Driver to drive?

Jim Thompson · Jan 24, 2012

On Tue, 24 Jan 2012 11:25:42 -0600, Lostgallifreyan
<no-one@nowhere.net> wrote:

Jim Thompson <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote in
newsspth79223jbsmjv1blovv8kpac0j6mlk7@4ax.com:

Google doesn't know $#^*

Well, that's kind of my point. > I'm citing their best output to encourage
you to come up with the good stuff.

I think analog accuracy has been underrated over many years. Never mind that
it doesn't get to the last bit-worth of accuracy, the SPEED of calculating
complex forms is second till none, at least until someone does it with
quantum computing. So much time might have been saved by analog computers,
leaving digital ones to refine the output.

I disagree (slightly)... It's hard to beat a uP for decision making
and using its outputs to control analog functions.

It's a rare chip I've designed in recent years that doesn't either
have an embedded uP in it (designed by or purchased IP embedded by a
subcontractor buddy of mine), or is controlled by an external uP. We
are, after all, living in a system on a chip (SOC) world.

...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice

Jim Thompson · Jan 24, 2012

On Mon, 23 Jan 2012 06:22:25 -0600, Lostgallifreyan
<no-one@nowhere.net> wrote:

Lostgallifreyan <no-one@nowhere.net> wrote in
news:Xns9FE1C0DDF59A8zoodlewurdle@216.196.109.145:

[snip]

John Larkin · Jan 24, 2012

On Tue, 24 Jan 2012 06:08:28 -0800, John Larkin
<jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Tue, 24 Jan 2012 20:25:32 +1100, "Phil Allison" <phil_a@tpg.com.au
wrote:

"Dishum"

I'm a hobbyist who doesn't have an ESR meter or (usually) a choice of
specific capacitor models and I'd like to have some idea of the kind of
ESR values one would expect from capacitors that are neither particularly
good nor particularly crappy in that respect. I'll really appreciate it if
you could cite some ballpark figures for -

1. 1uF/50V wet Al electrolytic
2. 100uF/25V wet Al electrolytic
3. 1000uF/50V wet Al electrolytic
4. 1uF/25V tantalum

** Measured from my parts bins:

1 = 3.1 ohms
2 = 0.5 ohms
3 = 0.06 ohms
4 = 4.2 ohms

In each case, the figure is for high frequency ESR or impedance at 100kHz.

ESR rises at low frequencies ( under 500Hz ) and falls with increasing
temperature.

Also, when an electro goes bad (ie dries out ) - ESR rises first followed
much later by a reduction in actual uF.

... Phil

That seems high for the tantalum. I seem to recall numbers like a
couple tenths of an ohm. I'll try a couple. I don't have an ESR meter,
but I can just apply a square wave from a 50 ohm generator and scope
the voltage.

Hmmm, both of the 1u caps have about the same ESR.

John

I tried a standard leaded gumdrop tantalum, 1 uF at 35 volts, and got
about 0.75 ohms.

22u 16v was about 0.35 ohms.

Tantalum ESR tends to be in sort of a sweet spot for taming voltage
regulators, both switchers and linear.

John

Lostgallifreyan · Jan 24, 2012

Jim Thompson <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote in
news:2bvth7l6oknup47dnmuk3c040pn14mgh64@4ax.com:

Ran your schematic... behavior is extraordinarily weird, output hangs
about 0.7V below ground when powered from +/-12V

Please recheck your netlist. Thanks!

I checked it, it's definitely wired ok, transistors are of correct type
and orientation, and the pins are correctly numbered according to node. I
can't model its parts so didn't check by running it.

I can think of a few things that might matter. I didn't assign a ground, so
there was no node 0, it was node 15, and maybe there has to be a node 0 in
all subcircuits. Also, I used European resistor value methods like 500R,
which LTspice can be told not to accept. Also, as LTspice uses M for MOSFET
and Q for bipolar, and I kept the numbering from the datasheet, the first M
value is 8, not 1, and there is no Q8. Maybe spice programs don't like
discontinuities like that.

