24V to 500-1000V, 20W floating DC-DC converter...

[Mikko OH2HVJ]

Matt B <matt.blessinger@gmail.com> writes:

I need to make a 24V to 500-1000V, 20W adjustable power supply for a pulser. It needs to be a floating or negative converter.

My specs are:
Input: 24-28V
Output voltage: 500-1000V adjustable via control signal
Output current: 10mA @ 500V (5W), 20mA @ 1000V (20W)

If this is a one-off or a few pieces, how about Ultravolt 1C24-N20 ?
We used that (actually positive version) to start a product line and later
designed a custom one.

In our case, we needed robust short circuit protection and cap charging,
which required some knowledge in the flyback control loop - mostly
reduced frequency for low end voltage. Other items to take care of were
correct snubbering and resistor chains to deal with 1kV on SMD. And
discharging on powerdown to avoid accidents!

Transformer is a custom one with 4 HV sections and a 2-turn foil primary
around them for down to 14V operation. Each HV section was rectified to
a cap and the DC was in series.

--
mikko

 

[bitrex]

On 7/22/2020 9:43 AM, jlarkin@highlandsniptechnology.com wrote:

On Wed, 22 Jul 2020 06:08:26 -0700 (PDT), Matt B
matt.blessinger@gmail.com> wrote:

On Tuesday, July 21, 2020 at 10:32:47 PM UTC-5, jla...@highlandsniptechnology.com wrote:
I you want to design the whole circuit, take a look at LT3803 and its
app notes. It just always seems to work.

Scribble some circuits and we can talk about them. LT Spice includes
the 3803 too. It seems to work just like it simulates.

I generally like to design or at least modify a current design so I understand what\'s going on. This is new territory for me, so I\'ll look into a start some simulations. Thanks for the tip on the LTC3803.

The pragmatic way to do this, at least for a one-off, is to buy a
dc/dc brick that has a control input, and be done. But as you say, you
wouldn\'t learn much about HV supplies.

LTC/ADI has some starting-point simulations for the 3803.

I\'m playing with an isolated flyback right now, probably about 200
volts at very low current, possibly with a voltage regulator on the
high side. When my customer gets un-confused about what he actually
needs, I\'ll build him a demo.

Does you piezo move slow or fast? Win Hill has posted some interesting
piezo drive ideas here recently.

Says he\'s got an implied 1000 volt piezo stack with a capacitance
of...5uF? that doesn\'t sound right. the answer to how fast it moves is
\"slow\" if that\'s really what it is and you\'ve got 20 watts

 

[John Larkin]

On Wed, 22 Jul 2020 12:52:25 -0400, bitrex <user@example.net> wrote:

On 7/22/2020 9:43 AM, jlarkin@highlandsniptechnology.com wrote:
On Wed, 22 Jul 2020 06:08:26 -0700 (PDT), Matt B
matt.blessinger@gmail.com> wrote:

On Tuesday, July 21, 2020 at 10:32:47 PM UTC-5, jla...@highlandsniptechnology.com wrote:
I you want to design the whole circuit, take a look at LT3803 and its
app notes. It just always seems to work.

Scribble some circuits and we can talk about them. LT Spice includes
the 3803 too. It seems to work just like it simulates.

I generally like to design or at least modify a current design so I understand what\'s going on. This is new territory for me, so I\'ll look into a start some simulations. Thanks for the tip on the LTC3803.

The pragmatic way to do this, at least for a one-off, is to buy a
dc/dc brick that has a control input, and be done. But as you say, you
wouldn\'t learn much about HV supplies.

LTC/ADI has some starting-point simulations for the 3803.

I\'m playing with an isolated flyback right now, probably about 200
volts at very low current, possibly with a voltage regulator on the
high side. When my customer gets un-confused about what he actually
needs, I\'ll build him a demo.

Does you piezo move slow or fast? Win Hill has posted some interesting
piezo drive ideas here recently.

Says he\'s got an implied 1000 volt piezo stack with a capacitance
of...5uF? that doesn\'t sound right.

That would lift trucks, or buildings.



the answer to how fast it moves is
>\"slow\" if that\'s really what it is and you\'ve got 20 watts

 

[John Larkin]

On Wed, 22 Jul 2020 12:52:25 -0400, bitrex <user@example.net> wrote:

On 7/22/2020 9:43 AM, jlarkin@highlandsniptechnology.com wrote:
On Wed, 22 Jul 2020 06:08:26 -0700 (PDT), Matt B
matt.blessinger@gmail.com> wrote:

On Tuesday, July 21, 2020 at 10:32:47 PM UTC-5, jla...@highlandsniptechnology.com wrote:
I you want to design the whole circuit, take a look at LT3803 and its
app notes. It just always seems to work.

Scribble some circuits and we can talk about them. LT Spice includes
the 3803 too. It seems to work just like it simulates.

I generally like to design or at least modify a current design so I understand what\'s going on. This is new territory for me, so I\'ll look into a start some simulations. Thanks for the tip on the LTC3803.

The pragmatic way to do this, at least for a one-off, is to buy a
dc/dc brick that has a control input, and be done. But as you say, you
wouldn\'t learn much about HV supplies.

LTC/ADI has some starting-point simulations for the 3803.

I\'m playing with an isolated flyback right now, probably about 200
volts at very low current, possibly with a voltage regulator on the
high side. When my customer gets un-confused about what he actually
needs, I\'ll build him a demo.

Does you piezo move slow or fast? Win Hill has posted some interesting
piezo drive ideas here recently.

