Sensing small inductances

[bitrex]

On 8/27/19 10:07 AM, jlarkin@highlandsniptechnology.com wrote:

On Tue, 27 Aug 2019 07:08:09 +0100, piglet <erichpwagner@hotmail.com
wrote:

On 26/08/2019 22:26, John Larkin wrote:
On Mon, 26 Aug 2019 08:33:48 +0100, piglet <erichpwagner@hotmail.com
A problem with measuring inductance by making the DUT part of an
oscillator tank and measuring frequency is the square root relationship
works against you by compressing sensitivity.

Only by 2:1. And it's easy to measure frequency to a part per million.


Yes of course but in one of his post's the OP was complaining their
hardware had problems measuring small differences in frequency.

piglet

He could buy a cheap counter and get to 1 PPM and be done.

Well the issue is the inductances I wanna measure, in the single digital
microhenries, to within a nanohenry, say, have a Q of about 0.3 in the
low MHz.

when you can get them to work as part of a standard oscillator tank that
oscillates at all, an octave and a half below their self-resonant
frequency, the stability is poor. if you have say a 4uH resonating with
a 10n cap to get 5MHz a 1 nH difference in the L is only a few hundred
Hz shift. But the oscillator is drifting around by several kHz over 20
minutes

 

[Jeroen Belleman]

Steve Wilson wrote:

Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:

On 8/27/19 4:07 PM, Steve Wilson wrote:
Where do you measure the negative resistance and how do you control it?

You measure it by watching the follower oscillate with no external
feedback. Put a small pot in series with the base and watch where the
oscillation stops.

Cheers

Phil Hobbs

Thanks. This has been very illuminating. First, I find the term
negative resistance has nothing to do with the classical definition,
where an increasing voltage causes decreasing current.

Well, it's negative under some conditions in the small
signal analysis.

The input impedance of an emitter follower is something
like Zb + Ze + ZbZe, with Zb=a'/S // Z(Cbe), the impedance
from base to emitter and Ze=Re // Z(Ce), the emitter load
impedance. Under certain conditions, this ZbZe term can go
negative enough to make the overall input impedance negative.

I include below a simple LTspice AC equivalent circuit model
of an emitter follower. Plot V(in)/I(Rin) in Cartesian
coordinates and you'll see the real part go negative around
200MHz.

Jeroen Belleman

==============================
Version 4
SHEET 1 880 820
WIRE 240 128 240 96
WIRE -128 144 -192 144
WIRE -16 144 -48 144
WIRE 32 144 -16 144
WIRE 192 144 32 144
WIRE 32 192 32 144
WIRE 64 192 32 192
WIRE 176 192 144 192
WIRE 192 192 176 192
WIRE -192 208 -192 144
WIRE 32 288 32 192
WIRE 64 288 32 288
WIRE 176 288 176 192
WIRE 176 288 128 288
WIRE 208 288 176 288
WIRE 240 288 240 208
WIRE 240 288 208 288
WIRE 368 288 240 288
WIRE -192 304 -192 288
WIRE 240 352 240 288
WIRE 368 368 368 288
WIRE 240 496 240 432
WIRE 368 496 368 432
FLAG -192 304 0
FLAG -16 144 in
FLAG 240 496 0
FLAG 208 288 e
FLAG 240 96 0
FLAG 368 496 0
SYMBOL voltage -192 192 R0
WINDOW 123 24 124 Left 2
WINDOW 39 0 0 Left 2
SYMATTR Value2 AC 1
SYMATTR InstName V1
SYMATTR Value SINE(0 1 1k)
SYMBOL res -144 128 M90
WINDOW 0 0 56 VBottom 2
WINDOW 3 32 56 VTop 2
SYMATTR InstName Rin
SYMATTR Value 50
SYMBOL g2 240 224 M180
SYMATTR InstName G1
SYMATTR Value .4
SYMBOL res 224 336 R0
SYMATTR InstName R1
SYMATTR Value 100
SYMBOL res 160 176 R90
WINDOW 0 0 56 VBottom 2
WINDOW 3 32 56 VTop 2
SYMATTR InstName R3
SYMATTR Value 250
SYMBOL cap 384 432 R180
WINDOW 0 24 56 Left 2
WINDOW 3 24 8 Left 2
SYMATTR InstName C1
SYMATTR Value 10p
SYMBOL cap 128 272 R90
WINDOW 0 0 32 VBottom 2
WINDOW 3 32 32 VTop 2
SYMATTR InstName C2
SYMATTR Value 10p
TEXT -264 48 Left 2 !.ac dec 100 10 1G

 

[bitrex]

On 8/28/19 10:03 AM, dagmargoodboat@yahoo.com wrote:

