"Standard parts" for rf amps?
 

On Tue, 22 Nov 2005 12:20:25 +0100, "Henry Kiefer"
wrote:

Thank Mark!

MC and Maxim is not new to me. I thought on list of cheap, easy getting
parts with reasonable performance.

- Henry

Minicircuits the MIMICs and mixers are of most interest and there
are inexpensive versions in the line up. What both offer is well
characterized RF 50 ohm port parts that behave well in circuits.
For example you can build a DBM, can you say for sure what the
characteristics will be from say 2-500mhz?

Other useful parts from there are VCOs and RF transformers.


Allison

"Standard parts" for rf amps?
 

"Henry Kiefer" wrote in message
...

Ok Jim. You're asked about the stray components hidden in the real
parts/circuit.

There is nothing hidden at all. Strays are a part of the real life at VHF
and above. Sometimes strays have more effect than the part itself.



If you take the values from the datasheets for die and/or
package or give it a guess if not available, and add values for the pcb
lines, you have a good simulation result.

Horsefeathers. What "values" do you add for the pcb lines?



The pcb by itself is not of much
interest if you simulate not much over 100MegHz. Even starting a
oscillator
with thermal noise is possible!
Sure, you must know what you're doing :-)

As with anything in the world.



Spice by itself can simulate almost all if you have the right models
included - even if you wanna simulate a mechanical system. There is no
frequency limit at least to 100GHz.

The problem is getting the right models, which is an art in and of itself.




Do you had bad result doing Spice?

To come back to your question: Spice will give you better results than
your
real circuit! You can connect a probe with no interaction to the circuit!
Try this with your real parts! Don't forget the component variations in
real
circuits coming from the manufacturing processes of the parts.

Again I ask the question: What does Spice say about a 1000 pf capacitor
with 1" total lead length at 150 MHz.? Don't give me the BS about Spice
being "better than a real circuit". Until you can give me a ferrite slug on
one end of a toothpick and a brass slug on the other end to increase or
decrease inductance in Spice, the physical circuit is the "real" circuit.

Jim

"Standard parts" for rf amps?
 

RST Engineering wrote:


"Henry Kiefer" wrote in message
...

Ok Jim. You're asked about the stray components hidden in the real
parts/circuit.

There is nothing hidden at all. Strays are a part of the real life at VHF
and above. Sometimes strays have more effect than the part itself.
You have to put the strays into the spice model if you want an accurate
result, just like you would have to if you work it out with a pencil and
paper.

If you take the values from the datasheets for die and/or
package or give it a guess if not available, and add values for the pcb
lines, you have a good simulation result.

Horsefeathers. What "values" do you add for the pcb lines?

You can work out the inductance, resistance (including skin effect) and
capacitance of arbitrary structures using the free programs FastCap and
FastHenry, though these consider inductive and capacitive effects
separately so you have to break the problem into several parts sometimes,
e.g. if the structure is a significant fraction of a wavelength in size.
Constructing the model files can be tedious, and you may end up having to
write some code to automate it. You can view the model files using a
viewer program which used to be and hopefully still is available from
www.fastfieldsolvers.com You could also buy one of the commercial
full-wave solvers if you have more spare money than time.

The pcb by itself is not of much
interest if you simulate not much over 100MegHz. Even starting a
oscillator
with thermal noise is possible!
Sure, you must know what you're doing :-)

As with anything in the world.



Spice by itself can simulate almost all if you have the right models
included - even if you wanna simulate a mechanical system. There is no
frequency limit at least to 100GHz.

The problem is getting the right models, which is an art in and of itself.
Right but you have to do that one way or another anyway even if the model is
inside your head, unless you design purely by trial and error.

Do you had bad result doing Spice?

To come back to your question: Spice will give you better results than
your
real circuit! You can connect a probe with no interaction to the circuit!
Try this with your real parts! Don't forget the component variations in
real
circuits coming from the manufacturing processes of the parts.

Again I ask the question: What does Spice say about a 1000 pf capacitor
with 1" total lead length at 150 MHz.? Don't give me the BS about Spice
being "better than a real circuit". Until you can give me a ferrite slug
on one end of a toothpick and a brass slug on the other end to increase or
decrease inductance in Spice, the physical circuit is the "real" circuit.

