Boost converter efficiency

[willib]

 

I used a fairly strong auto battery in series with a new 6V sla , on the output end, for 6V boost.

the donor battery is a 12V sla

I added a second inductor in parallel with the first .

At 60 percent efficiency i've been able get 53  Watts output

Maybe another diode in parallel with the first will get me better results?

at this point i'm stumpted , 100 Watts output seems real far away.

 output ------!--- input

 I - - V -----!--- I - - V .....E

 0.92(20.26) -!- 1.67(12.25).. 91 %

 1.44(21.91) -!- 3.13(12.10).. 83 %

 1.93(22.59) -!- 4.91(11.97).. 74 %

 2.27(23.37) -!- 7.44(11.81).. 60 %

for a description of the set up

http://www.fieldlines.com/story/2006/11/26/22831/362

The last time i did this I got 23 Watts output @ (84 %), this time i was able to get 32 Watts @ 83 %.

i think the 6V sla took the brunt of the voltage increase

[Nando]

What a rat nest !!

One of the reasons why the efficiency is lower !!

The efficiency is defined lower when the choke starts to get saturated with reduction of inductance and at the same time reduction of stored energy in the coil and the power supply current is incremented -- therefore lower efficiency.

To get out of "stumped" :

Show the circuit of the MosFet section, including the MosFet driving circuit to define the overall circuitry and to tell you what you could do to attain higher watts. and higher efficiencies

Supply values specially the coils, PWM frequency and Integrated Circuit.

Nando

[willib]

tomorrow i shall try boosting into 36V

if i can get the same current into 36 V then that would put me near 90 Watts.

[oztules]

Great to see you hard at it again Willib.

As you start to try for more power, you will have to pay more attention to the inductor. Saturation of the coil,... (fiddle with the air-gap and turns and thickness etc.)Also resistive losses in the coils as the current increases will quickly eat away at your efficiency. The waveform will suffer also.....are you using fast diodes yet?

If you re-read Flux's matching the load post, and perhaps the fun i had with my little converters, you'll find that it is only a switch a diode a coil and a pwm source.

Your pwm is controlled by you, the switch should be a low rds on fet, and your diode should be a fast shottky.... this leaves the coil to experiment on. sounds simple and mostly is. remember you may need some sort of snubber on the fet, and try to stop ringing in the output... waveform is everything and can tell you what is happening as you turn up the power. If the waveform falls to bits as you wind up the pulse width, then fets and coils will be heating up without much increase in output.

Nando correctly points out that the finely crafted network of wires that makes up the circuit of this fine device will contribute to your losses. This is of great importance in the power wiring in particular... try to keep the leads carrying the fast switching currents as short as practical...... not that I can preach in this area as my pics would indicate that mine wasn't much tidier.

Sorry for the absence, but my lousy net speed dwindled down to a trickle and became unusable.

Nice to see you are still at it

.....oztules

[willib]

no fast diodes yet, in fact i only have one diode attached to the circuit.

i was thinking two or three would be better?

just yesterday i looked at the pwm input to the gate and noticed a little ringing which as you can imagine produces a lot of ringing in the output, not a good thing when your fet is oscillating for part of the output.

i still have it mostly set up and can investigate the cause , i think i might need a driver on the gate , a little 2N7000 should do it? but like i said i havnt yet looked into what is causing the ringing

what do you think about using two fets, each with its own inductor and diode,it would be nice if i could get them to alternate , but i dont think thats possible, for me anyway.

at least my power output is out of the single digits  since i was able to get the PIC to produce 19Khz pwm

i used a monitor powersupply transformer( ferrite),dissambled it , and put in two coils in parallel , then i took another transformer and paralleled two more coils with the first two.

but lately i was working with my largest blades , i have them attached to a pair of rotors , which up to now was my test bed , for a hub i'm using a bicycle , with two roller bearings instead of the ball bearings that go where you would be pedaling from. very , very smooth .

i had to take some off the tips though , as it is now, they almost hit the floor.

the reason i'm fiddling with them is i noticed that when i changed the pitch on the small set i could slow them down a lot , for when it gets really windy.

essentially when they( the small ones) would be going 900 RPM , i believe i can cut that down to 450 RPM , with more torque,

so if it is a windy day i can decrease the pitch , if its not so windy i can set it to normal.

 it looks pretty cool and the whole alternator , less the stator ,is super light , i even thought of a way to use a saw blade as a disk brake ( they have them for bicycles now ! ),to stop it when needed.

nice to see you're back

william  .

[Flux]

I am inclined to suspect that the ringing you see on the gate drive is a consequence of the ringing on the drain circuit due to the rats nest. You need a low impedance gate drive to prevent the ringing on the drain circuit from destroying the gate insulation due to miller capacitance.

