high voltage charge pump

On Thu, 2 Jan 2020 09:00:59 -0800 (PST), Klaus Kragelund
<klauskvik@hotmail.com> wrote:

>On Thursday, January 2, 2020 at 5:44:39 PM UTC+1, jla...@highlandsniptechnology.com wrote:
>> On Thu, 2 Jan 2020 02:22:36 -0800 (PST), Klaus Kragelund
>> <klauskvik@hotmail.com> wrote:
>> 
>> >On Thursday, January 2, 2020 at 4:39:49 AM UTC+1, bitrex wrote:
>> >> On 1/1/20 11:18 AM, jlarkin@highlandsniptechnology.com wrote:
>> >> > On 1 Jan 2020 05:34:23 -0800, Winfield Hill <winfieldhill@yahoo.com>
>> >> > wrote:
>> >> > 
>> >> >> Bill Sloman wrote...
>> >> >>>
>> >> >>> On January 1, 2020, jlarkin wrote:
>> >> >>>>
>> >> >>>> We don't use power transformers much these days. We usually
>> >> >>>> buy fully standards-compliant, universal-input switching
>> >> >>>> power supplies, wall-warts or enclosed boxes.
>> >> >>>
>> >> >>> Which is fine until you need a power rail which isn't
>> >> >>> available off the shelf.
>> >> >>
>> >> >> A simple solution to that problem is to add a fixed-ratio
>> >> >> bus converter, see AoE x-Chapter 9x.16, "DC transformer".
>> >> > 
>> >> > Big switching supplies, for serious power busses, are available in
>> >> > most any voltage, and are available as adjustable.
>> >> > 
>> >> > I usually build a lot of secondary linear and switching regs, on my PC
>> >> > boards, that never get near the AC line, so I can generally design
>> >> > around a 24 volt wart or equivalent. We buy supplies with all the
>> >> > conformance stickers, real or maybe not, and that work anywhere in the
>> >> > world without tap switching and fuse changing hazards.
>> >> > 
>> >> > It's a lot easier to do a CE compliance doc if the power supply is
>> >> > already conforming.
>> >> > 
>> >> > A MeanWell 800 watt switcher box is a fraction of the weight and cost
>> >> > of an 800 watt transformer, and outputs clean DC. The heavy ugly
>> >> > transformer is just the start of a power supply.
>> >> > 
>> >> 
>> >> How do small companies that make e.g. tube guitar amps certify their 
>> >> products at reasonable cost? I don't think many of them are UL listed, 
>> >> too expensive. They don't have much choice but to use an internal 
>> >> off-line linear supply they designed themselves such as it is.
>> >
>> >That is dangerous
>> >
>> >In CE you can ship the product, if you are sure you comply with the regulations
>> >
>> >Trouble is, if they ask for the test results, you must be able to provide them with days of leadtime
>> >
>> >Cheers
>> >
>> >Klaus
>> 
>> But does that happen? And what if you ignore the request?
>> 
>I have only seen it once, in an earlier company I worked for, we were asked to provide the information. Luckily we had it all done, so could just send it out
>
>If you can't, I do not know exactly what kind of re-enforcement of the rules they have. I think you can be told to return all units
>
>Just did a search:
>
>https://www.cemarkingassociation.co.uk/how-is-the-ce-mark-enforced/
>
>Seems there are examples of prosecution of non compliance

My Brit colleagues laughed at me when I thought CE was serious.



-- 

John Larkin         Highland Technology, Inc

lunatic fringe electronics 

On Thu, 2 Jan 2020 09:08:52 -0800 (PST), Klaus Kragelund
<klauskvik@hotmail.com> wrote:

