Yo Larry

Jim Thompson · Mar 26, 2005

On Fri, 25 Mar 2005 23:48:59 -0800, Robert Monsen
<rcsurname@comcast.net> wrote:

Jim Thompson wrote:
On Fri, 25 Mar 2005 16:24:31 -0800, Robert Monsen
[snip]
I would imagine that LTspice can use my macros just ducky.

My experience is that neglecting impedances can shoot you in the foot.

...Jim Thompson

There is an LTSpice example for doing the current/voltage stimulation
thing you suggest in your paper. It actually gives very different
results from the big inductor.

480)460-2350 | |
| E-mail Address at Website Fax

Active8 · Mar 27, 2005

On Fri, 25 Mar 2005 20:40:57 GMT, Genome wrote:

"Larry Brasfield" <donotspam_larry_brasfield@hotmail.com> wrote in
message news:xTX0e.13$aL5.327@news.uswest.net...
"Genome" <ilike_spam@yahoo.co.uk> wrote in message
news:ttS0e.3931$ME3.649@newsfe1-gui.ntli.net...
[Stuff re games and loop gain cut.]
Hmmm, a little bit of a misunderstanding here. You are quite right
though, I am having a prod.

I would just as soon not play that game.

but

The closed loop response, when stable, can be
used as a measure of stability margin. Verifying
the absence of excessive peaking, or dips and
wobbles that occur over a narrow frequency
range, shows that no poles are very close to
the imaginary axis relative to their real part.
For simple feedback systems, where the loop
gain is relatively simple and has the common
narrowbanding compensation, that check is
enough.
--
--Larry Brasfield
email: donotspam_larry_brasfield@hotmail.com
Above views may belong only to me.

Interesting.

Perhaps you'd like to explain further.

If you google the hell out of control systems, there's at least one
paper out there I've read that discusses using closed loop response
to determine stability. It may have been based more on empirical
methods than theory and analysis, the latter of which I found more
interesting.
--
Best Regards,
Mike

Fred Bloggs · Mar 27, 2005

Active8 wrote:

On Fri, 25 Mar 2005 20:40:57 GMT, Genome wrote:

"Larry Brasfield" <donotspam_larry_brasfield@hotmail.com> wrote in
message news:xTX0e.13$aL5.327@news.uswest.net...

"Genome" <ilike_spam@yahoo.co.uk> wrote in message
news:ttS0e.3931$ME3.649@newsfe1-gui.ntli.net...
[Stuff re games and loop gain cut.]

Hmmm, a little bit of a misunderstanding here. You are quite right
though, I am having a prod.

I would just as soon not play that game.

but

The closed loop response, when stable, can be
used as a measure of stability margin. Verifying
the absence of excessive peaking, or dips and
wobbles that occur over a narrow frequency
range, shows that no poles are very close to
the imaginary axis relative to their real part.
For simple feedback systems, where the loop
gain is relatively simple and has the common
narrowbanding compensation, that check is
enough.
--
--Larry Brasfield
email: donotspam_larry_brasfield@hotmail.com
Above views may belong only to me.

Interesting.

Perhaps you'd like to explain further.

If you google the hell out of control systems, there's at least one
paper out there I've read that discusses using closed loop response
to determine stability. It may have been based more on empirical
methods than theory and analysis, the latter of which I found more
interesting.

This is all very elementary, like most of Brasfield's drivel, where it
is well-known that fundamentally the -3dB point of the closed-loop
response corresponds to the 0dB crossover of the open loop, and various
degrees of closed-loop peaking can be used to compute open-loop phase
and gain margin- and of course vice versa in that the open-loop
characterization predicts the closed-loop response.

John Larkin · Mar 27, 2005

On Sat, 26 Mar 2005 21:43:10 -0500, Active8 <reply2group@ndbbm.net>
wrote:

If you google the hell out of control systems, there's at least one
paper out there I've read that discusses using closed loop response
to determine stability.

Of course. If the closed-loop response is stable, the loop is stable.

John

Active8 · Mar 27, 2005

On Sat, 26 Mar 2005 21:09:20 -0800, John Larkin wrote:

On Sat, 26 Mar 2005 21:43:10 -0500, Active8 <reply2group@ndbbm.net
wrote:

If you google the hell out of control systems, there's at least one
paper out there I've read that discusses using closed loop response
to determine stability.

