Commanda's design

[Madscientist267]

If you ground the bases of the totem pole driver, the mosfet should switch OFF.

Yeah, sorry... it was late... after re-reading my post, I realized I wrote it backwards. It works as you describe as far as that goes.

The MOSFET I'm using is an IRFZ44N (or IRFZ44V, I have some of both)... far as I know, the only pertinent difference as far as this goes is some current handling ability... the V's are a little higher.

To run on 5 volts, the mosfet must be a logic level gate device.

I even tried it with a 7812...

I DID notice one thing that I had missed that was in your schematic that isn't in play in the 'pick-pocketed' driver board that has the MC34063 on it, and that is the resistor at pin 8.

After looking at it closer, I figure this could easily cause some extra current draw (and therefore heat) since the original board was using the internal output stage in a darlington emitter follower config, and so there's way too much collector (and following base) current flowing inside the chip using the output stage as an inverter.

So it sounds like I'm on track, just got a little confused. It's a little unorthodox for me in this configuration...  

More to come... thanks for all your help so far... I'll keep you posted.

Steve

Also - Would I be alright using 3 or 4 in hand for the toroid rather than 2 (so that it can tolerate panel upgrades)?

[ghurd]

Steve,
This is one of those cases where the characteristics of the parts need to be correctly known.
You are not using those freebie counterfiet fets in this, are you?
G-

[TomW]

You are not using those freebie counterfiet fets in this, are you?
G-

Ha, didn't know they made fake Fire Emitting Transistors?

Makes sense they fake everything else "over there".

Sorry to digress with my idea of humor.

Tom

[ghurd]

Ha, didn't know they made fake Fire Emitting Transistors?

Makes sense they fake everything else "over there".

Nobody reported any problems with them in low speed switching applications.
I can not say for 100% certain they are fake, but there are just too many inconsistiencies with parts I get from reputable suppliers (not ebay from China) to feel comfortable with this batch.
If it is too good to be true.....
G-

[commanda]

Just checked the datasheet. IRFZ44's should be fine.
No resistor on pin 8 explains why everything is getting hot.

Amanda

[Madscientist267]

Nope not the counterfeits... blew the only other one I got working with the forwarder... FWIW they didn't behave any differently than the real ones. Smoke, fire, lockup, pretty consistent no matter what I put in that design...

The other 'counterfeit' is actually still working very well in my first trial run with the dump controller... dump load is a 55W 12V halogen. Runs like a champ.

Tom - LOL some of my best 'performances' came from the real ones... the fake one just kinda went 'plick' and shorted (i think, either that or opened)... and that was it... I got a NICE show of fire and smoke out of 2 in the pictures on the other thread... almost made me proud (in some sick kinda way) hahahaha

This one I think I just had a couple oversights, and reading the schematic I wasn't entirely sure how it was to work, as it didn't really look like a whole lot else I have seen, and never saw any references to the floating output ground design anywhere in the datasheets (except for the voltage inversion type, but thats another thing...  )

I did at least come up with a way to stop smoking the MOSFETs until I have the driver sorted out... I have another 55W halogen that I am going to use in place of the toroid (with additional ballast) so I can see the waveforms on the scope and what not before I go hooking the rest of it up. Got the idea from somewhere after watching the ghurd controller do it's thing, and then earlier I was standing outside the pharmacy down the street and realized that my DRL on my car is just the high beams, PWM'd down to a percentage of their full power...

Insight comes from strange places.  

Steve

[commanda]

Also - Would I be alright using 3 or 4 in hand for the toroid rather than 2 (so that it can tolerate panel upgrades)?

The more the merrier. More parallel windings means less resistance = less losses.

Amanda

[Madscientist267]

Well, it's built. And it works. I think. It passed the smoke tests...

I'm not certain exactly what the specs are at the moment... I kinda finished it on the fly and have only ran a quick test, but it appears to be working properly.

No more heat, and the MOSFET is even running just fine without a heatsink. Don't think I'll leave it that way, but so far, it works better than any of the others I've messed with so far.

No MPPT in this one, just the buck... Gotta get more duct tape so I can wrap my head around everything that's happening in the schemo...