>On Thursday, January 2, 2020 at 5:23:05 PM UTC+1, jla...@highlandsniptechnology.com wrote:
>> On Thu, 2 Jan 2020 02:20:06 -0800 (PST), Klaus Kragelund
>> <klauskvik@hotmail.com> wrote:
>> 
>> >On Thursday, January 2, 2020 at 2:45:47 AM UTC+1, jla...@highlandsniptechnology.com wrote:
>> >> On Wed, 1 Jan 2020 16:45:53 -0800 (PST), Klaus Kragelund
>> >> <klauskvik@hotmail.com> wrote:
>> >> 
>> >> >On Thursday, January 2, 2020 at 12:39:54 AM UTC+1, jla...@highlandsniptechnology.com wrote:
>> >> >> On Wed, 1 Jan 2020 15:13:25 -0800 (PST), Klaus Kragelund
>> >> >> <klauskvik@hotmail.com> wrote:
>> >> >> 
>> >> >> >On Wednesday, January 1, 2020 at 5:18:56 PM UTC+1, jla...@highlandsniptechnology.com wrote:
>> >> >> >> On 1 Jan 2020 05:34:23 -0800, Winfield Hill <winfieldhill@yahoo.com>
>> >> >> >> wrote:
>> >> >> >> 
>> >> >> >> >Bill Sloman wrote...
>> >> >> >> >>
>> >> >> >> >> On January 1, 2020, jlarkin wrote:
>> >> >> >> >>> 
>> >> >> >> >>> We don't use power transformers much these days. We usually
>> >> >> >> >>> buy fully standards-compliant, universal-input switching
>> >> >> >> >>> power supplies, wall-warts or enclosed boxes.
>> >> >> >> >>
>> >> >> >> >> Which is fine until you need a power rail which isn't
>> >> >> >> >> available off the shelf. 
>> >> >> >> >
>> >> >> >> > A simple solution to that problem is to add a fixed-ratio
>> >> >> >> > bus converter, see AoE x-Chapter 9x.16, "DC transformer".
>> >> >> >> 
>> >> >> >> Big switching supplies, for serious power busses, are available in
>> >> >> >> most any voltage, and are available as adjustable. 
>> >> >> >> 
>> >> >> >
>> >> >> >True
>> >> >> >
>> >> >> >At our firm, we never use standard supplies. Price is king, so we need to find lowest cost. Even at low volumes, it pays to do a custom design
>> >> >> >
>> >> >> >
>> >> >> >> I usually build a lot of secondary linear and switching regs, on my PC
>> >> >> >> boards, that never get near the AC line, so I can generally design
>> >> >> >> around a 24 volt wart or equivalent. We buy supplies with all the
>> >> >> >> conformance stickers, real or maybe not, and that work anywhere in the
>> >> >> >> world without tap switching and fuse changing hazards.
>> >> >> >> 
>> >> >> >> It's a lot easier to do a CE compliance doc if the power supply is
>> >> >> >> already conforming.
>> >> >> >> 
>> >> >> >
>> >> >> >Only true if your power requirement is below 15W. If your unit draws more, or the external supply you use can supply more
>> >> >> >
>> >> >> >Above 15W all the rules with fire comes into play, so the approval process is pretty much the same as approving your own custom supply. And the cost for the approval is not that high, complete product might be 15k USD for an advanced product with several insulation systems
>> >> >> >
>> >> >> >The hurdle with insulation systems is that it seems UL by amazing coincidence have an idea to only approve components/parts that is not part of the IEC system, so that one is forced to make both US and EU variants
>> >> >> >
>> >> >> >> A MeanWell 800 watt switcher box is a fraction of the weight and cost
>> >> >> >> of an 800 watt transformer, and outputs clean DC. The heavy ugly
>> >> >> >> transformer is just the start of a power supply.
>> >> >> >> 
>> >> >> >Meanwell is expensive. But an easy solution. When you disassembly the Meanwell and competitor supplies, you find that they really are not that good. We did an internal products, just me and another guy, and in 12 months we had a solution that outprformed the Meanwell types in price and performance.
>> >> >> >
>> >> >> >Cheers
>> >> >> >
>> >> >> >Klaus
>> >> >> 
>> >> >> 
>> >> >> We have had excellent performance from MeanWell supplies. We've used
>> >> >> about 700 so far.
>> >> >> 
>> >> >
>> >> >Our lowest volume products are normaly 10k units
>> >> >
>> >> >Max volume 6 million
>> >> >
>> >> >> Can you make your own for this price? What did 24 man-months of
>> >> >> engineering cost?
>> >> >> 
>> >> >
>> >> >We can make it for 10 times lower that price, with efficiency orders of magnitude better (sub 90% efficiency is a joke, cannot see why do would sell such bad products)
>> >> 
>> >> 0.1 * $71 is $7.10. But without shipping, and in quantity, you'll have
>> >> to come in around $6 maybe. I sorta doubt that is possible.
>> >> 
>> >I have the luxury of having a a lot of high volume parts running in current products (many more than 20 mill pcs per year), so I can cherry pick the low cost ones and use in this supply
>> >
>> >For the custom part, you can do tricks on the pricing of the common mode inductors and transformers that will shock you. Those 2 parts are the most expensive parts in the construction. But it means you have to divert away from your COTS magnetics, and you do not seem to like that. You should reconsider, the price is a different ballgame when you go into deep dialogue with the manufactor
>> >
>> >Also you can do tricks that can remove certain parts that one would normally use, but I cannot disclose those details here for IP reasons
>> >
>> >> How can an efficiency be an order of magnitude above 90%?  900% ?
>> >> 
>> >
>> >When you are talking about magnitudes in efficiency and your are above 95%, any 1% increase is really a monumental effort and takes real creative engineering
>> >
>> >
>> >Cheers
>> >
>> >Klaus
>> 
>> If you can make a 600 watt offline PFC switcher, in a metal box, for
>> $6, why aren't you selling them on Digikey?
>> 
>
>It may be slightly larger than 7 USD, mechanics is hard to get in low volume cheap from the get go
>
>I am actually sort of working in that route. I do not have the funds yet to do a design like that, but I am working first on POL supplies. I have something in mind that I think would be sort of a gamechanger, but I do not want to reveal it here in a forum. Can I write you a private mail, for your opinion, since you seem to have used at least some number of different POLs over time?
>
>Cheers
>
>Klaus

