 0.1%.


Graham

"Mike Rivers"
Tony

Are you suggesting that they select the tempcos as well? or that it
always just happens that if a resistor is very close to its design
value, it automatically has a tighter tempco? Or that the makers lie
about the specs? Or have you just managed to find 0.1% resistors with
bad tempcos?

I have never been concerned with "tempcos" (I assume that's Australian
for temperature coefficient) for a device that will always be used at
room temperature.



** Resistors get very hot due to dissipation - you tenth witted, demented
parrot.

The term "tempco" is actually an Americanism.

Shame that does not extend to featherless poop droppers.



Isn't the temperature coefficient primarily a function of the material
the resitor is made of?


** Yep - and 0.1% types are not made from the same material as 1% types.



Educate me here, don't just challange my statement.


** No one can educate a ****wit parrot who thinks it already knows.


Squawk...squawk ....squawk .....



And explain, don't just point me to a web site.


** ROTFLMAO !!!!!!!



That way I'll be confident that you know
what you're saying.


** THAT has to be the dumbest post of the year !!!!


ROTFLMFAO !!!




....... Phil

Phil Allison wrote:

** THAT has to be the dumbest post of the year !!!!

I don't know, Phil. You have provided an awful lot of competition.

"Mike Rivers" wrote in message
news:znr1094259452k@trad
In article
writes:

Are you suggesting that they select the tempcos as well? or that it
always just happens that if a resistor is very close to its design
value, it automatically has a tighter tempco? Or that the makers lie
about the specs? Or have you just managed to find 0.1% resistors with
bad tempcos?

I have never been concerned with "tempcos" (I assume that's Australian
for temperature coefficient) for a device that will always be used at
room temperature.

Not to mention the fact that if the components of an attenuator are kept in
reasonbly close thermal contact with each other, their ratio, which sets the
attenuation, will remain the same, even if their values change with
temperature.

"Mike Rivers" wrote in message
news:znr1094259452k@trad
In article
writes:

Are you suggesting that they select the tempcos as well? or that it
always just happens that if a resistor is very close to its design
value, it automatically has a tighter tempco? Or that the makers lie
about the specs? Or have you just managed to find 0.1% resistors with
bad tempcos?

I have never been concerned with "tempcos" (I assume that's Australian
for temperature coefficient) for a device that will always be used at
room temperature.

Not to mention the fact that if the components of an attenuator are kept in
reasonbly close thermal contact with each other, their ratio, which sets the
attenuation, will remain the same, even if their values change with
temperature.

In article writes:

I have never been concerned with "tempcos" (I assume that's Australian
for temperature coefficient) for a device that will always be used at
room temperature.



** Resistors get very hot due to dissipation - you tenth witted, demented
parrot.


Mister-theory-practice-be-damned speaks wisely.

I'll keep that in mind the next time I connect a 5 watt microphone to
my home-made attenuator that uses 1 watt 1% hand selected resistors.































You dim-witted sheep masturbator!

--
I'm really Mike Rivers )
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me he double-m-eleven-double-zero at yahoo

In article writes:

I have never been concerned with "tempcos" (I assume that's Australian
for temperature coefficient) for a device that will always be used at
room temperature.



** Resistors get very hot due to dissipation - you tenth witted, demented
parrot.


Mister-theory-practice-be-damned speaks wisely.

I'll keep that in mind the next time I connect a 5 watt microphone to
my home-made attenuator that uses 1 watt 1% hand selected resistors.































You dim-witted sheep masturbator!

--
I'm really Mike Rivers )
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me he double-m-eleven-double-zero at yahoo

Pooh Bear wrote in message
I suspect that 0.1% resistors are generally manufactured on a special line. They are a specilaised high
cost item. In any event, the demand for them is inadequate to exhaust those 1% types that would be 0.1%.


Graham

When I brought that issue about .1% being tighter tolerance for better
common mode rejection, I did mention the tempco which is temperature
coefficient, and I pointed at 25ppm .1% as a "reasonable" way to go.
Why? Because you can get those for a buck each, and given that one is
considering going through all the trouble of buliding a rig, this is a
worth while thing to do.

But since people are asking questions, I would answer some and add a
point or two:

First, it is about materials, of course it is, and that is one of the
major factors for the resistor manufacturers to figure the proper mic
of materials to meet some given specifications. That does include both
abolute value, and temperature drift, and by the way, the aging is
already included in the tolerance, at least for the resistors I use.

Let me point out that I my sugestion to invest the couple of bucks was
reasonable, and I even pointed out you could buy them at Mouser or
Digikey. Those outlets will deal in small quantities, while the big
part houses will mostly not.

