Gapping Pot Core

[WJLServo]

Working on eval design for power inverter; will use a pot core for
transformer. Have picked a core size, based on allowable hysteresis
loss. Also have a pretty good idea what magnetization inductance
needed. Requires an AL value a good bit smaller than that of largest
gap core in vendor's catalog...

Thought of buying a few core sets with largest available gap, then
modifying them to test for proper indutance, and get vendor to make a
special order core, same gap, for production. But, damned if I can
figure out how to remove material from core center posts. Could just
use gap shims, I guess, but that would open up gap at periphery as
well as center of core, and I would not then be able to put core in a
shield cover without incurring eddy current losses in metal cover.

An E core would be easier, I think: just come in from the side with a
grinding wheel and shave down the center post, using a bit of flood
coolant to make a nice, smooth ground surface. Getting to the center
post of a pot core would almost seem to require a jig grinder, which I
don't have. Any suggestions?

Thanks!
W Letendre

 

[Joerg]

Working on eval design for power inverter; will use a pot core for
transformer. Have picked a core size, based on allowable hysteresis
loss. Also have a pretty good idea what magnetization inductance
needed. Requires an AL value a good bit smaller than that of largest
gap core in vendor's catalog...

Thought of buying a few core sets with largest available gap, then
modifying them to test for proper indutance, and get vendor to make a
special order core, same gap, for production. But, damned if I can
figure out how to remove material from core center posts. Could just
use gap shims, I guess, but that would open up gap at periphery as
well as center of core, and I would not then be able to put core in a
shield cover without incurring eddy current losses in metal cover.

An E core would be easier, I think: just come in from the side with a
grinding wheel and shave down the center post, using a bit of flood
coolant to make a nice, smooth ground surface. Getting to the center
post of a pot core would almost seem to require a jig grinder, which I
don't have. Any suggestions?

Well, I have used a drill press with a very good grinding bit in there.
But be very careful. It can easily start to vibrate and then you have
stuff flying about at high speed. If you use a Dremel in a drill press
stand don't run that at full speed. Use eye protection and make sure
nobody stands close. Of course the liquid carries it's own risks. So be
careful, proceed at your own risk ...

Oh, BTW, if the drill press is one of those el-cheapo deals and you can
feel the bearing slack forget it, then you need a better one.

 

[John Smith]

Working on eval design for power inverter; will use a pot core for
transformer. Have picked a core size, based on allowable hysteresis
loss. Also have a pretty good idea what magnetization inductance
needed. Requires an AL value a good bit smaller than that of largest
gap core in vendor's catalog...

Thought of buying a few core sets with largest available gap, then
modifying them to test for proper indutance, and get vendor to make a
special order core, same gap, for production. But, damned if I can
figure out how to remove material from core center posts. Could just
use gap shims, I guess, but that would open up gap at periphery as
well as center of core, and I would not then be able to put core in a
shield cover without incurring eddy current losses in metal cover.

An E core would be easier, I think: just come in from the side with a
grinding wheel and shave down the center post, using a bit of flood
coolant to make a nice, smooth ground surface. Getting to the center
post of a pot core would almost seem to require a jig grinder, which I
don't have. Any suggestions?

Thanks!
W Letendre

Dremmel tool with diamond wheel?

How steady is your hand? How good your eye?

How many cores you got to waste? :-|

Regards,
JS

 

[Phil Allison]

"WJLServo"

Working on eval design for power inverter; will use a pot core for
transformer. Have picked a core size, based on allowable hysteresis
loss. Also have a pretty good idea what magnetization inductance
needed. Requires an AL value a good bit smaller than that of largest
gap core in vendor's catalog...

Thought of buying a few core sets with largest available gap, then
modifying them to test for proper indutance, and get vendor to make a
special order core, same gap, for production. But, damned if I can
figure out how to remove material from core center posts. Could just
use gap shims, I guess, but that would open up gap at periphery as
well as center of core, and I would not then be able to put core in a
shield cover without incurring eddy current losses in metal cover.