I design circuit boards, which usually have a single DC input and
various switching and linear regulators here and there. I guess you
can call that POL regulation. A small board might have 4 or 5
different rails, and a big one can have a dozen or more. My new
alternator simulator uses a 48v 800w MeanWell as the prime mover,
feeding 7 different boards. The big power supply costs 0.6% of the
selling price... a different market from your high-volume stuff.

Sure, I'd like to take a look at your idea, maybe add suggestions.
Confidentially, I promise.



-- 

John Larkin         Highland Technology, Inc

lunatic fringe electronics 

On Friday, January 3, 2020 at 2:10:56 AM UTC+11, DecadentLinux...@decadence.org wrote:
> Bill Sloman <bill.sloman@ieee.org> wrote in
> news:23103e40-ea55-4d4b-b33c-946330e7c99b@googlegroups.com: 
> 
> > Specify, buy and use them. Or is the specification stage above
> > your pay grade? 
> > 
> 
>   COTS, idiot.  NOT 'specify'.  THAT would be 'custom', not COTS.

Wrong. Commercial off-the-shelf power supplies do have specification sheets, and they aren't all the same.
 
>   With COTS, one gets what the industry grows to accept. 

The market isn't a single monolithic block, and commercial organisations know how to serve niche markets.

> The high end,  super quiet stuff is expensive and at that point a custom is the way to go.

Merely quiet doesn't cost much extra. Horribly noisy has to be cheap to sell at all, and tends to get dropped as the market wakes up.

<snipped the usual idiocy>
 
>  I used to make power supplies that had ripple ratings you never ever 
> saw in your entire pathetic life.

When I worked at EMI Central Research, I put together our second prototype phased array ultrasound scanner - our "clinical trials machine".

The major part of the job was getting every last circuit diagram into a printed circuit board - the first machine had relied on hand-wired boards, which weren't all that reliable. We got a sub-contractor to lay out the boards, but checked every layout they produced, and had to clean up several of them - but I also dumped the linear power supplies which we used on the first prototype and replaced them with well-specified switching power supplies.

