Low-noise isolated DC/DC...

[server]

On Sun, 9 Aug 2020 10:55:59 -0700 (PDT), whit3rd <whit3rd@gmail.com>
wrote:

On Sunday, August 9, 2020 at 4:06:42 AM UTC-7, upsid...@downunder.com wrote:
On Sun, 09 Aug 2020 08:18:41 GMT, Jan Panteltje
pNaOnStPeAlMtje@yahoo.com> wrote:

On a sunny day (Sun, 9 Aug 2020 09:00:46 +0200) it happened Piotr Wyderski
peter.pan@neverland.mil> wrote in <rgo6ut$2p9ah$1@portraits.wsisiz.edu.pl>:

Hi,

I need to provide power to a precision ADC over a 5+kV isolation
barrier. V_IN=10-16V, V_OUT=5V, I_OUT=60mA.

60W Edisonb lightbulb -> solar panel.
No noise.

This is a bit so and so. With a standard sun (1000 W/m**2) about 0.5
dm**2 of solar power is required to provide 0.5 W of electricity. A 7
cm x 7 cm panel or 80 mm diameter cell is required.

The question is, can a 60 W incandescent bulb provide the same
illumination as the sun on that panel or cell area using concentrating
mirrors and lenses ?

The question is further complicated by different spectral distribution
of the sun, the solar panel and the incandescent lamp. The sun
contains a lot of short wave radiation, while majority of incandescent
lamp radiation is in the near IR. The panels quantum efficiency for IR
is low.

That\'s not right; the typical silicon solar cell peak efficiency is in near IR

https://www.pveducation.org/sites/default/files/PVCDROM/Solar-Cell-Operation/Images/SRREAL.gif

The peak wavelength for a 2400 K black body (incandescent lamp) is at
1.2 um, thus a large portion drops outside the silicon solar panel
response,

Admittedly a lot radiation falls also inside the panel response. Some
have used such light sources even for generating visible light with
very bad efficiency for moe than a century .

 

[John Doe]

Link doesn\'t work?

 

[Piotr Wyderski]

jlarkin@highlandsniptechnology.com wrote:

> I was doodling a sinewave-based converter

The problem with sine waveform is that the output voltage would depend
heavily on the load. Might be a problem if you depend on PSR. A bipolar
rectangular waveform will keep the V_OUT rock-stable.

An older design just switches the power converters off some
microseconds before the light pulse. I guess that\'s not a bad idea.

This is a multichannel sigma-delta, not exactly sure how to sync it with
anything.

Best regards, Piotr

 

[Piotr Wyderski]

bitrex wrote:

> If the load is constant an isolated Cuk converter could be an option.

Hmm, an isolated Cuk, I forgot about this one. Thanks!

I expect no more than +/-20% load change at 300mW, probably less.
Custom magnetics are not an issue, so stealing 200mw more for logic is
attractive.

Best regards, Piotr

 

[server]

On Mon, 10 Aug 2020 11:45:33 +0200, Piotr Wyderski
<peter.pan@neverland.mil> wrote:

jlarkin@highlandsniptechnology.com wrote:

I was doodling a sinewave-based converter

The problem with sine waveform is that the output voltage would depend
heavily on the load. Might be a problem if you depend on PSR. A bipolar
rectangular waveform will keep the V_OUT rock-stable.

There\'s not a lot of difference between rectifying a 12 volts p-p sine
vs rectifying 12 volts p-p square wave. A sine is pretty flat on top.
I\'d lost-regulate some anyhow.

An older design just switches the power converters off some
microseconds before the light pulse. I guess that\'s not a bad idea.

This is a multichannel sigma-delta, not exactly sure how to sync it with
anything.

Best regards, Piotr

I was also considering a more conventional push-pull drive into a
transformer, but with slow, trapezoidal edges. That pretty much
eliminates using any stock driver chips.

I have a couple of weird constant-amplitude sinewave power converters
I need to Spice. One is a power Colpitts, and one is a strange
resonant phase-shift oscillator.

I\'m reading about WWII torpedo development. Colpitts himself was an
important player in that. The Fido air-dropped, antisub,
semi-intelligent acoustic homing torpedo was impressive, for being
developed fast and using tubes, and for working at all.



--

John Larkin Highland Technology, Inc

Science teaches us to doubt.