Says he\'s got an implied 1000 volt piezo stack with a capacitance
of...5uF? that doesn\'t sound right.

That would lift trucks, or buildings.



the answer to how fast it moves is
>\"slow\" if that\'s really what it is and you\'ve got 20 watts

 

[bitrex]

On 7/22/2020 2:26 PM, John Larkin wrote:

On Wed, 22 Jul 2020 12:52:25 -0400, bitrex <user@example.net> wrote:

On 7/22/2020 9:43 AM, jlarkin@highlandsniptechnology.com wrote:
On Wed, 22 Jul 2020 06:08:26 -0700 (PDT), Matt B
matt.blessinger@gmail.com> wrote:

On Tuesday, July 21, 2020 at 10:32:47 PM UTC-5, jla...@highlandsniptechnology.com wrote:
I you want to design the whole circuit, take a look at LT3803 and its
app notes. It just always seems to work.

Scribble some circuits and we can talk about them. LT Spice includes
the 3803 too. It seems to work just like it simulates.

I generally like to design or at least modify a current design so I understand what\'s going on. This is new territory for me, so I\'ll look into a start some simulations. Thanks for the tip on the LTC3803.

The pragmatic way to do this, at least for a one-off, is to buy a
dc/dc brick that has a control input, and be done. But as you say, you
wouldn\'t learn much about HV supplies.

LTC/ADI has some starting-point simulations for the 3803.

I\'m playing with an isolated flyback right now, probably about 200
volts at very low current, possibly with a voltage regulator on the
high side. When my customer gets un-confused about what he actually
needs, I\'ll build him a demo.

Does you piezo move slow or fast? Win Hill has posted some interesting
piezo drive ideas here recently.

Says he\'s got an implied 1000 volt piezo stack with a capacitance
of...5uF? that doesn\'t sound right.

That would lift trucks, or buildings.

Yes, the original flyback in the EDN article has as 0.01uF 1500V cap as
the main output cap and 0.022u 500Vs as the CW boost caps. this is a 4uF
1000V cap, damn:

<https://i.ebayimg.com/images/g/LYEAAOSwiONYNMh1/s-l500.jpg>

the EDN circuit isn\'t in the right ballpark.

 

[bitrex]

On 7/22/2020 2:26 PM, John Larkin wrote:

On Wed, 22 Jul 2020 12:52:25 -0400, bitrex <user@example.net> wrote:

On 7/22/2020 9:43 AM, jlarkin@highlandsniptechnology.com wrote:
On Wed, 22 Jul 2020 06:08:26 -0700 (PDT), Matt B
matt.blessinger@gmail.com> wrote:

On Tuesday, July 21, 2020 at 10:32:47 PM UTC-5, jla...@highlandsniptechnology.com wrote:
I you want to design the whole circuit, take a look at LT3803 and its
app notes. It just always seems to work.

Scribble some circuits and we can talk about them. LT Spice includes
the 3803 too. It seems to work just like it simulates.

I generally like to design or at least modify a current design so I understand what\'s going on. This is new territory for me, so I\'ll look into a start some simulations. Thanks for the tip on the LTC3803.

The pragmatic way to do this, at least for a one-off, is to buy a
dc/dc brick that has a control input, and be done. But as you say, you
wouldn\'t learn much about HV supplies.

LTC/ADI has some starting-point simulations for the 3803.

I\'m playing with an isolated flyback right now, probably about 200
volts at very low current, possibly with a voltage regulator on the
high side. When my customer gets un-confused about what he actually
needs, I\'ll build him a demo.

Does you piezo move slow or fast? Win Hill has posted some interesting
piezo drive ideas here recently.

Says he\'s got an implied 1000 volt piezo stack with a capacitance
of...5uF? that doesn\'t sound right.

That would lift trucks, or buildings.

Yes, the original flyback in the EDN article has as 0.01uF 1500V cap as
the main output cap and 0.022u 500Vs as the CW boost caps. this is a 4uF
1000V cap, damn:

<https://i.ebayimg.com/images/g/LYEAAOSwiONYNMh1/s-l500.jpg>

the EDN circuit isn\'t in the right ballpark.

 

[bitrex]

On 7/22/2020 2:26 PM, John Larkin wrote:

On Wed, 22 Jul 2020 12:52:25 -0400, bitrex <user@example.net> wrote:

On 7/22/2020 9:43 AM, jlarkin@highlandsniptechnology.com wrote:
On Wed, 22 Jul 2020 06:08:26 -0700 (PDT), Matt B
matt.blessinger@gmail.com> wrote:

On Tuesday, July 21, 2020 at 10:32:47 PM UTC-5, jla...@highlandsniptechnology.com wrote:
I you want to design the whole circuit, take a look at LT3803 and its
app notes. It just always seems to work.

Scribble some circuits and we can talk about them. LT Spice includes
the 3803 too. It seems to work just like it simulates.

I generally like to design or at least modify a current design so I understand what\'s going on. This is new territory for me, so I\'ll look into a start some simulations. Thanks for the tip on the LTC3803.

The pragmatic way to do this, at least for a one-off, is to buy a
dc/dc brick that has a control input, and be done. But as you say, you
wouldn\'t learn much about HV supplies.

LTC/ADI has some starting-point simulations for the 3803.

I\'m playing with an isolated flyback right now, probably about 200
volts at very low current, possibly with a voltage regulator on the
high side. When my customer gets un-confused about what he actually
needs, I\'ll build him a demo.

Does you piezo move slow or fast? Win Hill has posted some interesting
piezo drive ideas here recently.

Says he\'s got an implied 1000 volt piezo stack with a capacitance
of...5uF? that doesn\'t sound right.