The job was done in parallel with the construction of a batch of pre-production prototypes by an EMI subsidiary. They used the original power supplies which weighed a lot more.

We got offered their printed-circuit layout for the 21 front-end amplifiers (one per element in the array ). We checked the layout, decided that it wouldn't work, modified it so that it could work and got a batch made (which did work).

When we told the pre-production team of our concerns, they said "too late, we've already had 210 boards made and loaded". They had to hand-modify all 210 to get something that more or less worked, which must have cost more than a new batch would have, and the hand-modified board weren't all that reliable in use.

We ran into one of the pre-production machines at an ultra-sound conference and found the seven of the channels weren't working ...

One of the publicity pictures showed a small nurse ostensibly pushing the pre-production machine towards a patient. In reality it took two hospital porters to get it to move. The clinical trials machine went into six foot cabinet on wheels, but at least it was light enough for one person to push around.

-- 
Bill Sloman, Sydney

On Friday, January 3, 2020 at 2:16:49 AM UTC+11, DecadentLinux...@decadence.org wrote:
> Bill Sloman <bill.sloman@ieee.org> wrote in
> news:23103e40-ea55-4d4b-b33c-946330e7c99b@googlegroups.com: 
> 
> > The filtering is between the output of the supply and device it is
> > powering. 
> 
>   Do not need a primer from an utter retard.

The utter retard here is you.
 
>   We were powering a $35k synthesizer by the mfgr specs.  Their 
> suggestion to use a linear instead of a switcher has to do with 
> detection of jitter during operation.  Incorporating the linear fixed 
> it.  Incorporating a filter on the switcher did not, because the EMI 
> flooded the chassis and the jitter got injected despite the rail 
> getting clean filtration.

I did mention that bad switchers tended to feed noise into every ground connection they could get at.

A Cambridge Instruments one switcher we used had to be wrapped in a grounded mesh screen to stop it radiating noise through free space, in addition to having common mode chokes on every set of connections.

We got a better one a year or so later.

> The linear and no post filter fixed it and 
> was cheaper than filtering.  It matters at tens of GHz.

Of course it does.
 
>   You are a 100% pure noise jackass.

Wrong again. The jackass here is you, retailing lessons learned where you hadn't hadn't actually understood the problem or the found the right solution to it.

-- 
Bill Sloman, Sydney

On Friday, January 3, 2020 at 2:28:27 AM UTC+11, DecadentLinux...@decadence.org wrote:
> Bill Sloman <bill.sloman@ieee.org> wrote in
> news:23103e40-ea55-4d4b-b33c-946330e7c99b@googlegroups.com: 
> 
> > Your original claim was that "  We had some very noise sensitive
> > devices that we were only able to power with linear supplies."
> 
>   Yep.  That is what we decided to use.  With millions of dollars 
> being invested by the DoD, we do what they tell us to do.  We 
> engineer into the system the proven circuits they want used.

The DoD does attract people who will do what they are told without thinking too hard about what they are being told to do.
 
>   When on makes a 1024 channel stimulator where all of them are 
> synched to within a nanosecond of each other, one wants repeatability 
> and cost effectiveness and precision.   Doing that across 16 racks of 
> gear takes real engineering, not some dork spouting off basic, fully 
> understood bullshit about how to filter a supply rail.

A 21-channel phased array ultrasound machine did pose some of the same kinds of problems. At 2MHz you wouldn't have though that a nanosecond would matter, but or real machine performed better than its computer simulations, until my boss got into the simulation software and used double precision on the timing data.

Single precision imposed a 1nsec granularity, which turned out to be too coarse.

The electron beam tester I worked on was supposed to provide 10psec granularity on timing. It was a rather silly requirement, granting that we never got an electron pulse narrower than 500psec (though we probably would have been able to offer a 100osec pulse if the development had gone to completion). We did deliver the 10psec increments anyway.
 