Of course. If the closed-loop response is stable, the loop is stable.

Until something changes. The peaking of the closed-loop response is

one thing I remember being used as an indicator of one margin or the
other.
--
Best Regards,
Mike

Larry Brasfield · Mar 27, 2005

"Active8" <reply2group@ndbbm.net> wrote in message
news:x68ld7tbglxu$.dlg@ID-222894.news.individual.net...

On Sat, 26 Mar 2005 21:09:20 -0800, John Larkin wrote:
On Sat, 26 Mar 2005 21:43:10 -0500, Active8 <reply2group@ndbbm.net
wrote:

If you google the hell out of control systems, there's at least one
paper out there I've read that discusses using closed loop response
to determine stability.

Of course. If the closed-loop response is stable, the loop is stable.

Until something changes. The peaking of the closed-loop response is
one thing I remember being used as an indicator of one margin or the
other.

Yep. That's why people like to see the phase margin
and gain margin, both characteristics of the loop gain.
When either of those goes bad, or near so, the closed
loop response poles end up in the RHP or too near it.
Either condition shows up in the frequency response,
and for simple systems, obviously so.

The information content equivalence, provided that
one can know, adequately estimate, or observe the
relevant internal variable G (loosely, "open loop gain")
is trivially apparent when one views the relevant math:
The equation for closed loop gain
Acl = G / (1 + GH)
can be rearranged to isolate loop gain
GH = G/Acl - 1
It is not hard to imagine this principle applied to the
result of a SPICE simulation, where those variables
_are_ visible. So, the old tricks people have used to
make loop gain easier to see in real hardware look
like a strange effort when a little arithmetic done on
an unmodified system so readily gets to the objective.

Getting to Genome's point, he imagined that he found
a contradiction and posted material to "support" that.
To that, I can only say there are no real winners in
that sort of game, but if he wants to get his jollies by
being declared "winner", he may consider it done.

--
--Larry Brasfield
email: donotspam_larry_brasfield@hotmail.com
Above views may belong only to me.

Fred Bloggs · Mar 27, 2005

Larry Brasfield wrote:

"Active8" <reply2group@ndbbm.net> wrote in message
news:x68ld7tbglxu$.dlg@ID-222894.news.individual.net...

On Sat, 26 Mar 2005 21:09:20 -0800, John Larkin wrote:

On Sat, 26 Mar 2005 21:43:10 -0500, Active8 <reply2group@ndbbm.net
wrote:

If you google the hell out of control systems, there's at least one
paper out there I've read that discusses using closed loop response
to determine stability.

Of course. If the closed-loop response is stable, the loop is stable.

Until something changes. The peaking of the closed-loop response is
one thing I remember being used as an indicator of one margin or the
other.

Yep. That's why people like to see the phase margin
and gain margin, both characteristics of the loop gain.
When either of those goes bad, or near so, the closed
loop response poles end up in the RHP or too near it.
Either condition shows up in the frequency response,
and for simple systems, obviously so.

The information content equivalence, provided that
one can know, adequately estimate, or observe the
relevant internal variable G (loosely, "open loop gain")
is trivially apparent when one views the relevant math:
The equation for closed loop gain
Acl = G / (1 + GH)
can be rearranged to isolate loop gain
GH = G/Acl - 1
It is not hard to imagine this principle applied to the
result of a SPICE simulation, where those variables
_are_ visible. So, the old tricks people have used to
make loop gain easier to see in real hardware look
like a strange effort when a little arithmetic done on
an unmodified system so readily gets to the objective.

Getting to Genome's point, he imagined that he found
a contradiction and posted material to "support" that.
To that, I can only say there are no real winners in
that sort of game, but if he wants to get his jollies by
being declared "winner", he may consider it done.

Unfortunately for you, the little s-domain block diagram does not work
too well because the *loaded* and interacting transfer functions are
generally not known. Seeing as how you can't even come up with a
credible design that works at DC, you would be the last person to listen
to about any dynamic performance prediction.

Larry Brasfield · Mar 27, 2005

"Fred Bloggs" <nospam@nospam.com> wrote in
message news:4246B70D.1000602@nospam.com...

