magnetics question

On Wed, 6 Apr 2022 18:45:43 +0200, Arie de Muijnck
<eternal.september@ademu.com> wrote:

>On 2022-04-06 18:33, jlarkin@highlandsniptechnology.com wrote:
>> On Wed, 6 Apr 2022 18:24:49 +0200, Arie de Muijnck
>> <eternal.september@ademu.com> wrote:
>> 
>>> On 2022-04-06 17:53, jlarkin@highlandsniptechnology.com wrote:
>>>> https://www.dropbox.com/s/jnb3jmw8rcmdeir/XfmrScatter.JPG?raw=1
>>>>
>>>> How does the mass of a transformer scale with frequency?
>>>>
>>>> I want to make a 120v 400 Hz power supply. I might boost my 48v up to
>>>> 200DC and use an isolated h-bridge out to the load, or we could put
>>>> the bridge down at 48v and boost with a biggish transformer.
>>>>
>>>> I'll have lots of air flow, so maybe I can push things some too.
>>>
>>> Weight is almost linear with power, and inversely with frequency.
>>>
>>> When I needed a 115V 400Hz for DO-160 aviation equipment development, I
>>> ordered a 100W amplifier set from Amplimo (https://www.amplimo.nl/):
>>> Mains transformer (toroid 115/230V in), rectifier, capacitors, amplifier
>>> module, and an extra mains transformer (used in reverse) for the 115V
>>> output.
>>> Added a 400 Hz oscillator with a timer controlled gain control (4051 MUX
>>> with some resistors) to create the required brownout and dropout
>>> profiles. The customer was so amused they ordered some for their
>>> production and test line (until then, they used a gigantic motor + 400Hz
>>> dynamo in the basement to get 400 Hz).
>>>
>>> Arie
>> 
>> I think so... for a given core, power scales linearly with frequency.
>> A simple scribble with a 4-winding transformer suggests that.
>> 
>> Thanks
>> 
>> John
>
>Yes, which is exactly why switching PSU's are now the standard. The cost 
>saving in iron (ferrite) and copper and capacitors, and all transport 
>costs, far outweighs (pun intended) the cost of the semiconductors.
>
>And why airplanes use 400 Hz - the tiny transformers I had designed in 
>in the product were amusing, 8 times smaller than the usual 50 hz versions.
>
>Arie

I think you're living in the past. 60-400Hz magnetics are 
routinely shipped around the world for local cost reduction, 
making 'iron' transport costs irrelevent.

Only end-use weight and volume remain signifigant.

RL

Piotr Wyderski <bombald@protonmail.com> wrote:

> Arie de Muijnck wrote:
> 
>> Yes, which is exactly why switching PSU's are now the standard. The
>> cost saving in iron (ferrite) and copper and capacitors, and all
>> transport costs, far outweighs (pun intended) the cost of the
>> semiconductors. 
> 
> Their regulation capability is far better than that of the more 
> conventional PSUs.
> 
>> And why airplanes use 400 Hz - the tiny transformers I had designed in 
>> in the product were amusing, 8 times smaller than the usual 50 hz
>> versions. 
> 
> But why 400 then? A typical scaling factor would be 10, so 500Hz should 
> be expected. Instead, they have selected the odd value of 8. Backward 
> compatibility with an arbitrarily selected frequency back in the 
> medieval times?
> 
>      Best regards, Piotr

Aircraft are moving away from the fixed 400Hz frequency to variable
360-800Hz. 

"On the other hand, the innovation of power supply system in aircraft
performance in the system of power supply: 360~800Hz large capacity
variable frequency AC power system is using gradually instead of the
constant frequency of 400Hz power supply on most of the aircraft[3,4]" 

https://www.atlantis-press.com/article/25862618.pdf




-- 
MRM

Mike Monett wrote:

> Aircraft are moving away from the fixed 400Hz frequency to variable
> 360-800Hz.
> 
> "On the other hand, the innovation of power supply system in aircraft
> performance in the system of power supply: 360~800Hz large capacity
> variable frequency AC power system is using gradually instead of the
> constant frequency of 400Hz power supply on most of the aircraft[3,4]"

Interesting. Why not DC then? It is going to be consumed by a switcher 
of some sort anyway, be it a PSU or motor controller.