Apart from that last point, which I left as I did it, so the numbers match
up, I changed the other details I mentioned for more likely compatibility,
and also reordered parts by stage as they appear in the datasheet schematic,
which should help with verifying their nodes are right.

*Intersil CA3140, basic model drafted from datasheet.
* 8, Strb -------------------------|
* 7, V+ -----------------------| |
* 6, Out --------------------| | |
* 5, null -----------------| | | |
* 4, Gnd --------------| | | | |
* 3, +In -----------| | | | | |
* 2, -In --------| | | | | | |
* 1, null -----| | | | | | | |
* | | | | | | | |
..SUBCKT CA3140 30 20 22 0 29 15 1 12
* BIAS CIRCUIT
D1 N001 N002 Diode
Q1 N005 N002 N001 0 PNP
Q6 N011 N005 N002 0 PNP
Q7 N005 N011 N016 0 NPN
R1 N016 0 8000
M8 N019 0 N011 N011 PMOS
D2 N019 0 Diode
* INPUT STAGE
Q2 N007 N002 N001 0 PNP
Q5 N023 N005 N007 0 PNP
D3 N021 N020 Zener
D4 N021 N022 Zener
D5 N021 N023 Zener
M9 N024 N020 N023 N023 PMOS
M10 N025 N022 N023 N023 PMOS
R2 N024 N028 500
R3 N025 N026 500
Q11 N028 N024 N029 0 NPN
Q12 N026 N024 N030 0 NPN
R4 N029 0 500
R5 N030 0 500
* SECOND STAGE
Q3 N006 N002 N001 0 PNP
Q4 N012 N005 N006 0 PNP
Q13 N012 N026 0 0 NPN
C1 N026 N012 12E-12
* OUTPUT STAGE
Q17 N001 N012 N017 0 NPN
R8 N017 N018 1000
Q18 N013 N018 N015 0 NPN
D7 N001 N004 Diode
R9 N004 N010 50
Q19 N006 N009 N013 0 NPN
R10 N009 N010 1000
R11 N010 N013 20
Q14 N017 N019 N031 0 NPN
R6 N031 0 50
Q15 N015 N019 0 0 NPN
* DYNAMIC CURRENT SINK
Q16 N015 N027 0 0 NPN
D6 N027 N032 Diode
R7 N032 0 30
Q20 N001 N003 N008 0 NPN
R12 N008 N014 12000
R13 N001 N003 5000
D8 P001 N003 Zener
R14 P001 0 20000
M21 N027 N015 N014 N014 PMOS
*
*MODELS NEEDED FOR CA3140 INNARDS
..model Diode D
..model Zener D
..lib E:\EDITORS\LTSPICE\lib\cmp\standard.dio
..model NPN NPN
..model PNP PNP
..lib E:\EDITORS\LTSPICE\lib\cmp\standard.bjt
..model NMOS NMOS
..model PMOS PMOS
..lib E:\EDITORS\LTSPICE\lib\cmp\standard.mos
..ENDS

Simon S Aysdie · Jan 24, 2012

On Jan 24, 11:59 am, John Larkin
<jjlar...@highNOTlandTHIStechnologyPART.com> wrote:

On Tue, 24 Jan 2012 06:08:28 -0800, John Larkin

jjlar...@highNOTlandTHIStechnologyPART.com> wrote:
On Tue, 24 Jan 2012 20:25:32 +1100, "Phil Allison" <phi...@tpg.com.au
wrote:

"Dishum"

I'm a hobbyist who doesn't have an ESR meter or (usually) a choice of
specific capacitor models and I'd like to have some idea of the kind of
ESR values one would expect from capacitors that are neither particularly
good nor particularly crappy in that respect. I'll really appreciate it if
you could cite some ballpark figures for -

1. 1uF/50V wet Al electrolytic
2. 100uF/25V wet Al electrolytic
3. 1000uF/50V wet Al electrolytic
4. 1uF/25V tantalum

** Measured from my parts bins:

1 = 3.1 ohms
2 = 0.5 ohms
3 = 0.06 ohms
4 = 4.2 ohms

In each case, the figure is for high frequency ESR or impedance at 100kHz.

