I want to make an inductor for this circuit, but my knowledge about electromagnetics is very limited. I was looking at some toroid cores at http://www.elfa.se/elfa-bin/dyndok.p...ok=2011480.htm - can anybody see which toroid, if any, of the ones described there can be used to make a suitable inductor for the ZLT++?
i was thinking of building a CCR2 sammie myself. I was reading up on some old and some recent ZLT discussions, lots of options described but no official "final parts list" thats proven to work well. it seems the Isense wire is very hard to adjust. that and inductor size are the biggest source of options and variation. i'm not sure if there is much choice as far as diodes or transistors, i think the FMMT618 was discussed but i'm not sure if anyone made the swap or the results.
If a final parts list can be produced, i can route the board and post it for those who need to. I'm thinking the indcutor can be fairly large cause i'll just route in onto the bottom of the heatsink on the other side. but i'll need to have the footprints for the components so i know how much area and volume i'm working with...
if MrAl or Newbie or pb wants to post a proven parts list and some alternatives for the inductor in varying but relatively close sizes in order of preference, i'll be more than happy to autoroute the PCB since i'm going to be doing that soon anyhow.
is there a good source for Coilcraft inductors yet? other than sample orders?
Actually I managed to get hold of quite a few parts for this circuit:
- zxsc300 chips
- fmmt617 transistors
- zhcs500 Schottky diodes
- some tiny x5r 10uF capacitors (ceramic)
For the inductor I was looking at the Ferroxcube TN 10/6/4 toroid core in the 3F3 material series, as it looks like it is small enough and has good permeability at ZLT circuit's operating frequency, 200kHz (?). I am not sure, however, that I understand the formula for inductance calculation. If N is the number of turns, does this mean that with 12 turns I get (12*12*740)nH = 107uH? How can I calculate the saturation current of an inductor made this way?
The toroid i used had a mu of about 5000 and it was 1/2
inch in diameter. If you select the 6000 mu toroid core
you'll probably be close enough.
The FMMT617 was used in most designs, along with the
Zetex 2000 Schottky diode (see their site).
As far as inductance and max sat current, you'll calculate
and inductance value but that will change because the tiny
core saturates for a short period of time.
The sat current will be fairly low for a small toroid like
this so it will look 'bad' at first, but the circuit
triggers on current so it's not quite as simple as with
a regular switcher with a linear inductor.
The hand wound inductors work great for me, I think someone mentioned larger capacitors are better? 50uf +? MrAl is the man to ask...
Do a search on posts from me "gpk", I got some great answers to my questions...
...yeah, what he said....where's the Rsense network? i can just add them to the schematics and layout pretty easily i think based on MrAl's thread.
but a question....is it necessary to use the very long and narrow toroidal inductor? given the same material and a ferrite bead core, would it be possible to substitute the core for something stubbier and maybe wider? how would that affect the number of windings and wire guage? are there "donut" shaped choices for winding that inductor? what's the reasoning behind the selection posted in the article?
I don't think that the shape of the inductor matters that much, at least not at these frequencies. I bet that both toroids and solenoids of different lengths will do as long as they do not saturate at or below the peak current, as MrAl noted. By the way, check out this page http://hyperphysics.phy-astr.gsu.edu...ic/indtor.html - it has inductance calculators for both types.
If you could make some layouts that include the two additional resistors that would be really sweet.
As for the inductor... I am not sure if this is complete nonsense, but isn't it true that the saturation current through a cored inductor with a given size and number of turns of wire is descending linear function of the relative permeability of the core (air-cored inductors never saturate?), while the inductance is an increasing linear function of the same? In that case choosing a core with lower mu value and adding a few more turns should help increase the sat. current while maintaining the same inductance? What do you think MrAl?
I have a couple formulas i'll present for your use.
I also have a new calculator on the way which was originally
intended for the EE course as i dont like the way most
available free calculators are set up. It doesnt do
everything yet, but it does quite a lot so i'll be
releasing one version in a day or two or three...
that way folks will be able to enter things in and get
quick answers to problems like these.
(or something similar)
will calculate the inductance from the toroid core
with a number of turns N. You'll have to enter in
the dimensions but that's all.
