Maker Pro
Maker Pro

Linear 3.3V power supply anywhere?

J

Joerg

Jan 1, 1970
0
Tried Digikey, Mouser, Newark and some others, could not find any: A
linear 3.3V/2A (or more) power supply. No switchers.

What we really need is 3.3V/2A, 5V/2A, 12V/4A and -12V/4A but I can
piece that together as needed. The 5V and 12V is, of course, no big
deal. Condor has some nice ones with the good old uA723 on there, so low
in noise. But for some reason no 3.3V. I want to avoid modifying them
but maybe we just have to :-(
 
D

D from BC

Jan 1, 1970
0
Tried Digikey, Mouser, Newark and some others, could not find any: A
linear 3.3V/2A (or more) power supply. No switchers.

What we really need is 3.3V/2A, 5V/2A, 12V/4A and -12V/4A but I can
piece that together as needed. The 5V and 12V is, of course, no big
deal. Condor has some nice ones with the good old uA723 on there, so low
in noise. But for some reason no 3.3V. I want to avoid modifying them
but maybe we just have to :-(

I can guess at why you're not finding one....
1) This is near the drop out voltage of typical regulators
2) Switchers may rule this area
3) At 3.3V ..%temp drift maybe a concern. The internal reference in a
hot linear reg may have too much annoying drift.
4) Low customer demand
5) low dropout linear reg + power transistors = any current you want

Here's some quirky wisdom...
If you can't find it...you're wrong! :)

(No offense meant..)
D from BC
 
R

Rene Tschaggelar

Jan 1, 1970
0
Joerg said:
Tried Digikey, Mouser, Newark and some others, could not find any: A
linear 3.3V/2A (or more) power supply. No switchers.

What we really need is 3.3V/2A, 5V/2A, 12V/4A and -12V/4A but I can
piece that together as needed. The 5V and 12V is, of course, no big
deal. Condor has some nice ones with the good old uA723 on there, so low
in noise. But for some reason no 3.3V. I want to avoid modifying them
but maybe we just have to :-(

Joerg,
I guess this ist the domain of controllers and
logic, so the manufacturers assume a switcher to
be sufficient for all applications.
Though there are no sole 3.3V switchers available
here either.

Rene
 
J

Joerg

Jan 1, 1970
0
D said:
I can guess at why you're not finding one....
1) This is near the drop out voltage of typical regulators


Drop-out doesn't matter. You just feed in enough volts. It's easy to buy
2V linear supplies, even 0.5V duals. But no 3.3V.

2) Switchers may rule this area


They sure do :-(

3) At 3.3V ..%temp drift maybe a concern. The internal reference in a
hot linear reg may have too much annoying drift.


No problem. Designed many low voltage regulators with ye olde uA723 or
semi-discrete. Rock stable. I just don't want to design another one,
this project has enough other work and the missus ain't going to be
happy if I tell her that instead of taking her out to our favorite Thai
place I'll design a power supply.

4) Low customer demand


Looks like it. Nobody is concerned about noise no more...

5) low dropout linear reg + power transistors = any current you want

No more LDOs in this here consulting office. They've given us too much
grief, undocumented pathology mode, explosions etc.

Here's some quirky wisdom...
If you can't find it...you're wrong! :)

(No offense meant..)


Aw, that hurts. It just piles on to John Larkin's accusation of heresy
because we are swinging a VME backplane at 3.3V. Plus we chopped up the
bus and run SPI over it, but shhhht, don't tell.
 
J

Joerg

Jan 1, 1970
0
Rene said:
Joerg,
I guess this ist the domain of controllers and
logic, so the manufacturers assume a switcher to
be sufficient for all applications.
Though there are no sole 3.3V switchers available
here either.

If you ever need any of these Digikey has tons of 3.3V single modules.
AFAIK from Astec but as you said they are all switchers. Digikey lists
some of them as linear but they aren't.

Thing is, yes, we have to supply logic but that sits in a card cage that
must remain super duper whisper quiet. Even the bus must be stopped
during signal acquisition.

So, I guess I'll roll my own. Again. Dang. Or modify a 5V but their
crude SCR crowbars aren't easy to fix at such low voltages.
 
J

John Larkin

Jan 1, 1970
0
Aw, that hurts. It just piles on to John Larkin's accusation of heresy
because we are swinging a VME backplane at 3.3V. Plus we chopped up the
bus and run SPI over it, but shhhht, don't tell.

I heard that! What idiot suggested that you use a VME backplane in the
first place?

Do you want me to design a 5v to 3.3v linear regulator for you? So you
don't miss the pad thai and twice-cooked duck?