ESR rises at low frequencies ( under 500Hz ) and falls with increasing
temperature.

Also, when an electro goes bad (ie dries out ) - ESR rises first followed
much later by a reduction in actual uF.

... Phil

That seems high for the tantalum. I seem to recall numbers like a
couple tenths of an ohm. I'll try a couple. I don't have an ESR meter,
but I can just apply a square wave from a 50 ohm generator and scope
the voltage.

Hmmm, both of the 1u caps have about the same ESR.

John

I tried a standard leaded gumdrop tantalum, 1 uF at 35 volts, and got
about 0.75 ohms.

22u 16v was about 0.35 ohms.

Tantalum ESR tends to be in sort of a sweet spot for taming voltage
regulators, both switchers and linear.

John- Hide quoted text -

- Show quoted text -

I haven't used these spice models, but maybe something can be gleaned
from them:

http://www.kemet.com/kemet/web/homepage/kechome.nsf/weben/kemsoftNetList

A brief Kemet paper:
http://www.kemet.com/kemet/web/homepage/kfbk3.nsf/vaFeedbackFAQ/53D5333B4453253585256BCD004EBC04/$file/TechTopics%20Vol4No5%20Sep94.PDF)

Lostgallifreyan · Jan 24, 2012

Jim Thompson <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote in
news:rnrth7hkj2icujq1bqgdf2vsi6j5shmmet@4ax.com:

I think analog accuracy has been underrated over many years. Never mind
that it doesn't get to the last bit-worth of accuracy, the SPEED of
calculating complex forms is second till none, at least until someone
does it with quantum computing. So much time might have been saved by
analog computers, leaving digital ones to refine the output.

I disagree (slightly)... It's hard to beat a uP for decision making
and using its outputs to control analog functions.

That's true. I guess it's the other perspective in a duality, the one I
didn't take. A good hybrid is likely the best of all possibles, so a really
effective computer might combine all methods in full intercommunication
between methods.

So the program can decide what to try, try analog where fast and appropriate,
then return to digital for accurate computations entirely ignoring methods
that analog methods quickly showed to be unacceptable. I'd like to see what
happens when quantum methods are added to a system like that.

Might be a
whole new kind of brain.

Lostgallifreyan · Jan 24, 2012

Jim Thompson <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote in
news:2bvth7l6oknup47dnmuk3c040pn14mgh64@4ax.com:

Ran your schematic... behavior is extraordinarily weird, output hangs
about 0.7V below ground when powered from +/-12V

Please recheck your netlist. Thanks!

I guess you did reorder the pin count and quantity.

(And possibly reversed
+in and -In in doing so by accident?) I used all 8, but I imagine my new
SUBCKT can be made standard for 5 pins +In -In V+ V- Out by doing this:
..SUBCKT CA3140 22 20 1 0 15
(Leaving the other 3 as no external connection).

Jim Thompson · Jan 24, 2012

On Tue, 24 Jan 2012 15:01:39 -0600, Lostgallifreyan
<no-one@nowhere.net> wrote:

Jim Thompson <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote in
news:2bvth7l6oknup47dnmuk3c040pn14mgh64@4ax.com:

Ran your schematic... behavior is extraordinarily weird, output hangs
about 0.7V below ground when powered from +/-12V

Please recheck your netlist. Thanks!

I checked it, it's definitely wired ok, transistors are of correct type
and orientation, and the pins are correctly numbered according to node. I
can't model its parts so didn't check by running it.