But before we get to the formulas, i just want to
stress that these calculations are not that important, really,
because the small toroid saturates for a short time
and it stops because the ic chip turns the transistor
off, so it's not quite the same as most switchers that
use a linear inductor. Not that you cant use a linear
inductor instead...it's been done with success...but
you'll have to try a few or else talk so someone who
found inductors that work well with this circuit.
I found some that worked but the efficiency wasnt as
high as with the hand wound coil on the small toroid.
That said, here's the formula for inductance:
L = 1.26*N*N*mu*Ae*1e-8 /(gap*mu+ml)
L is inductance in Henries
N is the number of turns
mu is the permeability of the core (typically 10 to 25000)
Ae is the area of the core in square cm
gap is the length of the gap (single gap in toroid) in cm
ml is the average (mean) magnetic path of the core in cm
Since your toroid will not have a gap, this reduces to:
L = 1.26*N*N*mu*Ae*1e-8 /(ml)
If you want to know about the saturation, you'll have to
look at the manufacturer's data sheet on the type of
material the core is made from. You'll find some level
of flux (B) that will represent the max allowing some
temperature rise. You can use the following formula
to determine if the inductor will saturate...
Bdc is the flux due to the dc current in gausses
Idc is the dc current in amps
N is the number of turns (wire goes through the center)
mu is the core material's permeability
gap is the gap length in cm
ml is the mean magnetic path in cm
With a gap of zero this reduces to:
After looking at the magnetic data you'll determine that
there is some max B (which we'll call Bmax) where the
core is saturated enough to be called in saturation.
Then, you can determine if the level of Bdc goes over Bmax
or not by simply comparing Bdc to Bmax. If Bdc is greater
than Bmax then it saturates.
If instead AL is given, you can calculate Bdc from:
where everything is the same as above, and
AL is the value of inductance (L) for 1000 turns on the core.
By examining the above formulas and doing a few calculations
we can show that it is true that a core with lower mu will
be harder to saturate, but it's also true that it will take
more turns to produce a given value of inductance.
Same with adding a small air gap.
Ok, now that we're done talking about the calculations
I'd just like to remind anyone who's interested that
the small toroid will most likely always saturate, but
that's not always a problem in a circuit that senses
inductor current directly (as with the Zetex 300 chip).
is there something wrong with using an inductor design like the following picture instead of the long inductor core that's mentioned in the article? can I use this kind of inductor (possibly with more windings) and get the same results or are there factors that prevent this (saturation problems maybe or available core materials)?
i figure it would be easier to make into a pill than the oddly-shaped Mostaccholi one. and since i'm gonna route the board for the ZLT+CCR-2, i want to see how i can space the pads.
To have something to experiment with I ordered an SMD inductor from Farnell (3631B220ML made by MEGGITT) - they have the datasheet here: http://www.farnell.com/datasheets/31236.pdf I will let you know if it works well for the ZLT... though it is not tiny, it may be worth the extra space if the circuit efficiency is decent.
Right now I can barely wait to get my hands on this and some other parts I ordered (zhcs2000 Schottkys, fmmt618 transistors and some resistors). I think I might start with the simpler ZLT design to test the inductors and then build the ZLT++.
Yes, the units for Bdc are gausses. 330mT would be 330 milliteslas, which equals
3300 gausses as you probably already know [img]/ubbthreads/images/graemlins/smile.gif[/img] (1 Tesla = 10,000 Gausses)
If you post all the dimensions of your core and the permeability i'll double
check your calculations if you like.
The efficiency was about 85 percent or maybe 86. All the other inductors
came in around 80 percent, but it may be that i didnt get to try enough
other ones to make a good comparison. The mu of that core was 5000.
I also recommend building up a standard Zetex 300 circuit before converting
it into a ZLT-2, just so you know everything is working beforehand (as
you already intended to do).
Couldnt get your link to work however...it said the file
was not there or something.
Hi again, im not sure what inductor your talking about, perhaps post a pic?
That pic you already posted looks more like what i used. The number of
turns depends on what the mu of your core is...do you know what kind of
material that core is made of (the one in your pic)? And the exact
Dimensions for the core are:
Height (as when layed flat on a table)
I'll try to find one of the cores i used and then i'll post the dimensions.
I have just built my first converter. I used a small toroid core left over from some earlier project. I have taken a picture and put it here: http://zhat.dk/pics/my1stzetex.jpg - I appologize for the poor resolution, my camera cannot focus at very short distance, that's why...
