Rail to Rail output stage for 2A 5V

[Winfield Hill]

Clive Arthur wrote...

Most downhole stuff is rated for 20kpsi, and 30kpsi ...

Clive, Schlumberger, with a large engineering group
near to me here in Cambridge, has hundreds of well-
experienced, highly-educated (PhDs, etc) engineers
and scientists developing downhole stuff. And they
have special parts to use that can handle 250C, maybe
higher. A good friend of mine works there. They DO
NOT tell anyone what they're doing or how they do it.


--
Thanks,
- Win

 

[Clive Arthur]

On 07/06/2019 12:37, Jasen Betts wrote:

<snip>

neither were designed to handle downhole pressure, but the box shape is
particularly vulnerable.

Most downhole stuff is rated for 20kpsi, and 30kpsi is not unusual.
That's 2000 atmospheres which needs a very thick-walled special steel tube.

Cheers
--
Clive

 

Clive Arthur <cliveta@nowaytoday.co.uk> wrote in
news:qddo5v$5ko$1@dont-email.me:

On 07/06/2019 12:37, Jasen Betts wrote:

snip

neither were designed to handle downhole pressure, but the box
shape is particularly vulnerable.

Most downhole stuff is rated for 20kpsi, and 30kpsi is not
unusual. That's 2000 atmospheres which needs a very thick-walled
special steel tube.

Cheers

Make a miniature "submarine" (a plumb bob) to enclose the array of
any tool you want down there.

What are the dimensions of such a box? IOW what are my physical
limits on the enclosure?

Steel vacuum cans that can also withstand the external pressure
you describe are quite possible, as I am sure you know.

And you also had a temp spec. How hot is it down there? If a gas
tube could be included in the tether, one could pump a cooling gas
down to take some heat off the bob.

Do these go behind a drill head or in place of it or? Is a
downhole probe an entirely different animal than the drill shaft?

 

[Clive Arthur]

On 07/06/2019 14:26, Winfield Hill wrote:

Clive Arthur wrote...

Most downhole stuff is rated for 20kpsi, and 30kpsi ...

Clive, Schlumberger, with a large engineering group
near to me here in Cambridge, has hundreds of well-
experienced, highly-educated (PhDs, etc) engineers
and scientists developing downhole stuff. And they
have special parts to use that can handle 250C, maybe
higher. A good friend of mine works there. They DO
NOT tell anyone what they're doing or how they do it.

Schlum stuff is usually good but *very* expensive. Wells are very
hostile environments and the risk of destruction or loss is high.

There are several other companies in the game, though mostly 'only' up
to 180'C, and anyway, Schlum staff don't stay there forever...

Cheers
--
Clive

 

[Marke]

Ti makes a high temp rail to rail op-amp they claim is good to 210C:
http://www.ti.com/lit/ds/symlink/ths4521-ht.pdf

Only 3.3V though...

 

[George Herold]

On Friday, June 7, 2019 at 11:46:39 AM UTC-4, Marke wrote:

Ti makes a high temp rail to rail op-amp they claim is good to 210C:
http://www.ti.com/lit/ds/symlink/ths4521-ht.pdf

Only 3.3V though...

Huh, Digikey lists 23 opamps that run at 210 (or even 225 C)
https://tinyurl.com/yyxbeas9


None of them are cheap.

George H.

 

[Lasse Langwadt Christense]

fredag den 7. juni 2019 kl. 15.50.53 UTC+2 skrev DecadentLinux...@decadence.org:

Clive Arthur <cliveta@nowaytoday.co.uk> wrote in
news:qddo5v$5ko$1@dont-email.me:

On 07/06/2019 12:37, Jasen Betts wrote:

snip

neither were designed to handle downhole pressure, but the box
shape is particularly vulnerable.

Most downhole stuff is rated for 20kpsi, and 30kpsi is not
unusual. That's 2000 atmospheres which needs a very thick-walled
special steel tube.

Cheers

Make a miniature "submarine" (a plumb bob) to enclose the array of
any tool you want down there.

What are the dimensions of such a box? IOW what are my physical
limits on the enclosure?