I can think of a few things that might matter. I didn't assign a ground, so
there was no node 0, it was node 15, and maybe there has to be a node 0 in
all subcircuits.

No. You don't need a Node Zero inside a subcircuit.

Also, I used European resistor value methods like 500R,

Fixed right away.

which LTspice can be told not to accept. Also, as LTspice uses M for MOSFET
and Q for bipolar, and I kept the numbering from the datasheet, the first M
value is 8, not 1, and there is no Q8. Maybe spice programs don't like
discontinuities like that.

Not a problem in PSpice. I did have to fix your headers, "15" =>
"N015", etc, otherwise everything "floats"

[snip]

It could be simply an issue with guessing the MOSFET parameters, for
example M(Q)8... probably a long channel device to act like a variable
resistor... how long, who knows? Plus that output current "sink"
structure is plain-ass somebody's wet dream ;-)

But the data sheet offers some clues, the strobe current is 220uA, so
that's the current in Q3/Q4.

Given the weirdness of that output stage I'm becoming inclined to
model it as a mix of behavioral blocks plus NPN's to match the data
sheet.

The data sheet indicates to me that it's not the world's gift to OpAmp
performance standards... why do you want to use it?

I invented that Q11, Q12, Q13 "turnaround"... see Tom Frederiksen's
book, "Intuitive IC Op Amps", page 14. The schematic in the data
sheet is NOT balanced... I suspect a typo... maybe there are more.

...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice

Jim Thompson · Jan 24, 2012

On Tue, 24 Jan 2012 15:08:47 -0600, Lostgallifreyan
<no-one@nowhere.net> wrote:

Jim Thompson <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote in
news:rnrth7hkj2icujq1bqgdf2vsi6j5shmmet@4ax.com:

I think analog accuracy has been underrated over many years. Never mind
that it doesn't get to the last bit-worth of accuracy, the SPEED of
calculating complex forms is second till none, at least until someone
does it with quantum computing. So much time might have been saved by
analog computers, leaving digital ones to refine the output.

I disagree (slightly)... It's hard to beat a uP for decision making
and using its outputs to control analog functions.

That's true. I guess it's the other perspective in a duality, the one I
didn't take. A good hybrid is likely the best of all possibles, so a really
effective computer might combine all methods in full intercommunication
between methods.

So the program can decide what to try, try analog where fast and appropriate,
then return to digital for accurate computations entirely ignoring methods
that analog methods quickly showed to be unacceptable. I'd like to see what
happens when quantum methods are added to a system like that. Might be a
whole new kind of brain.

Jim Thompson · Jan 24, 2012

On Tue, 24 Jan 2012 15:20:08 -0600, Lostgallifreyan
<no-one@nowhere.net> wrote:

Jim Thompson <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote in
news:2bvth7l6oknup47dnmuk3c040pn14mgh64@4ax.com:

Ran your schematic... behavior is extraordinarily weird, output hangs
about 0.7V below ground when powered from +/-12V

Please recheck your netlist. Thanks!

I guess you did reorder the pin count and quantity. (And possibly reversed
+in and -In in doing so by accident?) I used all 8, but I imagine my new
SUBCKT can be made standard for 5 pins +In -In V+ V- Out by doing this:
.SUBCKT CA3140 22 20 1 0 15
(Leaving the other 3 as no external connection).

Lostgallifreyan · Jan 24, 2012

Jim Thompson <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote in
news:ba9uh7pk732mi0etcmsca2bup8um3nvhrc@4ax.com:

Nope, no reorder... I symbolized it as an 8-pin block, pin order per
the data sheet... as you did.

Ok

Lostgallifreyan · Jan 24, 2012

Jim Thompson <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote in
news:u97uh7peo4he06h6gibgurcgme24odq8i6@4ax.com:

On Tue, 24 Jan 2012 15:01:39 -0600, Lostgallifreyan
no-one@nowhere.net> wrote:

Jim Thompson <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote
in news:2bvth7l6oknup47dnmuk3c040pn14mgh64@4ax.com:

Ran your schematic... behavior is extraordinarily weird, output hangs
about 0.7V below ground when powered from +/-12V

Please recheck your netlist. Thanks!