From some rough measurements with a DMM on the 10A scale (and the 20V scale for voltage) I calculated the efficiency to be 0.86, which rather surprised me. I believe that the core is MMG Neosid, material F19, I know I bought a few of these once... the mu is 1000 and the saturation flux density is 250 Gauss if the data I found is correct. So the inductor most definitely should saturate.
By the way, the capacitors on both in- and output are 10uF multilayer ceramic, and seem to suffice. Maybe even smaller input capacitor would do.
I forgot to mention that this is NOT the CCR-2 but more or less the standard circuit from Zetex Lighting Handbook design note 61.
I'll go over the calculations and post results here either
a little later tonight or tomorrow morning.
86 percent sounds good [img]/ubbthreads/images/graemlins/smile.gif[/img]
The sat flux being 250 on that material sounds way too low,
but im not familiar with F19 material so i'd have to
look that up too...i would have guessed 2500 instead or
in that region. The bottom line however is...
what is the efficiency...which appears to be very good for
your core. I'd use it! [img]/ubbthreads/images/graemlins/smile.gif[/img]
BTW very nice pic even though your camera isnt set up to
take close ups. Sometimes if you need a really close
pic if you have a magnifying glass large enough to
cover the entire lense of the camera you can tape it
on or something, then using a tripod, get a decent
close up that way. Color isnt 100 percent, but it
works to some degree. I took many close up pics of ic's
where the ic filled the whole photo frame using a normal
50mm camera and an old magnifying lense i had laying around.
Cant wait to hear your results using the CCR-2 too.
I'll post again later or tomorrow as early as possible.
Thank you MrAl, I will try the trick with the magnifying glass tonight.
You were right: I misread the F19 specs table - it says 260 mT and not 260 Gauss.
Just got my stuff from Farnell and the toroids from ELFA, can't wait to try those parts.
MrAl, do you have any comments on the capacitors? Can smaller input caps (e.g. 2.2 uF) be used? As I see it, the input caps improve efficiency by reducing current peaks through the battery, but how much does this matter in practice?
Oh ok, 260mT means 2600 gauss, so that sounds good.
I'd be very interested to find out what kind of results you
get from the new parts too. I'd even maybe like to start
a small database where we can store results from
experiments like these for future reference. A lot of
people like this circuit and it would be nice if they (we)
could just go look up the configuration and find parts
that work well without doing too much work. On the other
hand, experimentation is part of the fun [img]/ubbthreads/images/graemlins/smile.gif[/img]
Myself as well as others have run circuits without any
input cap at all. You're correct in that the input cap
smooths the pulses through the batteries and that could
make them last longer. Reducing the input pulse by half
would probably be a reasonable goal, so if you'd like
to try that too i can suggest a capacitor value that
will achieve that...i need to know your battery type and
how many in series, and your output current.
Reducing the peaks though the battery makes it easier on
the battery, but i think if your using NiCd's you might
not notice too much difference in run time as this type
handles pulses pretty well.
BTW, on the core calculations,
for a core
with an AL of 740 and 6 turns of wire the sat current
comes in around 0.88 amps. This is fairly close to
your estimate of 1 amp i guess.
designing for 1xCR123A, do post inductor and cap part numbers that work well. also, MrAl if you could post the part # of your inductor i woul dbe interested in getting a few off digikey to play with. and what gauge wire was used for 6 turns?
I have now built the CCR-2 circuit but I can't make it work very well. Maybe I need an R5 with greater resistance.
As for the inductors, I got very good results with the MEGGITT 3631B220ML inductor, but also with an axial inductor from BI Technologies, which you can find here: http://www.elfa.se/elfa-bin/setpage....ok=2015770.htm (I have the 22uH one). I got something like 0.85 efficiency using the same rough measurements as described earlier.
The hand wound inductors did not perform too well.
I also tried putting an extra 47uF cap on the output, but without much difference.
MrAl, I have an idea, but I'm not sure that it will work. It is to insert a zener diode with the drop of Vb (the low input voltage) in series with the R5 resistor to make the voltage across the bias network drop to zero at Vin = Vb. This way, the peak current through the inductor at Vb would only be determined by the value of Rsense. This would make the calibration procedure easier because (provided that R3 is set to 5R) the only resistor that needs to be adjusted is R5, and for this a variable (trimmer) resistor could be used. But again, I don't know if this would work at all.
