ZLT-CCR2 PCBs

[Leeoniya]

i've started autorouting on a 0.75" diameter board, and i have to say, the components look like bread crumbs on a dinnerplate. this circuit can get SMALL.

i'll post a routed board with the extra sense resistor network later tonight.

what mh does the inductor need to be? aside from being able to take 1A-1.5A current and have low DCR? can someone post a part # or several in different package sizes so I can plop them on just to get an idea of size ranges?

i remember an extra non-ceramic (maybe tant) input or output cap being discussed in parallel with a ceramic one to increase the uF but keep the ESR low?

i dont understand all this zener diode talk. (not a EE here, i just dabble.) is that instead of the sense resistors? (all i know is that zener diodes start to conduct backwards at a certain voltage but what does that do here?) and why was the LED used?

...so confused.

this is what i'm working with. do i need to add another tant cap on input or output (if so what uF)? what mH does the coil need to be? everything else look okay?

 

[MrAl]

Hi there Leon,

Geeze, one IMPORTANT thing missing...the LED !!!
Dont run the circuit without an LED connected or else
use the Luxeon simulator circuit (3 or 4 diodes of
type 1N4003 in series, also in series with a 1 ohm
resistor). Make absolutely sure the luxeon or the simulator
is connected so it gets forward biased when a voltage
appears across the output cap. Failure to do this could
result in SEVERAL parts blowing up... the output cap,
the transistor, the zetex diode, and the zetex chip.
I can stress this enough!

One more little note...
the input power should be applied right across the input
cap C1. The leads from the battery can connect directly to
that cap.

The inductor value ranges from about 20uH to about 100uH,
but i was hoping operator_smooth would post a part
number for the one he tried and got 85% efficiency with.
If not, Digikey has lots of small inductors that should
work good enough unless you feel like winding your own.
You may even get away with an air core, with about 20 turns
on a 1/4 inch form (not high efficiency however).
You dont need the zener really, but if you wish to use
a diode or zener in the bias circuit you would connect
it in series with the 1.1k resistor so that it gets
forward biased (anode connected to the batt + terminal).
You still need the resistors however.

Before you connect power for the first time please check
and recheck connections and parts polarities. If only one
part is not connected correctly it could take out other
parts with it if it blows out.

Whatever you do dont forget to connect the output LED or
simulator.

Hope this helps...

Take care,
Al

 

[operator_smooth]

Hi again,

MrAl, here is the link again:
http://www.elfa.se/elfa-bin/setpage.pl?http://www.elfa.se/elfa-bin/dyndok.pl?vat=0&dok=2015770.htm
According to PDF file linked to on that page the part number should be HM50-220K.

The other inductor is from MEGGITT and has part number 3631B220ML

If my calculations are correct, 10uH and 15uH inductors could be used as well - below that the operating frequency at 3V is getting close to the 200kHz limit that is recommended in the Zetex specs.

The idea with the extra diode is to make adjustment of the resistor network easier so that only Rsense has to be adjusted at Vb (the low input voltage) and then only R3 needs to be adjusted at Va (high input voltage). It also improves the efficiency at very low input voltages a tiny bit (but only a really TINY bit) by eliminating the network biasing current when input voltage is below Vb.

By the way, I was trying to visualize the effect of a linear peak current compensation for the decreasing input voltage on the output current so I made an HTML page for that purpose - you can even try with different voltage values and inductance: http://zhat.dk/cpf/zlt-ccr-2-graphs.html I don't know if the calculations are correct though.

Leeoniya: I look forward to seeing your PCB layout, I got really excited when I read that you can make it tiny.

 

[Leeoniya]

wow, didnt mean to scare you, hehe. i left the LED out cause i wasnt gonna use it for routing on the same side of the board, and i wasnt ever going to connect power without a load or testing it first. it will be put in for your comfort.

is it worth putting another tant cap in parallel with the input or output ceramic? i know this was brought up before.

also, what' the best way to wind the Isense wire? I presume away from the inductor and try not to coil it it back in on itself or does that not make too much a difference.

I have a really neat new pill design. not too sure of the details since i'll need to cad it out in a while. i'll post pictures tonight.

the design is not so new but it improves greatly on the circuit/pill design of the QIII. it incorporates a 0.102" thick silver disc for mounting the luxeon to. the back of that disk is arctic silvered to a 1/32" double sided PCB, the heatsink is drilled through for the V+ of the emitter and has a plastic insulative sleve. the disc itself is grounded to the flashlight body. the rest of it resembles a QIII setup with an aluminum sleeve around the board, possibly with the contact pad spring mounted for more universal applications.

i'll post cad later tonight.

Leon

 

[operator_smooth]

Leeoniya: now you've really woken my appetite!

I think that there is no real need for the big capacitors - off course this is always a compromise but the difference in terms of efficiency would probably not be that big if it was a 50uF cap instead of a 10uF one.