Hopefully soon, I'll have an actual usable day and can go outside with the real panels and see what kinda numbers she comes up with...

Thanks again Amanda!

(don't go anywhere just yet heheh)

Steve

MODS/ADMINS: "UPLOAD FOLDER IS FULL" - Steve

Pix to come when that gets straightened out...

[commanda]

Glad to hear it.  TO220 devices don't like more than a couple of watts without any heatsinking.  Even a couple square inches of flat sheet aluminium does wonders.

You beat me by a couple of hours. I just finished populating my pcb and gave it a quick run up as well.
Straight off, it just "sounds better" than the somewhat bodgy veroboard prototype.

Now for the heat sink, dodgy connection to the mppt controller section (still on veroboard), and pray for sunshine on the weekend.

Amanda

[Madscientist267]

Trying the pix again...












I ended up just unwinding the blue toroid to the point of 40 turns, then re-mounting it on the 'socket'...

It's not truly complete here... this was a quick and dirty mockup (in terms of feedback) for the smoke test. Still have to add the real voltage dividers, finish the traces, etc.

Went heavy with everything (as you can see) to avoid any more popping... It gets old replacing stuff hahaha

Steve

[Madscientist267]

Amanda -

You're my hero!

It works!

Granted the sky is not cooperating with me at ALL right now, but earlier I could just barely make out the solar disk and the panels were pushing about 100mA into the buck, and I was getting 220mA out...

I too found the sweet spot, so this confirms it works as designed!

Efficiency is going down as the light dissapears; it's completely overcast (as if it might rain/snow) but I am still pulling 70 or 80mA into the battery, with 50mA going in.

I can't wait to see how this thing does in full sun!

Steve

[Madscientist267]

Looks pretty good...

Hit one small smack of halfway decent sunshine and caught 1.24A out with .5A in!!!

As the cloud cover came and went, I took a bunch of pictures and then plotted up a graph, but it's hand drawn so I don't trust the 'curve'.  It looks fairly linear until it gets down to about 200mA input, and efficiency really goes to hell below 50mA input. At that point who cares.

So, this appears to be pretty efficient... I'm going to take a stab and say between 90 and 95% or so just doing some really quick math in my head. Everything runs cool to the touch even at the 1.24A! (still no heatsink on the MOSFET)

At a 1:2.5 current gain, I can't complain at all... At least the output hit over an amp, thats WAY more than that flaky piece of crap I started out with was doing in full sunlight, and dunno for sure, but appears that it would beat the forwarder by a long shot too...

How is the MPPT part coming? I'm not sure I entirely understand everything that is in the schematic (as far as following the signals)... is there that much to be gained warranting putting it together?

Kinda curious to see the performance difference between just the buck and the buck with the MPPT added... If it's only a couple %, I might just leave it out... ?

Steve

[commanda]

Steve,

So glad to hear it's finally happening for you.
I've learnt a whole heap with the last few days of playing around.

Heat kills solar panel output.
Heat in the toroid is a function of Vin/Vout. With the input connected to one panel, running at 15 volts in and 13 volts and 7 amps out, temperature rise is minimal.
Change the input to 2 panels in series, Vin at 27 volts, same output 13 volts and 7 amps,  I had a 42mm toroid, the biggest I had, at over 100degC. Thats about 70 above ambient. Typical commercial coils have the current rated for a 50degC rise, so I'm definitely over doing it.  I re-wound it with 3 conductors, to cut down resistance losses, and saved 10 degC. It's still too hot.

Vp of the panels changes markedly with temperature. One panel at the side of the house, several months ago when it was cooler, Vp was 17.6 volts. Now, with ambient temperatures in the high 20's, it's 15 volts. The 2 x 65 watt panels I wired in series and laid flat on the hot tin roof above the back verandah, Vp is averaging around 27 volts. That's 13.5 volts each. If I wired them in parallel, there would be nothing to gain from any dc-dc converter.

As for the full mppt controller, the cct I sent you is only a beta, and not fully optimised. Although I have just populated the proper pcb for it, and plugged it in, and it quickly ran the input back to 27 volts, where I had earlier tweaked the pot to for maximum output without the mppt.  So it appears to be working.