Steel vacuum cans that can also withstand the external pressure
you describe are quite possible, as I am sure you know.

there's a world of difference between 1 bar of vacuum and 2000bar pressure

 

Lasse Langwadt Christensen <langwadt@fonz.dk> wrote in
news:eb40ad41-053e-4b3c-8f86-4ec5d8643a32@googlegroups.com:

there's a world of difference between 1 bar of vacuum and 2000bar
pressure

I am merely saying that the circuit device must reside inside a
vacuum chamber that has that pressure on the outside. You are saying
that one out of 2000 is negligible, but the seal on that chamber needs
to handle it without failing

I know the difference. The gauge goes the other way and pegs the
needle. I get it. Sheesh... now I have peanut gallery commentary
crap.

 

[Jasen Betts]

On 2019-06-07, DecadentLinuxUserNumeroUno@decadence.org <DecadentLinuxUserNumeroUno@decadence.org> wrote:

Lasse Langwadt Christensen <langwadt@fonz.dk> wrote in
news:eb40ad41-053e-4b3c-8f86-4ec5d8643a32@googlegroups.com:

there's a world of difference between 1 bar of vacuum and 2000bar
pressure

I am merely saying that the circuit device must reside inside a
vacuum chamber that has that pressure on the outside. You are saying
that one out of 2000 is negligible, but the seal on that chamber needs
to handle it without failing

Ignoring the seals for the moment If you make the walls of the vessel
2000 times thicker it might not fit down the hole.

--
When I tried casting out nines I made a hash of it.

 

[John Larkin]

On Fri, 7 Jun 2019 10:03:06 -0700 (PDT), Lasse Langwadt Christensen
<langwadt@fonz.dk> wrote:

fredag den 7. juni 2019 kl. 15.50.53 UTC+2 skrev DecadentLinux...@decadence.org:
Clive Arthur <cliveta@nowaytoday.co.uk> wrote in
news:qddo5v$5ko$1@dont-email.me:

On 07/06/2019 12:37, Jasen Betts wrote:

snip

neither were designed to handle downhole pressure, but the box
shape is particularly vulnerable.

Most downhole stuff is rated for 20kpsi, and 30kpsi is not
unusual. That's 2000 atmospheres which needs a very thick-walled
special steel tube.

Cheers

Make a miniature "submarine" (a plumb bob) to enclose the array of
any tool you want down there.

What are the dimensions of such a box? IOW what are my physical
limits on the enclosure?

Steel vacuum cans that can also withstand the external pressure
you describe are quite possible, as I am sure you know.


there's a world of difference between 1 bar of vacuum and 2000bar pressure

I read a paper once, about using plastic ic's and transistors at high
ambient pressure, in an oil bath or potted. They worked but had DC
offsets. TO-can parts were crushed.




--

John Larkin Highland Technology, Inc
picosecond timing precision measurement

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

Jasen Betts <jasen@xnet.co.nz> wrote in
news:qdekj1$i1n$1@gonzo.revmaps.no-ip.org:

Ignoring the seals for the moment If you make the walls of the
vessel 2000 times thicker it might not fit down the hole.

I inquired as to the hole dimensions and max limits on a transducer
device shell.

The thing would not have to be 2000 times thicker as that medium's
strength goes up exponentially, not linearly, Consider an egg shape as
well, which also contributes to crush resistance.

 

[Clive Arthur]

On 05/06/2019 16:51, John Larkin wrote:

On Wed, 5 Jun 2019 15:27:13 +0100, Clive Arthur
cliveta@nowaytoday.co.uk> wrote:

I need a more-or-less (say <100mV) rail to rail output running from a 5V
supply and capable of sourcing and sinking 2A or so. Up to 50kHz.

Linear amplifier? Or switching to the rails?


It needs to work at >125'C and IC's like the OPA567 would get too hot.
I probably need op-amps around a discrete output stage, but what's the
best topology? Preferably no charge pumps used and no bias adjustments.

Cheers

Is 5V the only supply available?

Complementary follower mosfets would be easy but the gates would have
to be driven past the power rails. Flip the fets if only 5v is
available, but the bias gets a little trickier to avoid crossover
distortion.

I have a nice simulation, op-amps wrapped round complementary common
source MOSFETs, which looks good over a wide range of loads and to
within 250mV of the rails. Class B in fact, no Iq, but very little
crossover distortion - it's not meant to be Hi-Fi anyway.

But the LTspice .tran simulation at high loads running from a sine
source shows a few spikes on the output with the same sort of amplitude
as the signal. Zooming in shows these are between as a few tens of
femtoseconds long and a few picoseconds long. The alternate solver
doesn't show these.

Clearly these can't be real, but is the LTspice sim trying to tell me
something?