I could have "gone crazy" to sugest some really expansive parts, such
as bulk metal foil, and I have seen those go for $60 each at small
quantity. What do you get? .01% tolerance with 1ppm/degree tempco! I
did not sugest that, because I am a parctical design engineer, not
just someone wasting your time.
But the internet always seem to encourage one more guy to say one more
thing. I wish it were not so...

There is also another way to go, which may be just a bit more costly,
perhapse a couple of bucks more, and makes a lot of sense:

The idea is to have good matching, so the initial tolerance is
importent. But when it comes to drift due to temperature or againg, an
alternet way to go is to use 2 resistors made out of the exact same
material and to be sure that they are at the same temperature and
humidity. Who do you do that? You buy a resistor network (2 or more
resistors on the same substrate). It often looks like an IC (dual in
line - 2 rows, 8 pins, 14 pins, 16 pins...) or some are a SIP (single
in line package). All the resistors are made on the same substarte.
Often you deposite some material (they call it film), and then isolate
the areas to yield the desired resistance. Now, thick film is the
cheaper process. Thin film is the one to look for, it comes in veriety
of specs but tends to be good on tolerance and temperature.

If one wants to go that route, the initial tolerance is importent, but
the better spec is initial matching (for this attenuator case we are
discussing). Also, it is not the TC (temperature coefficient) that is
key. The TCT (temperature coefficient tracking) is the key. Why?
Because you really are not that interested the exact value. You want
to be dran shure that the pair will have the same value, out of the
box, and at different temperature and aging.
So you want initialy good match, than the variation to be the same on
both parts (tracking).

I have not look and will not look for a specific part for this case,
but with 30 years analog design, I bet you get what you want for less
than $3 for the pair.

Those that understood my comments about initial matching and tracking
of materials on the same substrate can now understand how resistor
based DAC's are designed. It is the same thing - good initial matching
(often done with laser triming of the material), and than all the
resistors are made of the same material, and all in a very small
space, thus same temperature and other conditions...

BR
Dan Lavry

Pooh Bear wrote in message
I suspect that 0.1% resistors are generally manufactured on a special line. They are a specilaised high
cost item. In any event, the demand for them is inadequate to exhaust those 1% types that would be 0.1%.


Graham

When I brought that issue about .1% being tighter tolerance for better
common mode rejection, I did mention the tempco which is temperature
coefficient, and I pointed at 25ppm .1% as a "reasonable" way to go.
Why? Because you can get those for a buck each, and given that one is
considering going through all the trouble of buliding a rig, this is a
worth while thing to do.

But since people are asking questions, I would answer some and add a
point or two:

First, it is about materials, of course it is, and that is one of the
major factors for the resistor manufacturers to figure the proper mic
of materials to meet some given specifications. That does include both
abolute value, and temperature drift, and by the way, the aging is
already included in the tolerance, at least for the resistors I use.

Let me point out that I my sugestion to invest the couple of bucks was
reasonable, and I even pointed out you could buy them at Mouser or
Digikey. Those outlets will deal in small quantities, while the big
part houses will mostly not.

I could have "gone crazy" to sugest some really expansive parts, such
as bulk metal foil, and I have seen those go for $60 each at small
quantity. What do you get? .01% tolerance with 1ppm/degree tempco! I
did not sugest that, because I am a parctical design engineer, not
just someone wasting your time.
But the internet always seem to encourage one more guy to say one more
thing. I wish it were not so...

There is also another way to go, which may be just a bit more costly,
perhapse a couple of bucks more, and makes a lot of sense:

The idea is to have good matching, so the initial tolerance is
importent. But when it comes to drift due to temperature or againg, an
alternet way to go is to use 2 resistors made out of the exact same
material and to be sure that they are at the same temperature and
humidity. Who do you do that? You buy a resistor network (2 or more
resistors on the same substrate). It often looks like an IC (dual in
line - 2 rows, 8 pins, 14 pins, 16 pins...) or some are a SIP (single
in line package). All the resistors are made on the same substarte.
Often you deposite some material (they call it film), and then isolate
the areas to yield the desired resistance. Now, thick film is the
cheaper process. Thin film is the one to look for, it comes in veriety
of specs but tends to be good on tolerance and temperature.

If one wants to go that route, the initial tolerance is importent, but
the better spec is initial matching (for this attenuator case we are
discussing). Also, it is not the TC (temperature coefficient) that is
key. The TCT (temperature coefficient tracking) is the key. Why?
Because you really are not that interested the exact value. You want
to be dran shure that the pair will have the same value, out of the
box, and at different temperature and aging.
So you want initialy good match, than the variation to be the same on
both parts (tracking).

