wide range supply...

[Phil Hobbs]

On 2020-08-22 12:47, Lasse Langwadt Christensen wrote:

lørdag den 22. august 2020 kl. 18.04.13 UTC+2 skrev Phil Hobbs:
On 2020-08-22 06:23, Klaus Kragelund wrote:
On Saturday, August 22, 2020 at 12:33:58 AM UTC+2, Phil Hobbs wrote:
On 2020-08-21 17:56, jlarkin@highlandsniptechnology.com wrote:
Given that I have 24 vdc, I want a supply adjustable from +1V to +100..
Stable, quiet, current limited. Roughly 10 watts out. I might want
several of these on a board, with independently programmable voltage.
Multi-channel pulse generator.

There are boost-buck chips around, but not to 100v, probably not very
quiet. This might be demanding for a sepic, too.

I was thinking that I could put a boost converter ahead of a linear
regulator. The boost output wouldn\'t go below 24ish, but that\'s ok...
the linear can work from that. I guess the boost output would always
stay 10 volts or so ahead of what the linear needs, kind of like the
trick of bootstrapping an LM317.

Could get a bit toasty providing +1V at 100 mA. (I assume that \"10
watts\" doesn\'t include 1V @ 10A.)

You could maybe put a buck before the boost, and at low voltage, let the
current pass through the boost inductor with the converter turned off.

A PFET in parallel with the buck would let you avoid loss of efficiency
at higher voltages, and right near +24, you could run both to avoid any
holes in the adjustment range due to the wall wart sagging a bit, for
instance.

I would do it with a boost first, then a buck

Boost has higher efficiency at high output voltage
Buck is easier to stabilize, so you can turn down the crossover for the boost


Do you have any fave bucks that will survive 100V on the supply pin, or
even half that?

LM5164 ?

Looks like a nice part, thanks. Have you used it?

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

 

[server]

On Sat, 22 Aug 2020 12:27:40 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2020-08-22 00:53, jlarkin@highlandsniptechnology.com wrote:
On Fri, 21 Aug 2020 22:19:19 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2020-08-21 21:01, dagmargoodboat@yahoo.com wrote:
On Friday, August 21, 2020 at 5:57:05 PM UTC-4, jla...@highlandsniptechnology.com wrote:
Given that I have 24 vdc, I want a supply adjustable from +1V to +100.
Stable, quiet, current limited. Roughly 10 watts out. I might want
several of these on a board, with independently programmable voltage.
Multi-channel pulse generator.

There are boost-buck chips around, but not to 100v, probably not very
quiet. This might be demanding for a sepic, too.

I was thinking that I could put a boost converter ahead of a linear
regulator. The boost output wouldn\'t go below 24ish, but that\'s ok...
the linear can work from that. I guess the boost output would always
stay 10 volts or so ahead of what the linear needs, kind of like the
trick of bootstrapping an LM317.

Phil\'s buck-into-a-boost is attractive. Here\'s a nuttier idea--you
could convert a buck into a boost with external switches when
needed, or revert to buck mode when that\'s appropriate.

(conceptual boost-mode operation)
Synchronous
Buck L1
.----. .-.-.-. D1
+24V >---| |--+--\' \' \' \'---+--->|---+---> +24..100V
| | | | |
| | | Q1 ||-\' ---
| | | ||<-. --- C1
\'----\' \'---------\'|--+ |
| | GND
GND GND

For buck mode, turn off Q1 and short D1 with a P-FET.

Cheers,
James Arthur


Fun!

I wonder if there are any sync bucks simple enough that that would work
stably. The feedback would have to be really slow to avoid the unstable
zero in CCM boost mode. ISTM it makes the volt-second problem a bit
worse, since in boost operation the left end of the inductor is at
ground instead of +24V.

We really need a fast synchronous version of the uA78S40 or MC34060 to
play these sorts of games with. I bet the inventor of the buck
regulator didn\'t think of the trick of grounding the output and letting
it pump its ground pin negative.

The cheap potted LM78xx replacements will do that. They are really
slick, for a few dollars.

I have persuaded the LTM8078 dual synchronous switcher brick to make
+24 into -2.5 and -5, or to make +5 and -5.

My original problem here seems to be non-trivial, especially if I want
to put maybe 5 on a board.


For a miniature laser driver board, I\'ve used AOZ1282CI-2 (400 mA, 1
MHz, SOT23) for an inverting buck to make -14 from +5, LMR23630 (2 MHz,
3A) for the class-H TEC driver, and TPS61175 (2 MHz, 3A) to make +15.
That boost won\'t take 24V input, but the whole works fit inside a 1\" x
1.5\" x 1/4\" board-level shield can (Leader Tech SMS-205 series).