Cheers
--
Clive

 

[Clive Arthur]

On 14/06/2019 17:13, Winfield Hill wrote:

Clive Arthur wrote...

I have a nice simulation, op-amps wrapped round complementary
common source MOSFETs, which looks good ...

Let me warn you that MOSFET SPICE models are horrible
at modeling linear operating at current well below
their rated Ron-limited switching current. They get
g_m vs Id completely wrong, by factors of up to 100.

You need bench, not SPICE measurements, before talking
about looking good.

Thanks for the heads up, Win.

Cheers
--
Clive

 

[Winfield Hill]

Clive Arthur wrote...

I have a nice simulation, op-amps wrapped round complementary
common source MOSFETs, which looks good ...

Let me warn you that MOSFET SPICE models are horrible
at modeling linear operating at current well below
their rated Ron-limited switching current. They get
g_m vs Id completely wrong, by factors of up to 100.

You need bench, not SPICE measurements, before talking
about looking good.


--
Thanks,
- Win

 

[Phil Hobbs]

On 6/14/19 11:44 AM, Clive Arthur wrote:

On 05/06/2019 16:51, John Larkin wrote:
On Wed, 5 Jun 2019 15:27:13 +0100, Clive Arthur
cliveta@nowaytoday.co.uk> wrote:

I need a more-or-less (say <100mV) rail to rail output running from a 5V
supply and capable of sourcing and sinking 2A or so.  Up to 50kHz.

Linear amplifier? Or switching to the rails?


It needs to work at >125'C and IC's like the OPA567 would get too hot.
I probably need op-amps around a discrete output stage, but what's the
best topology?  Preferably no charge pumps used and no bias adjustments.

Cheers

Is 5V the only supply available?

Complementary follower mosfets would be easy but the gates would have
to be driven past the power rails. Flip the fets if only 5v is
available, but the bias gets a little trickier to avoid crossover
distortion.

I have a nice simulation, op-amps wrapped round complementary common
source MOSFETs, which looks good over a wide range of loads and to
within 250mV of the rails.  Class B in fact, no Iq, but very little
crossover distortion - it's not meant to be Hi-Fi anyway.

But the LTspice .tran simulation at high loads running from a sine
source shows a few spikes on the output with the same sort of amplitude
as the signal.  Zooming in shows these are between as a few tens of
femtoseconds long and a few picoseconds long.  The alternate solver
doesn't show these.

Clearly these can't be real, but is the LTspice sim trying to tell me
something?

Cheers

You aren't using the alternate solver always, and you should.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[Winfield Hill]

Clive Arthur wrote...

On 14/06/2019 17:13, Winfield Hill wrote:
Clive Arthur wrote...

I have a nice simulation, op-amps wrapped round complementary
common source MOSFETs, which looks good ...

Let me warn you that MOSFET SPICE models are horrible
at modeling linear operating at current well below
their rated Ron-limited switching current. They get
g_m vs Id completely wrong, by factors of up to 100.

You need bench, not SPICE measurements, before talking
about looking good.

Thanks for the heads up, Win.

You can take simple bench measurements on your preferred
MOSFET types, and create accurate SPICE models. Here's a
crude draft of our AoE x-Chapter writeup on the subject,
read pages 33-39, and also some of the references there.
You can create a model that will match your subthreshold
measurements. Then SPICE will give the right answers.

(Don't worry about weird figure locations, that's the TEX
automated processing trying to decide where to put stuff.
When a publisher processes the file, a human does the job.)

https://www.dropbox.com/s/5lofu2chyc66kzr/3x.5_MOSFETs_Linear-Transistors_DRAFT_2019-04-20_WH-2.pdf?dl=0

Many of use prefer to take gm measurements in the middle
of the operating region, or a few regions, and do a set
of analytic small-signal calculations, to evaluate our
compensation circuitry, etc. Who needs SPICE anyway?
Bench measurements rule.


--
Thanks,
- Win

 

[John Larkin]

On Fri, 14 Jun 2019 16:44:39 +0100, Clive Arthur
<cliveta@nowaytoday.co.uk> wrote:

On 05/06/2019 16:51, John Larkin wrote:
On Wed, 5 Jun 2019 15:27:13 +0100, Clive Arthur
cliveta@nowaytoday.co.uk> wrote:

I need a more-or-less (say <100mV) rail to rail output running from a 5V
supply and capable of sourcing and sinking 2A or so. Up to 50kHz.