On Tuesday, August 27, 2019 at 11:17:04 PM UTC-4, jla...@highlandsniptechnology.com wrote:
On Tue, 27 Aug 2019 17:46:03 -0700 (PDT), dagmargoodboat@yahoo.com
wrote:

On Tuesday, August 27, 2019 at 8:08:45 PM UTC-4, John Larkin wrote:
On Tue, 27 Aug 2019 16:39:38 -0700 (PDT), dagmargoodboat@yahoo.com
wrote:

On Tuesday, August 27, 2019 at 4:50:59 PM UTC-4, John Larkin wrote:
On Tue, 27 Aug 2019 12:57:51 -0700 (PDT), dagmargoodboat@yahoo.com
wrote:

On Tuesday, August 27, 2019 at 3:10:58 PM UTC-4, John Larkin wrote:
On Tue, 27 Aug 2019 11:38:43 -0700 (PDT), dagmargoodboat@yahoo.com
wrote:

On Saturday, August 24, 2019 at 11:31:06 PM UTC-4, bitrex wrote:
Microcontroller-based strategies like this work OK for high-Q inductances:

http://www.pa3fwm.nl/technotes/tn11b.html

But don't work too good for little random-wire very lossy inductances,
of values around 0.5uH to 5uH, at the lower excitation frequencies that
microprocessors can easily provide from direct pin-switching system
clock-derived outputs. e.g. inductaors that have self-resonant
frequencies in the 100s of MHz.

I was thinking the small inductance could have its effective Q boosted
via boostrapping, perhaps (I'm kinda down on negative impedance
circuits, now, you can make some cute circuits with them but they all
obv. tend towards being unstable and are "fiddly" and I'm uncomfortable
using them in "real work")

and then you could measure a certain range of small inductances by
applying a clock to a tank circuit thru a resistor, and putting the
original clock plus the output from the tank into a phase detector a la
a PLL and look at the integrated leading or lagging phase "up/down"
signal to infer the inductance.

It might need little external hardware other than the Q-booster in some
implementation. Clock out to the tank and leading/lagging phase signal
back in to the uP to an onboard comparator/phase detector and integrator.

For my particular solution needs whatever form it takes, it would be
best to trade of absolute accuracy for precision/repeatability.

R.F. Design magazine featured a circuit some years ago that drives
an inductor with a 100kHz sinusoidal current, and measures the
resulting in-phase and quadrature voltages.

That method is robust, yielding independent measurements of inductance
and effective series resistance.

I have a scan of the original article somewhere(*), but this seems to
be a mostly-faithful reproduction:

https://www.qsl.net/va3iul/L_meter/L_meter.htm

You'll notice some interesting compound op-amp arrangements,
which the original article explains provide a 2nd-order
frequency compensation, which drastically improves the phase
response. (That's important, otherwise phase errors eat into
the I-Q scheme's accuracy.)

(*) Ah yes, here it is:
"Simple Digital Inductance Meter with 0.1nH Resolution"
Roger A. Williams, LTX Corp., R.F. Design, October 1987, p50-55


Cheers,
James Arthur

A more modern version could use a 50 ohm sine source, digitize at 4F,
and do some math.

50 ohms? The scheme needs a current source. That way the resistive
and reactive components are easily measured.

But 50 ohms is easier, and an accurate wideband 50 ohm resistor costs
under 1 cent. The ADC can measure stuff and a uP or PC can do the
math.

Actually, the ADC can sample at F or F/N, as long as its trigger phase
can be shifted around in steps of 90 degrees. That needs a couple
flipflops.

If you sample both ends of the 50 ohm resistor, you know the vector
current.

I'm not sure those numbers work. For 10nH, for example, the L/R
constant is 200ps.

But you could digitally drive a constant current source, that works.
20mA 100kHz sine into 1uH makes 25mV p-p, or 25uV for 1nH.

That would be a cool instrument, something that would plot the vector
impedance and equivalent R/L/C components vs frequency. A cheap little
USB thing. I'd like to go down to 1 Hz for power magnetics.

Cheers,
James

1 volt across 50 ohms is 20 mA.

I know. I was thinking the signal level would be too small for
an ADC and the timing differences too small to resolve -- ADCs
aren't good at high-res 1MHz 25uV measurements.


Signal average. Noise is cheap and plentiful.



But some appropriate gain stages fix that, and the 50 ohms is
enough larger than most interesting inductors' e.s.r. that 1V
drive becomes effectively a 40mA p-p current source. That's
not terrible.

One would of course want a sine wave drive at a few (or many)
different frequencies. Hz or KHz for power magnetics, many MHz for
nanohenry inductors.

The original scheme resolved 0.1nH with 100kHz x 20mA excitation.
That's pretty elegant. Not bad for two quad op-amps and one
MC1496.

