Favorite Buck Regulators

Fred Bartoli · Nov 19, 2011

Joerg a écrit :

legg wrote:
On Fri, 18 Nov 2011 17:01:22 -0600, John S <sophi.2@invalid.org
wrote:

On 11/18/2011 3:10 PM, Joerg wrote:
Klaus Kragelund wrote:
On 18 Nov., 00:56, Joerg<inva...@invalid.invalid> wrote:
Tim Wescott wrote:
On Thu, 17 Nov 2011 13:51:35 -0600, k...@att.bizzzzzzzzzzzz wrote:
On Thu, 17 Nov 2011 13:18:21 -0600, Tim Wescott<t...@seemywebsite.com
wrote:
24V --> 3.3V, around 250mA, good enough for digital.
There are a bazillion out there -- it seems like anyone who even
pretends to sell analog circuit parts sells these, with a wide variety
of price ranges, features, etc.
So -- who do you think is best, and why? I'm mostly looking for parts
that actually work the way that the data sheet says they will, and that
don't have any really bad "gotchas". If anyone has been particularly
bad about supply hiccups -- mention that, too.
LTC, if you can afford them. Simulation is a big factor. I've had good
luck with TI, too.
LT certainly is nice. I had a National apps engineer tell me flat out
that they didn't have behavioral SPICE models of their switching
controllers, and that I should just use their web tool -- which lead me
to tell him flat out that National had just lost a sale (had I been doing
something exactly in line with what the regulator was designed for I may
have had a different answer -- but I still didn't like the attitude of
"oh just use our design tools, you don't actually have to understand what
our products do").
Just for kicks I had used Webbench on a few of my projects. Every single
one of them came back as impossible to build. Yet oodles of them have
come out of various production facilities over the years

--
I second that.

On a project that needed to finish quickly it was the easy path.

We used the National LM2267x and KM267x simple switchers. They should
be called trouble switchers. Designed it according to the
recommendations, and they would overshoot at startup, overshoot at
shutdown too. At a certain input voltage the controller just lost
control of the FET and a big blurp.

I am never going to use national switcher parts again...

Be happy that it only went "blurp". I had a LDO from them, client
insisted on using it against my advice. Long story short it had an
undocumented flaw. I gradually raised the source impedance during
testing ... *KABLAM* ... a capacitor vaporized. Turns out it burst into
oscillation when you got above a certain limit. The datasheet was
completely silent about this.

And they haven't fixed it in roughly 10 years. We had an undocumented
problem as well. Not exactly the same as yours, but caused us to
eliminate the Simple Switcher as a contender. We never looked back.
Gentlemen. Could we please have some part numbers? They are printed on
the bodies of the parts in question.

I dispise gossip, but dislike buggy parts a lot worse.

It's been very long ago but IIRC mine was the LM2931. Another part where
I have experienced unexpected pathologies not so long ago (in 2006) was
the TPS71550, also related to input impedance issues. Considering that
this thing is advertized for ultra low power that is a rather sad state
of affairs.

Long story short I despise LDOs and advise clients not to use LDOs.
Occasionally I am overruled but then at least I've told them

What did you do that went wrong?
I've used those very parts in several designs without any issue...

--
Thanks,
Fred.

Jim Thompson · Nov 19, 2011

On Sat, 19 Nov 2011 21:43:34 +0100, Fred Bartoli <" "> wrote:

Joerg a écrit :
legg wrote:
On Fri, 18 Nov 2011 17:01:22 -0600, John S <sophi.2@invalid.org
wrote:

On 11/18/2011 3:10 PM, Joerg wrote:
Klaus Kragelund wrote:
On 18 Nov., 00:56, Joerg<inva...@invalid.invalid> wrote:
Tim Wescott wrote:
On Thu, 17 Nov 2011 13:51:35 -0600, k...@att.bizzzzzzzzzzzz wrote:
On Thu, 17 Nov 2011 13:18:21 -0600, Tim Wescott<t...@seemywebsite.com
wrote:
24V --> 3.3V, around 250mA, good enough for digital.
There are a bazillion out there -- it seems like anyone who even
pretends to sell analog circuit parts sells these, with a wide variety
of price ranges, features, etc.
So -- who do you think is best, and why? I'm mostly looking for parts
that actually work the way that the data sheet says they will, and that
don't have any really bad "gotchas". If anyone has been particularly
bad about supply hiccups -- mention that, too.
LTC, if you can afford them. Simulation is a big factor. I've had good
luck with TI, too.
LT certainly is nice. I had a National apps engineer tell me flat out
that they didn't have behavioral SPICE models of their switching
controllers, and that I should just use their web tool -- which lead me
to tell him flat out that National had just lost a sale (had I been doing
something exactly in line with what the regulator was designed for I may
have had a different answer -- but I still didn't like the attitude of
"oh just use our design tools, you don't actually have to understand what
our products do").
Just for kicks I had used Webbench on a few of my projects. Every single
one of them came back as impossible to build. Yet oodles of them have
come out of various production facilities over the years

--
I second that.