I have not look and will not look for a specific part for this case,
but with 30 years analog design, I bet you get what you want for less
than $3 for the pair.

Those that understood my comments about initial matching and tracking
of materials on the same substrate can now understand how resistor
based DAC's are designed. It is the same thing - good initial matching
(often done with laser triming of the material), and than all the
resistors are made of the same material, and all in a very small
space, thus same temperature and other conditions...

BR
Dan Lavry

"Mike Rivers"
Phil Allison

I have never been concerned with "tempcos" (I assume that's Australian
for temperature coefficient) for a device that will always be used at
room temperature.


** Resistors get very hot due to dissipation - you tenth witted,
demented
parrot.


Mister-theory-practice-be-damned speaks wisely.


** The lying, ****wit parrot has snipped and shifted the context again
!!!!!!!!!

Any resistor claimed to be of 0.1% precision by its **maker** must have
commensurate tempco and long term stability figures as well.

Selecting 0.1% tolerance examples from 1% stock will NOT give you that.

The ****wit parrot's ****wit notion is just as putrid ad ever.




............ Phil

"Mike Rivers"
Phil Allison

I have never been concerned with "tempcos" (I assume that's Australian
for temperature coefficient) for a device that will always be used at
room temperature.


** Resistors get very hot due to dissipation - you tenth witted,
demented
parrot.


Mister-theory-practice-be-damned speaks wisely.


** The lying, ****wit parrot has snipped and shifted the context again
!!!!!!!!!

Any resistor claimed to be of 0.1% precision by its **maker** must have
commensurate tempco and long term stability figures as well.

Selecting 0.1% tolerance examples from 1% stock will NOT give you that.

The ****wit parrot's ****wit notion is just as putrid ad ever.




............ Phil

"dan lavry"
Pooh Bear
I suspect that 0.1% resistors are generally manufactured on a special
line. They are a specilaised high
cost item. In any event, the demand for them is inadequate to exhaust
those 1% types that would be 0.1%.



When I brought that issue about .1% being tighter tolerance for better
common mode rejection,


** Which, as I demonstrated with real tests, is a total fallacy.


I did mention the tempco which is temperature
coefficient, and I pointed at 25ppm .1% as a "reasonable" way to go.


** It is a pedantic fool's idea.


Let me point out that I my sugestion to invest the couple of bucks was
reasonable,


** It is a pedantic fool's idea.


I could have "gone crazy" to sugest some really expansive parts


** That would be a pedantic ****wit's idea.



Those that understood my comments about initial matching and tracking
of materials on the same substrate can now understand how resistor
based DAC's are designed. It is the same thing - good initial matching
(often done with laser triming of the material), and than all the
resistors are made of the same material, and all in a very small
space, thus same temperature and other conditions...


** Massive red herring.



........... Phil

"dan lavry"
Pooh Bear
I suspect that 0.1% resistors are generally manufactured on a special
line. They are a specilaised high
cost item. In any event, the demand for them is inadequate to exhaust
those 1% types that would be 0.1%.



When I brought that issue about .1% being tighter tolerance for better
common mode rejection,


** Which, as I demonstrated with real tests, is a total fallacy.


I did mention the tempco which is temperature
coefficient, and I pointed at 25ppm .1% as a "reasonable" way to go.


** It is a pedantic fool's idea.


Let me point out that I my sugestion to invest the couple of bucks was
reasonable,


** It is a pedantic fool's idea.


I could have "gone crazy" to sugest some really expansive parts


** That would be a pedantic ****wit's idea.



Those that understood my comments about initial matching and tracking
of materials on the same substrate can now understand how resistor
based DAC's are designed. It is the same thing - good initial matching
(often done with laser triming of the material), and than all the
resistors are made of the same material, and all in a very small
space, thus same temperature and other conditions...


** Massive red herring.



........... Phil

On Sat, 4 Sep 2004 14:35:48 -0400, "Arny Krueger"
wrote:

"Mike Rivers" wrote in message
news:znr1094259452k@trad
In article
writes:

Are you suggesting that they select the tempcos as well? or that it
always just happens that if a resistor is very close to its design
value, it automatically has a tighter tempco? Or that the makers lie
about the specs? Or have you just managed to find 0.1% resistors with
bad tempcos?

I have never been concerned with "tempcos" (I assume that's Australian
for temperature coefficient) for a device that will always be used at
room temperature.