Larry Brasfield wrote:
"Active8" <reply2group@ndbbm.net> wrote in message
news:x68ld7tbglxu$.dlg@ID-222894.news.individual.net...

On Sat, 26 Mar 2005 21:09:20 -0800, John Larkin wrote:

On Sat, 26 Mar 2005 21:43:10 -0500, Active8 <reply2group@ndbbm.net
wrote:

If you google the hell out of control systems, there's at least one
paper out there I've read that discusses using closed loop response
to determine stability.

Of course. If the closed-loop response is stable, the loop is stable.

Until something changes. The peaking of the closed-loop response is
one thing I remember being used as an indicator of one margin or the
other.

Yep. That's why people like to see the phase margin
and gain margin, both characteristics of the loop gain.
When either of those goes bad, or near so, the closed
loop response poles end up in the RHP or too near it.
Either condition shows up in the frequency response,
and for simple systems, obviously so.

The information content equivalence, provided that
one can know, adequately estimate, or observe the
relevant internal variable G (loosely, "open loop gain")
is trivially apparent when one views the relevant math:
The equation for closed loop gain
Acl = G / (1 + GH)
can be rearranged to isolate loop gain
GH = G/Acl - 1
It is not hard to imagine this principle applied to the
result of a SPICE simulation, where those variables
_are_ visible. So, the old tricks people have used to
make loop gain easier to see in real hardware look
like a strange effort when a little arithmetic done on
an unmodified system so readily gets to the objective.

Getting to Genome's point, he imagined that he found
a contradiction and posted material to "support" that.
To that, I can only say there are no real winners in
that sort of game, but if he wants to get his jollies by
being declared "winner", he may consider it done.

[Irrelevant taunt cut.]

the little s-domain block diagram does not work too well because the *loaded* and interacting transfer functions are generally not
known.

That's a bit of an overstatement. In many circuits, they are readily
separated to an adequate degree. In the rest, the loading effects
*can be known* by instrumenting the interaction. If you want to
propose and publish a feedback circuit LTSpice simulation where
you believe this to be impossible, I'll be happy to show you how to
do it. Just keep it to one loop with defined input and output and
don't get carried away trying to entangle G and H. (I am willing
to grant right now that it is possible to make the G/H separation
arbitrarily difficult. But such situations are generally contrived.)

[Irrelevant crap cut.]
--
--Larry Brasfield
email: donotspam_larry_brasfield@hotmail.com
Above views may belong only to me.

Jim Thompson · Mar 27, 2005

On Sat, 26 Mar 2005 21:09:20 -0800, John Larkin
<jjSNIPlarkin@highTHISlandPLEASEtechnology.XXX> wrote:

On Sat, 26 Mar 2005 21:43:10 -0500, Active8 <reply2group@ndbbm.net
wrote:

If you google the hell out of control systems, there's at least one
paper out there I've read that discusses using closed loop response
to determine stability.

Of course. If the closed-loop response is stable, the loop is stable.

John

You also have to be on the lookout for loop misbehavior as you
power-up. Sometimes I've seen systems that lost their phase-margin at
certain voltages, and then couldn't break free of the oscillation.

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice

480)460-2350 | |
| E-mail Address at Website Fax

Fred Bloggs · Mar 27, 2005

Larry Brasfield wrote:

"Fred Bloggs" <nospam@nospam.com> wrote in
message news:4246B70D.1000602@nospam.com...

Larry Brasfield wrote:

"Active8" <reply2group@ndbbm.net> wrote in message
news:x68ld7tbglxu$.dlg@ID-222894.news.individual.net...

On Sat, 26 Mar 2005 21:09:20 -0800, John Larkin wrote:

On Sat, 26 Mar 2005 21:43:10 -0500, Active8 <reply2group@ndbbm.net
wrote:

If you google the hell out of control systems, there's at least one
paper out there I've read that discusses using closed loop response
to determine stability.

Of course. If the closed-loop response is stable, the loop is stable.

Until something changes. The peaking of the closed-loop response is
one thing I remember being used as an indicator of one margin or the
other.

Yep. That's why people like to see the phase margin
and gain margin, both characteristics of the loop gain.
When either of those goes bad, or near so, the closed
loop response poles end up in the RHP or too near it.
Either condition shows up in the frequency response,
and for simple systems, obviously so.