	Best regards, Piotr

Piotr Wyderski <bombald@protonmail.com> wrote:

> Mike Monett wrote:
> 
>> Aircraft are moving away from the fixed 400Hz frequency to variable
>> 360-800Hz. 
>> 
>> "On the other hand, the innovation of power supply system in aircraft
>> performance in the system of power supply: 360~800Hz large capacity
>> variable frequency AC power system is using gradually instead of the
>> constant frequency of 400Hz power supply on most of the aircraft[3,4]" 
> 
> Interesting. Why not DC then? It is going to be consumed by a switcher 
> of some sort anyway, be it a PSU or motor controller.
> 
>      Best regards, Piotr

My Piper Malibu distributed power in DC. Various instruments either used DC 
as is, or changed it to 400Hz. I believe larger A/C generate 3-phase.

A possible reason for limiting the frequency may be skin effect. When you 
have to run power from the engines under the wings all over the plane, the 
losses could become serious.



-- 
MRM

onsdag den 6. april 2022 kl. 23.26.01 UTC+2 skrev Piotr Wyderski:
> Mike Monett wrote: 
> 
> > Aircraft are moving away from the fixed 400Hz frequency to variable 
> > 360-800Hz. 
> > 
> > "On the other hand, the innovation of power supply system in aircraft 
> > performance in the system of power supply: 360~800Hz large capacity 
> > variable frequency AC power system is using gradually instead of the 
> > constant frequency of 400Hz power supply on most of the aircraft[3,4]"
> Interesting. Why not DC then? It is going to be consumed by a switcher 
> of some sort anyway, be it a PSU or motor controller. 

does relays and switches used for ~400Hz AC need to be beefed up like for DC?

On Wednesday, April 6, 2022 at 8:53:36 AM UTC-7, jla...@highlandsniptechnology.com wrote:
> https://www.dropbox.com/s/jnb3jmw8rcmdeir/XfmrScatter.JPG?raw=1 
> 
> How does the mass of a transformer scale with frequency? 

That's an interesting, and difficult, question.   I'm assuming power transformer with 
 unknown load on the secondary, there's current-transformer assumptions which are 'way different.

Firstly, you need the core  not to saturate at rated input voltage and
no load; that means the core material's B-H curve has to be considered to set core size.
This is about the magnetic flux distribution in the area (cross-section) of its path.

Second, you need the primary and secondary wire resistances to be low enough for
acceptable heat losses (usually, primary and secondary will have same loss, but that'll
have steps at available gage sizes...) which means every core shape will have differing
size according to the throat available for windings.  Variacs will have different
rules, because there's no separate primary and secondary wire gage,   This affects
the magnetic path length, not the area.

Third, you need the whole assembly to stay cool; we don't usually see heatsinks on
transformers, but every  pole pig IS certainly a kind of keep-it-cool problem, and 
for high frequency, core losses (and maybe even skin effect in conductors) will have to
be considered even for small items.

With only the first item considered, cross sectional area is constant for  peak current-turns
at no load, so  expect V /f proportional to area, thus weight  roughly scale by f**(-1.5).
It's a project that'll benefit from log-log and/or semilog graph paper.

onsdag den 6. april 2022 kl. 23.37.03 UTC+2 skrev whit3rd:
> On Wednesday, April 6, 2022 at 8:53:36 AM UTC-7, jla...@highlandsniptechnology.com wrote: 
> > https://www.dropbox.com/s/jnb3jmw8rcmdeir/XfmrScatter.JPG?raw=1 
> > 
> > How does the mass of a transformer scale with frequency?
> That's an interesting, and difficult, question. I'm assuming power transformer with 
> unknown load on the secondary, there's current-transformer assumptions which are 'way different. 
> 
> Firstly, you need the core not to saturate at rated input voltage and 
> no load; that means the core material's B-H curve has to be considered to set core size. 
> This is about the magnetic flux distribution in the area (cross-section) of its path. 
> 
> Second, you need the primary and secondary wire resistances to be low enough for 
> acceptable heat losses (usually, primary and secondary will have same loss, but that'll 
> have steps at available gage sizes...) which means every core shape will have differing 
> size according to the throat available for windings. Variacs will have different 
> rules, because there's no separate primary and secondary wire gage, This affects 
> the magnetic path length, not the area. 
> 
> Third, you need the whole assembly to stay cool; we don't usually see heatsinks on 
> transformers, but every pole pig IS certainly a kind of keep-it-cool problem, and 
> for high frequency, core losses (and maybe even skin effect in conductors) will have to 
> be considered even for small items. 
> 
> With only the first item considered, cross sectional area is constant for peak current-turns 
> at no load, so expect V /f proportional to area, thus weight roughly scale by f**(-1.5). 
> It's a project that'll benefit from log-log and/or semilog graph paper.