Not to mention the fact that if the components of an attenuator are kept in
reasonbly close thermal contact with each other, their ratio, which sets the
attenuation, will remain the same, even if their values change with
temperature.

That's very true, and certainly for the mic attenuator and many other
applications, tempcos would not be an issue. But my comment was
directed at Mike's assertion that 0.1% resistors were simply selected
from 1% production, which I still don't believe to be true.

Tony (remove the "_" to reply by email)

On Sat, 4 Sep 2004 14:35:48 -0400, "Arny Krueger"
wrote:

"Mike Rivers" wrote in message
news:znr1094259452k@trad
In article
writes:

Are you suggesting that they select the tempcos as well? or that it
always just happens that if a resistor is very close to its design
value, it automatically has a tighter tempco? Or that the makers lie
about the specs? Or have you just managed to find 0.1% resistors with
bad tempcos?

I have never been concerned with "tempcos" (I assume that's Australian
for temperature coefficient) for a device that will always be used at
room temperature.

Not to mention the fact that if the components of an attenuator are kept in
reasonbly close thermal contact with each other, their ratio, which sets the
attenuation, will remain the same, even if their values change with
temperature.

That's very true, and certainly for the mic attenuator and many other
applications, tempcos would not be an issue. But my comment was
directed at Mike's assertion that 0.1% resistors were simply selected
from 1% production, which I still don't believe to be true.

Tony (remove the "_" to reply by email)

On 4 Sep 2004 07:11:59 -0400, (Mike Rivers) wrote:
In article writes:

Are you suggesting that they select the tempcos as well? or that it
always just happens that if a resistor is very close to its design
value, it automatically has a tighter tempco? Or that the makers lie
about the specs? Or have you just managed to find 0.1% resistors with
bad tempcos?

I have never been concerned with "tempcos" (I assume that's Australian
for temperature coefficient) for a device that will always be used at
room temperature.

Yes, it's short for temperature coefficient, but not just Australian,
and yes, if you are only ever going to use a device at room
temperature, and you can select devices at the same "room
temperature", you don't need to be concerned with tempco. If you've
never considered tempcos, I guess I can see how you might think that
makers might select the "good" ones and rate them at 0.1%. But when
you think about the other "stability" ratings, that's just
impractical.

One other subtle point- some metal film resistors might have
+/-25ppm/K, others might have +150ppm/K nominal tempco, which doesn't
necessarily make them "bad", in fact it can be quite good in a tuned
circuit with a cap whose tempco - -150ppm/K.

Isn't the temperature coefficient primarily a function of the material
the resitor is made of? Unless (for example) there are no 0.1% metal
film resistors because the temperature coefficient can't support that
tolerance.

The material, the recipe tolerances, the precision of its preparation,
the precision of the control of thermal profiles, the care taken in
trimming, the lead attachment method, any annealing processes used,
etc, etc. As in everything, you can always find ways to do essentially
the same thing a little faster and/or cheaper, but you often pay for
it in some other way. If you want to rate a resistor as 0.1% with
similarly tighter tolerances for tempco, long-term drift, stability
under load and other drifts, you don't get to take so many short cuts.

Educate me here, don't just challange my statement. And explain, don't
just point me to a web site. That way I'll be confident that you know
what you're saying.

This IS only a public forum, in which no-one is really obliged to
always be an expert, or to not make mistakes, or even the occasional
major faux pas. We all just express our opinions. And occasionally
some of us may become a little too caught up in the debate. I'd like
to be able to expound the virtues of every resistor manufacturing
process used around the world, but I can't do that, or even suggest a
web site. But I CAN tell you that when ever I need to compare two
similar but different components, the data sheets usually have
everything I really need to know. Even the lowly resistor has quite a
story to tell; and don't get me started on capacitors...

Tony (remove the "_" to reply by email)

On 4 Sep 2004 07:11:59 -0400, (Mike Rivers) wrote:
In article writes:

Are you suggesting that they select the tempcos as well? or that it
always just happens that if a resistor is very close to its design
value, it automatically has a tighter tempco? Or that the makers lie
about the specs? Or have you just managed to find 0.1% resistors with
bad tempcos?

I have never been concerned with "tempcos" (I assume that's Australian
for temperature coefficient) for a device that will always be used at
room temperature.

Yes, it's short for temperature coefficient, but not just Australian,
and yes, if you are only ever going to use a device at room
temperature, and you can select devices at the same "room
temperature", you don't need to be concerned with tempco. If you've
never considered tempcos, I guess I can see how you might think that
makers might select the "good" ones and rate them at 0.1%. But when
you think about the other "stability" ratings, that's just
impractical.