The information content equivalence, provided that
one can know, adequately estimate, or observe the
relevant internal variable G (loosely, "open loop gain")
is trivially apparent when one views the relevant math:
The equation for closed loop gain
Acl = G / (1 + GH)
can be rearranged to isolate loop gain
GH = G/Acl - 1
It is not hard to imagine this principle applied to the
result of a SPICE simulation, where those variables
_are_ visible. So, the old tricks people have used to
make loop gain easier to see in real hardware look
like a strange effort when a little arithmetic done on
an unmodified system so readily gets to the objective.

Getting to Genome's point, he imagined that he found
a contradiction and posted material to "support" that.
To that, I can only say there are no real winners in
that sort of game, but if he wants to get his jollies by
being declared "winner", he may consider it done.

[Irrelevant taunt cut.]

the little s-domain block diagram does not work too well because the *loaded* and interacting transfer functions are generally not
known.

That's a bit of an overstatement. In many circuits, they are readily
separated to an adequate degree. In the rest, the loading effects
*can be known* by instrumenting the interaction. If you want to
propose and publish a feedback circuit LTSpice simulation where
you believe this to be impossible, I'll be happy to show you how to
do it. Just keep it to one loop with defined input and output and
don't get carried away trying to entangle G and H. (I am willing
to grant right now that it is possible to make the G/H separation
arbitrarily difficult. But such situations are generally contrived.)

[Irrelevant crap cut.]

It is clear from your response that you are an inexperienced and
ignorant idiot. So you think you know more about circuit analysis than
the people who have been inventing all these clever techniques for the
past thirty years?-And let's not forget that all this work was done
without motivation to improve the process, it was just for fun. Right,
idiot? And why do anything to an "adequate degree" when it can be done
precisely. Every time you open your pompous and jackass mouth you
confirm what nearly everyone knows about you by now.

John Larkin · Mar 27, 2005

On Sun, 27 Mar 2005 08:42:12 -0700, Jim Thompson
<thegreatone@example.com> wrote:

On Sat, 26 Mar 2005 21:09:20 -0800, John Larkin
jjSNIPlarkin@highTHISlandPLEASEtechnology.XXX> wrote:

On Sat, 26 Mar 2005 21:43:10 -0500, Active8 <reply2group@ndbbm.net
wrote:

If you google the hell out of control systems, there's at least one
paper out there I've read that discusses using closed loop response
to determine stability.

Of course. If the closed-loop response is stable, the loop is stable.

John

You also have to be on the lookout for loop misbehavior as you
power-up. Sometimes I've seen systems that lost their phase-margin at
certain voltages, and then couldn't break free of the oscillation.

...Jim Thompson

Nonlinear oscillations are legion and, as far as I know, pretty much
immune to analytics. Even if you have the tools to analyze a mode,
first you have to discover (or imagine) the mode.

My favorite example of your "couldn't break free" was in fact a
powerup reset circuit that couldn't power up:

+5
|
+--------+
| |
r |
| c
+------b 2N2222
| e
| |
cap +---------> ttl schmitt
| |
| r
| |
| |
gnd gnd

where the base just wouldn't pull up! Seems the 2222 was oscillating
at 100 MHz.

John

Jim Thompson · Mar 27, 2005

On Sun, 27 Mar 2005 08:47:26 -0800, John Larkin
<jjSNIPlarkin@highTHISlandPLEASEtechnology.XXX> wrote:

On Sun, 27 Mar 2005 08:42:12 -0700, Jim Thompson
thegreatone@example.com> wrote:

On Sat, 26 Mar 2005 21:09:20 -0800, John Larkin
jjSNIPlarkin@highTHISlandPLEASEtechnology.XXX> wrote:

On Sat, 26 Mar 2005 21:43:10 -0500, Active8 <reply2group@ndbbm.net
wrote:

If you google the hell out of control systems, there's at least one
paper out there I've read that discusses using closed loop response
to determine stability.

Of course. If the closed-loop response is stable, the loop is stable.

John

You also have to be on the lookout for loop misbehavior as you
power-up. Sometimes I've seen systems that lost their phase-margin at
certain voltages, and then couldn't break free of the oscillation.