Lotta stuff! How many layers?

The board is credit-card sized, stuffed both sides, and cooled via a 5
W/m/K gap pad.

I use some 8 w/mk stuff from 3G. About the color and consistancy of
used chewing gum.

With all those volt-seconds to deal with, your gizmo will probably need
a larger-volume inductor for the boost, and to make +100V I expect
you\'ll need a controller chip plus outboard FET, but on the plus side
you only need two supplies rather than three.

Cheers

Phil Hobbs

How about an LTC3803 autotransformer boost followed by a linear reg
with a current limit? The boost switcher would quit switching below 24
volts out. Above that, it would be programmed to keep maybe 10 volts
across the linear reg. Since my load is nominally 100 ohms (50 mine,
50 the customer\'s) the worst-case linear reg is when we make 12 volts
into 100 ohms, which only dissipates 1.4 watts in the linear reg. At
100 volts and 140 mA load, it also dissipates 1.4 watts.

I can use a SOT-89 depletion fet in the linear regulator. That solves
some gate drive problems. It resembles an old tube regulator circuit.

Of course, the customer can short my output too.




--

John Larkin Highland Technology, Inc

Science teaches us to doubt.

Claude Bernard

 

[server]

On Sat, 22 Aug 2020 12:41:28 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2020-08-22 12:31, jlarkin@highlandsniptechnology.com wrote:
On Sat, 22 Aug 2020 12:03:59 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2020-08-22 06:23, Klaus Kragelund wrote:
On Saturday, August 22, 2020 at 12:33:58 AM UTC+2, Phil Hobbs wrote:
On 2020-08-21 17:56, jlarkin@highlandsniptechnology.com wrote:
Given that I have 24 vdc, I want a supply adjustable from +1V to +100.
Stable, quiet, current limited. Roughly 10 watts out. I might want
several of these on a board, with independently programmable voltage.
Multi-channel pulse generator.

There are boost-buck chips around, but not to 100v, probably not very
quiet. This might be demanding for a sepic, too.

I was thinking that I could put a boost converter ahead of a linear
regulator. The boost output wouldn\'t go below 24ish, but that\'s ok...
the linear can work from that. I guess the boost output would always
stay 10 volts or so ahead of what the linear needs, kind of like the
trick of bootstrapping an LM317.

Could get a bit toasty providing +1V at 100 mA. (I assume that \"10
watts\" doesn\'t include 1V @ 10A.)

You could maybe put a buck before the boost, and at low voltage, let the
current pass through the boost inductor with the converter turned off.

A PFET in parallel with the buck would let you avoid loss of efficiency
at higher voltages, and right near +24, you could run both to avoid any
holes in the adjustment range due to the wall wart sagging a bit, for
instance.

I would do it with a boost first, then a buck

Boost has higher efficiency at high output voltage
Buck is easier to stabilize, so you can turn down the crossover for the boost


Do you have any fave bucks that will survive 100V on the supply pin, or
even half that?

I\'m frequently in need of a decent buck that\'ll run at 500 kHz or more
and handle >40V on the input, generally on account of wanting to make
-18 or so from +24. I usually buck down to ~+13, then invert from
there. (It\'s also an OK rail to make logic supplies from--mine usually
don\'t take a lot of current.)

Cheers

Phil Hobbs

Rather than battling the high voltage multiple times, I just go down
to 12 and split off from there. A couple of people make the little
3-pin things that will work from 72 volts.

Yeah, I\'ve come to the same conclusion. Those LMR23630s are pretty good
medicine if you don\'t mind the 650-ps edges. (I always put them in a
Faraday cage, so I don\'t care very much.)

Yikes. I\'ve seen some synchronous switchers that seemed optimized to
turn their substrate diodes into drift-step-recovery diodes. They
freaked out the front ends of opamps six inches away.

I have this little discrete Schmitt trigger circuit for startup
purposes--it runs the +3.3 linear supply for the MCU off +24 until the
low-voltage rail comes up. One dual NPN, one BSS84, a diode, and four
resistors. That way I can just have the micro deal with the supply
sequencing as it likes. (Since Simon has got so good at it, we put
Cortex M0+ or M4F/M0+ micros in lots of things that would have been
all-analogue previously.)

I like MAX809 for powerup reset. Sometimes I have it check a couple of
voltages and a thermistor.



--

John Larkin Highland Technology, Inc

Science teaches us to doubt.