I checked it, it's definitely wired ok, transistors are of correct type
and orientation, and the pins are correctly numbered according to node.
I can't model its parts so didn't check by running it.

I can think of a few things that might matter. I didn't assign a ground,
so there was no node 0, it was node 15, and maybe there has to be a node
0 in all subcircuits.

No. You don't need a Node Zero inside a subcircuit.

Ok. I thought not too, but reordered anyway, because I noticed that pin
orders must be contigous starting from 1 in a symbol, and I decided to
avoid similar trouble.

Also, I used European resistor value methods like 500R,

Fixed right away.

Ok. In mine I just used literal values of ohms, and 12E-12 for that cap.

which LTspice can be told not to accept. Also, as LTspice uses M for
MOSFET and Q for bipolar, and I kept the numbering from the datasheet,
the first M value is 8, not 1, and there is no Q8. Maybe spice programs
don't like discontinuities like that.

Not a problem in PSpice. I did have to fix your headers, "15" =
"N015", etc, otherwise everything "floats"

Yeah, I just noticed that! I chucked any model that looked like it would
fit from the standard libraries just to see if I can run this. it took a
while, chewing, but it DID run, and my output from that equal-divider
test shows correct 2.5V from 5V input.

With 'startup' in .tran it does funny things extremely fast at low
amplitude, but that might just be because I threw models recklessly
inside it. It still settled down eventually to a correct DC output.

For the record, this is what I threw inside:
*MODELS NEEDED FOR CA3140 INNARDS
..model Diode D(Is=2.52n Rs=.568 N=1.752 Cjo=4p M=.4 tt=20n Iave=200m Vpk=75 mfg=OnSemi type=silicon)
..model Zener D(Is=1.13E-14 N=1.103 Vpk=15 bv=15 Ibv=0.005 Rs=0.4798 Cjo=4.023E-11 Vj=0.6144 M=0.3297 FC=0.5 mfg=Philips type=Zener)
..model NPN NPN(Is=7.049f Xti=3 Eg=1.11 Vaf=23.89 Bf=493.2 Ise=99.2f Ne=1.829 Ikf=.1542 Nk=.6339 Xtb=1.5 Br=2.886 Isc=7.371p Nc=1.508 Ikr=5.426 Rc=1.175 Cjc=5.5p Mjc=.3132 Vjc=.4924 Fc=.5 Cje=11.5p Mje=.6558 Vje=.5 Tr=10n Tf=420.3p Itf=1.374 Xtf=39.42 Vtf=10 mfg=PHILIPS)
..model PNP PNP(Is=1.02f Xti=3 Eg=1.11 Vaf=34.62 Bf=401.6 Ise=38.26p Ne=5.635 Ikf=74.73m Nk=.512 Xtb=1.5 Br=9.011 Isc=1.517f Nc=1.831 Ikr=.1469 Rc=1.151 Cjc=9.81p Mjc=.332 Vjc=.4865 Fc=.5 Cje=30p Mje=.3333 Vje=.5 Tr=10n Tf=524p Itf=.9847 Xtf=17.71 Vtf=10 mfg=PHILIPS)
..model NMOS VDMOS(Rg=3 Vto=2.2 Rd=22m Rs=5.5m Rb=28m Kp=10 lambda=.01 Cgdmax=.4n Cgdmin=.1n Cgs=.64n Cjo=.2n Is=25p mfg=Fairchild Vds=60 Ron=55m Qg=12.5n)
..model PMOS VDMOS(pchan Rg=3 Vto=-2.5 Rd=42m Rs=10.5m Rb=53m Kp=9 lambda=.01 Cgdmax=.5n Cgdmin=.12n Cgs=.8n Cjo=.24n Is=30p mfg=Fairchild Vds=-60 Ron=105m Qg=15n)

It could be simply an issue with guessing the MOSFET parameters, for
example M(Q)8... probably a long channel device to act like a variable
resistor... how long, who knows? Plus that output current "sink"
structure is plain-ass somebody's wet dream ;-)

It's a mystery to me entirely.

