Modelling a signal generator's output impedence

[Paul Burridge]

Hi all,

Possibly a dumb question, but let's say my signal generator has
outputs for 600 and 50 ohms. Can I model this in Spice by the simple
addition of a 600R or 50R series resistance to an AC voltage source?
And is this impedance in real life for all practical purposes purely
resistive across the generators entire frequency output range?

thanks,

p.
--

"Windows [n.], A thirty-two bit extension and GUI shell to a sixteen bit patch
to an eight bit operating system originally coded for a four bit
microprocessor and produced by a two bit company."

 

[Don Pearce]

On Fri, 03 Oct 2003 11:45:33 +0100, Paul Burridge
<pb@osiris.notthisbit.co.uk> wrote:

Hi all,

Possibly a dumb question, but let's say my signal generator has
outputs for 600 and 50 ohms. Can I model this in Spice by the simple
addition of a 600R or 50R series resistance to an AC voltage source?
And is this impedance in real life for all practical purposes purely
resistive across the generators entire frequency output range?

thanks,

p.

Yes, this is exactly how you model a signal generator. Put a voltage
source in series with a resistor. Make the voltage double what you
want to appear at the output when working into a properly matched
load.

The model works because this is exactly how signal generators are made
(I can assure you of this as I have designed them - a low output
impedance amplifier followed by the levelling detector, followed in
turn by a 50 ohm resistor).

d

_____________________________

http://www.pearce.uk.com

 

[Dale Chisholm]

LTSpice allows you to specify a source's impedance as part of it's definition.

Dale

Paul Burridge <pb@osiris.notthisbit.co.uk> wrote in message news:<ggkqnv0po7d4ne13l8k1muubk1qgo3bfla@4ax.com>...

Hi all,

Possibly a dumb question, but let's say my signal generator has
outputs for 600 and 50 ohms. Can I model this in Spice by the simple
addition of a 600R or 50R series resistance to an AC voltage source?
And is this impedance in real life for all practical purposes purely
resistive across the generators entire frequency output range?

thanks,

p.

 

[Paul Burridge]

On 3 Oct 2003 12:58:25 -0700, dtchisholm@yahoo.com (Dale Chisholm)
wrote:

LTSpice allows you to specify a source's impedance as part of it's definition.

Yes, I know. But that wasn't quite what I was getting at.
--

"Windows [n.], A thirty-two bit extension and GUI shell to a sixteen bit patch
to an eight bit operating system originally coded for a four bit
microprocessor and produced by a two bit company."

 

[Paul Burridge]

On Fri, 03 Oct 2003 11:52:47 +0100, Don Pearce
<dxoxnxaxlxd@pearce.uk.com> wrote:

Yes, this is exactly how you model a signal generator. Put a voltage
source in series with a resistor. Make the voltage double what you
want to appear at the output when working into a properly matched
load.

The model works because this is exactly how signal generators are made
(I can assure you of this as I have designed them - a low output
impedance amplifier followed by the levelling detector, followed in
turn by a 50 ohm resistor).

Thanks. What I still find a bit odd is that these generators' output
impedences are so relatively high. I mean, 600 ohms?? The 'ideal'
signal generator would have zero Zout, would it not? Whilst the
theoretical ideal is not possible to implement in practice, one would
have thought that Zouts of just a few ohms would be perfectly
feasible. So why are their outputs so current limited?
<baffled>
--

"Windows [n.], A thirty-two bit extension and GUI shell to a sixteen bit patch
to an eight bit operating system originally coded for a four bit
microprocessor and produced by a two bit company."

 

[Active8]

In article <pcprnv89es9es5c1mekfn0615o5ptn51l8@4ax.com>,
pb@osiris.notthisbit.co.uk says...

On Fri, 03 Oct 2003 11:52:47 +0100, Don Pearce
dxoxnxaxlxd@pearce.uk.com> wrote:

Yes, this is exactly how you model a signal generator. Put a voltage
source in series with a resistor. Make the voltage double what you
want to appear at the output when working into a properly matched
load.

The model works because this is exactly how signal generators are made
(I can assure you of this as I have designed them - a low output
impedance amplifier followed by the levelling detector, followed in
turn by a 50 ohm resistor).