An E core would be easier, I think: just come in from the side with a
grinding wheel and shave down the center post, using a bit of flood
coolant to make a nice, smooth ground surface. Getting to the center
post of a pot core would almost seem to require a jig grinder, which I
don't have. Any suggestions?

** Consider using push pull drive to the transformer - obviates to need
to create gaps.



........ Phil

 

[Joerg]

"WJLServo"




** Consider using push pull drive to the transformer - obviates to need
to create gaps.

Yep, that's what I always do. And don't forget the inductor after the
diodes, especially if it's a big one.

 

[Winfield]

Working on eval design for power inverter; will use a pot core for
transformer. Have picked a core size, based on allowable hysteresis
loss. Also have a pretty good idea what magnetization inductance
needed. Requires an AL value a good bit smaller than that of largest
gap core in vendor's catalog...

Thought of buying a few core sets with largest available gap, then
modifying them to test for proper indutance, and get vendor to make
a special order core, same gap, for production. But, damned if I can
figure out how to remove material from core center posts. Could just
use gap shims, I guess, but that would open up gap at periphery as
well as center of core, and I would not then be able to put core in
a shield cover without incurring eddy current losses in metal cover.

An E core would be easier, I think: just come in from the side with a
grinding wheel and shave down the center post, using a bit of flood
coolant to make a nice, smooth ground surface. Getting to the center
post of a pot core would almost seem to require a jig grinder, which
I don't have. Any suggestions?

My advice: During the prototype phase, add the spacer
and don't worry about the flux leakage and eddy-current
losses with the shield -- they will be small. Go ahead
and measure the inverter's performance. Take heart in
the thought it might do better with a proper ground core.
Second, add the shield to get a handle on magnetic-field
leakage, etc., which at any rate will be less with the
production pot core. What's the problem? If you're
happy with the prototype design, tested this way, the
production one will be a bit better! So... what's the
big problem?

Oh, more advice: just forget grinding your ferrite core.

 

[Jim Thompson]

My advice: During the prototype phase, add the spacer
and don't worry about the flux leakage and eddy-current
losses with the shield -- they will be small. Go ahead
and measure the inverter's performance. Take heart in
the thought it might do better with a proper ground core.
Second, add the shield to get a handle on magnetic-field
leakage, etc., which at any rate will be less with the
production pot core. What's the problem? If you're
happy with the prototype design, tested this way, the
production one will be a bit better! So... what's the
big problem?

Oh, more advice: just forget grinding your ferrite core.

Yep, You can buy them already gapped.

...Jim Thompson

 

[WJLServo]

"WJLServo"






** Consider using push pull drive to the transformer - obviates to need
to create gaps.

....... Phil

Push pull drive would work, and work well. But, believe design I've
got in mind will be ~ $1.50 lower part cost. Target production volume
definitely high enough for part cost savings to pay off R&D labor.

In fact, had rather hoped that target volumes would be high enough so
that we could sweet talk vendor into making up an eval lot of cores
with assorted gaps "bracketing" intended AL value. Will still pursue
this, but, vendor is in Far East; they make good cores, and they make
them cheap, but they have not generally given us very good lead time
for prototype lots. So, thought I would try "fiddling" gap in our lab,
then giving gap spec to vendor for production order. May try Dremel
tool; do worry about cracking core!

W Letendre

 

[WJLServo]

My advice: During the prototype phase, add the spacer
and don't worry about the flux leakage and eddy-current
losses with the shield -- they will be small. Go ahead
and measure the inverter's performance. Take heart in
the thought it might do better with a proper ground core.
Second, add the shield to get a handle on magnetic-field
leakage, etc., which at any rate will be less with the
production pot core. What's the problem? If you're
happy with the prototype design, tested this way, the
production one will be a bit better! So... what's the
big problem?

Oh, more advice: just forget grinding your ferrite core.

Heh! Think your advice may be best! Only reservation is wondering how
easily the gap generated with spacer shim can be translated into a
production value. Had thought that a modified core with center post
ground to match desired AL could then be measured using dial indicator
to give an exact gap spec for vendor to use for production cores. At
the end of the day, though, may just use spacer, measure AL rather
than gap, and ask vendor to take AL value as their production spec...