On a project that needed to finish quickly it was the easy path.

We used the National LM2267x and KM267x simple switchers. They should
be called trouble switchers. Designed it according to the
recommendations, and they would overshoot at startup, overshoot at
shutdown too. At a certain input voltage the controller just lost
control of the FET and a big blurp.

I am never going to use national switcher parts again...

Be happy that it only went "blurp". I had a LDO from them, client
insisted on using it against my advice. Long story short it had an
undocumented flaw. I gradually raised the source impedance during
testing ... *KABLAM* ... a capacitor vaporized. Turns out it burst into
oscillation when you got above a certain limit. The datasheet was
completely silent about this.

And they haven't fixed it in roughly 10 years. We had an undocumented
problem as well. Not exactly the same as yours, but caused us to
eliminate the Simple Switcher as a contender. We never looked back.
Gentlemen. Could we please have some part numbers? They are printed on
the bodies of the parts in question.

I dispise gossip, but dislike buggy parts a lot worse.

It's been very long ago but IIRC mine was the LM2931. Another part where
I have experienced unexpected pathologies not so long ago (in 2006) was
the TPS71550, also related to input impedance issues. Considering that
this thing is advertized for ultra low power that is a rather sad state
of affairs.

Long story short I despise LDOs and advise clients not to use LDOs.
Occasionally I am overruled but then at least I've told them

What did you do that went wrong?
I've used those very parts in several designs without any issue...

Joerg · Nov 19, 2011

Fred Bartoli wrote:

Joerg a écrit :
legg wrote:
On Fri, 18 Nov 2011 17:01:22 -0600, John S <sophi.2@invalid.org
wrote:

On 11/18/2011 3:10 PM, Joerg wrote:
Klaus Kragelund wrote:
On 18 Nov., 00:56, Joerg<inva...@invalid.invalid> wrote:
Tim Wescott wrote:
On Thu, 17 Nov 2011 13:51:35 -0600, k...@att.bizzzzzzzzzzzz wrote:
On Thu, 17 Nov 2011 13:18:21 -0600, Tim
Wescott<t...@seemywebsite.com
wrote:
24V --> 3.3V, around 250mA, good enough for digital.
There are a bazillion out there -- it seems like anyone who even
pretends to sell analog circuit parts sells these, with a wide
variety
of price ranges, features, etc.
So -- who do you think is best, and why? I'm mostly looking
for parts
that actually work the way that the data sheet says they will,
and that
don't have any really bad "gotchas". If anyone has been
particularly
bad about supply hiccups -- mention that, too.
LTC, if you can afford them. Simulation is a big factor. I've
had good
luck with TI, too.
LT certainly is nice. I had a National apps engineer tell me
flat out
that they didn't have behavioral SPICE models of their switching
controllers, and that I should just use their web tool -- which
lead me
to tell him flat out that National had just lost a sale (had I
been doing
something exactly in line with what the regulator was designed
for I may
have had a different answer -- but I still didn't like the
attitude of
"oh just use our design tools, you don't actually have to
understand what
our products do").
Just for kicks I had used Webbench on a few of my projects. Every
single
one of them came back as impossible to build. Yet oodles of them
have
come out of various production facilities over the years

--
I second that.

On a project that needed to finish quickly it was the easy path.

We used the National LM2267x and KM267x simple switchers. They should
be called trouble switchers. Designed it according to the
recommendations, and they would overshoot at startup, overshoot at
shutdown too. At a certain input voltage the controller just lost
control of the FET and a big blurp.

I am never going to use national switcher parts again...

Be happy that it only went "blurp". I had a LDO from them, client
insisted on using it against my advice. Long story short it had an
undocumented flaw. I gradually raised the source impedance during
testing ... *KABLAM* ... a capacitor vaporized. Turns out it burst
into
oscillation when you got above a certain limit. The datasheet was
completely silent about this.