A 'digital' version with a DAC, ADC, and a uC still needs the
gain stages, but saves the multiplier and a few discrete hairballs.


If the frequency can be run up, you don't need so much gain.

Gain's cheap, speed isn't.

I have a power budget and a low supply voltage constraint (~3 volt) in
the project so speeding that oscillator up to where the test inductors
have a higher Q and I could just frequency count with a fast comparator
into a fast uP is gonna be a problem.

I feel my options are either to a) use a trick to boost the intrinsic
tank Q at lower frequency and use multiple measurements as Jan suggested
in his post, or use a suggestion like yours and not try to make the DUT
part of an oscillator circuit at all and do it indirectly.

I understand the urge to clean up all the discretes, but it seems a
bit campy to throw a million transistors + software at it.

I fell victim to that cleaning urge with my 'upgrade.' I drove the
inductor with a triangle-wave current excitation, since that was
stable, easily calibrated, and easily generated from my
variable-frequency digital source. No DAC required.

Triangular current-drive changes the inductor voltage to a squarewave
proportional to inductance, with e.s.r. ramps instead of flat tops
and bottoms.

The e.s.r. ramp starts at -i excitation and ends with +i excitation,
so if you in-phase demodulate, the e.s.r. component cancels and you're
left with the pure inductive component.

I replaced the original Jim Thompson(?) MC1496 analog multiplier with
CMOS switches. That saved a bunch of biasing and tweaking. De-modulating
in-phase eliminated the earlier design's quadrature phase-shifters and
associated adjustments.

(I'm basically in software hell at the moment, trying to get a
daisychain of undocumented, layered, script-kiddie IDE abstraction-heaps
going, so three analog ICs and a probe-able discrete hairball seems
pretty attractive at the moment, along with a carburetor, points, and
a distributor.)

Cheers,
James Arthur

 

[bitrex]

On 8/28/19 11:18 AM, bitrex wrote:

If the frequency can be run up, you don't need so much gain.

Gain's cheap, speed isn't.

I have a power budget and a low supply voltage constraint (~3 volt) in
the project so speeding that oscillator up to where the test inductors
have a higher Q and I could just frequency count with a fast comparator
into a fast uP is gonna be a problem.

I feel my options are either to a) use a trick to boost the intrinsic
tank Q at lower frequency and use multiple measurements as Jan suggested
in his post, or use a suggestion like yours and not try to make the DUT
part of an oscillator circuit at all and do it indirectly.

the good news is that gain is pretty cheap and gotten a lot cheaper
lately it seems, 20MHz RRIO CMOS op amps that go down to 1.8V supply, in
duals or quads are under a buck in small quantity

 

On Wed, 28 Aug 2019 11:03:40 -0400, bitrex <user@example.net> wrote:

On 8/27/19 10:07 AM, jlarkin@highlandsniptechnology.com wrote:
On Tue, 27 Aug 2019 07:08:09 +0100, piglet <erichpwagner@hotmail.com
wrote:

On 26/08/2019 22:26, John Larkin wrote:
On Mon, 26 Aug 2019 08:33:48 +0100, piglet <erichpwagner@hotmail.com
A problem with measuring inductance by making the DUT part of an
oscillator tank and measuring frequency is the square root relationship
works against you by compressing sensitivity.

Only by 2:1. And it's easy to measure frequency to a part per million.


Yes of course but in one of his post's the OP was complaining their
hardware had problems measuring small differences in frequency.

piglet

He could buy a cheap counter and get to 1 PPM and be done.




Well the issue is the inductances I wanna measure, in the single digital
microhenries, to within a nanohenry, say, have a Q of about 0.3 in the
low MHz.

when you can get them to work as part of a standard oscillator tank that
oscillates at all, an octave and a half below their self-resonant
frequency, the stability is poor. if you have say a 4uH resonating with
a 10n cap to get 5MHz a 1 nH difference in the L is only a few hundred
Hz shift. But the oscillator is drifting around by several kHz over 20
minutes

Inductors are goofy parts, so it only makes sense to measure their
inductance near the frequency they will be used at. An easy way to do
that is to build an oscillator and measure the frequency. The
stability is usually dominated by the inductor tempco, often in the
+100 PPM/K range or so for sensibly constructed air cores.

Try heating and cooling the inductor a bit to see if the inductance is
actually changing.

 

[John Larkin]

On Wed, 28 Aug 2019 13:14:42 GMT, Steve Wilson <no@spam.com> wrote:

Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:

On 8/27/19 4:07 PM, Steve Wilson wrote:
Where do you measure the negative resistance and how do you control it?

You measure it by watching the follower oscillate with no external
feedback. Put a small pot in series with the base and watch where the
oscillation stops.