> > You had enough control over what you used to reject switching
> > power supplies.
> 
>   Design choices were 100% our engineering group's scope.  Us and the 
> DoD.

With the DoD in the driving seat.

<snip>
 
> > you could have rejected the noisier switching power supplies.
> 
>   We rejected switchers altogether for that application,

You just said the DoD did that for you. 

<snipped DLUNU bitching about his impotence to do anything about the DoD's choices>

-- 
Bill Sloman, Sydney

Bill Sloman <bill.sloman@ieee.org> wrote in
news:9f0613df-f3f6-47b7-8fcd-8ac14962f464@googlegroups.com: 

>>   COTS, idiot.  NOT 'specify'.  THAT would be 'custom', not COTS.
> 
> Wrong.

  Not wrong, you fucking idiot.

> Commercial off-the-shelf power supplies do have
> specification sheets, and they aren't all the same. 

  Exactly.  As in NOT specify, but CHOOSE.

  You lose...  AGAIN.

On Saturday, January 4, 2020 at 5:06:03 PM UTC+11, DecadentLinux...@decadence.org wrote:
> Bill Sloman <bill.sloman@ieee.org> wrote in
> news:9f0613df-f3f6-47b7-8fcd-8ac14962f464@googlegroups.com: 
> 
> >>   COTS, idiot.  NOT 'specify'.  THAT would be 'custom', not COTS.
> > 
> > Wrong.
> 
>   Not wrong, <snip>
> 
> > Commercial off-the-shelf power supplies do have
> > specification sheets, and they aren't all the same. 
> 
>   Exactly.  As in NOT specify, but CHOOSE.

You ought to specify the performance you need, and compare what you require with what's commercially available. You choose after you have worked out what you need. This might involve getting hold of a few samples on loan and sticking them into some kind of test bed, if the manufacturer's specifications aren't tight enough.

Waving a a wand and saying only linear supplies will do isn't the way it works anywhere where the engineers are taken seriously.

Your approach seems to have been to do what the DoD told you

>   You lose...  AGAIN.

I may have failed to make sufficient allowance for your remoteness from the people who make those kinds of choices.

You do seem have an odd idea of what might be involved in them.

-- 
Bill Sloman, Sydney

Bill Sloman <bill.sloman@ieee.org> wrote in
news:5522ca67-3bb3-4152-b37d-da11a6afe4df@googlegroups.com: 

> On Saturday, January 4, 2020 at 5:06:03 PM UTC+11,
> DecadentLinux...@decadence.org wrote: 
>> Bill Sloman <bill.sloman@ieee.org> wrote in
>> news:9f0613df-f3f6-47b7-8fcd-8ac14962f464@googlegroups.com: 
>> 
>> >>   COTS, idiot.  NOT 'specify'.  THAT would be 'custom', not
>> >>   COTS. 
>> > 
>> > Wrong.
>> 
>>   Not wrong, <snip>

 SnipTard strikes again!  Pussy!

>> 
>> > Commercial off-the-shelf power supplies do have
>> > specification sheets, and they aren't all the same. 
>> 
>>   Exactly.  As in NOT specify, but CHOOSE.
> 
> You ought to specify the performance you need,

  No, idiot.  An engineer DECIDES what is needed and what is 
tolerable based on what is available.

> and compare what
> you require with what's commercially available.

  I know how to incorporate an available product into one of my 
designs, you stupid fuck.

> You choose after
> you have worked out what you need.

  I was doing requirements analysis before you even knew the term.  
Oh...  that's right...  looks like you don't.

  Billy Sloman the SloTard is an idiot because he thinks everyone 
else is.

> This might involve getting hold
> of a few samples on loan and sticking them into some kind of test
> bed, if the manufacturer's specifications aren't tight enough. 

  You are an idiot.  The process is well known.  We do not need a 
dweeb reject like you spouting off about something your 'recent' 
experience with was thirty five years ago.

> Waving a a wand and saying only linear supplies will do isn't the
> way it works anywhere where the engineers are taken seriously.

  There was none of that, you presumptuous fucktard!  Nice try 
though, loser.
 