...Jim Thompson

Nonlinear oscillations are legion and, as far as I know, pretty much
immune to analytics. Even if you have the tools to analyze a mode,
first you have to discover (or imagine) the mode.

My favorite example of your "couldn't break free" was in fact a
powerup reset circuit that couldn't power up:

+5
|
+--------+
| |
r |
| c
+------b 2N2222
| e
| |
cap +---------> ttl schmitt
| |
| r
| |
| |
gnd gnd

where the base just wouldn't pull up! Seems the 2222 was oscillating
at 100 MHz.

John

480)460-2350 | |
| E-mail Address at Website Fax

Genome · Mar 27, 2005

"Larry Brasfield" <donotspam_larry_brasfield@hotmail.com> wrote in
message news:114d62v64ftgvff@news.supernews.com...

"Active8" <reply2group@ndbbm.net> wrote in message
news:x68ld7tbglxu$.dlg@ID-222894.news.individual.net...
On Sat, 26 Mar 2005 21:09:20 -0800, John Larkin wrote:
On Sat, 26 Mar 2005 21:43:10 -0500, Active8
reply2group@ndbbm.net
wrote:

If you google the hell out of control systems, there's at least
one
paper out there I've read that discusses using closed loop
response
to determine stability.

Of course. If the closed-loop response is stable, the loop is
stable.

Until something changes. The peaking of the closed-loop response
is
one thing I remember being used as an indicator of one margin or
the
other.

Yep. That's why people like to see the phase margin
and gain margin, both characteristics of the loop gain.
When either of those goes bad, or near so, the closed
loop response poles end up in the RHP or too near it.
Either condition shows up in the frequency response,
and for simple systems, obviously so.

The information content equivalence, provided that
one can know, adequately estimate, or observe the
relevant internal variable G (loosely, "open loop gain")
is trivially apparent when one views the relevant math:
The equation for closed loop gain
Acl = G / (1 + GH)
can be rearranged to isolate loop gain
GH = G/Acl - 1
It is not hard to imagine this principle applied to the
result of a SPICE simulation, where those variables
_are_ visible. So, the old tricks people have used to
make loop gain easier to see in real hardware look
like a strange effort when a little arithmetic done on
an unmodified system so readily gets to the objective.

Getting to Genome's point, he imagined that he found
a contradiction and posted material to "support" that.
To that, I can only say there are no real winners in
that sort of game, but if he wants to get his jollies by
being declared "winner", he may consider it done.

--
--Larry Brasfield
email: donotspam_larry_brasfield@hotmail.com
Above views may belong only to me.

No I didn't.

I said you were plotting closed loop gain and gave you a method for
plotting loop gain.

DNA

Harry Dellamano · Mar 27, 2005

"Jim Thompson" <thegreatone@example.com> wrote in message
news:khpd41pmbe562ktdl8a17tf3hvfe1nlj9k@4ax.com...

On Sun, 27 Mar 2005 08:47:26 -0800, John Larkin
jjSNIPlarkin@highTHISlandPLEASEtechnology.XXX> wrote:

On Sun, 27 Mar 2005 08:42:12 -0700, Jim Thompson
thegreatone@example.com> wrote:

On Sat, 26 Mar 2005 21:09:20 -0800, John Larkin
jjSNIPlarkin@highTHISlandPLEASEtechnology.XXX> wrote:

You also have to be on the lookout for loop misbehavior as you
power-up. Sometimes I've seen systems that lost their phase-margin at
certain voltages, and then couldn't break free of the oscillation.

...Jim Thompson

Nonlinear oscillations are legion and, as far as I know, pretty much
immune to analytics. Even if you have the tools to analyze a mode,
first you have to discover (or imagine) the mode.

My favorite example of your "couldn't break free" was in fact a
powerup reset circuit that couldn't power up:

+5
|
+--------+
| |
r |
| c
+------b 2N2222
| e
| |
cap +---------> ttl schmitt
| |
| r
| |
| |
gnd gnd

where the base just wouldn't pull up! Seems the 2222 was oscillating
at 100 MHz.

John

Ahhh! Long line (capacitance) on the emitter?