One other subtle point- some metal film resistors might have
+/-25ppm/K, others might have +150ppm/K nominal tempco, which doesn't
necessarily make them "bad", in fact it can be quite good in a tuned
circuit with a cap whose tempco - -150ppm/K.

Isn't the temperature coefficient primarily a function of the material
the resitor is made of? Unless (for example) there are no 0.1% metal
film resistors because the temperature coefficient can't support that
tolerance.

The material, the recipe tolerances, the precision of its preparation,
the precision of the control of thermal profiles, the care taken in
trimming, the lead attachment method, any annealing processes used,
etc, etc. As in everything, you can always find ways to do essentially
the same thing a little faster and/or cheaper, but you often pay for
it in some other way. If you want to rate a resistor as 0.1% with
similarly tighter tolerances for tempco, long-term drift, stability
under load and other drifts, you don't get to take so many short cuts.

Educate me here, don't just challange my statement. And explain, don't
just point me to a web site. That way I'll be confident that you know
what you're saying.

This IS only a public forum, in which no-one is really obliged to
always be an expert, or to not make mistakes, or even the occasional
major faux pas. We all just express our opinions. And occasionally
some of us may become a little too caught up in the debate. I'd like
to be able to expound the virtues of every resistor manufacturing
process used around the world, but I can't do that, or even suggest a
web site. But I CAN tell you that when ever I need to compare two
similar but different components, the data sheets usually have
everything I really need to know. Even the lowly resistor has quite a
story to tell; and don't get me started on capacitors...

Tony (remove the "_" to reply by email)

On Sun, 05 Sep 2004 15:49:31 +1000, Tony wrote:

On Sat, 4 Sep 2004 14:35:48 -0400, "Arny Krueger"
wrote:

"Mike Rivers" wrote in message
news:znr1094259452k@trad
In article
writes:

Are you suggesting that they select the tempcos as well? or that it
always just happens that if a resistor is very close to its design
value, it automatically has a tighter tempco? Or that the makers lie
about the specs? Or have you just managed to find 0.1% resistors with
bad tempcos?

I have never been concerned with "tempcos" (I assume that's Australian
for temperature coefficient) for a device that will always be used at
room temperature.

Not to mention the fact that if the components of an attenuator are kept in
reasonbly close thermal contact with each other, their ratio, which sets the
attenuation, will remain the same, even if their values change with
temperature.

That's very true, and certainly for the mic attenuator and many other
applications, tempcos would not be an issue. But my comment was
directed at Mike's assertion that 0.1% resistors were simply selected
from 1% production, which I still don't believe to be true.

Tony (remove the "_" to reply by email)

You are right - it certainly isn't true. If it were, all resistors
other than the very best would have a bimodal distribution - a hole in
the middle where the best ones had been removed. They don't.

Typically resistor tolerance describes the width of the three-sigma
level of distribution, and that distribution tends to be normal. In
other words, resistors are "made" to the appropriate tolerance.

d

Pearce Consulting
http://www.pearce.uk.com

On Sun, 05 Sep 2004 15:49:31 +1000, Tony wrote:

On Sat, 4 Sep 2004 14:35:48 -0400, "Arny Krueger"
wrote:

"Mike Rivers" wrote in message
news:znr1094259452k@trad
In article
writes:

Are you suggesting that they select the tempcos as well? or that it
always just happens that if a resistor is very close to its design
value, it automatically has a tighter tempco? Or that the makers lie
about the specs? Or have you just managed to find 0.1% resistors with
bad tempcos?

I have never been concerned with "tempcos" (I assume that's Australian
for temperature coefficient) for a device that will always be used at
room temperature.

Not to mention the fact that if the components of an attenuator are kept in
reasonbly close thermal contact with each other, their ratio, which sets the
attenuation, will remain the same, even if their values change with
temperature.

That's very true, and certainly for the mic attenuator and many other
applications, tempcos would not be an issue. But my comment was
directed at Mike's assertion that 0.1% resistors were simply selected
from 1% production, which I still don't believe to be true.

Tony (remove the "_" to reply by email)

You are right - it certainly isn't true. If it were, all resistors
other than the very best would have a bimodal distribution - a hole in
the middle where the best ones had been removed. They don't.

Typically resistor tolerance describes the width of the three-sigma
level of distribution, and that distribution tends to be normal. In
other words, resistors are "made" to the appropriate tolerance.

d

Pearce Consulting
http://www.pearce.uk.com