Claude Bernard

 

[server]

On Mon, 10 Aug 2020 11:50:15 +0200, Piotr Wyderski
<peter.pan@neverland.mil> wrote:

bitrex wrote:

If the load is constant an isolated Cuk converter could be an option.

Hmm, an isolated Cuk, I forgot about this one. Thanks!

I expect no more than +/-20% load change at 300mW, probably less.
Custom magnetics are not an issue, so stealing 200mw more for logic is
attractive.

Best regards, Piotr

TI has some cool logic isolators with power isolation included, with
highside power enough for other gadgets.



--

John Larkin Highland Technology, Inc

Science teaches us to doubt.

Claude Bernard

 

[Piotr Wyderski]

jlarkin@highlandsniptechnology.com wrote:

I was also considering a more conventional push-pull drive into a
transformer, but with slow, trapezoidal edges. That pretty much
eliminates using any stock driver chips.

This is what the LT3439 does.

I have a couple of weird constant-amplitude sinewave power converters
I need to Spice. One is a power Colpitts, and one is a strange
resonant phase-shift oscillator.

I have good experience with the two-transistor current-fed Royer. The
output was a sine, at least visually on the scope -- I didn\'t bother to
measure its quality any deeper.

I\'m reading about WWII torpedo development. Colpitts himself was an
important player in that. The Fido air-dropped, antisub,
semi-intelligent acoustic homing torpedo was impressive, for being
developed fast and using tubes, and for working at all.

Some reading on the German V2 might be interesting to you too. Those
guys achieved perfection with magamps.

Best regards, Piotr

 

[bitrex]

On 8/9/2020 11:03 PM, John Doe wrote:

Link doesn\'t work?

Don\'t know, seems to work ok for me. try the general manufacturer
product page there are many variants:

<https://www.silabs.com/isolation/isolated-gate-drivers>

Powerful isolated gate drivers and the prices are pretty reasonable;
among other things it makes doing stuff where you want to say drive an
H-bridge from a bipolar high-voltage supply from a microcontroller/logic
level easy, there\'s an isolation gap so you don\'t have to worry about
level-shifting schemes.

 

[bitrex]

On 8/10/2020 2:49 PM, bitrex wrote:

On 8/9/2020 11:03 PM, John Doe wrote:
Link doesn\'t work?


Don\'t know, seems to work ok for me. try the general manufacturer
product page there are many variants:

https://www.silabs.com/isolation/isolated-gate-drivers

Powerful isolated gate drivers and the prices are pretty reasonable;
among other things it makes doing stuff where you want to say drive an
H-bridge from a bipolar high-voltage supply from a microcontroller/logic
level easy, there\'s an isolation gap so you don\'t have to worry about
level-shifting schemes.

Only downside is all gate drivers that have isolation internally on the
chip seem to have moisture sensitivity, should be soldered in
immediately after opening the pack or baked if they\'re exposed to
moisture, or performance isn\'t guaranteed.

 

[server]

On Mon, 10 Aug 2020 20:25:12 +0200, Piotr Wyderski
<peter.pan@neverland.mil> wrote:

jlarkin@highlandsniptechnology.com wrote:

I was also considering a more conventional push-pull drive into a
transformer, but with slow, trapezoidal edges. That pretty much
eliminates using any stock driver chips.

This is what the LT3439 does.

That\'s interesting. It might even be worth $8.73. Plus the cost of a
custom transformer.

I have a couple of weird constant-amplitude sinewave power converters
I need to Spice. One is a power Colpitts, and one is a strange
resonant phase-shift oscillator.

I have good experience with the two-transistor current-fed Royer. The
output was a sine, at least visually on the scope -- I didn\'t bother to
measure its quality any deeper.

I\'m reading about WWII torpedo development. Colpitts himself was an
important player in that. The Fido air-dropped, antisub,
semi-intelligent acoustic homing torpedo was impressive, for being
developed fast and using tubes, and for working at all.

Some reading on the German V2 might be interesting to you too. Those
guys achieved perfection with magamps.

Best regards, Piotr

--

John Larkin Highland Technology, Inc

Science teaches us to doubt.

Claude Bernard

 

[Piotr Wyderski]

jlarkin@highlandsniptechnology.com wrote:

This is what the LT3439 does.

That\'s interesting. It might even be worth $8.73. Plus the cost of a
custom transformer.

If you need duty cycle control, then the LT3999 would be an option too.
Same principle, just more functionality.

Best regards, Piotr