300w boost converter ..take 2

[oztules]

No I'll download it wen I go to Whitemark later this week. Connection to slow here.

.............oztules

[willib]

nevermind i found it -- 54 turns of 1.8mm wire on some sort of core material, sorry for dumb questions

[willib]

i'm running at 1250 hz which is the highest i can go without reprogramming , so 1250 it is for now

i think i'm gonna need a larger inductance?

dissregard the 20mA i got before here is the new data..

input I . .output I . .output V (current is in amps)

(.1) . . . . . (.07 ) . . . . .(12.37)

(.2)  . . . . . (.14) . . . . .(12.39)

(.5) . . . . . .(.28) . . . . . (12.42)

(1.0) . . . . . (.43) . . . . . (12.47)

(2.0) . . . . . (.6) . . . . .  (12.5)

(3.0) . . . . . (.64) . . . . . (12.56)

i didnt want to go above 3A because of my flimsy wireing

what do you think about the inductor i should be using for this frequency

[oztules]

efficiency dropped away after 1A input. What was the input V

Well done first attempt... see it does work and it is easy.

Either your flimsy wiring was absorbing the extra input current, or your inductor efficiency fell off.

I don't know which, but i assume that with your low freq, more turns on the inductor and airgap testing may be the key here.

Quick experiment will show the way.

Check the air gap as this will saturate at higher currents, and not give you any return on the output voltage.

So.

 1  Fiddle the airgap and note the difference.

 2  throw on some more turns and note the difference.

Not very scientific I know but from here it the best I can do.

The inductor is the key to output i have found. My setup is not ideal, but it works. You will do well to invest time into testing every whichway with the inductor and airgap as that is the key to your efficiency over the range.

A lousy inductor works efficiently for a very small current, then falls away as you pour the juice on.

A good inductor works well for the small current, but keeps the faith for larger currents as well. Eventually you will saturate any inductor, but it's all in the coil.

have you checked the wave form across the fet yet.. Is is nice and square?

If the fets are staying dead cold at these currents, the wave form must be good. At 3A with poor waveform, heat will turn up very quickly.

Once again, well done, will this solve your stepup problem?

.................oztules

[oztules]

Had another thought about this. Plot the Watts in v's Watts out. I'm wondering what your input source was. Last night when I used the Makita batt, and input current wentfrom 10A but  up to 30A it was also dropping from 9v to 3v so power in stayed the same....then went batt went flat.......rats!

Is this happening here ? or is the input voltage staying pretty constant, and the efficiency dropping.

If the voltage is dropping rapidly, maybe the converter is not so bad..ie 1A - 3A but volts dropped from 9v - 3v I don't have this info... but you do.

............oztules

[willib]

this is the voltage across the FET, not exactly a nice looking wave yet

http://www.otherpower.com/images/scimages/2965/IM002047.JPG

left meter input current= 2.96A

middle meter input voltage 11.41V

right  meter output current 0.58A

http://www.otherpower.com/images/scimages/2965/IM002048.JPG

this is just after  i let some smoke out of that green wire there, not a problem.

the input battery has allready recovered to 11.93V

thats my coil/inductor there and diode

the fets did get warm but the coil did not , the wires didnt fare as well though

http://www.otherpower.com/images/scimages/2965/IM002049.JPG

[willib]

what air gap? i'm not getting what you mean by air gap , if this was wound on one on those inductor cylinders where would the air gap be?

[oztules]

My uneducated guess is that if that is the coil (aircoil) then this is why you will only have any efficiency at very low current. You will probably get a vast improvement just by putting some steel in the middle....even a screw driver..... move it in and out and watch the meters, that should give you some clues to follow.

At the freq you are using, that inductance will be very poor. So there is plenty of room for improvement... that is a good thing, coz thats where the learning will come from.

It is worth looking at the waveform at the fet input as well, if it's good, then fiddle with the inductor until the output one is as good at the loads you want.

At least it worked, now for the improvements.....