That would lift trucks, or buildings.

Yes, the original flyback in the EDN article has as 0.01uF 1500V cap as
the main output cap and 0.022u 500Vs as the CW boost caps. this is a 4uF
1000V cap, damn:

<https://i.ebayimg.com/images/g/LYEAAOSwiONYNMh1/s-l500.jpg>

the EDN circuit isn\'t in the right ballpark.

 

[bitrex]

On 7/22/2020 3:02 PM, bitrex wrote:

On 7/22/2020 2:26 PM, John Larkin wrote:
On Wed, 22 Jul 2020 12:52:25 -0400, bitrex <user@example.net> wrote:

On 7/22/2020 9:43 AM, jlarkin@highlandsniptechnology.com wrote:
On Wed, 22 Jul 2020 06:08:26 -0700 (PDT), Matt B
matt.blessinger@gmail.com> wrote:

On Tuesday, July 21, 2020 at 10:32:47 PM UTC-5,
jla...@highlandsniptechnology.com wrote:
I you want to design the whole circuit, take a look at LT3803 and its
app notes. It just always seems to work.

Scribble some circuits and we can talk about them. LT Spice includes
the 3803 too. It seems to work just like it simulates.

I generally like to design or at least modify a current design so I
understand what\'s going on. This is new territory for me, so I\'ll
look into a start some simulations. Thanks for the tip on the LTC3803.

The pragmatic way to do this, at least for a one-off, is to buy a
dc/dc brick that has a control input, and be done. But as you say, you
wouldn\'t learn much about HV supplies.

LTC/ADI has some starting-point simulations for the 3803.

I\'m playing with an isolated flyback right now, probably about 200
volts at very low current, possibly with a voltage regulator on the
high side. When my customer gets un-confused about what he actually
needs, I\'ll build him a demo.

Does you piezo move slow or fast? Win Hill has posted some interesting
piezo drive ideas here recently.

Says he\'s got an implied 1000 volt piezo stack with a capacitance
of...5uF? that doesn\'t sound right.

That would lift trucks, or buildings.


Yes, the original flyback in the EDN article has as 0.01uF 1500V cap as
the main output cap and 0.022u 500Vs as the CW boost caps. this is a 4uF
1000V cap, damn:

https://i.ebayimg.com/images/g/LYEAAOSwiONYNMh1/s-l500.jpg

the EDN circuit isn\'t in the right ballpark.

Hoping he actually meant 5n or something

 

[bitrex]

On 7/22/2020 3:02 PM, bitrex wrote:

On 7/22/2020 2:26 PM, John Larkin wrote:
On Wed, 22 Jul 2020 12:52:25 -0400, bitrex <user@example.net> wrote:

On 7/22/2020 9:43 AM, jlarkin@highlandsniptechnology.com wrote:
On Wed, 22 Jul 2020 06:08:26 -0700 (PDT), Matt B
matt.blessinger@gmail.com> wrote:

On Tuesday, July 21, 2020 at 10:32:47 PM UTC-5,
jla...@highlandsniptechnology.com wrote:
I you want to design the whole circuit, take a look at LT3803 and its
app notes. It just always seems to work.

Scribble some circuits and we can talk about them. LT Spice includes
the 3803 too. It seems to work just like it simulates.

I generally like to design or at least modify a current design so I
understand what\'s going on. This is new territory for me, so I\'ll
look into a start some simulations. Thanks for the tip on the LTC3803.

The pragmatic way to do this, at least for a one-off, is to buy a
dc/dc brick that has a control input, and be done. But as you say, you
wouldn\'t learn much about HV supplies.

LTC/ADI has some starting-point simulations for the 3803.

I\'m playing with an isolated flyback right now, probably about 200
volts at very low current, possibly with a voltage regulator on the
high side. When my customer gets un-confused about what he actually
needs, I\'ll build him a demo.

Does you piezo move slow or fast? Win Hill has posted some interesting
piezo drive ideas here recently.

Says he\'s got an implied 1000 volt piezo stack with a capacitance
of...5uF? that doesn\'t sound right.

That would lift trucks, or buildings.


Yes, the original flyback in the EDN article has as 0.01uF 1500V cap as
the main output cap and 0.022u 500Vs as the CW boost caps. this is a 4uF
1000V cap, damn:

https://i.ebayimg.com/images/g/LYEAAOSwiONYNMh1/s-l500.jpg

the EDN circuit isn\'t in the right ballpark.

Hoping he actually meant 5n or something

 

[bitrex]

On 7/22/2020 3:02 PM, bitrex wrote:

On 7/22/2020 2:26 PM, John Larkin wrote:
On Wed, 22 Jul 2020 12:52:25 -0400, bitrex <user@example.net> wrote:

On 7/22/2020 9:43 AM, jlarkin@highlandsniptechnology.com wrote:
On Wed, 22 Jul 2020 06:08:26 -0700 (PDT), Matt B
matt.blessinger@gmail.com> wrote:

On Tuesday, July 21, 2020 at 10:32:47 PM UTC-5,
jla...@highlandsniptechnology.com wrote:
I you want to design the whole circuit, take a look at LT3803 and its
app notes. It just always seems to work.

Scribble some circuits and we can talk about them. LT Spice includes
the 3803 too. It seems to work just like it simulates.

I generally like to design or at least modify a current design so I
understand what\'s going on. This is new territory for me, so I\'ll
look into a start some simulations. Thanks for the tip on the LTC3803.