...Jim Thompson
--

Or about 1.0" of lead inductance in the base will yield a 250MHz
oscillator. Clean up the wiring and/or add some base resistance to D-Q the
circuit will fix the problem. Just had a tech spend a week trying to fix a
system with this hidden problem.
Regards,
Harry

Jim Thompson · Mar 27, 2005

On Sun, 27 Mar 2005 17:18:17 GMT, "Genome" <ilike_spam@yahoo.co.uk>
wrote:

"Larry Brasfield" <donotspam_larry_brasfield@hotmail.com> wrote in
message news:114d62v64ftgvff@news.supernews.com...
"Active8" <reply2group@ndbbm.net> wrote in message
news:x68ld7tbglxu$.dlg@ID-222894.news.individual.net...
On Sat, 26 Mar 2005 21:09:20 -0800, John Larkin wrote:
On Sat, 26 Mar 2005 21:43:10 -0500, Active8
reply2group@ndbbm.net
wrote:

If you google the hell out of control systems, there's at least
one
paper out there I've read that discusses using closed loop
response
to determine stability.

Of course. If the closed-loop response is stable, the loop is
stable.

Until something changes. The peaking of the closed-loop response
is
one thing I remember being used as an indicator of one margin or
the
other.

Yep. That's why people like to see the phase margin
and gain margin, both characteristics of the loop gain.
When either of those goes bad, or near so, the closed
loop response poles end up in the RHP or too near it.
Either condition shows up in the frequency response,
and for simple systems, obviously so.

The information content equivalence, provided that
one can know, adequately estimate, or observe the
relevant internal variable G (loosely, "open loop gain")
is trivially apparent when one views the relevant math:
The equation for closed loop gain
Acl = G / (1 + GH)
can be rearranged to isolate loop gain
GH = G/Acl - 1
It is not hard to imagine this principle applied to the
result of a SPICE simulation, where those variables
_are_ visible. So, the old tricks people have used to
make loop gain easier to see in real hardware look
like a strange effort when a little arithmetic done on
an unmodified system so readily gets to the objective.

Getting to Genome's point, he imagined that he found
a contradiction and posted material to "support" that.
To that, I can only say there are no real winners in
that sort of game, but if he wants to get his jollies by
being declared "winner", he may consider it done.

--
--Larry Brasfield
email: donotspam_larry_brasfield@hotmail.com
Above views may belong only to me.

No I didn't.

I said you were plotting closed loop gain and gave you a method for
plotting loop gain.

DNA

480)460-2350 | |
| E-mail Address at Website Fax

Active8 · Mar 27, 2005

On Sun, 27 Mar 2005 10:35:10 -0700, Jim Thompson wrote:

<snip>

Does anyone use Nichols Charts? I vaguely remember them from college
course work.

Would that be a useful automatic tool for PSpice?

It's still being taught. Dunno if anyone uses them in real life.
--
Best Regards,
Mike

John Larkin · Mar 27, 2005

On Sun, 27 Mar 2005 09:59:41 -0700, Jim Thompson
<thegreatone@example.com> wrote:

On Sun, 27 Mar 2005 08:47:26 -0800, John Larkin
jjSNIPlarkin@highTHISlandPLEASEtechnology.XXX> wrote:

On Sun, 27 Mar 2005 08:42:12 -0700, Jim Thompson
thegreatone@example.com> wrote:

On Sat, 26 Mar 2005 21:09:20 -0800, John Larkin
jjSNIPlarkin@highTHISlandPLEASEtechnology.XXX> wrote:

On Sat, 26 Mar 2005 21:43:10 -0500, Active8 <reply2group@ndbbm.net
wrote:

If you google the hell out of control systems, there's at least one
paper out there I've read that discusses using closed loop response
to determine stability.

Of course. If the closed-loop response is stable, the loop is stable.

John

You also have to be on the lookout for loop misbehavior as you
power-up. Sometimes I've seen systems that lost their phase-margin at
certain voltages, and then couldn't break free of the oscillation.

...Jim Thompson

Nonlinear oscillations are legion and, as far as I know, pretty much
immune to analytics. Even if you have the tools to analyze a mode,
first you have to discover (or imagine) the mode.