All I could make of it was a weird
feedback potential but there is meant to be feedback anyway. For now
I'll have to accept my ignorance on this one. Likewise most of what follows...

But the data sheet offers some clues, the strobe current is 220uA, so
that's the current in Q3/Q4.

Given the weirdness of that output stage I'm becoming inclined to
model it as a mix of behavioral blocks plus NPN's to match the data
sheet.

The data sheet indicates to me that it's not the world's gift to OpAmp
performance standards... why do you want to use it?

I found it performs very well in a laser diode analpog modulator I designed.
Better than other hobbyists are sellign boards to punters are acheiving. I
had to use LT1215 to sigificantly beat it. (And if you can recommend other
amps likely to share the essential qualities of those two (single rail,
fast slew rate), please do).

Also, it's been in general use for so long, and is always fairly easy to find.
Sometimes it might be better than using something newer (those hobbyists were
using new video amps and getting less performance, though my circuit might
suceed where theirs fail by using the amp(s) at unity gain or less and doing
the gain with a voltage regulator (of all things, but it DOES work...).

In short, I learned to like it a lot. I found it worked so often in various
things that I'd like to model it as first base in many new ideas.

I invented that Q11, Q12, Q13 "turnaround"... see Tom Frederiksen's
book, "Intuitive IC Op Amps", page 14. The schematic in the data
sheet is NOT balanced... I suspect a typo... maybe there are more.

Nasty possibility... If so, it persisted from the Harris original.

> ...Jim Thompson

Lostgallifreyan · Jan 24, 2012

Jim Thompson <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote in
news:d69uh7hsm9gpkg0tuhls48u5bidqqjh9cj@4ax.com:

PSpice _is_ a mixed-signal simulator. I can throw in digital
primitives right along with analog stuff. I often simply use 74HC
behaviorals to speed up my design cycle, then convert to device-level.

Hmm, logically I ought to apply that idea to my switches.

If I needed
enough of them I might, for speed. As it is I used the voltage source and the
voltage controlled switch, and used my own arrangement of arguments to input
to the voltage used as a pulse, then subcircuited all, with appropriate
symbols. It works but leaves me thinking there ought to be a better way.

Joerg · Jan 25, 2012

John Larkin wrote:

On Tue, 24 Jan 2012 06:08:28 -0800, John Larkin
jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Tue, 24 Jan 2012 20:25:32 +1100, "Phil Allison" <phil_a@tpg.com.au
wrote:

"Dishum"
I'm a hobbyist who doesn't have an ESR meter or (usually) a choice of
specific capacitor models and I'd like to have some idea of the kind of
ESR values one would expect from capacitors that are neither particularly
good nor particularly crappy in that respect. I'll really appreciate it if
you could cite some ballpark figures for -

1. 1uF/50V wet Al electrolytic
2. 100uF/25V wet Al electrolytic
3. 1000uF/50V wet Al electrolytic
4. 1uF/25V tantalum

** Measured from my parts bins:

1 = 3.1 ohms
2 = 0.5 ohms
3 = 0.06 ohms
4 = 4.2 ohms

In each case, the figure is for high frequency ESR or impedance at 100kHz.

ESR rises at low frequencies ( under 500Hz ) and falls with increasing
temperature.

Also, when an electro goes bad (ie dries out ) - ESR rises first followed
much later by a reduction in actual uF.

... Phil

That seems high for the tantalum. I seem to recall numbers like a
couple tenths of an ohm. I'll try a couple. I don't have an ESR meter,
but I can just apply a square wave from a 50 ohm generator and scope
the voltage.