4 discretes should do it, maybe 3.

John
 
J

Joerg

Jan 1, 1970
0
John said:
I heard that! What idiot suggested that you use a VME backplane in the
first place?

Guilty. We needed a quick prototype, I spotted a fancy Schroff box with
all the bling on EBay same day, the client's engineer pokered by bidding
one buck above min and, bingo. I promise, in the final version we'll
call it something other than VME. But might still use a VME backplane.
If it's any comfort the really unorthodox matrix stuff happens on P2 :)

Do you want me to design a 5v to 3.3v linear regulator for you? So you
don't miss the pad thai and twice-cooked duck?

4 discretes should do it, maybe 3.

Thanks, but I've already butchered a PowerOne 5V supply. The only pain
is the crowbar. They use a SCR plus zener. It strays from 0.6-1.5V and
the trigger current from 3-32mA. And zeners in the 3V range aren't
exactly that great either. Maybe I'll just use a string of diodes.

The nice thing about these PowerOne supplies is that you can write a
nice mod procedure for production. Two screws and the circuit board
comes off. But tape the transistor in place beforehand.

This Thai restaurant is a tiny place that only the locals can find. But
quite genuine and always busy. If I order medium it drives the sweat
beads onto my scalp although I am used to very spicy food. Then there is
two more grades, hot, and Thai-hot. Oh, and the Singha beer...
 
D

D from BC

Jan 1, 1970
0
D from BC wrote:
[snip]
I can guess at why you're not finding one....
1) This is near the drop out voltage of typical regulators


Drop-out doesn't matter. You just feed in enough volts. It's easy to buy
2V linear supplies, even 0.5V duals. But no 3.3V.
[snip]

Ohh come on.... it does matter a bit..
Let's say the drop out voltage is 3Volts..
Regulator dissipation is 2A*3V=6 watts (or more)
A monolithic regulator might have some yucky drift too at 6Watts..
That's nearly = to the load dissipation at 6.6Watts.

I'm saying that a large dropout voltage results in a less efficient
linear regulator.
D from BC
 
J

Joerg

Jan 1, 1970
0
D said:
D from BC wrote:

[snip]
I can guess at why you're not finding one....
1) This is near the drop out voltage of typical regulators


Drop-out doesn't matter. You just feed in enough volts. It's easy to buy
2V linear supplies, even 0.5V duals. But no 3.3V.

[snip]

Ohh come on.... it does matter a bit..
Let's say the drop out voltage is 3Volts..
Regulator dissipation is 2A*3V=6 watts (or more)
A monolithic regulator might have some yucky drift too at 6Watts..
That's nearly = to the load dissipation at 6.6Watts.

I'm saying that a large dropout voltage results in a less efficient
linear regulator.


Sure but we live in Caleefohniah where the power can be anywhere between
110V and 130V, or sometimes 0V. So you need dropout just to accommodate
that. Dissipation doesn't matter right now and I am not a friend of
monolithic regulators for anything above a few hundred mA. The LM317 is
pretty much as far as I will go in that domain. I like the uA723 a lot,
plus some nice fat pass transistors.
 
R

Rich Grise

Jan 1, 1970
0
This Thai restaurant is a tiny place that only the locals can find. But
quite genuine and always busy. If I order medium it drives the sweat
beads onto my scalp although I am used to very spicy food. Then there is
two more grades, hot, and Thai-hot. Oh, and the Singha beer...

I'd like to find someone who imports Oscar "Port-Wine-type" "Champagne". ;-)

It tastes like soda pop, but works like Thunderbird. ;-)

Cheers!
Rich
 
J

John Larkin

Jan 1, 1970
0

Oh. I thought it might have been me.
We needed a quick prototype, I spotted a fancy Schroff box with
all the bling on EBay same day, the client's engineer pokered by bidding
one buck above min and, bingo. I promise, in the final version we'll
call it something other than VME. But might still use a VME backplane.
If it's any comfort the really unorthodox matrix stuff happens on P2 :)

There's a lot of cheap VME on ebay.
Thanks, but I've already butchered a PowerOne 5V supply. The only pain
is the crowbar. They use a SCR plus zener. It strays from 0.6-1.5V and
the trigger current from 3-32mA. And zeners in the 3V range aren't
exactly that great either. Maybe I'll just use a string of diodes.

The nice thing about these PowerOne supplies is that you can write a
nice mod procedure for production. Two screws and the circuit board
comes off. But tape the transistor in place beforehand.