Cheers

Phil Hobbs

Thanks. This has been very illuminating. First, I find the term
negative resistance has nothing to do with the classical definition,
where an increasing voltage causes decreasing current. An example is
tunnel diode oscillators.

In this application, there is absolutely no negative resistance in
the classical sense. The term is a complete misnomer in this useage.

Second, I find a base resistance of 14.96 Ohms in the circuit I gave
you is sufficient to basically stop the oscillations.

Third, adding Darlington increases the required resistance to 46.58
Ohms.

These findings are of tremendous importance in everyday electronics.

It explains why a small bead or resistor in the base of a transistor
is so effective at stopping parasitic oscillations. It also explains
why parasitic oscillations are so hard to kill in Darlingstons.

The next problem is to find out exactly how the small base
resistance works. This opens a completely new field of investigation
where I am certain the new knowledge gained will be worth the
effort.

Thanks again.

Randy Rhea has written a lot about negative resistance oscillators.

http://tinyurl.com/y2f68w7v

 

[John Larkin]

On Wed, 28 Aug 2019 07:03:30 -0700 (PDT), dagmargoodboat@yahoo.com
wrote:

On Tuesday, August 27, 2019 at 11:17:04 PM UTC-4, jla...@highlandsniptechnology.com wrote:
On Tue, 27 Aug 2019 17:46:03 -0700 (PDT), dagmargoodboat@yahoo.com
wrote:

On Tuesday, August 27, 2019 at 8:08:45 PM UTC-4, John Larkin wrote:
On Tue, 27 Aug 2019 16:39:38 -0700 (PDT), dagmargoodboat@yahoo.com
wrote:

On Tuesday, August 27, 2019 at 4:50:59 PM UTC-4, John Larkin wrote:
On Tue, 27 Aug 2019 12:57:51 -0700 (PDT), dagmargoodboat@yahoo.com
wrote:

On Tuesday, August 27, 2019 at 3:10:58 PM UTC-4, John Larkin wrote:
On Tue, 27 Aug 2019 11:38:43 -0700 (PDT), dagmargoodboat@yahoo.com
wrote:

On Saturday, August 24, 2019 at 11:31:06 PM UTC-4, bitrex wrote:
Microcontroller-based strategies like this work OK for high-Q inductances:

http://www.pa3fwm.nl/technotes/tn11b.html

But don't work too good for little random-wire very lossy inductances,
of values around 0.5uH to 5uH, at the lower excitation frequencies that
microprocessors can easily provide from direct pin-switching system
clock-derived outputs. e.g. inductaors that have self-resonant
frequencies in the 100s of MHz.

I was thinking the small inductance could have its effective Q boosted
via boostrapping, perhaps (I'm kinda down on negative impedance
circuits, now, you can make some cute circuits with them but they all
obv. tend towards being unstable and are "fiddly" and I'm uncomfortable
using them in "real work")

and then you could measure a certain range of small inductances by
applying a clock to a tank circuit thru a resistor, and putting the
original clock plus the output from the tank into a phase detector a la
a PLL and look at the integrated leading or lagging phase "up/down"
signal to infer the inductance.

It might need little external hardware other than the Q-booster in some
implementation. Clock out to the tank and leading/lagging phase signal
back in to the uP to an onboard comparator/phase detector and integrator.

For my particular solution needs whatever form it takes, it would be
best to trade of absolute accuracy for precision/repeatability.

R.F. Design magazine featured a circuit some years ago that drives
an inductor with a 100kHz sinusoidal current, and measures the
resulting in-phase and quadrature voltages.

That method is robust, yielding independent measurements of inductance
and effective series resistance.

I have a scan of the original article somewhere(*), but this seems to
be a mostly-faithful reproduction:

https://www.qsl.net/va3iul/L_meter/L_meter.htm

You'll notice some interesting compound op-amp arrangements,
which the original article explains provide a 2nd-order
frequency compensation, which drastically improves the phase
response. (That's important, otherwise phase errors eat into
the I-Q scheme's accuracy.)

(*) Ah yes, here it is:
"Simple Digital Inductance Meter with 0.1nH Resolution"
Roger A. Williams, LTX Corp., R.F. Design, October 1987, p50-55


Cheers,
James Arthur

A more modern version could use a 50 ohm sine source, digitize at 4F,
and do some math.

50 ohms? The scheme needs a current source. That way the resistive
and reactive components are easily measured.

But 50 ohms is easier, and an accurate wideband 50 ohm resistor costs
under 1 cent. The ADC can measure stuff and a uP or PC can do the
math.

Actually, the ADC can sample at F or F/N, as long as its trigger phase
can be shifted around in steps of 90 degrees. That needs a couple
flipflops.

If you sample both ends of the 50 ohm resistor, you know the vector
current.