Claude Bernard

 

[Phil Hobbs]

On 2020-08-22 13:03, jlarkin@highlandsniptechnology.com wrote:

On Sat, 22 Aug 2020 12:27:40 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2020-08-22 00:53, jlarkin@highlandsniptechnology.com wrote:
On Fri, 21 Aug 2020 22:19:19 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2020-08-21 21:01, dagmargoodboat@yahoo.com wrote:
On Friday, August 21, 2020 at 5:57:05 PM UTC-4, jla...@highlandsniptechnology.com wrote:
Given that I have 24 vdc, I want a supply adjustable from +1V to +100.
Stable, quiet, current limited. Roughly 10 watts out. I might want
several of these on a board, with independently programmable voltage.
Multi-channel pulse generator.

There are boost-buck chips around, but not to 100v, probably not very
quiet. This might be demanding for a sepic, too.

I was thinking that I could put a boost converter ahead of a linear
regulator. The boost output wouldn\'t go below 24ish, but that\'s ok...
the linear can work from that. I guess the boost output would always
stay 10 volts or so ahead of what the linear needs, kind of like the
trick of bootstrapping an LM317.

Phil\'s buck-into-a-boost is attractive. Here\'s a nuttier idea--you
could convert a buck into a boost with external switches when
needed, or revert to buck mode when that\'s appropriate.

(conceptual boost-mode operation)
Synchronous
Buck L1
.----. .-.-.-. D1
+24V >---| |--+--\' \' \' \'---+--->|---+---> +24..100V
| | | | |
| | | Q1 ||-\' ---
| | | ||<-. --- C1
\'----\' \'---------\'|--+ |
| | GND
GND GND

For buck mode, turn off Q1 and short D1 with a P-FET.

Cheers,
James Arthur


Fun!

I wonder if there are any sync bucks simple enough that that would work
stably. The feedback would have to be really slow to avoid the unstable
zero in CCM boost mode. ISTM it makes the volt-second problem a bit
worse, since in boost operation the left end of the inductor is at
ground instead of +24V.

We really need a fast synchronous version of the uA78S40 or MC34060 to
play these sorts of games with. I bet the inventor of the buck
regulator didn\'t think of the trick of grounding the output and letting
it pump its ground pin negative.

The cheap potted LM78xx replacements will do that. They are really
slick, for a few dollars.

I have persuaded the LTM8078 dual synchronous switcher brick to make
+24 into -2.5 and -5, or to make +5 and -5.

My original problem here seems to be non-trivial, especially if I want
to put maybe 5 on a board.


For a miniature laser driver board, I\'ve used AOZ1282CI-2 (400 mA, 1
MHz, SOT23) for an inverting buck to make -14 from +5, LMR23630 (2 MHz,
3A) for the class-H TEC driver, and TPS61175 (2 MHz, 3A) to make +15.
That boost won\'t take 24V input, but the whole works fit inside a 1\" x
1.5\" x 1/4\" board-level shield can (Leader Tech SMS-205 series).

Lotta stuff! How many layers?

Six, with blind vias so that we can put noncritical analogue stuff on
the back side of the SMPS box. On this board that\'s the TEC driver
stuff, since we can use big bypass caps on the TEC leads with no
worries. The ultraquiet laser driver section goes opposite the MCU,
which doesn\'t cause any magnetic worries.

Even though we\'re pulling amps from a +5V supply, there\'s no schmutz on
the diode laser bias even with zero dB attenuation on the SA and a 10-Hz
resolution bandwidth. The laser bias current is way sub-Poissonian, so
that\'s pretty indicative. Mr. Faraday was a very smart man, as others
have previously observed.

The board is credit-card sized, stuffed both sides, and cooled via a 5
W/m/K gap pad.

I use some 8 w/mk stuff from 3G. About the color and consistancy of
used chewing gum.

Same sort of stuff here. We just need it not to flow under thermal
cycling, to prevent gaps opening up. (That is, it has to be a soft
solid and not a thick liquid, so that we can maintain a nonzero preload.)

With all those volt-seconds to deal with, your gizmo will probably need
a larger-volume inductor for the boost, and to make +100V I expect
you\'ll need a controller chip plus outboard FET, but on the plus side
you only need two supplies rather than three.

How about an LTC3803 autotransformer boost followed by a linear reg
with a current limit? The boost switcher would quit switching below 24
volts out. Above that, it would be programmed to keep maybe 10 volts
across the linear reg. Since my load is nominally 100 ohms (50 mine,
50 the customer\'s) the worst-case linear reg is when we make 12 volts
into 100 ohms, which only dissipates 1.4 watts in the linear reg. At
100 volts and 140 mA load, it also dissipates 1.4 watts.