This Thai restaurant is a tiny place that only the locals can find. But
quite genuine and always busy. If I order medium it drives the sweat
beads onto my scalp although I am used to very spicy food. Then there is
two more grades, hot, and Thai-hot. Oh, and the Singha beer...

Singha Gold is wonderful beer, but hard to find. The regular Singha is
pretty good, too.

We like Regent Thai on Church Street, right where they filmed "Sister
Act." It's run by all women who wear gorgeous silk Thai dresses. Great
food and cheap.

John
 
J

Joerg

Jan 1, 1970
0
John said:
Oh. I thought it might have been me.




There's a lot of cheap VME on ebay.




Singha Gold is wonderful beer, but hard to find. The regular Singha is
pretty good, too.

We like Regent Thai on Church Street, right where they filmed "Sister
Act." It's run by all women who wear gorgeous silk Thai dresses. Great
food and cheap.

Yeah, Sister Mary Clarence, a.k.a. Whoopi Goldberg. One of the best
movies. Often critics call it just a "comedy" but that doesn't do it
justice. There are a lot of civic (and religious) lessons in there.
Everyone can do great things, they just have to be "motivated".

Hey, why are places like Digikey out of stock on so much stuff these
days? Couldn't even get a decent paired fuse holder there, for crying
out loud. Even for blue 24AWG hookup wire there is a whopping supply of
one lone spool left. Arrgh.
 
D

D from BC

Jan 1, 1970
0
D said:
D from BC wrote:

[snip]

I can guess at why you're not finding one....
1) This is near the drop out voltage of typical regulators


Drop-out doesn't matter. You just feed in enough volts. It's easy to buy
2V linear supplies, even 0.5V duals. But no 3.3V.

[snip]

Ohh come on.... it does matter a bit..
Let's say the drop out voltage is 3Volts..
Regulator dissipation is 2A*3V=6 watts (or more)
A monolithic regulator might have some yucky drift too at 6Watts..
That's nearly = to the load dissipation at 6.6Watts.

I'm saying that a large dropout voltage results in a less efficient
linear regulator.


Sure but we live in Caleefohniah where the power can be anywhere between
110V and 130V, or sometimes 0V. So you need dropout just to accommodate
that. Dissipation doesn't matter right now and I am not a friend of
monolithic regulators for anything above a few hundred mA. The LM317 is
pretty much as far as I will go in that domain. I like the uA723 a lot,
plus some nice fat pass transistors.

The latest linear regulator I've played with is the LP2951.
Datasheet says a 3.3V version is available.
Of course pass transistors will be needed for 2A.
For example:
Digikey LP2950ACZ-3.3-ND is in stock.
D from BC
 
T

Tam/WB2TT

Jan 1, 1970
0
Joerg said:
If you ever need any of these Digikey has tons of 3.3V single modules.
AFAIK from Astec but as you said they are all switchers. Digikey lists
some of them as linear but they aren't.

Thing is, yes, we have to supply logic but that sits in a card cage that
must remain super duper whisper quiet. Even the bus must be stopped during
signal acquisition.

So, I guess I'll roll my own. Again. Dang. Or modify a 5V but their crude
SCR crowbars aren't easy to fix at such low voltages.

We ended up using an LDO regulator off the 5V. Board mounted. Not sure if
the 2 Amps will be a problem.

Tam
 
J

Joerg

Jan 1, 1970
0
D said:
D from BC wrote:

On Thu, 26 Apr 2007 13:29:50 -0700, Joerg



D from BC wrote:


[snip]


I can guess at why you're not finding one....
1) This is near the drop out voltage of typical regulators


Drop-out doesn't matter. You just feed in enough volts. It's easy to buy
2V linear supplies, even 0.5V duals. But no 3.3V.


[snip]

Ohh come on.... it does matter a bit..
Let's say the drop out voltage is 3Volts..
Regulator dissipation is 2A*3V=6 watts (or more)
A monolithic regulator might have some yucky drift too at 6Watts..
That's nearly = to the load dissipation at 6.6Watts.

I'm saying that a large dropout voltage results in a less efficient
linear regulator.


Sure but we live in Caleefohniah where the power can be anywhere between
110V and 130V, or sometimes 0V. So you need dropout just to accommodate
that. Dissipation doesn't matter right now and I am not a friend of
monolithic regulators for anything above a few hundred mA. The LM317 is
pretty much as far as I will go in that domain. I like the uA723 a lot,
plus some nice fat pass transistors.


The latest linear regulator I've played with is the LP2951.
Datasheet says a 3.3V version is available.
Of course pass transistors will be needed for 2A.
For example:
Digikey LP2950ACZ-3.3-ND is in stock.