.............oztules

are we a happy puppy or what:)

[willib]

yeah this is pretty neat.

just did what you sugested , i stuck a socket in the middle and wow..

the output current stayed the same @ .51A But the input current went down from 1.42A to 1.0A.

it also dimished that  ringing (after that tall spike ) , down to a little blip

 

[willib]

just tried a 17ga. coil , a rather large one , lol

dont know the turns off hand around 100 -150 .

the input current was 1.55A and the output was 1.0A getting better !!

the wave looked a lot  better , the only thing out of place , was that very high spike and a little ringing right after ,i was all set to take a photo, but the camera batts need recharging

[willib]

this is the improved wave at 1.0A output and 1.57A input, the coil has about 200 turns of 17 ga. wire, with a socket in the center

same wave , no flash

[oztules]

Now we are seeing some results from experimentation. You always learn more when things start out bad.

Remember, that at some stage you will loose more in the winding than you gain from the extra turns. This is where you will start to use coil designs with less turns but better core material. The same rules apply to the magnetising effects as apply to the coils in your windmill.

Remember when everyone used steel strips to increase their magnetic fields in their stator design.... it was to concentrate the flux.  Now its time to do the same thing. Using an E core ferrite, you can magnify the effects dramatically, and leave the socket behind.....but it proved a valuable point.

The cross sectional area is important (assuming the freq stays the same). It will be a product of cross section by turns (on a given ferrite compound) that will give you the inductance.

The reluctance of the socket former is probably helping spike the wave form unnecessarily, but your on the way.....keep testing .turns, former material, shape, etc

The meter leads will cause you grief as the current rises, and will distort the waveform as well. You may have to make short leads or use analogue meters, or a .1R resistance and measure the voltage over it to calculate the amps or some combination of each.

At some point, you will decide that more power is best done by thicker wire (or litz thicker) on a smaller former at a higher frequency....Plenty of variables for you to work with now....

As you go up in freq, you will need to be a bit more mindful of the layout.

notice how with each small step forward, things fall more into place....

................oztules

[oztules]

Also check the diode is a highspeed one. It must be able to track the transients. It looks like a 1N400X .if it is it's too slow, may help explain the peak at the start of the wave. (not turning on fast enough so the voltage peaks before it comes on.)...just a wild guess.

If it is not a fast diode, pinch one out of a computer power supply. The small TO220 rectifier on the  heatsink will do fine.(the 12v one) (the bigger one will be the 5v rectifier).

...........oztules

[commanda]

John,

You have the patience of a saint.

willib;

A few general rules of thumb;

Commercial SMPS (switch mode power supply) run at 20KHz plus. Computer power supplies run at 30-50KHz. Some small exotic supplies run at anything up to 1MHz.

Ferrite is your friend.

generally, the higher the frequency, the smaller the coil & core required.

I think the best advice anyone could give you right now would be;

1. : Increase your operating frequency. 10-20 KHz is where you want to be.
   
2. : That coil needs to be wound on a ferrite former.
   
   

As a last resort, grab an old computer power supply. Rip out the biggest transformer. Soak it in paint thinner overnight, then strip it. Use normal precautions when playing with paint stripper.

Amanda

[oztules]

Sorry GaryD Didn't see your question till now..

Well by now Willib has got one up and running, and is in the throws of increasing its efficiency and power handling. So It can be done.  The oatley electronics thing is cheap and worthwile to experiment with.

It only needs a coil and a diode and to get running as a basic booster. I assume that in the USA they will have similar kits to this (or just order it from them).

When your happy controlling motors, you can move on to boost testing by adding the coil and the diode and then hook up to your mill. It's a good way of getting a fix on what PWM stuff really is all about. It is not mysterious..although it can get as complicated as your imagination allows.

There is a fair bit of information with the kits. You won't get smoke with the controller. It is well designed and easy to build...easier than breadboard when you are ready to jump.