The Rsense could also be an SMD resistor, so I suggest that you layout the pads as if it was a 0805-size resistor. The best way is not to wind it but twist it and bend it - this way we avoid coupling from unshielded inductor and avoid making an inductor of it by winding a coil.

 

[Leeoniya]

okay, did some searching. since i'm in the states, digikey is where i shop.

check out these specs on the 22uH ones and tell me what you guys think:

axial with 0.05 DCR:
http://www.jwmiller.com/pdf2/5800.pdf
5800-220

and this one with a higher DCR (0.085) but higher current:
http://www.jwmiller.com/pdf2/PM3316.pdf
PM3316-220M ... this is probably too large for my liking and my pill design. also has a rel. high DCR compared to the cheaper following one:

this one looks good as well with 0.07 DCR:
http://www.jwmiller.com/pdf2/6000.pdf
6000-220K

this last radial one looks good, it's only $1.51.

let me know what you guys think, i'm gonna order all the parts tonight or tomorrow. probably along with a few FMMT618 for testing.

thanks,
Leon.

CAD will be posted in an hour or so.

 

[operator_smooth]

I think that all of the inductors that you mentioned should work very well - I'd go for the one with the smallest DCR.

 

[Leeoniya]

the problems i see right now with my design. is that the components take up too much space. for example, the diode, which really only needs 2 pads, instead uses 6. also, if the one of the 1206 caps could be moved to the contact PCB, that would help with space as well.

video here:
http://216.170.179.171/assem.avi

the weird looking thing is that 5800 series inductor. scarily big.

the biggest problem becomes the inductor. the lowest DCR one of the ones i picked has a max size that is bigger than any dimension on my pill.... my pill OD is 0.625" and the max length of the inductor is speced at 0.7" I was wondering if anyone knew whether or not this is dependent on what uH inductor you buy from that series (i asssume more coils would indicate that), or if it's the same size for every cap and it's just a tolerance? this is important to know for me cause it would make a huge difference.

the actual electronics are being routed inside the aluminum spacer, the ID of it is 0.495" (OD is 0.625)

as you can see from the pic, I got lazy and didnt take out the proper pins for all the SOT-23 packages, but i'll do that later. the point is that the diode has a lot of wasted space. since it only needs 2 contacts.

i put together a small vid. for those who have high-speed. it's 10 megs. it shows how the assembly goes together for the most part.

is there another diode I can use that only uses 2 contact pads? and is that inductor size the same for that entire series or does it change with uH?

thanks,
Leon.

 

[operator_smooth]

Your design really looks nice.
Could you please post the copper path layouts?

 

[MrAl]

Hi again,

operator:
Hey that graph is cool. One thing though, is that the ledgend that tells what
the colors mean is overlapping the input boxes in my browser. If you could move
the colors indicator box down about 2 inches that would be great.
I think i'll add graphing to my calculator next.

Leon:
Hee hee, no prob. I just like to make sure people know about the 'disconnected LED'
problem with boost circuits because i've heard way too many stories of how they
turned the circuit on only to blow out a multitude of parts. I hate to see that
happen, especially when this might be someones first boost circuit.
BTW the extra pins on the diode are to take more heat away from the die. That part
is rated for 2 amps, which at full steam could be heating up. The benefit to using
that particular diode is that it's got lower than usual forward V drop, which
increases eff.
Nice video...how did you make that? What software?
Yeah some of the inductors are bigggggg.


Take care,
Al

 

[operator_smooth]

I have set fixed widths on the form and legend boxes, I hope that it helps. I have only Mozilla and MSIE6 to test the page so I am not too surprised if other browsers render it differently. Did the graphs look correct?

Is there any reason not to use a smaller inductor, e.g. 10uH? I think that it would be easier to find or make a small 10uH inductor that can handle 2.5A than it is to find a 22uH inductor with almost the same current rating.

 

[MrAl]

Hi again operator,

The graphs at:
http://zhat.dk/cpf/zlt-ccr-2-graphs.html
look much better now in IE browser.
Before you couldnt even see the 'ok' button, much less
click it /ubbthreads/images/graemlins/smile.gif Now it's great.
The graphs look correct (although i havent gone over
the calculations yet...sorry) but it would help a lot
if you added grid lines...even a train of dots separated
by 2 dots each would be nice for each grid line, or even
just one dot at the corner of each grid intersection.

Using a smaller inductor must be possible, because i think
Zetex has one circuit showing an unusually small value.
When i first calculated everything, it looked like this
would increase output ripple, but the output cap just
has to be sized properly.
One thing about the hand wound coil i used is that the
inductance swings from low to high, and when it's low it's
very low...like 20uH.

Maybe it's time to go over the calculations to try to make
some new judgements about this circuit.

Some initial observations:
A 10uH inductor will charge twice as fast as a 20uH, so
it will play closer to the switching time of the transistor,
which could lower efficiency.
The 'on' duty cycle might be longer, reducing efficiency
also. With the hand wound coil the 'on' duty cycle is
very short.