And they haven't fixed it in roughly 10 years. We had an
undocumented problem as well. Not exactly the same as yours, but
caused us to eliminate the Simple Switcher as a contender. We never
looked back.
Gentlemen. Could we please have some part numbers? They are printed on
the bodies of the parts in question.

I dispise gossip, but dislike buggy parts a lot worse.

It's been very long ago but IIRC mine was the LM2931. Another part where
I have experienced unexpected pathologies not so long ago (in 2006) was
the TPS71550, also related to input impedance issues. Considering that
this thing is advertized for ultra low power that is a rather sad state
of affairs.

Long story short I despise LDOs and advise clients not to use LDOs.
Occasionally I am overruled but then at least I've told them

What did you do that went wrong?
I've used those very parts in several designs without any issue...

We had a battery that provided around 20V and it was a small one.
Towards the end of the discharge cycle the natural battery thing
occurred, the impedance goes up. If you then switched on the line to the
battery ... *POP* ... the chip blew a crater. Communication with TI did
not provide a solution and so I designed it out.

If you are using them at lower voltages and/or with a low impedance
power source they may be ok. But I will not use them.

--
Regards, Joerg

http://www.analogconsultants.com/

Joerg · Nov 19, 2011

Jim Thompson wrote:

On Sat, 19 Nov 2011 21:43:34 +0100, Fred Bartoli <" "> wrote:

Joerg a écrit :
legg wrote:
On Fri, 18 Nov 2011 17:01:22 -0600, John S <sophi.2@invalid.org
wrote:

On 11/18/2011 3:10 PM, Joerg wrote:
Klaus Kragelund wrote:
On 18 Nov., 00:56, Joerg<inva...@invalid.invalid> wrote:
Tim Wescott wrote:
On Thu, 17 Nov 2011 13:51:35 -0600, k...@att.bizzzzzzzzzzzz wrote:
On Thu, 17 Nov 2011 13:18:21 -0600, Tim Wescott<t...@seemywebsite.com
wrote:
24V --> 3.3V, around 250mA, good enough for digital.
There are a bazillion out there -- it seems like anyone who even
pretends to sell analog circuit parts sells these, with a wide variety
of price ranges, features, etc.
So -- who do you think is best, and why? I'm mostly looking for parts
that actually work the way that the data sheet says they will, and that
don't have any really bad "gotchas". If anyone has been particularly
bad about supply hiccups -- mention that, too.
LTC, if you can afford them. Simulation is a big factor. I've had good
luck with TI, too.
LT certainly is nice. I had a National apps engineer tell me flat out
that they didn't have behavioral SPICE models of their switching
controllers, and that I should just use their web tool -- which lead me
to tell him flat out that National had just lost a sale (had I been doing
something exactly in line with what the regulator was designed for I may
have had a different answer -- but I still didn't like the attitude of
"oh just use our design tools, you don't actually have to understand what
our products do").
Just for kicks I had used Webbench on a few of my projects. Every single
one of them came back as impossible to build. Yet oodles of them have
come out of various production facilities over the years

--
I second that.

On a project that needed to finish quickly it was the easy path.

We used the National LM2267x and KM267x simple switchers. They should
be called trouble switchers. Designed it according to the
recommendations, and they would overshoot at startup, overshoot at
shutdown too. At a certain input voltage the controller just lost
control of the FET and a big blurp.

I am never going to use national switcher parts again...

Be happy that it only went "blurp". I had a LDO from them, client
insisted on using it against my advice. Long story short it had an
undocumented flaw. I gradually raised the source impedance during
testing ... *KABLAM* ... a capacitor vaporized. Turns out it burst into
oscillation when you got above a certain limit. The datasheet was
completely silent about this.

And they haven't fixed it in roughly 10 years. We had an undocumented
problem as well. Not exactly the same as yours, but caused us to
eliminate the Simple Switcher as a contender. We never looked back.
Gentlemen. Could we please have some part numbers? They are printed on
the bodies of the parts in question.

I dispise gossip, but dislike buggy parts a lot worse.

It's been very long ago but IIRC mine was the LM2931. Another part where
I have experienced unexpected pathologies not so long ago (in 2006) was
the TPS71550, also related to input impedance issues. Considering that
this thing is advertized for ultra low power that is a rather sad state
of affairs.

Long story short I despise LDOs and advise clients not to use LDOs.
Occasionally I am overruled but then at least I've told them

What did you do that went wrong?
I've used those very parts in several designs without any issue...