The pragmatic way to do this, at least for a one-off, is to buy a
dc/dc brick that has a control input, and be done. But as you say, you
wouldn\'t learn much about HV supplies.

LTC/ADI has some starting-point simulations for the 3803.

I\'m playing with an isolated flyback right now, probably about 200
volts at very low current, possibly with a voltage regulator on the
high side. When my customer gets un-confused about what he actually
needs, I\'ll build him a demo.

Does you piezo move slow or fast? Win Hill has posted some interesting
piezo drive ideas here recently.

Says he\'s got an implied 1000 volt piezo stack with a capacitance
of...5uF? that doesn\'t sound right.

That would lift trucks, or buildings.


Yes, the original flyback in the EDN article has as 0.01uF 1500V cap as
the main output cap and 0.022u 500Vs as the CW boost caps. this is a 4uF
1000V cap, damn:

https://i.ebayimg.com/images/g/LYEAAOSwiONYNMh1/s-l500.jpg

the EDN circuit isn\'t in the right ballpark.

Hoping he actually meant 5n or something

 

[George Herold]

On Wednesday, July 22, 2020 at 12:24:16 PM UTC-4, bitrex wrote:

On 7/22/2020 10:21 AM, George Herold wrote:
On Tuesday, July 21, 2020 at 6:51:49 PM UTC-4, Matt B wrote:
I need to make a 24V to 500-1000V, 20W adjustable power supply for a pulser. It needs to be a floating or negative converter.

My specs are:
Input: 24-28V
Output voltage: 500-1000V adjustable via control signal
Output current: 10mA @ 500V (5W), 20mA @ 1000V (20W)

I\'ve been reading through threads on different HV designs. So far the only one I\'ve come across that is isolated is Figure 50 of LT Application Note 29 (https://www.analog.com/media/en/technical-documentation/application-notes/an29f.pdf). It is specified for 1000V, 5W so not beefy enough. The transformer is also obsolete, but this comment mentions potential replacements. https://groups.google.com/d/msg/sci.electronics.design/F120QpxWWkc/74ljdPZ_BQAJ

There\'s also this design (https://www.edn.com/1-kv-power-supply-produces-a-continuous-arc/) that is 1kV, 20W. I just don\'t know enough about making it adjustable and isolated.

Thoughts on how to proceed?

If you are going to try building it yourself, then I\'d start with the
500V 5 mA one first. I built my first flyback with the LT3083,

(2018... thread here
https://groups.google.com/forum/?fromgroups=&hl=en#!searchin/sci.electronics.design/my$20first$20flyback%7Csort:date/sci.electronics.design/FQAIY1VaYBY/mLvGKJC2AgAJ

Good luck. Oh buying a ready made HV supply would be a good thing.
Gives you something to compare to if/when you have problems putting
to whole system together.

George H.


Yep. OP did not like that I disparaged/was skeptical of his plan to
build a 20 watt 1000 volt converter out of the gate but it strikes me as
a certain kind of madness as a first project if OP has not designed and
built a flyback of any type before.

Since you bring it up, it bothered me when you said that too.
I assume people mostly know what they are talking about.
Why not be nice?

BTW 5 uF for a piezo stack doesn\'t strike me as impossible.
I used this little multilayer stacks (for motion) that were ~0.1 uf
(only 150 V.)
https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=61

George H.

Even if someone here were able to provide a perfect custom design that
fit the bill the probability of something not working right on a first
prototype is high, and the probability of being able to figure it out
without a real working reference design or experience building more
modest switchers, is low

https://thedailywtf.com/articles/plz-email-me-teh-codez

 

[George Herold]

On Wednesday, July 22, 2020 at 12:24:16 PM UTC-4, bitrex wrote:

On 7/22/2020 10:21 AM, George Herold wrote:
On Tuesday, July 21, 2020 at 6:51:49 PM UTC-4, Matt B wrote:
I need to make a 24V to 500-1000V, 20W adjustable power supply for a pulser. It needs to be a floating or negative converter.

My specs are:
Input: 24-28V
Output voltage: 500-1000V adjustable via control signal
Output current: 10mA @ 500V (5W), 20mA @ 1000V (20W)

I\'ve been reading through threads on different HV designs. So far the only one I\'ve come across that is isolated is Figure 50 of LT Application Note 29 (https://www.analog.com/media/en/technical-documentation/application-notes/an29f.pdf). It is specified for 1000V, 5W so not beefy enough. The transformer is also obsolete, but this comment mentions potential replacements. https://groups.google.com/d/msg/sci.electronics.design/F120QpxWWkc/74ljdPZ_BQAJ

There\'s also this design (https://www.edn.com/1-kv-power-supply-produces-a-continuous-arc/) that is 1kV, 20W. I just don\'t know enough about making it adjustable and isolated.

Thoughts on how to proceed?

If you are going to try building it yourself, then I\'d start with the
500V 5 mA one first. I built my first flyback with the LT3083,

(2018... thread here
https://groups.google.com/forum/?fromgroups=&hl=en#!searchin/sci.electronics.design/my$20first$20flyback%7Csort:date/sci.electronics.design/FQAIY1VaYBY/mLvGKJC2AgAJ

Good luck. Oh buying a ready made HV supply would be a good thing.
Gives you something to compare to if/when you have problems putting
to whole system together.

George H.


Yep. OP did not like that I disparaged/was skeptical of his plan to
build a 20 watt 1000 volt converter out of the gate but it strikes me as
a certain kind of madness as a first project if OP has not designed and
built a flyback of any type before.

Since you bring it up, it bothered me when you said that too.
I assume people mostly know what they are talking about.
Why not be nice?