Could get a bit toasty if all five channels are doing that at once.

I\'ve never done a flyback supply, but the suggested application circuit
looks like an EMI failure waiting to happen. How about the Vlad
Memorial antiseptic converter, that does its snubbing by returning
turn-off transient to the output?

I can use a SOT-89 depletion fet in the linear regulator. That solves
some gate drive problems. It resembles an old tube regulator circuit.

Of course, the customer can short my output too.

Yeah, that\'s a problem.

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

 

[Phil Hobbs]

On 2020-08-22 13:11, jlarkin@highlandsniptechnology.com wrote:

On Sat, 22 Aug 2020 12:41:28 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2020-08-22 12:31, jlarkin@highlandsniptechnology.com wrote:
On Sat, 22 Aug 2020 12:03:59 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2020-08-22 06:23, Klaus Kragelund wrote:
On Saturday, August 22, 2020 at 12:33:58 AM UTC+2, Phil Hobbs wrote:
On 2020-08-21 17:56, jlarkin@highlandsniptechnology.com wrote:
Given that I have 24 vdc, I want a supply adjustable from +1V to +100.
Stable, quiet, current limited. Roughly 10 watts out. I might want
several of these on a board, with independently programmable voltage.
Multi-channel pulse generator.

There are boost-buck chips around, but not to 100v, probably not very
quiet. This might be demanding for a sepic, too.

I was thinking that I could put a boost converter ahead of a linear
regulator. The boost output wouldn\'t go below 24ish, but that\'s ok...
the linear can work from that. I guess the boost output would always
stay 10 volts or so ahead of what the linear needs, kind of like the
trick of bootstrapping an LM317.

Could get a bit toasty providing +1V at 100 mA. (I assume that \"10
watts\" doesn\'t include 1V @ 10A.)

You could maybe put a buck before the boost, and at low voltage, let the
current pass through the boost inductor with the converter turned off.

A PFET in parallel with the buck would let you avoid loss of efficiency
at higher voltages, and right near +24, you could run both to avoid any
holes in the adjustment range due to the wall wart sagging a bit, for
instance.

I would do it with a boost first, then a buck

Boost has higher efficiency at high output voltage
Buck is easier to stabilize, so you can turn down the crossover for the boost


Do you have any fave bucks that will survive 100V on the supply pin, or
even half that?

I\'m frequently in need of a decent buck that\'ll run at 500 kHz or more
and handle >40V on the input, generally on account of wanting to make
-18 or so from +24. I usually buck down to ~+13, then invert from
there. (It\'s also an OK rail to make logic supplies from--mine usually
don\'t take a lot of current.)


Rather than battling the high voltage multiple times, I just go down
to 12 and split off from there. A couple of people make the little
3-pin things that will work from 72 volts.

Yeah, I\'ve come to the same conclusion. Those LMR23630s are pretty good
medicine if you don\'t mind the 650-ps edges. (I always put them in a
Faraday cage, so I don\'t care very much.

Yikes. I\'ve seen some synchronous switchers that seemed optimized to
turn their substrate diodes into drift-step-recovery diodes. They
freaked out the front ends of opamps six inches away.

I remember. LM3103 or something?

I have this little discrete Schmitt trigger circuit for startup
purposes--it runs the +3.3 linear supply for the MCU off +24 until the
low-voltage rail comes up. One dual NPN, one BSS84, a diode, and four
resistors. That way I can just have the micro deal with the supply
sequencing as it likes. (Since Simon has got so good at it, we put
Cortex M0+ or M4F/M0+ micros in lots of things that would have been
all-analogue previously.)

I like MAX809 for powerup reset. Sometimes I have it check a couple of
voltages and a thermistor.

This is for the next step in the process, when the MCU has told the SMPS
to start but it hasn\'t come up yet, and for those unfortunate occasions
when something happens to the lower-voltage rail--the MCU switches back
to running of +24 with no muss, fuss, reboots, etc.

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

 

[server]

On Sat, 22 Aug 2020 13:25:37 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2020-08-22 13:03, jlarkin@highlandsniptechnology.com wrote:
On Sat, 22 Aug 2020 12:27:40 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2020-08-22 00:53, jlarkin@highlandsniptechnology.com wrote:
On Fri, 21 Aug 2020 22:19:19 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2020-08-21 21:01, dagmargoodboat@yahoo.com wrote:
On Friday, August 21, 2020 at 5:57:05 PM UTC-4, jla...@highlandsniptechnology.com wrote:
Given that I have 24 vdc, I want a supply adjustable from +1V to +100.
Stable, quiet, current limited. Roughly 10 watts out. I might want
several of these on a board, with independently programmable voltage.
Multi-channel pulse generator.