The National LDO series is where I got my first big black eye. I was
already prejudiced against LDO but a client really liked it so they said
we should use it. The client is king, so I had to. Fired up the
prototype, the output voltage came up sluggishly and a "high amperage
smell" developed. Hung a scope onto the rail and it was singing the
blues, oscillating like crazy. Turned out it didn't like it when the
source impedance got too high but we had no choice. Naturally, that was
not mentioned in the data sheet. App Engineering told us more or less
"You are right, it cannot handle hi-Z, sorry about that". Great. I
forgot the exact part number. IIRC its wasn't LP but something like
LM2931. I forgot because I vowed never to design one in again. And I
never did, threw away the datasheet.

Not that I want to diss one particular vendor here. I've come across
more than one hardcore LDO problem. For example a TPS71550 that "didn't
like it" when the supply came up too fast. Phsssst ... BANG. Was that
mentioned anywhere? Oh no. TI wanted scope plots and what not. I had
sent them the rather simple schematic but they were not willing to throw
it onto SPICE. Would have been easy but not for the customer since they
do not release the innards anymore. So I voted that one off the island
as well ;-)
 
D

D from BC

Jan 1, 1970
0
[snip]
The National LDO series is where I got my first big black eye. I was
already prejudiced against LDO but a client really liked it so they said
we should use it. The client is king, so I had to. Fired up the
prototype, the output voltage came up sluggishly and a "high amperage
smell" developed. Hung a scope onto the rail and it was singing the
blues, oscillating like crazy. Turned out it didn't like it when the
source impedance got too high but we had no choice. Naturally, that was
not mentioned in the data sheet. App Engineering told us more or less
"You are right, it cannot handle hi-Z, sorry about that". Great. I
forgot the exact part number. IIRC its wasn't LP but something like
LM2931. I forgot because I vowed never to design one in again. And I
never did, threw away the datasheet.

Not that I want to diss one particular vendor here. I've come across
more than one hardcore LDO problem. For example a TPS71550 that "didn't
like it" when the supply came up too fast. Phsssst ... BANG. Was that
mentioned anywhere? Oh no. TI wanted scope plots and what not. I had
sent them the rather simple schematic but they were not willing to throw
it onto SPICE. Would have been easy but not for the customer since they
do not release the innards anymore. So I voted that one off the island
as well ;-)

What's the difference between an audio amplifier and a linear
regulator?

(I'll be working on the punch line...)
D from BC
 
M

Michael A. Terrell

Jan 1, 1970
0
D said:
What's the difference between an audio amplifier and a linear
regulator?

(I'll be working on the punch line...)


There are no "Regulatorphooles"?


--
Service to my country? Been there, Done that, and I've got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida
 
T

Tam/WB2TT

Jan 1, 1970
0
Joerg said:
D said:
D from BC wrote:


On Thu, 26 Apr 2007 13:29:50 -0700, Joerg



D from BC wrote:


[snip]


I can guess at why you're not finding one....
1) This is near the drop out voltage of typical regulators


Drop-out doesn't matter. You just feed in enough volts. It's easy to
buy 2V linear supplies, even 0.5V duals. But no 3.3V.


[snip]

Ohh come on.... it does matter a bit..
Let's say the drop out voltage is 3Volts..
Regulator dissipation is 2A*3V=6 watts (or more)
A monolithic regulator might have some yucky drift too at 6Watts..
That's nearly = to the load dissipation at 6.6Watts.

I'm saying that a large dropout voltage results in a less efficient
linear regulator.


Sure but we live in Caleefohniah where the power can be anywhere between
110V and 130V, or sometimes 0V. So you need dropout just to accommodate
that. Dissipation doesn't matter right now and I am not a friend of
monolithic regulators for anything above a few hundred mA. The LM317 is
pretty much as far as I will go in that domain. I like the uA723 a lot,
plus some nice fat pass transistors.


The latest linear regulator I've played with is the LP2951.
Datasheet says a 3.3V version is available.
Of course pass transistors will be needed for 2A.
For example:
Digikey LP2950ACZ-3.3-ND is in stock.


The National LDO series is where I got my first big black eye. I was
already prejudiced against LDO but a client really liked it so they said
we should use it. The client is king, so I had to. Fired up the prototype,
the output voltage came up sluggishly and a "high amperage smell"
developed. Hung a scope onto the rail and it was singing the blues,
oscillating like crazy. Turned out it didn't like it when the source
impedance got too high but we had no choice. Naturally, that was not
mentioned in the data sheet. App Engineering told us more or less "You are
right, it cannot handle hi-Z, sorry about that". Great. I forgot the exact
part number. IIRC its wasn't LP but something like LM2931. I forgot
because I vowed never to design one in again. And I never did, threw away
the datasheet.