Linear amplifier? Or switching to the rails?


It needs to work at >125'C and IC's like the OPA567 would get too hot.
I probably need op-amps around a discrete output stage, but what's the
best topology? Preferably no charge pumps used and no bias adjustments.

Cheers

Is 5V the only supply available?

Complementary follower mosfets would be easy but the gates would have
to be driven past the power rails. Flip the fets if only 5v is
available, but the bias gets a little trickier to avoid crossover
distortion.

I have a nice simulation, op-amps wrapped round complementary common
source MOSFETs, which looks good over a wide range of loads and to
within 250mV of the rails. Class B in fact, no Iq, but very little
crossover distortion - it's not meant to be Hi-Fi anyway.

But the LTspice .tran simulation at high loads running from a sine
source shows a few spikes on the output with the same sort of amplitude
as the signal. Zooming in shows these are between as a few tens of
femtoseconds long and a few picoseconds long. The alternate solver
doesn't show these.

Clearly these can't be real, but is the LTspice sim trying to tell me
something?

Cheers

Sometimes Spice does goofy things. Your circuit is probably OK.


--

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

[Clifford Heath]

On 15/6/19 12:00 pm, Winfield Hill wrote:

Clive Arthur wrote...
On 14/06/2019 17:13, Winfield Hill wrote:
Let me warn you that MOSFET SPICE models are horrible...
You need bench, not SPICE measurements, before talking
about looking good.
You can take simple bench measurements on your preferred
MOSFET types, and create accurate SPICE models. > Bench measurements rule.

Winfield, many thanks for sharing.

How would you go about modeling the large RF LDMOS FETs? Would these
simple bench measurements be useful at all? We are reverse engineering
the PCB design of an old VHF (225MHz) linear amp used for analog TV,
which used BLF278s. The aim is to create a SPICE model that will allow
us to scale it for better operation at 144MHz (though it works ok there
already).

Clifford Heath.

 

[Phil Hobbs]

On 6/14/19 10:00 PM, Winfield Hill wrote:

Clive Arthur wrote...

On 14/06/2019 17:13, Winfield Hill wrote:
Clive Arthur wrote...

I have a nice simulation, op-amps wrapped round complementary
common source MOSFETs, which looks good ...

Let me warn you that MOSFET SPICE models are horrible
at modeling linear operating at current well below
their rated Ron-limited switching current. They get
g_m vs Id completely wrong, by factors of up to 100.

You need bench, not SPICE measurements, before talking
about looking good.

Thanks for the heads up, Win.

You can take simple bench measurements on your preferred
MOSFET types, and create accurate SPICE models. Here's a
crude draft of our AoE x-Chapter writeup on the subject,
read pages 33-39, and also some of the references there.
You can create a model that will match your subthreshold
measurements. Then SPICE will give the right answers.

(Don't worry about weird figure locations, that's the TEX
automated processing trying to decide where to put stuff.
When a publisher processes the file, a human does the job.)

https://www.dropbox.com/s/5lofu2chyc66kzr/3x.5_MOSFETs_Linear-Transistors_DRAFT_2019-04-20_WH-2.pdf?dl=0

Interesting stuff, thanks. You may have been spending too much time in
Massachusetts, though--on P. 32 you have you have a LaTeX description
environment

Avalanche. Although the very term “avalanche” evokes images of mayhem
and destruction,(25) MOSFETs are not damaged by such insults as long as
the peak and repetitive avalanche ratings of the device are respected."
(and so forth).

Footnote 25 reads, "(25) Which bothers some men. The word itself makes
some men uncomfortable."

Actual guys like nothing better than blowing up stuff.

Cheers

Phil Hobbs

Many of use prefer to take gm measurements in the middle
of the operating region, or a few regions, and do a set
of analytic small-signal calculations, to evaluate our
compensation circuitry, etc. Who needs SPICE anyway?
Bench measurements rule.

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote in
news:Us-dnZlbHpn2CpnAnZ2dnUU7-SPNnZ2d@supernews.com:

snip

Avalanche. Although the very term “avalanche” evokes images
of mayhem and destruction,(25) MOSFETs are not damaged by such
insults as long as the peak and repetitive avalanche ratings of
the device are respected." (and so forth).

Footnote 25 reads, "(25) Which bothers some men. The word itself
makes some men uncomfortable."

Actual guys like nothing better than blowing up stuff.

Even God likes smoke release events.

He may not like some of the labels we use either though.