Just add a resistor for testing when you try boost. The only bits your able to kill is the fet and maybe the driver transistors...couple of dollars only.

This thing should be turning itself off at around 15mph, so the high windspeeds it won't even see or have any part of. (not in the off the self format though) Flux's use of the np25 and opamp to drive the pulse width down just needs to be incorporated. I will probably do this at a later date.... like when the fibreglass turns up and I can start my stator.....still waiting.

have fun........oztules

[oztules]

$1.56/litre down here in Bass Strait.... oil fields are in Bass Strait too.....makes it hurt even worse...... bet it's not so bad up there all of a sudden..

.............oztules

[oztules]

Inductor cylinder??? what is this? The E cores are two pieces of ferrite in the shape of an E  if the E's face each other and press together we have the usual transformer shape. We wind a cylindrical coil on a coil former (plastic) and slide the central E' arms into it. If they meet in the middle there is no air gap seperating the two E's. This will lead to a condition of saturation if we push decent current into it.

So we pull them apart just a little bit (approx 2mm for a start), and this allows us to not saturate the inductor at the current loads we intend.

So the inductance (turns and cross section) and the airgap give us our best output.

Do as Commanda suggests..hoik  the biggest tranny from a computer supply.

(don't laugh) I boiled the transformers in a pot of water on the stove  for 5 minutes and unglued them that way. (Wife wasn't home that day..didn't think of the stripper)

Wind a coil on the former and put it together so that you can slide the E's in and out. Turn on and see what happens. When pressed together, the input current goes up but the output doesn't. The coil is saturating.

As you slide them apart, the efficiency improves until as you go further, it degrades again. This will give you the best airgap for your best current transfer. (at least that how i did it).

..........oztules

[commanda]

<tic>But you don't have far to drive. Go about 10K's in any direction and you'd fall off the edge of the island, wouldn't you?.</tic>

(that's "tongue in cheek" if you haven't seen that tag before).

I burn up more petrol sitting at traffic lights than I do actually going anywhere.

Amanda

[oztules]

.... out my front gate.... turn left.....900 meters...... fall off island. who needs 10k's

glub glub glub........oztules

[willib]

Yes it's a cylindrical ferrite core wound with litz wire and found in some old pc monitors i've come across

[oztules]

Sounds like a choke... open at each end.

litz in a monitor..... probably fairly light capacity. ?

You will want fairly heavy for any current capacity.

Go for a transformer design. that way you can close the magnetic loop totally and then work back with an airgap.

Because the freq is so low, it will be a fair sized inductor, which will limit your current somewhat. but depends on how much your aiming for.

maybe a small picture.

........oztules

[willib]

picture? of ? the choke?

i tried to find one but in all the stuff i have around here i couldnt find one to take a picture of

i have a bunch of transformers though large and not so large.

amanda mentioned soaking them , have never tried that ..

how is it done?

[oztules]

A picture of the litz coil (so we can see if it is a choke etc.

Commanda propaply meant that in order to get the baked enamel off the ferrite transformers from a psu supply, first you must get the baked enamel off the and out of the core so you can slide the e sections apart. Some have e and i sections.

You must soften the lacquer up so you can get it apart. She soakes them in stripper, I boil them up in water. Hot air gun works to (hot to handle too).

Once apart you can rewind them for your purposes.

In your case with the low frequency, ant the usually small transformers in the 250w or so supplies, they may not be big enough until you up the freq a bit.

Experiment with whatever you have available. Litz at your freq is not much of an advantage. only starts to be useful at higher frequencies.

how much power are you after?

off to bed... 2.15am here

.........oztules

[willib]

above is a small choke and an ecore that i could possibly use , the ecore is laminated , the slots are around 1/2" wide

the choke ? is about an inch long with two terminals

[commanda]

Use the e-core. The choke will work for lower power levels, possibly 20 watts.

What you could do is up the frequency on the pic, and use the choke; that'll get you several steps closer to where you want to get.