Not that I want to diss one particular vendor here. I've come across more
than one hardcore LDO problem. For example a TPS71550 that "didn't like
it" when the supply came up too fast. Phsssst ... BANG. Was that mentioned
anywhere? Oh no. TI wanted scope plots and what not. I had sent them the
rather simple schematic but they were not willing to throw it onto SPICE.
Would have been easy but not for the customer since they do not release
the innards anymore. So I voted that one off the island as well ;-)

I don't think it is as bad as all that. We had two National LP3964
regulators on the board, with no problems. There were 10 uF Tantalums across
the input and output, plus a .22 on the output. In the one case, there was
even an RF choke in series with the 5 V feed, on the other side of the input
cap, of course. They do tell you that you need a bypass cap with an SER of
very close to 1 Ohm.

Tam
 
J

John Larkin

Jan 1, 1970
0
Joerg said:
D said:
On Fri, 27 Apr 2007 00:10:02 GMT, Joerg


D from BC wrote:


On Thu, 26 Apr 2007 13:29:50 -0700, Joerg



D from BC wrote:


[snip]


I can guess at why you're not finding one....
1) This is near the drop out voltage of typical regulators


Drop-out doesn't matter. You just feed in enough volts. It's easy to
buy 2V linear supplies, even 0.5V duals. But no 3.3V.


[snip]

Ohh come on.... it does matter a bit..
Let's say the drop out voltage is 3Volts..
Regulator dissipation is 2A*3V=6 watts (or more)
A monolithic regulator might have some yucky drift too at 6Watts..
That's nearly = to the load dissipation at 6.6Watts.

I'm saying that a large dropout voltage results in a less efficient
linear regulator.


Sure but we live in Caleefohniah where the power can be anywhere between
110V and 130V, or sometimes 0V. So you need dropout just to accommodate
that. Dissipation doesn't matter right now and I am not a friend of
monolithic regulators for anything above a few hundred mA. The LM317 is
pretty much as far as I will go in that domain. I like the uA723 a lot,
plus some nice fat pass transistors.


The latest linear regulator I've played with is the LP2951.
Datasheet says a 3.3V version is available.
Of course pass transistors will be needed for 2A.
For example:
Digikey LP2950ACZ-3.3-ND is in stock.


The National LDO series is where I got my first big black eye. I was
already prejudiced against LDO but a client really liked it so they said
we should use it. The client is king, so I had to. Fired up the prototype,
the output voltage came up sluggishly and a "high amperage smell"
developed. Hung a scope onto the rail and it was singing the blues,
oscillating like crazy. Turned out it didn't like it when the source
impedance got too high but we had no choice. Naturally, that was not
mentioned in the data sheet. App Engineering told us more or less "You are
right, it cannot handle hi-Z, sorry about that". Great. I forgot the exact
part number. IIRC its wasn't LP but something like LM2931. I forgot
because I vowed never to design one in again. And I never did, threw away
the datasheet.

Not that I want to diss one particular vendor here. I've come across more
than one hardcore LDO problem. For example a TPS71550 that "didn't like
it" when the supply came up too fast. Phsssst ... BANG. Was that mentioned
anywhere? Oh no. TI wanted scope plots and what not. I had sent them the
rather simple schematic but they were not willing to throw it onto SPICE.
Would have been easy but not for the customer since they do not release
the innards anymore. So I voted that one off the island as well ;-)

I don't think it is as bad as all that. We had two National LP3964
regulators on the board, with no problems. There were 10 uF Tantalums across
the input and output, plus a .22 on the output. In the one case, there was
even an RF choke in series with the 5 V feed, on the other side of the input
cap, of course. They do tell you that you need a bypass cap with an SER of
very close to 1 Ohm.

Tam

But tantalums explode. And the capacitor you see on the schematic,
right next to the ldo output pin, isn't usually the only capacitor on
that node. There could well be another bunch of microfarads of ceramic
bypasses all over the board, with a net esr in the milliohms.

And if you use an aluminum cap, and the temperature drops, the esr
goes to hell.

Any not-insane regulator should tolerate esr's down to zero.

John
 
R

Robert Latest

Jan 1, 1970
0
Joerg said:
No problem. Designed many low voltage regulators with ye olde uA723 or
semi-discrete. Rock stable. I just don't want to design another one,

Just how much design is a 3.3V linear regulator? How about a couple of
paralleled 317's?

robert
 
Top