BTW 5 uF for a piezo stack doesn\'t strike me as impossible.
I used this little multilayer stacks (for motion) that were ~0.1 uf
(only 150 V.)
https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=61

George H.

Even if someone here were able to provide a perfect custom design that
fit the bill the probability of something not working right on a first
prototype is high, and the probability of being able to figure it out
without a real working reference design or experience building more
modest switchers, is low

https://thedailywtf.com/articles/plz-email-me-teh-codez

 

[Matt B]

On Wednesday, July 22, 2020 at 2:04:25 PM UTC-5, bitrex wrote:

On 7/22/2020 3:02 PM, bitrex wrote:
On 7/22/2020 2:26 PM, John Larkin wrote:
On Wed, 22 Jul 2020 12:52:25 -0400, bitrex <user@example.net> wrote:

On 7/22/2020 9:43 AM, jlarkin@highlandsniptechnology.com wrote:
On Wed, 22 Jul 2020 06:08:26 -0700 (PDT), Matt B
m...@gmail.com> wrote:

On Tuesday, July 21, 2020 at 10:32:47 PM UTC-5,
jla...@highlandsniptechnology.com wrote:
I you want to design the whole circuit, take a look at LT3803 and its
app notes. It just always seems to work.

Scribble some circuits and we can talk about them. LT Spice includes
the 3803 too. It seems to work just like it simulates.

I generally like to design or at least modify a current design so I
understand what\'s going on. This is new territory for me, so I\'ll
look into a start some simulations. Thanks for the tip on the LTC3803.

The pragmatic way to do this, at least for a one-off, is to buy a
dc/dc brick that has a control input, and be done. But as you say, you
wouldn\'t learn much about HV supplies.

LTC/ADI has some starting-point simulations for the 3803.

I\'m playing with an isolated flyback right now, probably about 200
volts at very low current, possibly with a voltage regulator on the
high side. When my customer gets un-confused about what he actually
needs, I\'ll build him a demo.

Does you piezo move slow or fast? Win Hill has posted some interesting
piezo drive ideas here recently.

Says he\'s got an implied 1000 volt piezo stack with a capacitance
of...5uF? that doesn\'t sound right.

That would lift trucks, or buildings.


Yes, the original flyback in the EDN article has as 0.01uF 1500V cap as
the main output cap and 0.022u 500Vs as the CW boost caps. this is a 4uF
1000V cap, damn:

https://i.ebayimg.com/images/g/LYEAAOSwiONYNMh1/s-l500.jpg

the EDN circuit isn\'t in the right ballpark.



Hoping he actually meant 5n or something

The 5uF is the bulk capacitance of the pulser circuit, not the piezo element, to prevent droop during pulses. So the caps are not fully charged and discharged every cycle that would need to fast cap charge capability of the Ultravolt unit. The bulk capacitance will also be \"separated\" from the power supply with a resistor to slow charge the caps.

 

[Matt B]

On Wednesday, July 22, 2020 at 2:04:25 PM UTC-5, bitrex wrote:

On 7/22/2020 3:02 PM, bitrex wrote:
On 7/22/2020 2:26 PM, John Larkin wrote:
On Wed, 22 Jul 2020 12:52:25 -0400, bitrex <user@example.net> wrote:

On 7/22/2020 9:43 AM, jlarkin@highlandsniptechnology.com wrote:
On Wed, 22 Jul 2020 06:08:26 -0700 (PDT), Matt B
m...@gmail.com> wrote:

On Tuesday, July 21, 2020 at 10:32:47 PM UTC-5,
jla...@highlandsniptechnology.com wrote:
I you want to design the whole circuit, take a look at LT3803 and its
app notes. It just always seems to work.

Scribble some circuits and we can talk about them. LT Spice includes
the 3803 too. It seems to work just like it simulates.

I generally like to design or at least modify a current design so I
understand what\'s going on. This is new territory for me, so I\'ll
look into a start some simulations. Thanks for the tip on the LTC3803.

The pragmatic way to do this, at least for a one-off, is to buy a
dc/dc brick that has a control input, and be done. But as you say, you
wouldn\'t learn much about HV supplies.

LTC/ADI has some starting-point simulations for the 3803.

I\'m playing with an isolated flyback right now, probably about 200
volts at very low current, possibly with a voltage regulator on the
high side. When my customer gets un-confused about what he actually
needs, I\'ll build him a demo.

Does you piezo move slow or fast? Win Hill has posted some interesting
piezo drive ideas here recently.

Says he\'s got an implied 1000 volt piezo stack with a capacitance
of...5uF? that doesn\'t sound right.

That would lift trucks, or buildings.


Yes, the original flyback in the EDN article has as 0.01uF 1500V cap as
the main output cap and 0.022u 500Vs as the CW boost caps. this is a 4uF
1000V cap, damn:

https://i.ebayimg.com/images/g/LYEAAOSwiONYNMh1/s-l500.jpg

the EDN circuit isn\'t in the right ballpark.



Hoping he actually meant 5n or something

The 5uF is the bulk capacitance of the pulser circuit, not the piezo element, to prevent droop during pulses. So the caps are not fully charged and discharged every cycle that would need to fast cap charge capability of the Ultravolt unit. The bulk capacitance will also be \"separated\" from the power supply with a resistor to slow charge the caps.