Thanks. What I still find a bit odd is that these generators' output
impedences are so relatively high. I mean, 600 ohms?? The 'ideal'
signal generator would have zero Zout, would it not? Whilst the
theoretical ideal is not possible to implement in practice, one would
have thought that Zouts of just a few ohms would be perfectly
feasible. So why are their outputs so current limited?
baffled

Well, as you may know, RF is generally a 50 ohm deal, so you wouldn't

want your sig gen mismatched to the line or the circuit you're injecting
wouldn't get the same sig as that going into the coax.

600 ohm is standard audio line level impedance depending on who you ask.
680 ohms is also quite common. you do want your stereo amplifier to see
the same source impedance it will expect to see in real life, right?

it's a matter of convienience to have selectable impedances on a sig
gen. no need to add resistors.

mike

 

[Dale Chisholm]

(A bit of a SWAG on my part, but perhaps somebody whose hair is grayer
than mine will come along and correct the fuzziness between my ears.)

Standardized resistive input and output impedances for electronic
equipment were well established at least 75 years ago. I once tended a
radio station which still had a few pieces of audio test gear laying
around from the early 1930's, and 600 ohms seems to have been well
established by then. The 50 (and 75) ohm standards for RF equipment may
have come a bit later, around 1940 when coax cable replaced open wire
transmission line. But that doesn't really answer "Why?".

I suspect it's a convergence of several things. There's no one big
reason, but rather several small reasons. They were probably first
understood by the telephone industry, which was the driving force behind
a lot of early developments in electronics:

1. Active devices don't like to see reactive loads. Forcing (with the
addition of discrete resistance, if necessary) the generator's (or
amplifier's) internal impedance to a particular value reduces the
likelihood of damaging the generator or distorting the output signal if
the equipment is connected to a less-than-nice load. It also reduces the
possibility that the thing will oscillate at God-only-knows-what
frequency if presented with a reactive load. In this day of penny
transistors it's hard to believe that as recently as the 1960's a simple
table radio could have a complement (5, to be exact) of vacuum tubes
whose total cost exceeded the value of groceries that an individual
could carry to his car! In that environment the added output resistance
helped protect high-value components.

2. Early active circuits couldn't come anywhere near as close to zero
output impedance/infinite input impedance as we can today. For instance,
the role of feedback - ubiquitous in just about every modern circuit
with more than half a dozen parts - doesn't seem to have been well
understood until the late 1930's. Since (as the Preacher told me last
Sunday) nobody can be perfect, there's value in everybody behaving in
the same imperfect way.

3. Passive networks' frequency- and transient responses are similarly
much more predictable and nicely behaved when driven by, and/or
terminated with, resistive loads. A well-chosen value of source/load
resistance will suppress any tendancy toward resonance; unintended
resonance will affect the performance of the equipment or the accuracy
of the measurement.

4. Any practical passive signal processing network (filter, splitter,
mixer, transducer, etc) will have a measurable (neither zero nor
infinite) impedance at its input(s) and/or output(s). Once you accept
the fact that sources, in general, will not be zero impedance and loads
will generally not be infinite, it makes sense to standardize on some
particular value. If sources and loads are all standardized to the SAME
value, then any piece of equipment or any test instrument may be
inserted at any point without altering the behavior of the network in
some unintended way.

5. The particular values you cite - 600 ohms and 50 ohms - may have had
some justificationin the physical world. Like when you draw copper wire
to a diameter small enough to not be a massive load, but still large
enough to not stretch under it's own weight when supported at. say, 100
foot intervals (possibly about AWG18?); and space a pair of these far
enough apart that they won't tangle in the wind (about 3"?); then
the characteristic impedance is about 600 ohms. (This is about the
arrangement used for early telephone lines.) ANY signal coming out of
several miles of this transmission line will be:
a.) Much lower in amplitude than where it started;
b.) Have an apparent source impedance much closer
to 600 ohms, than to the source impedance of the
actual generator.
To minimize corruption of this weak signal by noise in the processing
circuit, the equipment it connects to should appear as a 600 ohm
resistive load.

6. The coaxial standards may not be as straightforward, but I seem to
recall once seeing a derivation of the total losses in a coaxial cable.
If the inner conductor is much smaller than the outer conductor, the
cable has a high characteristic impedance and the losses are dominated
by E-field losses in the dielectric. If the inner conductor is almost
the same size as the outer conductor, the characteristic impedance is
low and the losses are dominated by resistive (skin effect) losses in
the conductors themselves. The optimal (minimum total loss) arrangement
occurs when the dielectric and resistive losses are equal. If I recall
correctly, this occurs when the coax's impedance is about 50 ohms if the
dielectric is solid polyethylene. (Polyethylene was one of the first
plastic insulation materials that could be manufactured cheaply, circa
1940.) Or, if you construct a cable by supporting the center conductor
with a phenolic spacer every few inches, the minimum loss case
corresponds to a characteristic impedance of 75 ohms. I have no idea
why radar video (and eventually, all TV-related applications)
standardized on the 75 ohm versions and most other radio-related
applications use 50 ohms.