Bottom line however is still this: Does the efficiency
increase or decrease with the use of a lower inductor?
If it decreases much, can the decrease be tolerated in order
to obtain the benefit of reduced size?

If i can find a little time i'll go over the calculations
again.


Take care for now,
Al

 

[operator_smooth]

Gridlines added.
Also added calculation of R3 and Rsense.

 

[Leeoniya]

i use autodesk inventor 9 professional. Its exploded assembly animation is somewhat more limited in parameters than i would like (you can only control how many frames are used to record the entire animation and what video codec to use, all activity is captured during animation, like panning or rotating. it's sort of tricky to group components together to move at the same time also (like PCB and components), but it's quick and easy and i dont feel like whipping out lightwave, maya, or 3ds max.)

so i guess no one knows if those axial inductor sizes are the same for all mH values in that 5800 series?

thanks,
Leon.

will post pcb tonight. i want to find a useable inductor for my pill! grrrrr.

 

[MrAl]

Hi again guys,

operator:
Very nice! The grid lines are good, i was wondering if you
could get maybe each increment of say 1 amp to be a bolder
color, say for 1 amp, 2 amp, and for voltage maybe
0.5 volts, 1 volt, 1.5 volt, 2 volt, etc., which would
make it easier to read. Very nice however in any case.
What formulas are you using...i'll take a second look.

Leon:
Sorry i couldnt find anything but what you probably already
know...
make length is 0.7 inchs, make diameter is 0.275, min
lead length is 1 inch of #22 wire.
With this info you at least know it will mount within
a hole pattern of say 1.00 inches, no matter which
inductance you order.
Maybe i should try to come up with a set of inductors
that work for sale or something at cost + shipping.
People are always asking for inductors i guess because
that's the only strange part in the whole circuit.

Would be nice to say..."ok here... send me $1.00", or
something like that /ubbthreads/images/graemlins/smile.gif
This way people dont have to hunt them down and wonder if
they're going to work or not.

For now, order 1 or two parts to start with and see if you
like the results.

Let me know how it works if you can...

Take care,
Al

 

[operator_smooth]

I think the sizes are the same for all the inductors in that series. They probably use the same bobbin but with different gauge wire and number of turns.

 

[operator_smooth]

BTW, is there any reason why you prefer toroidal inductors, MrAl?

 

[MrAl]

Hi there operator,


The only reason i would prefer toroid inductors is if the
efficiency was higher in a given circuit using a toroid.
Since all the measurements i did back in 2002 showed that
the small toroid (with maybe 6 turns) gave me the highest
efficiency of any that i tried so i tended to like that.

There's also the cost...i think i got the cores surplus
pretty cheap so each inductor cost less than a dollar.
Maybe i'll look into getting some more and wind a few
so people can try them out. Shouldnt be too difficult /ubbthreads/images/graemlins/smile.gif


Take care,
Al

 

[operator_smooth]

I have made a simple design for the ZLT++ circuit - without the diode. I have built it and apart from being a little more challenging to make with a knife it seems to work.

To get regulation to work with a LED current around 320mA and input voltage 2-3V I found (by experiment) that the R3 should be something like 800 ohms - this means (assuming my calculations are correct) that the sense voltage is 21mV or a bit higher. I will try to improve the model on http://zhat.dk/cpf/zlt-ccr-2-graphs.html so it can help pick the right resistor values.

 

[Leeoniya]

okay, here is the deal. i have been battling my size restrictions for quite some time now. at last i am victorious. the casualties, however, were the sense wire, the inductor, and the negative emitter lead.

i had to get them out of the picture to have enough space to route the rest of the board in an 0.5" diameter circle. however. I still dont want to rely on the fact that the heatsink connects to the body which then connects to the aluminum spacer for grounding the luxeon. so i might still play with it and find a way to get the luxeon grounded to the PCB through the heatsink.

it was a challeneg, but i finally worked it out with lots of tweaking.

you may have noticed that this board is of course made for PNP blue, and not for an exacto knife. and at this size, i dont think you would have much success with one. the lack of copper may lead to concerns about heat dissipation, but at this size i would expect to pot this circuit in thermally conductive epoxy. thats the idea behind the thick aluminum surrounding the circuit and the cilver heatsink mated to the other side of the PCB.

this will NOT get any smaller. believe me, there is absolutley no space...all the traces are routed with 20/15 rules. meaning all the widths are 0.020" and the min spacing between them is 0.015"

the whole board is 0.625 diam, and the components and traces are inside the aluminum spacer, the ID of which is 0.495"

let me know what you guys think. i had to do a lot of troubleshooting between schamatic pin designators and the pin number-to-pad designator links on the footprint layouts. so if i fudged something, let me know.

i'll post a more neat version in a bit. i'm just happy to see "100% completion" on my router after so much work.

Leon