W Letendre

 

[Terry Given]

Heh! Think your advice may be best! Only reservation is wondering how
easily the gap generated with spacer shim can be translated into a
production value. Had thought that a modified core with center post
ground to match desired AL could then be measured using dial indicator
to give an exact gap spec for vendor to use for production cores. At
the end of the day, though, may just use spacer, measure AL rather
than gap, and ask vendor to take AL value as their production spec...

W Letendre

firstly, if you gap across the whole core face then the total gap is 2x
the gap thickness (this is apparent if you think about the path the flux
traverses, it crosses the gap in the centre leg, and again in the outer
limbs). so centre-leg-gap = 2 x shim-thickness

Secondly, you can always calculate it, if you know the material
permeability. its pretty easy to do.

Cheers
Terry

 

[WJLServo]

firstly, if you gap across the whole core face then the total gap is 2x
the gap thickness (this is apparent if you think about the path the flux
traverses, it crosses the gap in the centre leg, and again in the outer
limbs). so centre-leg-gap = 2 x shim-thickness

Secondly, you can always calculate it, if you know the material
permeability. its pretty easy to do.

Cheers
Terry

Tried calculation, first up. Values I came up with were close to
vendor's published values, AL versus gap, for the half dozen stock
gaps they offer. Close... but not close enough for design use,
especially as I need to go ~ 2X out from largest stock gap. Think I do
need to test design with actual gap.

2X rule of thumb (center post versus dore face gap) may also be close
enough; would like to think that it is. Think, though, that I will
trust the measurements over the math!

W Letendre

 

[Eeyore]

Tried calculation, first up. Values I came up with were close to
vendor's published values, AL versus gap, for the half dozen stock
gaps they offer. Close... but not close enough for design use,
especially as I need to go ~ 2X out from largest stock gap. Think I do
need to test design with actual gap.

Get yourself a copy of Epcos's 'Ferrite Magnetic Designer' program. It makes
life so easy.

Why aren't you shimming rather than gapping ? Even low volume manufacturing
(and some not so low volume) uses shims rather than gaps.

Graham

 

[Terry Given]

WJLServo wrote:






Get yourself a copy of Epcos's 'Ferrite Magnetic Designer' program. It makes
life so easy.

Why aren't you shimming rather than gapping ? Even low volume manufacturing
(and some not so low volume) uses shims rather than gaps.

he already explained that - gapping across the core allows fringing flux
to escape, which defeats the purpose of using pot cores, which is to
keep the fields contained.

Graham

Cheers
Terry

 

[Terry Given]

Tried calculation, first up. Values I came up with were close to
vendor's published values, AL versus gap, for the half dozen stock
gaps they offer. Close... but not close enough for design use,
especially as I need to go ~ 2X out from largest stock gap. Think I do
need to test design with actual gap.

for "large" gaps (> 0.1mm or so) then assuming all the energy is
concentrated in the gap is a very good approximation. for small gaps,
this is not very accurate, but again (assuming you have access to
Snellings Ferrites book) this is fairly easy to calculate.

because you want a big gap, you may also want to take fringing into
account, which increases the effective pole area.

even without the finer considerations, the all-energy-in-gap approach
will get you very close to the desired gap.

2X rule of thumb (center post versus dore face gap) may also be close
enough; would like to think that it is. Think, though, that I will
trust the measurements over the math!

W Letendre

its not a rule of thumb. follow the flux path, and you see where the 2x
comes from. the maths is based on some pretty sound (and well
understood) physics.

for power converter work, IME the exact value of L isnt *that* critical,
provided you allow for sufficient margin in the overall design to suck
up tolerances. Heck, just look at a graph of relative permeability vs
temperature! or look at the spec'd tolerances in pre-gapped magnetics.....