Is that more like the answer you're looking for?

Dale


Paul Burridge <pb@osiris.notthisbit.co.uk> wrote in message news:<ggkqnv0po7d4ne13l8k1muubk1qgo3bfla@4ax.com>...

Hi all,

Possibly a dumb question, but let's say my signal generator has
outputs for 600 and 50 ohms. Can I model this in Spice by the simple
addition of a 600R or 50R series resistance to an AC voltage source?
And is this impedance in real life for all practical purposes purely
resistive across the generators entire frequency output range?

thanks,

p.

 

[Paul Burridge]

On 3 Oct 2003 21:42:34 -0700, dtchisholm@yahoo.com (Dale Chisholm)
wrote:

[pretty exhaustive explanation snipped!]

Is that more like the answer you're looking for?

Very much so, Dale. And many thanks indeed for taking the trouble to
explain it at such length. I've copied the message over to permanent
storage for future reference!
--

"Windows [n.], A thirty-two bit extension and GUI shell to a sixteen bit patch
to an eight bit operating system originally coded for a four bit
microprocessor and produced by a two bit company."

 

[Active8]

[snip]

6. The coaxial standards may not be as straightforward, but I seem to
recall once seeing a derivation of the total losses in a coaxial cable.
If the inner conductor is much smaller than the outer conductor, the
cable has a high characteristic impedance and the losses are dominated
by E-field losses in the dielectric. If the inner conductor is almost
the same size as the outer conductor, the characteristic impedance is
low and the losses are dominated by resistive (skin effect) losses in
the conductors themselves. The optimal (minimum total loss) arrangement
occurs when the dielectric and resistive losses are equal. If I recall
correctly, this occurs when the coax's impedance is about 50 ohms if the
dielectric is solid polyethylene. (Polyethylene was one of the first
plastic insulation materials that could be manufactured cheaply, circa
1940.) Or, if you construct a cable by supporting the center conductor
with a phenolic spacer every few inches, the minimum loss case
corresponds to a characteristic impedance of 75 ohms. I have no idea
why radar video (and eventually, all TV-related applications)
standardized on the 75 ohm versions and most other radio-related
applications use 50 ohms.

good question. you know not all TV coax is disc type, right. was the
original stuff really disc?

most published formulas for coax impedance that i've seen have the
resistance term dropped and the dielectric constant left in.

FYI, the old crap was a bunch of discs that, after prepping the cable,
you could mash the disks inward to get the connector on. now it's called
fused disc. there's a plastic duct inside the aluminum sheath, a plastic
duct/coating around the center conductor, with fused discs in between.
Trilogy makes MC^2 which has a VP of .93. Commscope makes regular P3 and
Quantum Reach which has a better VP than P3 (closed-cell polyethylene)
IIRC it's .86 or better.

hard to say why 75 ohm was used, but if the old foam stuff had a lower
VP which is approximately 1/sqrt(u.e) or c/sqrt(ur.er) - IOW the VP is
dependant on the dielectric - then that means it had more loss than the
air dielectric or disc supported stuff.

so maybe the first 75 ohm cable was made with discs and once a bunch of
equipment was made, it was too late. eventually, when they needed to go
farther distances, cable makers started doing it with PE for cost
reasons. the goal then would have been to get close to disc performance
with foam.

that fused disc is some pretty good stuff. the quantum reach has good
points and bad.

http://www.trilogycoax.com/products_catv.html
http://www.trilogycoax.com/pdf/MC2Brochure.pdf

mike

Is that more like the answer you're looking for?

Dale


Paul Burridge <pb@osiris.notthisbit.co.uk> wrote in message news:<ggkqnv0po7d4ne13l8k1muubk1qgo3bfla@4ax.com>...
Hi all,

Possibly a dumb question, but let's say my signal generator has
outputs for 600 and 50 ohms. Can I model this in Spice by the simple
addition of a 600R or 50R series resistance to an AC voltage source?
And is this impedance in real life for all practical purposes purely
resistive across the generators entire frequency output range?

thanks,

p.

 

[JeffM]

and produced by a two bit company

that doesn't like one bit of competition.