There are boost-buck chips around, but not to 100v, probably not very
quiet. This might be demanding for a sepic, too.

I was thinking that I could put a boost converter ahead of a linear
regulator. The boost output wouldn\'t go below 24ish, but that\'s ok...
the linear can work from that. I guess the boost output would always
stay 10 volts or so ahead of what the linear needs, kind of like the
trick of bootstrapping an LM317.

Phil\'s buck-into-a-boost is attractive. Here\'s a nuttier idea--you
could convert a buck into a boost with external switches when
needed, or revert to buck mode when that\'s appropriate.

(conceptual boost-mode operation)
Synchronous
Buck L1
.----. .-.-.-. D1
+24V >---| |--+--\' \' \' \'---+--->|---+---> +24..100V
| | | | |
| | | Q1 ||-\' ---
| | | ||<-. --- C1
\'----\' \'---------\'|--+ |
| | GND
GND GND

For buck mode, turn off Q1 and short D1 with a P-FET.

Cheers,
James Arthur


Fun!

I wonder if there are any sync bucks simple enough that that would work
stably. The feedback would have to be really slow to avoid the unstable
zero in CCM boost mode. ISTM it makes the volt-second problem a bit
worse, since in boost operation the left end of the inductor is at
ground instead of +24V.

We really need a fast synchronous version of the uA78S40 or MC34060 to
play these sorts of games with. I bet the inventor of the buck
regulator didn\'t think of the trick of grounding the output and letting
it pump its ground pin negative.

The cheap potted LM78xx replacements will do that. They are really
slick, for a few dollars.

I have persuaded the LTM8078 dual synchronous switcher brick to make
+24 into -2.5 and -5, or to make +5 and -5.

My original problem here seems to be non-trivial, especially if I want
to put maybe 5 on a board.


For a miniature laser driver board, I\'ve used AOZ1282CI-2 (400 mA, 1
MHz, SOT23) for an inverting buck to make -14 from +5, LMR23630 (2 MHz,
3A) for the class-H TEC driver, and TPS61175 (2 MHz, 3A) to make +15.
That boost won\'t take 24V input, but the whole works fit inside a 1\" x
1.5\" x 1/4\" board-level shield can (Leader Tech SMS-205 series).

Lotta stuff! How many layers?

Six, with blind vias so that we can put noncritical analogue stuff on
the back side of the SMPS box. On this board that\'s the TEC driver
stuff, since we can use big bypass caps on the TEC leads with no
worries. The ultraquiet laser driver section goes opposite the MCU,
which doesn\'t cause any magnetic worries.

Even though we\'re pulling amps from a +5V supply, there\'s no schmutz on
the diode laser bias even with zero dB attenuation on the SA and a 10-Hz
resolution bandwidth. The laser bias current is way sub-Poissonian, so
that\'s pretty indicative. Mr. Faraday was a very smart man, as others
have previously observed.

The board is credit-card sized, stuffed both sides, and cooled via a 5
W/m/K gap pad.

I use some 8 w/mk stuff from 3G. About the color and consistancy of
used chewing gum.

Same sort of stuff here. We just need it not to flow under thermal
cycling, to prevent gaps opening up. (That is, it has to be a soft
solid and not a thick liquid, so that we can maintain a nonzero preload.)


With all those volt-seconds to deal with, your gizmo will probably need
a larger-volume inductor for the boost, and to make +100V I expect
you\'ll need a controller chip plus outboard FET, but on the plus side
you only need two supplies rather than three.

How about an LTC3803 autotransformer boost followed by a linear reg
with a current limit? The boost switcher would quit switching below 24
volts out. Above that, it would be programmed to keep maybe 10 volts
across the linear reg. Since my load is nominally 100 ohms (50 mine,
50 the customer\'s) the worst-case linear reg is when we make 12 volts
into 100 ohms, which only dissipates 1.4 watts in the linear reg. At
100 volts and 140 mA load, it also dissipates 1.4 watts.

Could get a bit toasty if all five channels are doing that at once.

I\'ve never done a flyback supply, but the suggested application circuit
looks like an EMI failure waiting to happen. How about the Vlad
Memorial antiseptic converter, that does its snubbing by returning
turn-off transient to the output?

I can use a SOT-89 depletion fet in the linear regulator. That solves
some gate drive problems. It resembles an old tube regulator circuit.

Of course, the customer can short my output too.

Yeah, that\'s a problem.