I'm not sure those numbers work. For 10nH, for example, the L/R
constant is 200ps.

But you could digitally drive a constant current source, that works.
20mA 100kHz sine into 1uH makes 25mV p-p, or 25uV for 1nH.

That would be a cool instrument, something that would plot the vector
impedance and equivalent R/L/C components vs frequency. A cheap little
USB thing. I'd like to go down to 1 Hz for power magnetics.

Cheers,
James

1 volt across 50 ohms is 20 mA.

I know. I was thinking the signal level would be too small for
an ADC and the timing differences too small to resolve -- ADCs
aren't good at high-res 1MHz 25uV measurements.


Signal average. Noise is cheap and plentiful.



But some appropriate gain stages fix that, and the 50 ohms is
enough larger than most interesting inductors' e.s.r. that 1V
drive becomes effectively a 40mA p-p current source. That's
not terrible.

One would of course want a sine wave drive at a few (or many)
different frequencies. Hz or KHz for power magnetics, many MHz for
nanohenry inductors.

The original scheme resolved 0.1nH with 100kHz x 20mA excitation.
That's pretty elegant. Not bad for two quad op-amps and one
MC1496.

A 'digital' version with a DAC, ADC, and a uC still needs the
gain stages, but saves the multiplier and a few discrete hairballs.


If the frequency can be run up, you don't need so much gain.

Gain's cheap, speed isn't.

I understand the urge to clean up all the discretes, but it seems a
bit campy to throw a million transistors + software at it.

A small USB thing could do the measurements, and then let a PC take
over.

Analog Devices has some nice DDS synthesizers and differential-input
analog multipliers, which would be another way to go. And the network
analyzer chip that someone mentioned.

All it takes is a little trig. I have a nice trig textbook, printed in
1868. It was bought by George P Lents, for $1.25, in 1872.

 

[John Larkin]

On Wed, 28 Aug 2019 19:52:50 GMT, Steve Wilson <no@spam.com> wrote:

jlarkin@highlandsniptechnology.com wrote:

On Wed, 28 Aug 2019 13:14:42 GMT, Steve Wilson <no@spam.com> wrote:
Thanks. This has been very illuminating. First, I find the term
negative resistance has nothing to do with the classical definition,
where an increasing voltage causes decreasing current. An example is
tunnel diode oscillators.

In this application, there is absolutely no negative resistance in the
classical sense. The term is a complete misnomer in this useage.

Given an LC tank or an equivalent 1-port passive resonator, only
seeing a negative resistance will make it oscillate. There is a class
of such oscillators that are analyzed based on negative resistance.

Basically most, if not all oscillators can be analyzed using a series
resistor at the input.

All oscillators can be analyzed using the Barkhausen criterion.

Second, I find a base resistance of 14.96 Ohms in the circuit I gave
you is sufficient to basically stop the oscillations.

Third, adding Darlington increases the required resistance to 46.58
Ohms.

These findings are of tremendous importance in everyday electronics.

It explains why a small bead or resistor in the base of a transistor
is so effective at stopping parasitic oscillations. It also explains
why parasitic oscillations are so hard to kill in Darlingstons.

The Mini-Circuits type MMICS are unconditionally stable. They are
Darlingtons.

Sure. They have minimal inductance in the base, grounded emitters,
multilayer pcbs with good ground plane, 50 Ohms in and out, good layout and
bypassing, and so on.

I meant this sort of thing:

https://www.minicircuits.com/WebStore/dashboard.html?model=ERA-5XSM%2B

 

[Steve Wilson]

John Larkin <jlarkin@highland_atwork_technology.com> wrote:

Randy Rhea has written a lot about negative resistance oscillators.

http://tinyurl.com/y2f68w7v

Looks like he is analyzing a standard Clapp oscillator and omitting the base-
emitter and stray capacitances.

Most, if not all oscillators can be analyzed with a resistor at the input.
All oscillators can be analyzed using the Barkhausen criterion.

 

[Steve Wilson]

Jeroen Belleman <jeroen@nospam.please> wrote:

Steve Wilson wrote:

Thanks. This has been very illuminating. First, I find the term
negative resistance has nothing to do with the classical definition,
where an increasing voltage causes decreasing current.

Well, it's negative under some conditions in the small
signal analysis.

The input impedance of an emitter follower is something
like Zb + Ze + ZbZe, with Zb=a'/S // Z(Cbe), the impedance
from base to emitter and Ze=Re // Z(Ce), the emitter load
impedance. Under certain conditions, this ZbZe term can go
negative enough to make the overall input impedance negative.

I include below a simple LTspice AC equivalent circuit model
of an emitter follower. Plot V(in)/I(Rin) in Cartesian
coordinates and you'll see the real part go negative around
200MHz.