Jim
The real advantage of spice is in situations where trial and error is more
expensive than getting it right the first time (where you can justify
spending a long time making good models). He does also have a point about
being able to probe components inside a circuit that you could never probe
on a real one due to loading effects of the real probe. It seems to me
that your real objection is not with computer simulation of circuits, but
rather with poor models for components. Fair enough, garbage in garbage
out, but I would consider using bad models to be a form of 'user error'.

Chris

"Standard parts" for rf amps?
 

In article ,
RST Engineering wrote:

"Henry Kiefer" wrote in message
...

Ok Jim. You're asked about the stray components hidden in the real
parts/circuit.

There is nothing hidden at all. Strays are a part of the real life at VHF
and above. Sometimes strays have more effect than the part itself.

Heck, that can happen before you hit 1MHz. Check out the ESL and ESR of a
47,000uF capacitor some time.

--
--
forging knowledge

"Standard parts" for rf amps?
 

Ken Smith wrote:

Heck, that can happen before you hit 1MHz. Check out the ESL and ESR of a
47,000uF capacitor some time.


How about dielectric absorption ("soak")? I first encountered this in an
electrolytic capacitor being used for generating a sweep of about a
minute duration. The capacitor had to be kept shorted for several
seconds between sweeps in order for the charge to adequately empty, an
equivalent frequency of less than one Hz.

Roy Lewallen, W7EL

"Standard parts" for rf amps?
 

"Ken Smith" wrote in message
...
Heck, that can happen before you hit 1MHz. Check out the ESL and ESR of a
47,000uF capacitor some time.

For some of the university work I've done, some of our reference papers are
from people modeling power lines as full-fledged transmission lines that are
many wavelengths long at 50Hz. No simple equivalent circuit model works for
them either!

"Standard parts" for rf amps?
 

"Joel Kolstad" wrote in message
...
For some of the university work I've done, some of our reference papers are
from people modeling power lines as full-fledged transmission lines that are
many wavelengths long at 50Hz. No simple equivalent circuit model works for
them either!

OK, I did get a little over-excited there -- it should say "that are
significant fractions of a wavelength long," (e.g., more than 1/6th) such that
you can't use a simple circuit model for them.

"Standard parts" for rf amps?
 

On Mon, 21 Nov 2005 23:55:26 -0800, "RST Engineering"
wrote:

Well, just for starters, what does Spice say about a 1000 pf capacitor with
either ½" lead lengths or a total of 1" of PCB trace at 150 MHz.?

Jim



When they matter, just poke in the appropriate series L or a bit of
transmission line. I use Spice now and then to sim picosecond stuff.
Even if the simulation isn't highly accurate, it helps train your
instincts, shows you which parasitics will have which effects, so when
you build the real thing you have a jump on the complexity.

Hell, 150 MHz is slow.

John

"Standard parts" for rf amps?
 

On Tue, 22 Nov 2005 10:05:02 -0800, "RST Engineering"
wrote:


"Henry Kiefer" wrote in message
...

Ok Jim. You're asked about the stray components hidden in the real
parts/circuit.

There is nothing hidden at all. Strays are a part of the real life at VHF
and above. Sometimes strays have more effect than the part itself.



If you take the values from the datasheets for die and/or
package or give it a guess if not available, and add values for the pcb
lines, you have a good simulation result.

Horsefeathers. What "values" do you add for the pcb lines?



Lumped L and C if things are slow, or a hunk of transmission line when
things get fast. Works fine.

John

"Standard parts" for rf amps?
 

Joel Kolstad wrote:
"Joel Kolstad" wrote in message
...

For some of the university work I've done, some of our reference papers are
from people modeling power lines as full-fledged transmission lines that are
many wavelengths long at 50Hz. No simple equivalent circuit model works for
them either!


OK, I did get a little over-excited there -- it should say "that are
significant fractions of a wavelength long," (e.g., more than 1/6th) such that
you can't use a simple circuit model for them.


That's interesting -- SPICE has models for (lossless) transmission
lines, and lines of that length can also be adequately modeled with a
few LC sections. I can imagine line loss could be tricky, though. I've
modeled skin effect loss in the process of designing time-domain
circuits to compensate for delay line loss over a wide time range (and
therefore broad frequency range). But loss due to interaction of the
fields with ground underneath the wires might be more difficult. At
least you've only got one frequency to deal with -- unless harmonic
content is high enough to worry about. Perhaps you've also got to deal
with loss due to radiation?

Roy Lewallen, W7EL