Fast diodes and extremely good layout are absolutely essential if you want to operate at any significant power level without smoke.

Flux

[willib]

Flux i tried to keep the resistance of both sides as low as possible

since i'm not willing to solder the ammeter connections , whats a person to do?

[maker of toys]

rule 0:  make only one change at a time when troubleshooting an exisiting circuit.  It may be worthwhile to keep a log of what you did and what effects it had.

physical volume is your enemy; capacitive and inductive coupling are problems, too.  However, inline inductance is useful in RFI/EMI suppression, and can help knock down the amplitude of ringing IF USED PROPERLY.  in general, short leads and small footprints couple less strongly than large ones.

look at your circuit mesh, and try to group each module tightly on the bread-board/PCB. keep the digital/HF switching stuff away from the analog.

Don't let wires for separate parts of the circuit run togeather. especially avoid letting power and signal lines run close to each-other. violate the 'shorter is better' rule as needed to achieve this goal. where they must cross, make it 90' if you can.

try to connect only one power lead and one common from each module to any of the other modules; best is if there is a single power distribution point and a single common point where all modules interface with the power and ground buses. (this last can be tricky to integrate with some other suggestions; do what you can.) in any case, keep analog/instrumentation power as separate as possible

solderable/ field terminatable banana plugs that will fit the common sort of test equipment/DVM are available at just about any electronics store. use these to make custom leads to replace the ones that are supplied with the equipment; and make the new leads as sort as possible. adaptors for BNC-terminated coax are available, too. Pomona makes lots of stuff for this market that are reasonably priced. (with coax, you introduce more capacitance, which might make ringing and ground loop problems worse. . . )

eliminate loops.  Be ruthless! use shunts and voltmeters instead of ammeters if you can. . .  you want the absolute minimum amount of wire carrying significant currents.

ground only one end of the shield in shielded cable.  (sometime exception: RF in coax cable)

Make absolutely sure there are no loops in the conductors carrying FET gate drive current, and keep those conductors away from the power conductors and all inductors.

twist the leads to/from each meter together as tightly as you can. (reduces RF coupling into and out of instrumentation cabling; helps with cross-talk, too.  clip-on ferrites around lead pairs may help damp some of the ringing, but i would consider them bandaids rather than real fixes.

Hope some of the above helps!

-Dan

[willib]

all good sugestions thanks Dan

Most of whats on the breadboard has nothing to do with the PWM, the pwm driving the Fet is at the far left , the whole right side is my data collecting PIC,which is on hold ATM.

i was thinking of socketing the PWM (PIC) controller , because the lead comming from it is about 8" long , and i could just run the power down to it , thusly shortening the drive lead to less than 2"

there are no ground loops, because i've since soldered a large solid copper wire to the fet board, which handles both battery ground points right at the FET.

the pwm ground is connected to this point also

i like the idea of using shunts for current measurement, as soon as i get back to it, at the moment i'm in the middle of hooking that VCR head up to my new mini gen.

the vcr head is an astounding piece of workmanship , ever look at one?

the heart of it in my opinion is the concentric loops of very fine wire embeded in two oposing ferrite disks.

they never touch but somehow transfer the tape signals to the vcr?

and what powers the heads if there is no physical connection to them?

i'm not asking you perse' just commenting on them..

anyway those disks are 2.1" approx by about an eigth of an inch, might be usefull to someone interrested in experimenting with a new type of alternator " the ferrite core axial alt " hmmm i think i smell another experiment ,lol

Anyway by this time tomorrow i should have the new minigen rotors mounted on the VCR head, i hope

[maker of toys]

glad that some of my little brain-dump sparked some ideas for you.  

IMHO, the shorter you can get the gate drive leads, the better things will work for you.  FETS, with their high gate impedance, are quite sensitive to stray capacitance in their drive circuits.  So, YES, socket that PIC. keep the gate lines away from the lines between the PIC and the rails, if you can. (you knew that. . .)

Your big copper common is the right idea.  (some of you may note that I differentiate between 'common' and 'ground' and 'negative bus.' making these sorts of distinctions becomes important when you ARE chasing ground loops and other unwanted oscillations and interferences. (fascinating/frustrating occupation, that. . .)) You can get 'common loops', too, though probably not with a nice single-point common like Willib is using.

-Dan

[willib]

i got it done.

its mounted to the stator support plate.

i should have taken shots of it during construction ,but i have to dissassemble it again so i will take photos









here i have the hub(head) inverted , amazeingly there were holes , just the right size to tap , to mount the adapter plate to

it spins like silk

as you can see in the first shot i've also got a new blade in the works