Cheers

Phil Hobbs

I reserve the right to burn his fingers, to teach him a lesson.

Well, I did that to myself, on a hot SMA connector.

https://www.dropbox.com/s/caagy8kh8czd3fz/J270_On_Towel.jpg?raw=1

It gets pretty hot trying to put 44 volt pulses into a 1 ohm load,
without any cooling.



--

John Larkin Highland Technology, Inc

Science teaches us to doubt.

Claude Bernard

 

[server]

On Saturday, August 22, 2020 at 11:17:49 AM UTC-4, jla...@highlandsniptechnology.com wrote:

On Sat, 22 Aug 2020 04:04:01 -0700 (PDT), blocher@columbus.rr.com
wrote:

On Friday, August 21, 2020 at 5:57:05 PM UTC-4, jla...@highlandsniptechnology.com wrote:
Given that I have 24 vdc, I want a supply adjustable from +1V to +100.
Stable, quiet, current limited. Roughly 10 watts out. I might want
several of these on a board, with independently programmable voltage.
Multi-channel pulse generator.

There are boost-buck chips around, but not to 100v, probably not very
quiet. This might be demanding for a sepic, too.

I was thinking that I could put a boost converter ahead of a linear
regulator. The boost output wouldn\'t go below 24ish, but that\'s ok...
the linear can work from that. I guess the boost output would always
stay 10 volts or so ahead of what the linear needs, kind of like the
trick of bootstrapping an LM317.





--

John Larkin Highland Technology, Inc

Science teaches us to doubt.

Claude Bernard

If you start off the AC lines and move down?

We very rarely connect to the AC lines. Most of our stuff runs off
24VDC from a wall-wart. The better warts are cheap and work really
well and are plastered with compliance stickers, presumably real. It\'s
easier to write up a CE conformance report if you use a wart.

Our new rackmount AC source uses a MeanWell 48V, 1KW metal-box supply.
I hope no customer asks for a CE sticker on that one! That would take
some creative writing.

My new tiny pulse generator uses a 48V wart. The wart accepts
worldwide line voltages and comes with a set of international plug
adapters. I just linear regulate down from 48, to get +44 to +1, but
that\'s inefficient and won\'t scale for voltage or more channels.

The abstract problem is, what\'s a good way to get quiet 1 to 100 volts
DC in a couple of square inches of PCB area?


In the lab supplies I use I hear clicking sounds when certain voltage points are crossed. They must have more than one internal configuration

Some of ours do that. They are the heavy ones that probably have a
line-frequency power transformer, and they are switching taps with
relays.



--

John Larkin Highland Technology, Inc

Science teaches us to doubt.

Claude Bernard

I am not sure how much current you can get out of a voltage doubler, but maybe you regulate to some point and then double. you could probably find a way to regulate it even

 

[Phil Hobbs]

On 2020-08-22 12:27, Phil Hobbs wrote:

On 2020-08-22 00:53, jlarkin@highlandsniptechnology.com wrote:
On Fri, 21 Aug 2020 22:19:19 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2020-08-21 21:01, dagmargoodboat@yahoo.com wrote:
On Friday, August 21, 2020 at 5:57:05 PM UTC-4,
jla...@highlandsniptechnology.com wrote:
Given that I have 24 vdc, I want a supply adjustable from +1V to +100.
Stable, quiet, current limited. Roughly 10 watts out. I might want
several of these on a board, with independently programmable voltage.
Multi-channel pulse generator.

There are boost-buck chips around, but not to 100v, probably not very
quiet. This might be demanding for a sepic, too.

I was thinking that I could put a boost converter ahead of a linear
regulator. The boost output wouldn\'t go below 24ish, but that\'s ok...
the linear can work from that. I guess the boost output would always
stay 10 volts or so ahead of what the linear needs, kind of like the
trick of bootstrapping an LM317.
Phil\'s buck-into-a-boost is attractive.  Here\'s a nuttier idea--you
could convert a buck into a boost with external switches when
needed, or revert to buck mode when that\'s appropriate.

(conceptual boost-mode operation)
           Synchronous
              Buck        L1
             .----.     .-.-.-.       D1
    +24V >---|    |--+--\' \' \' \'---+--->|---+---> +24..100V
             |    |  |            |        |
             |    |  |     Q1  ||-\'       ---
             |    |  |         ||<-.      --- C1
             \'----\'  \'---------\'|--+       |
                |                  |      GND
               GND                GND

For buck mode, turn off Q1 and short D1 with a P-FET.

Cheers,
James Arthur


Fun!