Amanda

[oztules]

The thing to remember is that at 3khz I used about 54 turns of heavy (1.8mm) wire, and put it on a core slightly bigger than the e core you have there. The wire fills the winding area. (and I feel it could do with more.)It is just a learning tool but can easily do in excess of 300W with 1 coil.

To emulate that at your frequency, you would want over 150 turns of heavier gauge wire just to come close.... this is where your problem lies in getting any more efficiency at higher currents.

At 100khz, you may only require 5 turns of fencing wire to achieve the same thing on the same core. this you can fit..... but these frequencies bring their own bag full of problems..... most notably the layout.

The choke is a no starter unless your going to shoot for very much higher frequencies, and then your current capacity will be governed by the wire thickness and it's ability to carry current without significant R loss. If you have too thin a wire on the inductor, you will start to develop phase shifts in the coil which will lower your efficiency again. The more resistance, the higher the phase shift will be. (This effect is what you use to start split phase induction motors.... the start field is wound with different thickness wire, and so phase shifts are introduced to start the rotor moving..... (and then there is the cap start where the shift is effected by the capacitor).

So the long and the short of it is that at your frequencies, the ecore with at least 150 turns will (at an uneducated guess) get you close to good efficiency, but this will deteriate quickly as the wire size to achieve this may hold you back to 100W or so.

The choke will give you low power outputs before it gets very lossy.

Another rule of thumb I found is that if the coil gets hot, but the fet does not, then too many turns. If the fet gets hot but the coil does not...too little turns.

If they both warm up at the same rate(ish) then the power transmission is close to even between the fet and the inductor. Now this may be totally wrong for the theorists, but it is what I have found thus far.

just my guess

.........oztules

[willib]

wow , thanks

i read it and understand compleatly , i would really like to reprogram the pic ...

but my programmer isnt hooked up ( i built it myself), i could make another one but other projects are higher on my to do list, like finishing my awesome turbine blades..

and this link below..

if you havnt seen it, would you look at it.?

http://www.fieldlines.com/story/2006/5/6/72313/63838

[willib]

Thanks for your opinion , i value it

[oztules]

One last thing on the frequency/inductor issue.

I went out to the shed and put a pot in place of the 10k timing resistor. I then turned the input up to around 120w and twiddled the frequency to try and match whatever the inductor was, to the best output and so by definition best effeciency for that setup.

So try this:

If you check the figures on the meters, you will find that within a messy range inaccuracy, the efficiency appears to be very high.

You will also notice the air-gap.... it's overlarge, so the extra frequency had to compensate for this and so a match was obtained at 18.4khz.

The E cores are held apart with 3/16" plastic disks pushed down the inside of the former, holding the coils apart.

So it works just as we discussed.... nothing like testing it out.

best of luck

.............oztules

[willib]

1.8mm is only 0.0709" hmmm thats about a 13 ga. wire..

your fets are on the other side of the board ?

how many amps is that reading?

looks like 10A on the right one?

[oztules]

The input is about 9.5Ampish @ 11.86v=about 113w

the output is  about 8amps @ 13.19  = about 105w or around 93%

The fets are as per the overlay . Here I'm only using one fet on the board only, so testing doesn't blow up too many at once. It has no heatsink, and after a few minutes at this power level there is only a luke warm feel to the fet. So it must be efficient. Some losses will be in the wiring (pretty flimsy for 10A), some in the meter shunts, some in the coil and some accross the single fet. When the six of them are in the resistance across this junction will be very small, with only 1 fet it is quite high for these 100v 28A units.

So by arbitry  opening of the air gap, we can find a frequency which best suits it. These dumb systems are very flexible.

Most sensible people wind the inductor to suit their calculated parameters I guess... but I haven't been accused of being sensible recently.

Do you want a different pic of the board to check the output side?

..........oztules

[willib]

sure , any other pictures you have , i'd be interested in seeing..

[willib]

oz how did you  get those traces to handle 10A , or did you just not use traces on the power handling circuit