 

[Matt B]

On Wednesday, July 22, 2020 at 2:04:25 PM UTC-5, bitrex wrote:

On 7/22/2020 3:02 PM, bitrex wrote:
On 7/22/2020 2:26 PM, John Larkin wrote:
On Wed, 22 Jul 2020 12:52:25 -0400, bitrex <user@example.net> wrote:

On 7/22/2020 9:43 AM, jlarkin@highlandsniptechnology.com wrote:
On Wed, 22 Jul 2020 06:08:26 -0700 (PDT), Matt B
m...@gmail.com> wrote:

On Tuesday, July 21, 2020 at 10:32:47 PM UTC-5,
jla...@highlandsniptechnology.com wrote:
I you want to design the whole circuit, take a look at LT3803 and its
app notes. It just always seems to work.

Scribble some circuits and we can talk about them. LT Spice includes
the 3803 too. It seems to work just like it simulates.

I generally like to design or at least modify a current design so I
understand what\'s going on. This is new territory for me, so I\'ll
look into a start some simulations. Thanks for the tip on the LTC3803.

The pragmatic way to do this, at least for a one-off, is to buy a
dc/dc brick that has a control input, and be done. But as you say, you
wouldn\'t learn much about HV supplies.

LTC/ADI has some starting-point simulations for the 3803.

I\'m playing with an isolated flyback right now, probably about 200
volts at very low current, possibly with a voltage regulator on the
high side. When my customer gets un-confused about what he actually
needs, I\'ll build him a demo.

Does you piezo move slow or fast? Win Hill has posted some interesting
piezo drive ideas here recently.

Says he\'s got an implied 1000 volt piezo stack with a capacitance
of...5uF? that doesn\'t sound right.

That would lift trucks, or buildings.


Yes, the original flyback in the EDN article has as 0.01uF 1500V cap as
the main output cap and 0.022u 500Vs as the CW boost caps. this is a 4uF
1000V cap, damn:

https://i.ebayimg.com/images/g/LYEAAOSwiONYNMh1/s-l500.jpg

the EDN circuit isn\'t in the right ballpark.



Hoping he actually meant 5n or something

The 5uF is the bulk capacitance of the pulser circuit, not the piezo element, to prevent droop during pulses. So the caps are not fully charged and discharged every cycle that would need to fast cap charge capability of the Ultravolt unit. The bulk capacitance will also be \"separated\" from the power supply with a resistor to slow charge the caps.

 

[bitrex]

On 7/22/2020 3:30 PM, George Herold wrote:

On Wednesday, July 22, 2020 at 12:24:16 PM UTC-4, bitrex wrote:
On 7/22/2020 10:21 AM, George Herold wrote:
On Tuesday, July 21, 2020 at 6:51:49 PM UTC-4, Matt B wrote:
I need to make a 24V to 500-1000V, 20W adjustable power supply for a pulser. It needs to be a floating or negative converter.

My specs are:
Input: 24-28V
Output voltage: 500-1000V adjustable via control signal
Output current: 10mA @ 500V (5W), 20mA @ 1000V (20W)

I\'ve been reading through threads on different HV designs. So far the only one I\'ve come across that is isolated is Figure 50 of LT Application Note 29 (https://www.analog.com/media/en/technical-documentation/application-notes/an29f.pdf). It is specified for 1000V, 5W so not beefy enough. The transformer is also obsolete, but this comment mentions potential replacements. https://groups.google.com/d/msg/sci.electronics.design/F120QpxWWkc/74ljdPZ_BQAJ

There\'s also this design (https://www.edn.com/1-kv-power-supply-produces-a-continuous-arc/) that is 1kV, 20W. I just don\'t know enough about making it adjustable and isolated.

Thoughts on how to proceed?

If you are going to try building it yourself, then I\'d start with the
500V 5 mA one first. I built my first flyback with the LT3083,

(2018... thread here
https://groups.google.com/forum/?fromgroups=&hl=en#!searchin/sci.electronics.design/my$20first$20flyback%7Csort:date/sci.electronics.design/FQAIY1VaYBY/mLvGKJC2AgAJ

Good luck. Oh buying a ready made HV supply would be a good thing.
Gives you something to compare to if/when you have problems putting
to whole system together.

George H.


Yep. OP did not like that I disparaged/was skeptical of his plan to
build a 20 watt 1000 volt converter out of the gate but it strikes me as
a certain kind of madness as a first project if OP has not designed and
built a flyback of any type before.

Since you bring it up, it bothered me when you said that too.
I assume people mostly know what they are talking about.
Why not be nice?

BTW 5 uF for a piezo stack doesn\'t strike me as impossible.
I used this little multilayer stacks (for motion) that were ~0.1 uf
(only 150 V.)
https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=61

George H.

Do the math on how much peak current you\'d need to slew a 1000 volt 5uF
piezo stack full-range to have a sub-ms response time mah dude! with a
20 watt converter it must not be doing anything fast. the V^2 in
0.5*C*V^2 is a bitch

Even if someone here were able to provide a perfect custom design that
fit the bill the probability of something not working right on a first
prototype is high, and the probability of being able to figure it out
without a real working reference design or experience building more
modest switchers, is low

https://thedailywtf.com/articles/plz-email-me-teh-codez

 

[bitrex]

On 7/22/2020 3:30 PM, George Herold wrote:

On Wednesday, July 22, 2020 at 12:24:16 PM UTC-4, bitrex wrote:
On 7/22/2020 10:21 AM, George Herold wrote:
On Tuesday, July 21, 2020 at 6:51:49 PM UTC-4, Matt B wrote:
I need to make a 24V to 500-1000V, 20W adjustable power supply for a pulser. It needs to be a floating or negative converter.