> 
> Your approach seems to have been to do what the DoD told you

  Your approach is decidedly one of someone unaware of what 
contractor specification means.  You are nothing more than an 
industry interloper with zero experience and even less practical 
knowledge, much less hands on grasp of any process.

>>   You lose...  AGAIN.
> 
> I may have failed to make sufficient allowance for your remoteness
> from the people who make those kinds of choices. 

  Bill Sloman is an abject idiot.  You would last less than a day.
> 
> You do seem have an odd idea of what might be involved in them.

  You do seem to be a presumptuous fucktard at every fucking turn.  
You have ZERO clue about it and even les clue about me.  Nice try 
though, punk.

On Sunday, January 5, 2020 at 1:22:00 AM UTC+11, DecadentLinux...@decadence.org wrote:
> Bill Sloman <bill.sloman@ieee.org> wrote in
> news:5522ca67-3bb3-4152-b37d-da11a6afe4df@googlegroups.com: 
> 
> > On Saturday, January 4, 2020 at 5:06:03 PM UTC+11,
> > DecadentLinux...@decadence.org wrote: 
> >> Bill Sloman <bill.sloman@ieee.org> wrote in
> >> news:9f0613df-f3f6-47b7-8fcd-8ac14962f464@googlegroups.com: 
> >> 
> >> >>   COTS, idiot.  NOT 'specify'.  THAT would be 'custom', not
> >> >>   COTS. 
> >> > 
> >> > Wrong.
> >> 
> >>   Not wrong, <snip>

<snip>

> >> > Commercial off-the-shelf power supplies do have
> >> > specification sheets, and they aren't all the same. 
> >> 
> >>   Exactly.  As in NOT specify, but CHOOSE.

You can't choose until you have worked out what you need.

> > You ought to specify the performance you need,
> 
>   No, idiot.  An engineer DECIDES what is needed and what is 
> tolerable based on what is available.

And they'd better do it before they've looked hard at what is available. 

It's tempting to pick on the best of what's available, even when that isn't good enough.
 
> > and compare what
> > you require with what's commercially available.
> 
>   I know how to incorporate an available product into one of my 
> designs <snip>.

Unless the DoD tells you not to bother, by requiring you to use a linear power supply.

> > You choose after you have worked out what you need.
> 
>   I was doing requirements analysis before you even knew the term.  
> Oh...  that's right...  looks like you don't.

Who cares what it is called - apart from military clowns. I was doing requirements analysis when I was putting together the system I used to do the experimental part of my Ph.D. project.My colleagues called me "gadget happy" but my gadgets worked, because they did what the experiment required them to.

I got the Ph.D. in 1970.
 
<snipped insulting speculation>
 
> > This might involve getting hold
> > of a few samples on loan and sticking them into some kind of test
> > bed, if the manufacturer's specifications aren't tight enough. 
 
<snipped another pointless insult>
 
> > Waving a a wand and saying only linear supplies will do isn't the
> > way it works anywhere where the engineers are taken seriously.

<snipped another unconvincing claim>

> > Your approach seems to have been to do what the DoD told you

<and another>
 
> >>   You lose...  AGAIN.
> > 
> > I may have failed to make sufficient allowance for your remoteness
> > from the people who make those kinds of choices. 

<and another> 

> > You do seem have an odd idea of what might be involved in them.

<and another>

DLUNU does seem to have to dive deep into his imagination to find anything to say at all.

-- 
Bill Sloman, Sydney

Bill Sloman <bill.sloman@ieee.org> wrote in news:45891de0-c0de-4c51-
a3d7-b07d1b21db81@googlegroups.com:

> 
> Who cares what it is called - apart from military clowns.

  The term "requirements analysis" has been around as long as computers 
have because application programmers use it, database administrators 
use it, PCB layout engineers use it, colleges use it.  The list is 
pretty darn big, and not related to military.

  Only Billy Sloman wants to downplay proper terminology.  Too bad, 
Billy.  Everybody uses it.  Nice try, classless clown.