The inductor i used was a core from All Electronics and 6 turns of #24 wire.
They arent hard to make really. If you want to purchase one from Digikey
you can look for something maybe 50uH or better and a sat current of 2 amps
or better. Operator_smooth seems to have some good results with his
purchased inductors too.
If you add 100uf to the input the peak current though the battery will
be much less. That should increase the run time.
The 15m ohm resistor needs to be adjusted for your circuit. We've always
used a short length of thin wire (#32) starting with about 4 inches, then
making it shorter to increase output current to the desired level.
5.1 ohms should work fine for the 5 ohm value.
The R5 resistor really needs to be adjusted also, but your idea of the
zener sounds interesting too. All i can say without doing any calculations
is that you'd have to try it. You'll probably have to use a zener that is
below the Vb low point because zener voltages arent spec'd that closely,
but with a series resistor you should get some results. As the input voltage
rises, the inductor peak current is reduced which is what we want.
I'd start with about 1k and work it down, checking output current at
low batt, medium batt, and high batt voltage.
I do not have a zener diode with the voltage I need, but I connected two 1N4001 in series and an used adjustable 1kR resistor and got a very good result. The total voltage across the two diodes was around 1.3V. I measured 0.32A @ 2.88V, 0.33A @ 2.33V and 0.25A @ 1.3V.
Wow, that does sound pretty good. I like the idea of the
two diodes too, although i'd like to see some data
over a temperature range of say 20 deg C to maybe 40 deg C.
The diodes drop (and zener also) will be a bit temperature
sensitive, but it should still work to some degree.
BTW, the calculator i was talking about is now available
as a free download. I've posted it in the other thread.
You can do calculations with inductors and cores and
stuff like that, just by entering the parameters.
For example, suppose you wish to calculate the
magnetic path of a core who's dimensions are
given by it's OD and ID, one of the formulas included in
the file "EE_Formulas.dat" (which loads via menu open)
You just copy that text (after loading and clicking "Show")
and replace OD with the actual size and ID with the actual
ID....so for that core with 10.6mm OD and 5.2mm ID you would
Then highlight it and click "In Place Eval"
and the result will be placed after it with an equals sign.
Note the values of OD and ID have to be divided by
10 because the formula takes dimensions of cm, but if you
wanted to you could first convert the OD and ID using the
conversion formulas also in that file.
MrAl, would you please post the link to that thread, I would really like to check out your code.
I have been working on a peak current calculator for the ZLT circuit, which takes the following parameters: Vin, Vled, Iout and L and makes some assumptions about the off time (Toff = 1.7us) and the schottky forward voltage (Vschottky = 400mV).
Do any of the following equations look correct?
Inductor "charging" time: Ton = Ipeak * L / Vin
Inductor energy: E = Ipeak * Ipeak * L / 2
Power transfered to output: Pout = E / (Ton+Toff) * n, where n is estimated efficiency, e.g. 0.85.
If it is assumed that Pout = Iout * (Vschottkey + Vled)
then it should be easy to solve the equations (iteratively or explicitly) to find Ipeak for the given values of L, Iout, Vin and Vled?
To calculate Rsense some assumptions must be made about the sense voltage Vsense. Looking at the charts in the zxsc300 specs I think that a good linear approximation would be: Vsense = 15mV + 0.02mV/V * Vin
Hey you've got some pretty good ideas there.
I like the idea about calculating the Vsense too, but
it may be that we would have to calculate two values:
an upper limit and a lower limit.
In any case, it would be simple to write out an equation
for the calculator something like this (for Vsense):
and for inductor energy:
These two (as well as any more you care to add) can
be saved to a file called for example 'ZLT_Formulas.dat'
for later use by clicking on the menu File/SaveAs. The
entire set of formulas can then be reloaded by clicking
on the menu File/Open.
Sound simple enough?
When the person who wishes to calculate a value goes to
use the calc, they would first open the formulas file
by clicking on the File/Open menu, then click "Show"
then select the formula they wish to use. Let's say
They would then copy and paste
then replace 'Vin' with their input voltage like:
then highlight that line and click "In Place Eval".
The answer then appears to the right of the
with an equals sign.
It's easier to do then to explain :-)