My specs are:
Input: 24-28V
Output voltage: 500-1000V adjustable via control signal
Output current: 10mA @ 500V (5W), 20mA @ 1000V (20W)

I\'ve been reading through threads on different HV designs. So far the only one I\'ve come across that is isolated is Figure 50 of LT Application Note 29 (https://www.analog.com/media/en/technical-documentation/application-notes/an29f.pdf). It is specified for 1000V, 5W so not beefy enough. The transformer is also obsolete, but this comment mentions potential replacements. https://groups.google.com/d/msg/sci.electronics.design/F120QpxWWkc/74ljdPZ_BQAJ

There\'s also this design (https://www.edn.com/1-kv-power-supply-produces-a-continuous-arc/) that is 1kV, 20W. I just don\'t know enough about making it adjustable and isolated.

Thoughts on how to proceed?

If you are going to try building it yourself, then I\'d start with the
500V 5 mA one first. I built my first flyback with the LT3083,

(2018... thread here
https://groups.google.com/forum/?fromgroups=&hl=en#!searchin/sci.electronics.design/my$20first$20flyback%7Csort:date/sci.electronics.design/FQAIY1VaYBY/mLvGKJC2AgAJ

Good luck. Oh buying a ready made HV supply would be a good thing.
Gives you something to compare to if/when you have problems putting
to whole system together.

George H.


Yep. OP did not like that I disparaged/was skeptical of his plan to
build a 20 watt 1000 volt converter out of the gate but it strikes me as
a certain kind of madness as a first project if OP has not designed and
built a flyback of any type before.

Since you bring it up, it bothered me when you said that too.
I assume people mostly know what they are talking about.
Why not be nice?

BTW 5 uF for a piezo stack doesn\'t strike me as impossible.
I used this little multilayer stacks (for motion) that were ~0.1 uf
(only 150 V.)
https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=61

George H.

Do the math on how much peak current you\'d need to slew a 1000 volt 5uF
piezo stack full-range to have a sub-ms response time mah dude! with a
20 watt converter it must not be doing anything fast. the V^2 in
0.5*C*V^2 is a bitch

Even if someone here were able to provide a perfect custom design that
fit the bill the probability of something not working right on a first
prototype is high, and the probability of being able to figure it out
without a real working reference design or experience building more
modest switchers, is low

https://thedailywtf.com/articles/plz-email-me-teh-codez

 

[bitrex]

On 7/22/2020 3:30 PM, George Herold wrote:

On Wednesday, July 22, 2020 at 12:24:16 PM UTC-4, bitrex wrote:
On 7/22/2020 10:21 AM, George Herold wrote:
On Tuesday, July 21, 2020 at 6:51:49 PM UTC-4, Matt B wrote:
I need to make a 24V to 500-1000V, 20W adjustable power supply for a pulser. It needs to be a floating or negative converter.

My specs are:
Input: 24-28V
Output voltage: 500-1000V adjustable via control signal
Output current: 10mA @ 500V (5W), 20mA @ 1000V (20W)

I\'ve been reading through threads on different HV designs. So far the only one I\'ve come across that is isolated is Figure 50 of LT Application Note 29 (https://www.analog.com/media/en/technical-documentation/application-notes/an29f.pdf). It is specified for 1000V, 5W so not beefy enough. The transformer is also obsolete, but this comment mentions potential replacements. https://groups.google.com/d/msg/sci.electronics.design/F120QpxWWkc/74ljdPZ_BQAJ

There\'s also this design (https://www.edn.com/1-kv-power-supply-produces-a-continuous-arc/) that is 1kV, 20W. I just don\'t know enough about making it adjustable and isolated.

Thoughts on how to proceed?

If you are going to try building it yourself, then I\'d start with the
500V 5 mA one first. I built my first flyback with the LT3083,

(2018... thread here
https://groups.google.com/forum/?fromgroups=&hl=en#!searchin/sci.electronics.design/my$20first$20flyback%7Csort:date/sci.electronics.design/FQAIY1VaYBY/mLvGKJC2AgAJ

Good luck. Oh buying a ready made HV supply would be a good thing.
Gives you something to compare to if/when you have problems putting
to whole system together.

George H.


Yep. OP did not like that I disparaged/was skeptical of his plan to
build a 20 watt 1000 volt converter out of the gate but it strikes me as
a certain kind of madness as a first project if OP has not designed and
built a flyback of any type before.

Since you bring it up, it bothered me when you said that too.
I assume people mostly know what they are talking about.
Why not be nice?

BTW 5 uF for a piezo stack doesn\'t strike me as impossible.
I used this little multilayer stacks (for motion) that were ~0.1 uf
(only 150 V.)
https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=61

George H.

Do the math on how much peak current you\'d need to slew a 1000 volt 5uF
piezo stack full-range to have a sub-ms response time mah dude! with a
20 watt converter it must not be doing anything fast. the V^2 in
0.5*C*V^2 is a bitch

Even if someone here were able to provide a perfect custom design that
fit the bill the probability of something not working right on a first
prototype is high, and the probability of being able to figure it out
without a real working reference design or experience building more
modest switchers, is low

https://thedailywtf.com/articles/plz-email-me-teh-codez

 

[bitrex]

On 7/22/2020 3:35 PM, Matt B wrote:

On Wednesday, July 22, 2020 at 2:04:25 PM UTC-5, bitrex wrote:
On 7/22/2020 3:02 PM, bitrex wrote:
On 7/22/2020 2:26 PM, John Larkin wrote:
On Wed, 22 Jul 2020 12:52:25 -0400, bitrex <user@example.net> wrote:

On 7/22/2020 9:43 AM, jlarkin@highlandsniptechnology.com wrote:
On Wed, 22 Jul 2020 06:08:26 -0700 (PDT), Matt B
m...@gmail.com> wrote:

On Tuesday, July 21, 2020 at 10:32:47 PM UTC-5,
jla...@highlandsniptechnology.com wrote:
I you want to design the whole circuit, take a look at LT3803 and its
app notes. It just always seems to work.