Jeroen Belleman

How do you get LTspice to plot in Cartesian coordinates?

 

[bitrex]

On 8/28/19 12:56 PM, jlarkin@highlandsniptechnology.com wrote:

On Wed, 28 Aug 2019 11:03:40 -0400, bitrex <user@example.net> wrote:

On 8/27/19 10:07 AM, jlarkin@highlandsniptechnology.com wrote:
On Tue, 27 Aug 2019 07:08:09 +0100, piglet <erichpwagner@hotmail.com
wrote:

On 26/08/2019 22:26, John Larkin wrote:
On Mon, 26 Aug 2019 08:33:48 +0100, piglet <erichpwagner@hotmail.com
A problem with measuring inductance by making the DUT part of an
oscillator tank and measuring frequency is the square root relationship
works against you by compressing sensitivity.

Only by 2:1. And it's easy to measure frequency to a part per million.


Yes of course but in one of his post's the OP was complaining their
hardware had problems measuring small differences in frequency.

piglet

He could buy a cheap counter and get to 1 PPM and be done.




Well the issue is the inductances I wanna measure, in the single digital
microhenries, to within a nanohenry, say, have a Q of about 0.3 in the
low MHz.

when you can get them to work as part of a standard oscillator tank that
oscillates at all, an octave and a half below their self-resonant
frequency, the stability is poor. if you have say a 4uH resonating with
a 10n cap to get 5MHz a 1 nH difference in the L is only a few hundred
Hz shift. But the oscillator is drifting around by several kHz over 20
minutes

Inductors are goofy parts, so it only makes sense to measure their
inductance near the frequency they will be used at. An easy way to do
that is to build an oscillator and measure the frequency. The
stability is usually dominated by the inductor tempco, often in the
+100 PPM/K range or so for sensibly constructed air cores.

Try heating and cooling the inductor a bit to see if the inductance is
actually changing.

the inductor in this case is functioning as a strain gauge/transducer
(as opposed to resistivity strain gauge) so the circuit that it's in, is
the capacity that it's being used in, as an indirect measurement of
displacement

 

[Steve Wilson]

jlarkin@highlandsniptechnology.com wrote:

On Wed, 28 Aug 2019 13:14:42 GMT, Steve Wilson <no@spam.com> wrote:
Thanks. This has been very illuminating. First, I find the term
negative resistance has nothing to do with the classical definition,
where an increasing voltage causes decreasing current. An example is
tunnel diode oscillators.

In this application, there is absolutely no negative resistance in the
classical sense. The term is a complete misnomer in this useage.

Given an LC tank or an equivalent 1-port passive resonator, only
seeing a negative resistance will make it oscillate. There is a class
of such oscillators that are analyzed based on negative resistance.

Basically most, if not all oscillators can be analyzed using a series
resistor at the input.

All oscillators can be analyzed using the Barkhausen criterion.

Second, I find a base resistance of 14.96 Ohms in the circuit I gave
you is sufficient to basically stop the oscillations.

Third, adding Darlington increases the required resistance to 46.58
Ohms.

These findings are of tremendous importance in everyday electronics.

It explains why a small bead or resistor in the base of a transistor
is so effective at stopping parasitic oscillations. It also explains
why parasitic oscillations are so hard to kill in Darlingstons.

The Mini-Circuits type MMICS are unconditionally stable. They are
Darlingtons.

Sure. They have minimal inductance in the base, grounded emitters,
multilayer pcbs with good ground plane, 50 Ohms in and out, good layout and
bypassing, and so on.

I never said all Darlingtons oscillate. But I explained when you find one
that does, why it is so difficult to kill the oscillation. I suspect part
of the reason is the higher input impedance. The other reasons the base
resistor kills the oscillation is a subject for further study.

 

[Steve Wilson]

John Larkin <jlarkin@highland_atwork_technology.com> wrote:

On Wed, 28 Aug 2019 19:52:50 GMT, Steve Wilson <no@spam.com> wrote:

The Mini-Circuits type MMICS are unconditionally stable. They are
Darlingtons.

Sure. They have minimal inductance in the base, grounded emitters,
multilayer pcbs with good ground plane, 50 Ohms in and out, good layout
and bypassing, and so on.

I meant this sort of thing:

https://www.minicircuits.com/WebStore/dashboard.html?model=ERA-5XSM%2B

Same thing. Emitters go to ground with recommended layout to minimize ground
inductance. 50 Ohm environment. Multilayer pcb with ground plane and good
layout prectises required.

Again, I never said that all Darlintons oscillate.

 

[Steve Wilson]

John Larkin <jlarkin@highland_atwork_technology.com> wrote:

On Wed, 28 Aug 2019 17:03:04 +0200, Jeroen Belleman
jeroen@nospam.please> wrote:

Steve Wilson wrote:
Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:

On 8/27/19 4:07 PM, Steve Wilson wrote:
Where do you measure the negative resistance and how do you control
it?