My favorite example of your "couldn't break free" was in fact a
powerup reset circuit that couldn't power up:

+5
|
+--------+
| |
r |
| c
+------b 2N2222
| e
| |
cap +---------> ttl schmitt
| |
| r
| |
| |
gnd gnd

where the base just wouldn't pull up! Seems the 2222 was oscillating
at 100 MHz.

John

Ahhh! Long line (capacitance) on the emitter?

...Jim Thompson

No, this was pretty tight. Wirebonds are all you need to make a 2222
(or most any other fastish ss NPN) emitter follower oscillate. As
Harry says, add a base resistor.

I recently did a current source, opamp closing the loop around a
BCX71. Like a plebe, I left out the base resistor, and it oscillated
big time... just the transistor, not the loop. The kluge was to use
one of those "digital transistor" things with internal resistors.

Oh, about open-loop response: if in a simulation you inject a signal
anywhere in the "closed" loop, just to have something to work with,
can't you just observe the vector ratio of A/B, where A and B are the
voltages at any two arbitrary nodes? That's the "open loop" transfer
function from B to A. After all, in Spice there's no noise, and if you
need a gain of 1e8 to resolve things, just do it.

John

Jim Thompson · Mar 27, 2005

On Sun, 27 Mar 2005 10:22:24 -0800, John Larkin
<jjSNIPlarkin@highTHISlandPLEASEtechnology.XXX> wrote:

[snip]

Oh, about open-loop response: if in a simulation you inject a signal
anywhere in the "closed" loop, just to have something to work with,
can't you just observe the vector ratio of A/B, where A and B are the
voltages at any two arbitrary nodes? That's the "open loop" transfer
function from B to A. After all, in Spice there's no noise, and if you
need a gain of 1e8 to resolve things, just do it.

John

480)460-2350 | |
| E-mail Address at Website Fax

Active8 · Mar 27, 2005

On Sun, 27 Mar 2005 11:39:41 -0700, Jim Thompson wrote:

On Sun, 27 Mar 2005 10:22:24 -0800, John Larkin
jjSNIPlarkin@highTHISlandPLEASEtechnology.XXX> wrote:

[snip]

Oh, about open-loop response: if in a simulation you inject a signal
anywhere in the "closed" loop, just to have something to work with,
can't you just observe the vector ratio of A/B, where A and B are the
voltages at any two arbitrary nodes? That's the "open loop" transfer
function from B to A. After all, in Spice there's no noise, and if you
need a gain of 1e8 to resolve things, just do it.

John

No. Read Middlebrook's paper to understand the issue.

That the paper at your site that goes with your loop cutter macros,
right?
--
Best Regards,
Mike

John Larkin · Mar 27, 2005

On Sun, 27 Mar 2005 11:39:41 -0700, Jim Thompson
<thegreatone@example.com> wrote:

On Sun, 27 Mar 2005 10:22:24 -0800, John Larkin
jjSNIPlarkin@highTHISlandPLEASEtechnology.XXX> wrote:

[snip]

Oh, about open-loop response: if in a simulation you inject a signal
anywhere in the "closed" loop, just to have something to work with,
can't you just observe the vector ratio of A/B, where A and B are the
voltages at any two arbitrary nodes? That's the "open loop" transfer
function from B to A. After all, in Spice there's no noise, and if you
need a gain of 1e8 to resolve things, just do it.

John

No. Read Middlebrook's paper to understand the issue.

...Jim Thompson

OK; say I have a gain stage or two, and draw a little dotted line
around that, with A being the input node and B being the output.
There's a sine wave at A and a resulting sine wave at B. How's this
section of the system aware that it's operating inside an open loop or
a closed one? How will its transfer function change?

John

Yo Larry

Jim Thompson

Guest

Active8

Guest

Fred Bloggs

Guest

John Larkin

Guest

Active8

Guest

Larry Brasfield

Guest

Fred Bloggs

Guest

Larry Brasfield

Guest

Jim Thompson

Guest

Fred Bloggs

Guest

John Larkin

Guest

Jim Thompson

Guest

Genome

Guest

Harry Dellamano

Guest

Jim Thompson

Guest

Active8

Guest

John Larkin

Guest

Jim Thompson

Guest

Active8

Guest

John Larkin

Guest

Log in

Welcome to EDABoard.com

Sponsor