Joerg has continual *KABLAM* issues. ...

No, this one was more like .. tzt ... *POP* ... very fast.

... I think he has a black thumb ;-)

For some reason almost every time this happens the support folks scratch
their heads, things come to light that weren't quite made clear in the
datasheet, claims that were in the datasheets appeared to be written by
Marketing, and lots of other things depending on the case. The topper
was one large group that assembled at the mfg, the big old chip design
drawings came onto the table, and finally one guy exclaimed "Ooooh
S..T!". I had inadvertently unearthed yet another gotcha but this one
was serious. The group then went into a sort of damage control mode.

--
Regards, Joerg

http://www.analogconsultants.com/

Michael A. Terrell · Nov 20, 2011

Jim Thompson wrote:

Joerg has continual *KABLAM* issues. I think he has a black thumb ;-)

It's from that cheap, twisted hammer (with a broken face) that he
brought with him from Europe! ;-)

--
You can't have a sense of humor, if you have no sense.

Nico Coesel · Nov 20, 2011

Joerg <invalid@invalid.invalid> wrote:

For some reason almost every time this happens the support folks scratch
their heads, things come to light that weren't quite made clear in the
datasheet, claims that were in the datasheets appeared to be written by
Marketing, and lots of other things depending on the case. The topper
was one large group that assembled at the mfg, the big old chip design
drawings came onto the table, and finally one guy exclaimed "Ooooh
S..T!". I had inadvertently unearthed yet another gotcha but this one
was serious. The group then went into a sort of damage control mode.

Just be happy that you actually got some people together that can tell
you. Some components that I (have to) work with have undocumented
features or no complete specs at all. I'm currently writing some
software to control a camera module with a lot of image processing
features. I have a datasheet for the camera chip under NDA but it
seemed the software that comes with the development kit writes a lot
of undocumented registers as well. The manufacturer won't disclose
what those registers do. The lense focus control is another story.
There is no documentation at all

I managed to find out which chip
it uses and how its connected. But AD doesn't provide the datasheet. I
simply tried if it would work the same as a similar chip from AD and
it does.

--
Failure does not prove something is impossible, failure simply
indicates you are not using the right tools...
nico@nctdevpuntnl (punt=.)
--------------------------------------------------------------

Jamie · Nov 20, 2011

Joerg wrote:

Fred Bartoli wrote:

Joerg a écrit :

legg wrote:

On Fri, 18 Nov 2011 17:01:22 -0600, John S <sophi.2@invalid.org
wrote:

On 11/18/2011 3:10 PM, Joerg wrote:

Klaus Kragelund wrote:

On 18 Nov., 00:56, Joerg<inva...@invalid.invalid> wrote:

Tim Wescott wrote:

On Thu, 17 Nov 2011 13:51:35 -0600, k...@att.bizzzzzzzzzzzz wrote:

On Thu, 17 Nov 2011 13:18:21 -0600, Tim
Wescott<t...@seemywebsite.com
wrote:

24V --> 3.3V, around 250mA, good enough for digital.
There are a bazillion out there -- it seems like anyone who even
pretends to sell analog circuit parts sells these, with a wide
variety
of price ranges, features, etc.
So -- who do you think is best, and why? I'm mostly looking
for parts
that actually work the way that the data sheet says they will,
and that
don't have any really bad "gotchas". If anyone has been
particularly
bad about supply hiccups -- mention that, too.

LTC, if you can afford them. Simulation is a big factor. I've
had good
luck with TI, too.

LT certainly is nice. I had a National apps engineer tell me
flat out
that they didn't have behavioral SPICE models of their switching
controllers, and that I should just use their web tool -- which
lead me
to tell him flat out that National had just lost a sale (had I
been doing
something exactly in line with what the regulator was designed
for I may
have had a different answer -- but I still didn't like the
attitude of
"oh just use our design tools, you don't actually have to
understand what
our products do").

Just for kicks I had used Webbench on a few of my projects. Every
single
one of them came back as impossible to build. Yet oodles of them
have
come out of various production facilities over the years

--

I second that.

On a project that needed to finish quickly it was the easy path.

We used the National LM2267x and KM267x simple switchers. They should
be called trouble switchers. Designed it according to the
recommendations, and they would overshoot at startup, overshoot at
shutdown too. At a certain input voltage the controller just lost
control of the FET and a big blurp.

I am never going to use national switcher parts again...