You measure it by watching the follower oscillate with no external
feedback. Put a small pot in series with the base and watch where
the oscillation stops.

Cheers

Phil Hobbs

Thanks. This has been very illuminating. First, I find the term
negative resistance has nothing to do with the classical definition,
where an increasing voltage causes decreasing current.

Well, it's negative under some conditions in the small
signal analysis.

The input impedance of an emitter follower is something
like Zb + Ze + ZbZe, with Zb=a'/S // Z(Cbe), the impedance
from base to emitter and Ze=Re // Z(Ce), the emitter load
impedance. Under certain conditions, this ZbZe term can go
negative enough to make the overall input impedance negative.

I include below a simple LTspice AC equivalent circuit model
of an emitter follower. Plot V(in)/I(Rin) in Cartesian
coordinates and you'll see the real part go negative around 200MHz.

That's interesting in a time-domain sim, with a 1 GHz 1 volt source.
There is voltage gain at the right-hand end of the 50 ohm resistor,
about 1.6 volts peak with a little phase shift. The generator clearly
sees a negative resistor.

That's his circcuit. Try a Colpitts.

 

[John Larkin]

On Wed, 28 Aug 2019 17:03:04 +0200, Jeroen Belleman
<jeroen@nospam.please> wrote:

Steve Wilson wrote:
Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:

On 8/27/19 4:07 PM, Steve Wilson wrote:
Where do you measure the negative resistance and how do you control it?

You measure it by watching the follower oscillate with no external
feedback. Put a small pot in series with the base and watch where the
oscillation stops.

Cheers

Phil Hobbs

Thanks. This has been very illuminating. First, I find the term
negative resistance has nothing to do with the classical definition,
where an increasing voltage causes decreasing current.

Well, it's negative under some conditions in the small
signal analysis.

The input impedance of an emitter follower is something
like Zb + Ze + ZbZe, with Zb=a'/S // Z(Cbe), the impedance
from base to emitter and Ze=Re // Z(Ce), the emitter load
impedance. Under certain conditions, this ZbZe term can go
negative enough to make the overall input impedance negative.

I include below a simple LTspice AC equivalent circuit model
of an emitter follower. Plot V(in)/I(Rin) in Cartesian
coordinates and you'll see the real part go negative around
200MHz.

That's interesting in a time-domain sim, with a 1 GHz 1 volt source.
There is voltage gain at the right-hand end of the 50 ohm resistor,
about 1.6 volts peak with a little phase shift. The generator clearly
sees a negative resistor.

 

[Steve Wilson]

Jeroen Belleman <jeroen@nospam.please> wrote:

Steve Wilson wrote:

Thanks. This has been very illuminating. First, I find the term
negative resistance has nothing to do with the classical definition,
where an increasing voltage causes decreasing current.

Well, it's negative under some conditions in the small
signal analysis.

The input impedance of an emitter follower is something
like Zb + Ze + ZbZe, with Zb=a'/S // Z(Cbe), the impedance
from base to emitter and Ze=Re // Z(Ce), the emitter load
impedance. Under certain conditions, this ZbZe term can go
negative enough to make the overall input impedance negative.

I include below a simple LTspice AC equivalent circuit model
of an emitter follower. Plot V(in)/I(Rin) in Cartesian
coordinates and you'll see the real part go negative around
200MHz.

Jeroen Belleman

I found how to plot in Cartesian coordinates. Simple.

Your plot shows the impedance going negative above 10 MHz. I think that may
be an artifact of your circuit. It sure doesn't look like a Colpitts to me.

I have carefully examined the input impedance under small and large signal
conditions. I was unable to find any region that exhibited negative input
impedance.

The term "negative resistance" is a complete misnomer. It has nothing to do
with the classical definition of negative resistance, where the current
decreases as the voltage increases. The term is simply ill-chosen.

The term merely illustrates the value of resistance needed at the input of
an oscillator to kill oscillations.

 

[John Larkin]

On Wed, 28 Aug 2019 22:35:44 GMT, Steve Wilson <no@spam.com> wrote:

John Larkin <jlarkin@highland_atwork_technology.com> wrote:

On Wed, 28 Aug 2019 17:03:04 +0200, Jeroen Belleman
jeroen@nospam.please> wrote:

Steve Wilson wrote:
Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:

On 8/27/19 4:07 PM, Steve Wilson wrote:
Where do you measure the negative resistance and how do you control
it?

You measure it by watching the follower oscillate with no external
feedback. Put a small pot in series with the base and watch where
the oscillation stops.