Hmmm, both of the 1u caps have about the same ESR.

John

I tried a standard leaded gumdrop tantalum, 1 uF at 35 volts, and got
about 0.75 ohms.

22u 16v was about 0.35 ohms.

Tantalum ESR tends to be in sort of a sweet spot for taming voltage
regulators, both switchers and linear.

Until they have exploded, that is

--
Regards, Joerg

http://www.analogconsultants.com/

John Larkin · Jan 25, 2012

On Tue, 24 Jan 2012 17:02:23 -0800, Joerg <invalid@invalid.invalid>
wrote:

John Larkin wrote:
On Tue, 24 Jan 2012 06:08:28 -0800, John Larkin
jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Tue, 24 Jan 2012 20:25:32 +1100, "Phil Allison" <phil_a@tpg.com.au
wrote:

"Dishum"
I'm a hobbyist who doesn't have an ESR meter or (usually) a choice of
specific capacitor models and I'd like to have some idea of the kind of
ESR values one would expect from capacitors that are neither particularly
good nor particularly crappy in that respect. I'll really appreciate it if
you could cite some ballpark figures for -

1. 1uF/50V wet Al electrolytic
2. 100uF/25V wet Al electrolytic
3. 1000uF/50V wet Al electrolytic
4. 1uF/25V tantalum

** Measured from my parts bins:

1 = 3.1 ohms
2 = 0.5 ohms
3 = 0.06 ohms
4 = 4.2 ohms

In each case, the figure is for high frequency ESR or impedance at 100kHz.

ESR rises at low frequencies ( under 500Hz ) and falls with increasing
temperature.

Also, when an electro goes bad (ie dries out ) - ESR rises first followed
much later by a reduction in actual uF.

... Phil

That seems high for the tantalum. I seem to recall numbers like a
couple tenths of an ohm. I'll try a couple. I don't have an ESR meter,
but I can just apply a square wave from a 50 ohm generator and scope
the voltage.

Hmmm, both of the 1u caps have about the same ESR.

John

I tried a standard leaded gumdrop tantalum, 1 uF at 35 volts, and got
about 0.75 ohms.

22u 16v was about 0.35 ohms.

Tantalum ESR tends to be in sort of a sweet spot for taming voltage
regulators, both switchers and linear.

Until they have exploded, that is

Tantalums can be very reliable, much better than aluminums. You just
have to handle them properly.

John

John Larkin · Jan 25, 2012

On Tue, 24 Jan 2012 11:59:47 -0800, John Larkin
<jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Tue, 24 Jan 2012 06:08:28 -0800, John Larkin
jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Tue, 24 Jan 2012 20:25:32 +1100, "Phil Allison" <phil_a@tpg.com.au
wrote:

"Dishum"

I'm a hobbyist who doesn't have an ESR meter or (usually) a choice of
specific capacitor models and I'd like to have some idea of the kind of
ESR values one would expect from capacitors that are neither particularly
good nor particularly crappy in that respect. I'll really appreciate it if
you could cite some ballpark figures for -

1. 1uF/50V wet Al electrolytic
2. 100uF/25V wet Al electrolytic
3. 1000uF/50V wet Al electrolytic
4. 1uF/25V tantalum

** Measured from my parts bins:

1 = 3.1 ohms
2 = 0.5 ohms
3 = 0.06 ohms
4 = 4.2 ohms

In each case, the figure is for high frequency ESR or impedance at 100kHz.

ESR rises at low frequencies ( under 500Hz ) and falls with increasing
temperature.

Also, when an electro goes bad (ie dries out ) - ESR rises first followed
much later by a reduction in actual uF.

... Phil

That seems high for the tantalum. I seem to recall numbers like a
couple tenths of an ohm. I'll try a couple. I don't have an ESR meter,
but I can just apply a square wave from a 50 ohm generator and scope
the voltage.

Hmmm, both of the 1u caps have about the same ESR.