can't it be more or less simplified to: 
if you take a suitable transformer and run it at 8 times the frequency you can also 
increase the voltage 8 times, and thus get 8 times the power 

On 7/4/22 7:00 am, Piotr Wyderski wrote:
> Arie de Muijnck wrote:
>> And why airplanes use 400 Hz - the tiny transformers I had designed in 
>> in the product were amusing, 8 times smaller than the usual 50 hz 
>> versions.
> 
> But why 400 then?

Probably to do with still using rotating converters and other 
synchronous things. 24000RPM is feasible in 1950s technology, but much 
more gets hard.

On Wed, 6 Apr 2022 21:19:49 -0000 (UTC), Mike Monett <spamme@not.com>
wrote:

>Piotr Wyderski <bombald@protonmail.com> wrote:
>
>> Arie de Muijnck wrote:
>> 
>>> Yes, which is exactly why switching PSU's are now the standard. The
>>> cost saving in iron (ferrite) and copper and capacitors, and all
>>> transport costs, far outweighs (pun intended) the cost of the
>>> semiconductors. 
>> 
>> Their regulation capability is far better than that of the more 
>> conventional PSUs.
>> 
>>> And why airplanes use 400 Hz - the tiny transformers I had designed in 
>>> in the product were amusing, 8 times smaller than the usual 50 hz
>>> versions. 
>> 
>> But why 400 then? A typical scaling factor would be 10, so 500Hz should 
>> be expected. Instead, they have selected the odd value of 8. Backward 
>> compatibility with an arbitrarily selected frequency back in the 
>> medieval times?
>> 
>>      Best regards, Piotr
>
>Aircraft are moving away from the fixed 400Hz frequency to variable
>360-800Hz. 
>
>"On the other hand, the innovation of power supply system in aircraft
>performance in the system of power supply: 360~800Hz large capacity
>variable frequency AC power system is using gradually instead of the
>constant frequency of 400Hz power supply on most of the aircraft[3,4]" 
>
>https://www.atlantis-press.com/article/25862618.pdf

There are constant-frequency generators that always make 400 Hz. I
don't know how they work.

360 (sometimes 250) to 800 Hz is "Wild power", what you can get from
anywhere to stay alive. A ram air turbine is "the thing you never want
to see used."

-- 

If a man will begin with certainties, he shall end with doubts, 
but if he will be content to begin with doubts he shall end in certainties.
Francis Bacon

On Wednesday, April 6, 2022 at 2:51:50 PM UTC-7, lang...@fonz.dk wrote:
> onsdag den 6. april 2022 kl. 23.37.03 UTC+2 skrev whit3rd: 
> > On Wednesday, April 6, 2022 at 8:53:36 AM UTC-7, jla...@highlandsniptechnology.com wrote: 
> > > https://www.dropbox.com/s/jnb3jmw8rcmdeir/XfmrScatter.JPG?raw=1 
> > > 
> > > How does the mass of a transformer scale with frequency? 
> > That's an interesting, and difficult, question. I'm assuming power transformer with 
> > unknown load on the secondary, there's current-transformer assumptions which are 'way different. 
> > 
> > Firstly, you need the core not to saturate at rated input voltage and 
> > no load; that means...

> > With only the first item considered, cross sectional area is constant for peak current-turns 
> > at no load, so expect V /f proportional to area, thus weight roughly scale by f**(-1.5). 
> > It's a project that'll benefit from log-log and/or semilog graph paper.

> can't it be more or less simplified to: 
> if you take a suitable transformer and run it at 8 times the frequency you can also 
> increase the voltage 8 times, and thus get 8 times the power

Yes, that's correct.   I was thinking, though, of resizing the core which would shorten
the wire length required, thus allow thinner wire with similar resistance; a redesign of
the transformer for the higher frequency is different from using the same transformer.
So, my scaling assumes a transformer reconfiguration in shape.  It doesn't get into
the correct way to do that wire re-dimensioning, because that includes dissipation of heat
changing with size... and heat can be shed by conduction, or convection, with different
power laws.