Hi Vic,
Sorry to hear about your extra shipping charges, that must be a pain.
Makes it harder to get stuff from some places. I see now why you are
having a problem obtaining a scope.
Also sorry for the long post, but i figured i better get this
stuff out there while i was still thinking about it
There are other ways to determine if the inductor is working
correctly. Most notable is if the switchers series transistor
starts to heat up rapidly. Since you may be working with a chip
that has a built in transistor, you may have to keep a finger
(carefully) on the chip to see if it heats up quickly. This might
be a good time to ask what chip did you decide on? It would be
nice to know if it has built in current limit or whatever.
The series transistor is the first thing to blow out if the
inductor doesnt work. It might overheat and blow quickly
so you have to be careful unless the ic chip provides
peak current limit.
IN circuit testing is a little more reliable, since after all your
test setup is not an abstract of what you 'intend' to do with the coil,
but rather the exact application. This is one reason why i recommend
this approach over others. Also, we still arent sure how exactly
you LCR meter works -- does it put out an ac voltage and measure
ac current, or does it put out an ac current and measure a voltage,
or does it put out a pulse and measure something at the end of the
pulse? To know for sure i think we would have to look into the
LCR meter then go from there, or else have someone check out the
meter with a scope and determine how it works exactly. Otherwise
i would be afraid to advise on the outcomes of tests used with
this meter. It 'should' work the way Doug and i were discussing
but im still a little wary about assuming it works exactly the
way we hope it does. Maybe it does, but i think we should use
its measurements as a 'backup' and not as a specification test
procedure until we know more. It would be nice to see the
results anyway.
Since by now you know that inductor saturation is the first and
most important thing to look for in a new design, ways of finding
this out without using a scope also include looking at the peak
current through the inductor. Im thinking that perhaps another
way to do this is to use a peak detector such as that used with
the Zetex circuit to detect output ripple. With a 0.1 ohm resistor
in series with the inductor perhaps we could look across this
resistor with a peak detector. At first we would measure a small
peak voltage--around 0.5 volt perhaps. Then as the input voltage
is increased this voltage would either stay the same or increase
slowly. If it increases very sharply we know the inductor is
saturating beyond that which is useable.
If instead the chip starts to cut back the current as to limit
the max output current, then the condition is easily detected
by noting a decrease in output voltage. This should be
easy to detect.
We also have to remember that this core doesnt have an air gap
so it's most likely going to saturate at some point. Where it
saturates and how bad its effect is will determine just how
much output current you can get out of the switcher with this
core, and possibly input output differential max too.
Questions:
1. Multi-strand wire
Dont bother yourself with this right now, as you dont need
super great efficiency with a test unit like this. If you
wish to try this later there's nothing wrong with it though.
Perhaps two strands would be enough just so you get the two
windings for testing the LCR meter idea (for short quick
tests so the wire doesnt overheat).
2. LCR meter protection...
If you decide to do these tests also, you
should connect the current source first (with binding posts or
something that wont become disconnected even for a split second).
Set the current then connect the meter. Disconnect the meter
before changing the current setting also. This protects the meter
from high surges in voltage. Remember when connecting and disconnecting
a source from an inductor could give rise to very high, short bursts
of energy in the form of a high voltage which could cause nasty
shocks too. Dont touch the coil at all when connecting/disconnecting
and make sure the connections cant come loose.
3. Acceptable percent inductance change...
Rather then specify a particular percent change, look for a minimum
inductance value that will still be useable with your input voltage
and ripple current requirements. Lets say 1 amp peak max ripple.
At 60kHz, with 30v input max and 15 volt output (switcher at
50% duty cycle) the minimum acceptable inductance is 62.5uH.
Since we want 3 amps dc output with 1 amp peak ripple this brings
the total peak current to 4 amps. With 4 amps inductor current
the inductance should be at least 62.5uH. If the inductance is
less then that the ripple current increases. At 31uH for example,
the ripple current will be twice that as 62uH which comes out to
about 2 amps peak instead of 1 amp peak for a total of 5 amps peak.
This also says something about the state of the inductor...
If we watch the ripple current if it rises too much we know
the core is no longer providing enough permeability to
give enough inductance for our application. Interestingly,
since the ripple current rises the output ripple voltage
also rises so this is another thing we can watch in order
to keep an eye on the dynamic state of the inductor.
Perhaps you would want to build up a peak detector to
check these measurements.
For these in circuit tests it may be possible to get the
ic chip to put out a 50% duty cycle switching pattern
regardless of the input voltage level. This would give
us a good way to test the inductor. This should be rather
easy to do by setting the output voltage for half of the
input voltage. This would require a voltage divider
with a pot to adjust output voltage with, and im sure
you can come up with something yourself to do this?
The output dc current would also have to be adjusted
for the same current (probably 3 amps).
This means the test procedure would go something like this...
[Test #1]
Starting with 10v input and 5v output loaded to 100ma (50 ohms).
Check output ripple voltage and possibly peak current
through inductor. Increase load in steps such as
200ma, 400ma, 1amp, 2amps, etc, up to 3 amps.
Check transistor/ic overheating, keep an eye on
ripple measurements. Increase input voltage if
needed to maintain 10v input and 5v output.
If you reach a point in output current where a problem
comes up, note the best input/output conditions you can
get just before the problem appears.
[Test #2]
Increase input to 15v and output to 7.5v loaded to 100ma.
Make measurements and check overheating as in Test #1.
[Test #3 and more]
Continue to gradually increase the input voltage and
output voltage making measurements and checking for
overheating.
In each test remember that as each quantity is slowly
changed each measurement should slowly change also
or not change much at all.
An abrupt change in a measurement indicates something
is going wrong, probably the inductor saturates beyond
that which is useable.
When increasing input voltage, try to adjust all the
other variables (output voltage and output current)
smoothly by increasing by 1 volt, then adjusting
output voltage and output current, then 1 more
volt, etc., etc.
If you only get up to maybe 20 volts before something
starts to increase quickly or something overheats,
then you either need to change something about the
inductor or get another core and start with that.
A core with a larger cross section would be good,
or perhaps even two cores side by side with turns
around both would work up to 30 volts where one core
couldnt make it.
Before that you could try adding 5 more turns and
see if that gets you up to a higher input voltage
before something goes wrong. If so, another
5 turns (10 more total) might get you up to the
full 30 volts.
Another idea if you have lots of these cores around is
to carefully break the core by placing it in a vise
after wrapping with maybe aluminum foil. Glue it back
together with super glue. This is kinda fun like a
3d puzzle
The glue will act as a small
air gap in two or more places in the magnetic path.
You'll then need to wrap turns and test with your LCR
meter as you did before. We can use the results of
this test to determine if we can use the broken core
or not, and if so how many turns are needed.
If it works, perhaps a higher temperature glue
would be better for long term use.
Hopefully the broken core wont require too many
turns to make it again act as a useable inductor
for this app.
Possibly more ideas in the future.
Take care,
Al