John

I tried a standard leaded gumdrop tantalum, 1 uF at 35 volts, and got
about 0.75 ohms.

22u 16v was about 0.35 ohms.

Tantalum ESR tends to be in sort of a sweet spot for taming voltage
regulators, both switchers and linear.

John

ftp://jjlarkin.lmi.net/Tant_ESR_Rig.JPG

ftp://jjlarkin.lmi.net/Tant_ESR_Scope.JPG

John

Tim Williams · Jan 25, 2012

"John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in
message news:7n0vh7do0pn8unfpn8p5gvcuth54ktpk4b@4ax.com...

Tantalum ESR tends to be in sort of a sweet spot for taming voltage
regulators, both switchers and linear.

ftp://jjlarkin.lmi.net/Tant_ESR_Rig.JPG

ftp://jjlarkin.lmi.net/Tant_ESR_Scope.JPG

Yes, suffice it to say, even though (dry slug) tantalums tend to have ESR
comparable to (higher grade) electrolytics, they are far, far simpler:
while lytics suffer from complicated ESL effects, tantalums are
essentially the inductance of the body length and that's it. The
equivalent circuit of C, ESR and ESL is very representative, and this
makes ripple voltage much more manageble.

And of course, polymers are just about ideal, yadda yadda.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

Phil Allison · Jan 25, 2012

"Tim Williams is so full of shit "

Yes, suffice it to say, even though (dry slug) tantalums tend to have ESR
comparable to (higher grade) electrolytics, they are far, far simpler:

** Their failure modes are many and failures far more common too.

while lytics suffer from complicated ESL effects,

** Horse poo.

tantalums are essentially the inductance of the body length and that's it.

** Same goes for most electros too.

.... Phil

John Larkin · Jan 25, 2012

On Tue, 24 Jan 2012 22:30:27 -0600, "Tim Williams"
<tmoranwms@charter.net> wrote:

"John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in
message news:7n0vh7do0pn8unfpn8p5gvcuth54ktpk4b@4ax.com...
Tantalum ESR tends to be in sort of a sweet spot for taming voltage
regulators, both switchers and linear.

ftp://jjlarkin.lmi.net/Tant_ESR_Rig.JPG

ftp://jjlarkin.lmi.net/Tant_ESR_Scope.JPG

Yes, suffice it to say, even though (dry slug) tantalums tend to have ESR
comparable to (higher grade) electrolytics, they are far, far simpler:
while lytics suffer from complicated ESL effects, tantalums are
essentially the inductance of the body length and that's it. The
equivalent circuit of C, ESR and ESL is very representative, and this
makes ripple voltage much more manageble.

Yup. On the gumdrops the leads are the most inductive part.

And of course, polymers are just about ideal, yadda yadda.

Yupyup.

John

John Larkin · Jan 25, 2012

On Wed, 25 Jan 2012 15:49:22 +1100, "Phil Allison" <phil_a@tpg.com.au>
wrote:

"Tim Williams is so full of shit "

Yes, suffice it to say, even though (dry slug) tantalums tend to have ESR
comparable to (higher grade) electrolytics, they are far, far simpler:

** Their failure modes are many and failures far more common too.

If you keep dV/dT down, they are very reliable.

John

Driver to drive?

Jim Thompson

Guest

Jim Thompson

Guest

John Larkin

Guest

Lostgallifreyan

Guest

Simon S Aysdie

Guest

Lostgallifreyan

Guest

Lostgallifreyan

Guest

Jim Thompson

Guest

Jim Thompson

Guest

Jim Thompson

Guest

Lostgallifreyan

Guest

Lostgallifreyan

Guest

Lostgallifreyan

Guest

Joerg

Guest

John Larkin

Guest

John Larkin

Guest

Tim Williams

Guest

Phil Allison

Guest

John Larkin

Guest

John Larkin

Guest

Log in

Welcome to EDABoard.com

Sponsor