Cheers
Terry

 

[Winfield]

Heh! Think your advice may be best! Only reservation is wondering how
easily the gap generated with spacer shim can be translated into a
production value. Had thought that a modified core with center post
ground to match desired AL could then be measured using dial indicator
to give an exact gap spec for vendor to use for production cores. At
the end of the day, though, may just use spacer, measure AL rather
than gap, and ask vendor to take AL value as their production spec...

That's right, except I suggest you begin comparing
your A_L measurements to the limited standard A_L
values available for that ferrite, and regap/rewind
your core until you match one, then test your setup.
Continue this process until you like your design,
then order that AL ferrite core.

Measure A_L, not gaps, and compare to the catalog.

A_L is easy to measure, wind 5 - 10 turns and measure
the inductance. A_L = L / N^2, where L is in nH.

 

[Joerg]

Push pull drive would work, and work well. But, believe design I've
got in mind will be ~ $1.50 lower part cost. Target production volume
definitely high enough for part cost savings to pay off R&D labor.

In fact, had rather hoped that target volumes would be high enough so
that we could sweet talk vendor into making up an eval lot of cores
with assorted gaps "bracketing" intended AL value. Will still pursue
this, but, vendor is in Far East; they make good cores, and they make
them cheap, but they have not generally given us very good lead time
for prototype lots. So, thought I would try "fiddling" gap in our lab,
then giving gap spec to vendor for production order. May try Dremel
tool; do worry about cracking core!

Then just give it to a really good machine shop. They'll do it
professionally and precise.

 

[WJLServo]

Then just give it to a really good machine shop. They'll do it
professionally and precise.

--
Regards, Joerg

http://www.analogconsultants.com/- Hide quoted text -

- Show quoted text -

True enough. Does require a shop with at least a little experience
with brittle, high Rockwell materials, though. Ferrite, as a ceramic,
almost certainly requires grinding rather than machining. Would hate
to give a ferrite core to a machinist who might try to cut it with an
end mill!

W Letendre

 

[Spehro Pefhany]

True enough. Does require a shop with at least a little experience
with brittle, high Rockwell materials, though. Ferrite, as a ceramic,
almost certainly requires grinding rather than machining. Would hate
to give a ferrite core to a machinist who might try to cut it with an
end mill!

W Letendre

I doubt that would happen- most machinists are pretty canny. If it's
an RM type pot core your machinist may be able to handle it on an
ordinary surface grinder with a narrow wheel and a magnetic chuck.

Best regards,
Spehro Pefhany

 

[Joerg]

True enough. Does require a shop with at least a little experience
with brittle, high Rockwell materials, though. Ferrite, as a ceramic,
almost certainly requires grinding rather than machining. Would hate
to give a ferrite core to a machinist who might try to cut it with an
end mill!

Use a shop that is experienced in handling ceramics. They should be able
to grind it down precisely to within a few micrometers.

 

[qrk]

Heh! Think your advice may be best! Only reservation is wondering how
easily the gap generated with spacer shim can be translated into a
production value. Had thought that a modified core with center post
ground to match desired AL could then be measured using dial indicator
to give an exact gap spec for vendor to use for production cores. At
the end of the day, though, may just use spacer, measure AL rather
than gap, and ask vendor to take AL value as their production spec...

W Letendre

Win's method is what I have used in the past. Beware that the junk
spitting out from the core can get into other electronics and cause
problems if your gap is large enough. I've used business cards as
shims. Seems to be the best use for them. For proto purposes, you can
use black electrical tape to hold the core halves together. Since you
are gapping your cores, the offical hardware isn't necessary for your
experimental stage.

You need diamond tools to gap your own ferrite. Best to use the shim
method.

Once you have a gap you like, simply calculate the AL of your core set
by measuring the inductance of a coil with a known number of turns. Be
sure your bobbin is nearly full as the AL value is more sensitive to
winding height with lower AL vaules (a 0.2 relative winding height
will give an AL value 6% lower at and AL of 160, 2% lower at an AL of
600 for many pot cores). When ordering custom gapped cores, all you
need to give is the AL value you want and they do the rest. However,
see if you can use a standard gap as that makes life much easier and
cheaper.