Be happy that it only went "blurp". I had a LDO from them, client
insisted on using it against my advice. Long story short it had an
undocumented flaw. I gradually raised the source impedance during
testing ... *KABLAM* ... a capacitor vaporized. Turns out it burst
into
oscillation when you got above a certain limit. The datasheet was
completely silent about this.

And they haven't fixed it in roughly 10 years. We had an
undocumented problem as well. Not exactly the same as yours, but
caused us to eliminate the Simple Switcher as a contender. We never
looked back.

Gentlemen. Could we please have some part numbers? They are printed on
the bodies of the parts in question.

I dispise gossip, but dislike buggy parts a lot worse.

It's been very long ago but IIRC mine was the LM2931. Another part where
I have experienced unexpected pathologies not so long ago (in 2006) was
the TPS71550, also related to input impedance issues. Considering that
this thing is advertized for ultra low power that is a rather sad state
of affairs.

Long story short I despise LDOs and advise clients not to use LDOs.
Occasionally I am overruled but then at least I've told them

What did you do that went wrong?
I've used those very parts in several designs without any issue...

We had a battery that provided around 20V and it was a small one.
Towards the end of the discharge cycle the natural battery thing
occurred, the impedance goes up. If you then switched on the line to the
battery ... *POP* ... the chip blew a crater. Communication with TI did
not provide a solution and so I designed it out.

If you are using them at lower voltages and/or with a low impedance
power source they may be ok. But I will not use them.

sounds like the battery was acting as a shunt clamp. Sounds like a job

for a snubber/Zener..

Jamie

Joerg · Nov 20, 2011

Jamie wrote:

Joerg wrote:

[...]

We had a battery that provided around 20V and it was a small one.
Towards the end of the discharge cycle the natural battery thing
occurred, the impedance goes up. If you then switched on the line to the
battery ... *POP* ... the chip blew a crater. Communication with TI did
not provide a solution and so I designed it out.

If you are using them at lower voltages and/or with a low impedance
power source they may be ok. But I will not use them.

sounds like the battery was acting as a shunt clamp. Sounds like a job
for a snubber/Zener..

Well, I'd expect such needs to be mentioned in the datasheet. Even more
so as the TI part was (still is ...) advertised for the micro-power market.

The least I'd expect is that after such episodes someone revisits the
chip design and adds the essential information to the datasheet.

--
Regards, Joerg

http://www.analogconsultants.com/

Jon Elson · Nov 21, 2011

On 11/19/2011 01:00 PM, John S wrote:

Sorry I don't have a part number. As I said, that was 10 years ago. I
just remember that it was a three-terminal plastic package similar to a
to-220 and was called Simple Switcher.

The LM2575 is a 5-terminal TO-220-sized package, and they call that the
"simple Switcher'. Maybe they had an earlier version, but the general
scheme makes me think it would be hard to do in 3 terminals.
You need input, ground and output, and a sense connection, so it seems
it needs at least 4 terminals.

I have been using the LM2575S-5.0 in a product for a while, and have had
no problems with it so far.

Jon

John S · Nov 21, 2011

On 11/21/2011 1:50 PM, Jon Elson wrote:

On 11/19/2011 01:00 PM, John S wrote:

Sorry I don't have a part number. As I said, that was 10 years ago. I
just remember that it was a three-terminal plastic package similar to a
to-220 and was called Simple Switcher.

The LM2575 is a 5-terminal TO-220-sized package, and they call that the
"simple Switcher'. Maybe they had an earlier version, but the general
scheme makes me think it would be hard to do in 3 terminals.
You need input, ground and output, and a sense connection, so it seems
it needs at least 4 terminals.

I have been using the LM2575S-5.0 in a product for a while, and have had
no problems with it so far.

Jon

You are correct, of course. Too much to-220 stuff floating around in my
head.

Favorite Buck Regulators

Fred Bartoli

Guest

Jim Thompson

Guest

Joerg

Guest

Joerg

Guest

Michael A. Terrell

Guest

Nico Coesel

Guest

Jamie

Guest

Joerg

Guest

Jon Elson

Guest

John S

Guest

Welcome to EDABoard.com

Sponsor

Online statistics

Forum statistics

Favorite Buck Regulators

Fred Bartoli

Guest

Jim Thompson

Guest

Joerg

Guest

Joerg

Guest

Michael A. Terrell

Guest

Nico Coesel

Guest

Jamie

Guest

Joerg

Guest

Jon Elson

Guest

John S

Guest

Log in

Welcome to EDABoard.com

Sponsor