Scribble some circuits and we can talk about them. LT Spice includes
the 3803 too. It seems to work just like it simulates.

I generally like to design or at least modify a current design so I
understand what\'s going on. This is new territory for me, so I\'ll
look into a start some simulations. Thanks for the tip on the LTC3803.

The pragmatic way to do this, at least for a one-off, is to buy a
dc/dc brick that has a control input, and be done. But as you say, you
wouldn\'t learn much about HV supplies.

LTC/ADI has some starting-point simulations for the 3803.

I\'m playing with an isolated flyback right now, probably about 200
volts at very low current, possibly with a voltage regulator on the
high side. When my customer gets un-confused about what he actually
needs, I\'ll build him a demo.

Does you piezo move slow or fast? Win Hill has posted some interesting
piezo drive ideas here recently.

Says he\'s got an implied 1000 volt piezo stack with a capacitance
of...5uF? that doesn\'t sound right.

That would lift trucks, or buildings.


Yes, the original flyback in the EDN article has as 0.01uF 1500V cap as
the main output cap and 0.022u 500Vs as the CW boost caps. this is a 4uF
1000V cap, damn:

https://i.ebayimg.com/images/g/LYEAAOSwiONYNMh1/s-l500.jpg

the EDN circuit isn\'t in the right ballpark.



Hoping he actually meant 5n or something

The 5uF is the bulk capacitance of the pulser circuit, not the piezo element, to prevent droop during pulses. So the caps are not fully charged and discharged every cycle that would need to fast cap charge capability of the Ultravolt unit. The bulk capacitance will also be \"separated\" from the power supply with a resistor to slow charge the caps.

Jeepers that\'s a large bulk capacitance.

If all you really need is an expedient ~1kV supply to charge cap bank
with that energy that can power up off a 12 or 24 volt supply you could
use a sine inverter off the shelf and a couple of appropriately-rated
line transformers like:

<https://www.dropbox.com/s/7byu9dkapx7npcm/p4bBqD70.jpeg?dl=0>

but please don\'t quote me that I got those dots correct, or actually
recommended any of this, damn son.

If what you need is like that but compact and very efficient and
infinitely adjustable I\'ll have to defer. GL

 

[bitrex]

On 7/22/2020 3:35 PM, Matt B wrote:

On Wednesday, July 22, 2020 at 2:04:25 PM UTC-5, bitrex wrote:
On 7/22/2020 3:02 PM, bitrex wrote:
On 7/22/2020 2:26 PM, John Larkin wrote:
On Wed, 22 Jul 2020 12:52:25 -0400, bitrex <user@example.net> wrote:

On 7/22/2020 9:43 AM, jlarkin@highlandsniptechnology.com wrote:
On Wed, 22 Jul 2020 06:08:26 -0700 (PDT), Matt B
m...@gmail.com> wrote:

On Tuesday, July 21, 2020 at 10:32:47 PM UTC-5,
jla...@highlandsniptechnology.com wrote:
I you want to design the whole circuit, take a look at LT3803 and its
app notes. It just always seems to work.

Scribble some circuits and we can talk about them. LT Spice includes
the 3803 too. It seems to work just like it simulates.

I generally like to design or at least modify a current design so I
understand what\'s going on. This is new territory for me, so I\'ll
look into a start some simulations. Thanks for the tip on the LTC3803.

The pragmatic way to do this, at least for a one-off, is to buy a
dc/dc brick that has a control input, and be done. But as you say, you
wouldn\'t learn much about HV supplies.

LTC/ADI has some starting-point simulations for the 3803.

I\'m playing with an isolated flyback right now, probably about 200
volts at very low current, possibly with a voltage regulator on the
high side. When my customer gets un-confused about what he actually
needs, I\'ll build him a demo.

Does you piezo move slow or fast? Win Hill has posted some interesting
piezo drive ideas here recently.

Says he\'s got an implied 1000 volt piezo stack with a capacitance
of...5uF? that doesn\'t sound right.

That would lift trucks, or buildings.


Yes, the original flyback in the EDN article has as 0.01uF 1500V cap as
the main output cap and 0.022u 500Vs as the CW boost caps. this is a 4uF
1000V cap, damn:

https://i.ebayimg.com/images/g/LYEAAOSwiONYNMh1/s-l500.jpg

the EDN circuit isn\'t in the right ballpark.



Hoping he actually meant 5n or something

The 5uF is the bulk capacitance of the pulser circuit, not the piezo element, to prevent droop during pulses. So the caps are not fully charged and discharged every cycle that would need to fast cap charge capability of the Ultravolt unit. The bulk capacitance will also be \"separated\" from the power supply with a resistor to slow charge the caps.

Jeepers that\'s a large bulk capacitance.

If all you really need is an expedient ~1kV supply to charge cap bank
with that energy that can power up off a 12 or 24 volt supply you could
use a sine inverter off the shelf and a couple of appropriately-rated
line transformers like:

<https://www.dropbox.com/s/7byu9dkapx7npcm/p4bBqD70.jpeg?dl=0>

but please don\'t quote me that I got those dots correct, or actually
recommended any of this, damn son.

If what you need is like that but compact and very efficient and
infinitely adjustable I\'ll have to defer. GL