I wonder if there are any sync bucks simple enough that that would work
stably.  The feedback would have to be really slow to avoid the unstable
zero in CCM boost mode.  ISTM it makes the volt-second problem a bit
worse, since in boost operation the left end of the inductor is at
ground instead of +24V.

We really need a fast synchronous version of the uA78S40 or MC34060 to
play these sorts of games with.  I bet the inventor of the buck
regulator didn\'t think of the trick of grounding the output and letting
it pump its ground pin negative.

The cheap potted LM78xx replacements will do that. They are really
slick, for a few dollars.

I have persuaded the LTM8078 dual synchronous switcher brick to make
+24 into -2.5 and -5, or to make +5 and -5.

My original problem here seems to be non-trivial, especially if I want
to put maybe 5 on a board.


For a miniature laser driver board, I\'ve used AOZ1282CI-2 (400 mA, 1
MHz, SOT23) for an inverting buck to make -14 from +5, LMR23630 (2 MHz,
3A) for the class-H TEC driver, and TPS61175 (2 MHz, 3A) to make +15.
That boost won\'t take 24V input, but the whole works fit inside a 1\" x
1.5\" x 1/4\" board-level shield can (Leader Tech SMS-205 series).

The board is credit-card sized, stuffed both sides, and cooled via a 5
W/m/K gap pad.

With all those volt-seconds to deal with, your gizmo will probably need
a larger-volume inductor for the boost, and to make +100V I expect
you\'ll need a controller chip plus outboard FET, but on the plus side
you only need two supplies rather than three.

I lied. I double-checked the schematic, and it turns out that there\'s
no AOZ inverter on this board, but the two others do fit nicely under
that 1.5 square inch shield.

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

 

[Michael_A_Terrell]

blocher@columbus.rr.com wrote:

I am not sure how much current you can get out of a voltage doubler, but maybe you regulate to some point and then double. you could probably find a way to regulate it even

Computer power supplies use a doubler when run from 1120 VAC. Some put
out over a KW. That is over eight amps through the doubler at the input.


--
Never piss off an Engineer!

They don\'t get mad.

They don\'t get even.

They go for over unity! ;-)

 

[Reinhardt Behm]

On 8/23/20 12:03 AM, Phil Hobbs wrote:

On 2020-08-22 06:23, Klaus Kragelund wrote:
On Saturday, August 22, 2020 at 12:33:58 AM UTC+2, Phil Hobbs wrote:
On 2020-08-21 17:56, jlarkin@highlandsniptechnology.com wrote:
Given that I have 24 vdc, I want a supply adjustable from +1V to +100.
Stable, quiet, current limited. Roughly 10 watts out. I might want
several of these on a board, with independently programmable voltage.
Multi-channel pulse generator.

There are boost-buck chips around, but not to 100v, probably not very
quiet. This might be demanding for a sepic, too.

I was thinking that I could put a boost converter ahead of a linear
regulator. The boost output wouldn\'t go below 24ish, but that\'s ok...
the linear can work from that. I guess the boost output would always
stay 10 volts or so ahead of what the linear needs, kind of like the
trick of bootstrapping an LM317.

Could get a bit toasty providing +1V at 100 mA.  (I assume that \"10
watts\" doesn\'t include 1V @ 10A.)

You could maybe put a buck before the boost, and at low voltage, let the
current pass through the boost inductor with the converter turned off.

A PFET in parallel with the buck would let you avoid loss of efficiency
at higher voltages, and right near +24, you could run both to avoid any
holes in the adjustment range due to the wall wart sagging a bit, for
instance.

I would do it with a boost first, then a buck

Boost has higher efficiency at high output voltage
Buck is easier to stabilize, so you can turn down the crossover for
the boost


Do you have any fave bucks that will survive 100V on the supply pin, or
even half that?

Half of that, or to be more exactly up to 65V: LT8620. Can supply 2A. I
used it for several avionics products where supply surge up to 60V are
common. Got it quiet enough with conducted emissions, too.

 

[Phil Hobbs]

On 2020-08-26 02:01, Reinhardt Behm wrote:

On 8/23/20 12:03 AM, Phil Hobbs wrote:
On 2020-08-22 06:23, Klaus Kragelund wrote:
On Saturday, August 22, 2020 at 12:33:58 AM UTC+2, Phil Hobbs wrote:
On 2020-08-21 17:56, jlarkin@highlandsniptechnology.com wrote:
Given that I have 24 vdc, I want a supply adjustable from +1V to +100.
Stable, quiet, current limited. Roughly 10 watts out. I might want
several of these on a board, with independently programmable voltage.
Multi-channel pulse generator.