Cheers

Phil Hobbs

Thanks. This has been very illuminating. First, I find the term
negative resistance has nothing to do with the classical definition,
where an increasing voltage causes decreasing current.

Well, it's negative under some conditions in the small
signal analysis.

The input impedance of an emitter follower is something
like Zb + Ze + ZbZe, with Zb=a'/S // Z(Cbe), the impedance
from base to emitter and Ze=Re // Z(Ce), the emitter load
impedance. Under certain conditions, this ZbZe term can go
negative enough to make the overall input impedance negative.

I include below a simple LTspice AC equivalent circuit model
of an emitter follower. Plot V(in)/I(Rin) in Cartesian
coordinates and you'll see the real part go negative around 200MHz.

That's interesting in a time-domain sim, with a 1 GHz 1 volt source.
There is voltage gain at the right-hand end of the 50 ohm resistor,
about 1.6 volts peak with a little phase shift. The generator clearly
sees a negative resistor.

That's his circcuit. Try a Colpitts.

Any passive resonator that oscillates is seeing a negative resistance.
Energy is conserved, and the resonator loss has to be made up for.

 

[Steve Wilson]

John Larkin <jlarkin@highland_atwork_technology.com> wrote:

On Wed, 28 Aug 2019 22:35:44 GMT, Steve Wilson <no@spam.com> wrote:
That's his circcuit. Try a Colpitts.

Any passive resonator that oscillates is seeing a negative resistance.
Energy is conserved, and the resonator loss has to be made up for.

Where did you get that? Try measuring the input impedance of a Colpitts. It
is always positive. The negative resistance is a complete misnomer. It has
nothing to do with the voltage increasing and the current decreasing. It is
only the resistance needed to stop oscillations.

The Barkhausen criteria states the phase shift has to be zero or multiples of
360 degrees, and the loop gain has to be equal to or greater than 1. No
negative resistance is needed.

In a cc Colpitts, you control the energy loss by adjusting the emitter
resistance and thus the energy delivered to the tank. In a Pierce, you adjust
the resistance from the output to the tank.

See Oscillators.zip at

https://drive.google.com/open?id=1ZsbpkV0aaKS5LURIb1dfu_ndshsSaYtf

 

[bitrex]

On 8/28/19 9:29 PM, Steve Wilson wrote:

John Larkin <jlarkin@highland_atwork_technology.com> wrote:

On Wed, 28 Aug 2019 22:35:44 GMT, Steve Wilson <no@spam.com> wrote:
That's his circcuit. Try a Colpitts.

Any passive resonator that oscillates is seeing a negative resistance.
Energy is conserved, and the resonator loss has to be made up for.

Where did you get that? Try measuring the input impedance of a Colpitts. It
is always positive. The negative resistance is a complete misnomer. It has
nothing to do with the voltage increasing and the current decreasing. It is
only the resistance needed to stop oscillations.

The Barkhausen criteria states the phase shift has to be zero or multiples of
360 degrees, and the loop gain has to be equal to or greater than 1. No
negative resistance is needed.

In a cc Colpitts, you control the energy loss by adjusting the emitter
resistance and thus the energy delivered to the tank. In a Pierce, you adjust
the resistance from the output to the tank.

See Oscillators.zip at

https://drive.google.com/open?id=1ZsbpkV0aaKS5LURIb1dfu_ndshsSaYtf

I don't know if it's entirely relevant but the Barkhausen criteria are
necessary criteria for oscillation, but they don't imply that they are
sufficient conditions.

I don't believe anyone's ever come up with a set of universal criteria
that act as sufficient conditions for predicting oscillations, in all
electronic positive feedback loop structures that it's possible to
theoretically construct.

So whether negative resistance appears in some oscillating system or
not, and whether it's required to be there or not for it to oscillate,
likely depends entirely on the particular system.

That is to say the mathematical theory of large-signal periodic
oscillations in electronic circuits is pretty complicated, there are a
number of table-breaking books written on the subject, and at the very
least I think it's difficult to make any un-qualified "all oscillators
do this" or "all oscillators do that"-type statements about them.

 

[bitrex]

On 8/28/19 4:17 PM, Steve Wilson wrote:

John Larkin <jlarkin@highland_atwork_technology.com> wrote:

Randy Rhea has written a lot about negative resistance oscillators.

http://tinyurl.com/y2f68w7v

Looks like he is analyzing a standard Clapp oscillator and omitting the base-
emitter and stray capacitances.

Most, if not all oscillators can be analyzed with a resistor at the input.
All oscillators can be analyzed using the Barkhausen criterion.

There's only so much you can say about them using small-signal analysis;
oscillations are inherently large-signal phenomena.

You can only analyze circuits with small-signal techniques for stability
or not-stability, and if the analysis says not-stability it doesn't
immediately imply oscillation.