There are boost-buck chips around, but not to 100v, probably not very
quiet. This might be demanding for a sepic, too.

I was thinking that I could put a boost converter ahead of a linear
regulator. The boost output wouldn\'t go below 24ish, but that\'s ok...
the linear can work from that. I guess the boost output would always
stay 10 volts or so ahead of what the linear needs, kind of like the
trick of bootstrapping an LM317.

Could get a bit toasty providing +1V at 100 mA.  (I assume that \"10
watts\" doesn\'t include 1V @ 10A.)

You could maybe put a buck before the boost, and at low voltage, let
the
current pass through the boost inductor with the converter turned off.

A PFET in parallel with the buck would let you avoid loss of efficiency
at higher voltages, and right near +24, you could run both to avoid any
holes in the adjustment range due to the wall wart sagging a bit, for
instance.

I would do it with a boost first, then a buck

Boost has higher efficiency at high output voltage
Buck is easier to stabilize, so you can turn down the crossover for
the boost


Do you have any fave bucks that will survive 100V on the supply pin,
or even half that?

Half of  that, or to be more exactly up to 65V: LT8620. Can supply 2A. I
used it for several avionics products where supply surge up to 60V are
common. Got it quiet enough with conducted emissions, too.

Yikes, six bucks in reels. Looks like a nice part though.

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

 

[server]

On Wed, 26 Aug 2020 14:01:55 +0800, Reinhardt Behm
<rbehm@hushmail.com> wrote:

On 8/23/20 12:03 AM, Phil Hobbs wrote:
On 2020-08-22 06:23, Klaus Kragelund wrote:
On Saturday, August 22, 2020 at 12:33:58 AM UTC+2, Phil Hobbs wrote:
On 2020-08-21 17:56, jlarkin@highlandsniptechnology.com wrote:
Given that I have 24 vdc, I want a supply adjustable from +1V to +100.
Stable, quiet, current limited. Roughly 10 watts out. I might want
several of these on a board, with independently programmable voltage.
Multi-channel pulse generator.

There are boost-buck chips around, but not to 100v, probably not very
quiet. This might be demanding for a sepic, too.

I was thinking that I could put a boost converter ahead of a linear
regulator. The boost output wouldn\'t go below 24ish, but that\'s ok...
the linear can work from that. I guess the boost output would always
stay 10 volts or so ahead of what the linear needs, kind of like the
trick of bootstrapping an LM317.

Could get a bit toasty providing +1V at 100 mA.  (I assume that \"10
watts\" doesn\'t include 1V @ 10A.)

You could maybe put a buck before the boost, and at low voltage, let the
current pass through the boost inductor with the converter turned off.

A PFET in parallel with the buck would let you avoid loss of efficiency
at higher voltages, and right near +24, you could run both to avoid any
holes in the adjustment range due to the wall wart sagging a bit, for
instance.

I would do it with a boost first, then a buck

Boost has higher efficiency at high output voltage
Buck is easier to stabilize, so you can turn down the crossover for
the boost


Do you have any fave bucks that will survive 100V on the supply pin, or
even half that?

Half of that, or to be more exactly up to 65V: LT8620. Can supply 2A. I
used it for several avionics products where supply surge up to 60V are
common. Got it quiet enough with conducted emissions, too.

There a circuit, flyback converter driving a voltage doubler, that in
theory outputs zero to infinite voltage. Well, sort of theory.



--

John Larkin Highland Technology, Inc

Science teaches us to doubt.

Claude Bernard

 

[Bill Sloman]

On Thursday, August 27, 2020 at 1:13:09 AM UTC+10, jla...@highlandsniptechnology.com wrote:

On Wed, 26 Aug 2020 14:01:55 +0800, Reinhardt Behm
rb...@hushmail.com> wrote:

On 8/23/20 12:03 AM, Phil Hobbs wrote:
On 2020-08-22 06:23, Klaus Kragelund wrote:
On Saturday, August 22, 2020 at 12:33:58 AM UTC+2, Phil Hobbs wrote:
On 2020-08-21 17:56, jla...@highlandsniptechnology.com wrote:

<snip>

There a circuit, flyback converter driving a voltage doubler, that in
theory outputs zero to infinite voltage. Well, sort of theory.

Only if you can find a flyback transformer with no inter-winding capacitance.

The energy stored in the inductor is finite, and once you have used all of it to charge the interwinding capacitance, you\'ve got as much voltage swing as you can get. The voltage doubler is just a bit more capacitance you have to charge.

The \"sort of theory\" involved seems to be the sort that\'s imperfectly understood.

--
Bill Sloman, Sydney