Because the peak of the power curve is relatively flat, using just a dc-dc converter will get you most of the way there.  Tweaking the pot for seasonal temperature variations would be the next refinement. If one had large temperature variations across the course of the day, this may be sub-optimal.

My next move will be to a gapped ferrite E-core.

Amanda

[commanda]

So, this is where I'm at.





I've tacked the new core on the back, just for testing purposes. The ferrite is 42 x 42 x 15 mm, with a 1mm air gap. Number of turns is unknown, as I wound this some time ago.  But it measures 50uH. Looks like 18 turns of 1.2mm in one layer, counting by feel through the tape.

Aluminium heatsink is 25 x 12 x 1 mm angle, 75 mm long.

The add-on boards, standing vertically, from the left are:
output isolation diode, LTC4357 & fet, 20mV forward voltage drop.
current sensor, ACS712.
mppt controller, CD4046 & LM358. connects to main board with pin header & socket for easy removal.

Using an IR thermometer, temperature rise is no more than 5 degC, when running 27 volts input, and 7 amps at 13 volts output.
That's the heatsink and the inductor.

An interesting side effect of all this, the better this thing works, the faster the batteries get charged, and the less time there is for testing.

Amanda

[DamonHD]

An interesting side effect of all this, the better this thing works, the faster the batteries get charged, and the less time there is for testing.

So does that get filed under "the cussedness of things" or "nature's way of telling you that you're close enough"?

I'm deeply envious: that all looks so grown-up!

Rgds

Damon

[Madscientist267]

That's definitely true...

The other day, I ended up breaking out an old 60W incandescent plugged into a small MSW to dump the surplus as it was coming in.

I'd have to go with "the cussedness of things"... LOL

Very nice layout, I must say. How does the isolation diode/FET thing work? Went around in a few circles to try and eliminate it from the design, and haven't actually included it in the protoboard version, but I do see why it was part of your design... I get a small backfeed into the buck from the battery, although doesn't seem to be much (several mA if I recall), but would add up of course over time.

Due to parts available when I made mine, I have another minor problem that needs addressing... The OTV at the output rises above safe values for [some of] the output caps with the battery disconnected.

The temporary fix will be a 5 and 12V zener in series driving an optocoupler that will try to pull pin 1 of the 34063 low. The effect will be that when the output tries to rise above ~19-20V, the totem pole (and therefore MOSFET) stops getting the drive signal and shuts it down.

A modified ghurd controller sits on the battery side of the output, which is set up to open it's MOSFET when the voltage at the output goes above threshold.

I don't foresee any problems with heat due to any of this, because by the time the buck is triggering shutdown, there will be no load on it's output.

Steve

[Opera House]

"try to pull pin 1 of the 34063 low"

I would like to see how you do that.  If you went into that mode with a battery connected the fet would overheat due to low turn on voltage.  It would be just as easy to pull the voltage of the voltage sense pin high.

"take a stab and say between 90 and 95%"

Congratulations on getting it to work, but I can guarantee that you are closer to 86%. 

[Madscientist267]

If you went into that mode with a battery connected the fet would overheat

There would be no load on it when it goes into shutdown. When the battery is reconnected, the opto would let go immediately and the FET would be driven at full voltage. It's all or nothing, no in-between to worry about insufficient drive.

If I had a load on it while it was in transition around 18-20V, it would most certainly overheat. Something would have to go wrong with the charge controller in order for this to happen. If I had a battery at 19V, I'd have more than an overheating MOSFET to worry about.

Can't go high on pin 4 because the duty cycle will never see 0%. At best, it goes down to ~12% duty, so this is the only way to shut it down.

If you look carefully at the design, the 'input' gets inverted at the output, so that a higher voltage at the input drives it harder. It's not the typical use case for this chip.

Steve

[commanda]

Efficiency: with careful choice of components, a straight buck converter can get up to 95%.  I've got one in a project for work doing a confirmed 95%. To get above 95%, you need to go to a synchronous buck.  86% is typical of a push-pull converter.

The full battery problem: I would envisage the usual course, with lead-acid batteries, would be to just dump the excess.

In my case, I'm charging Lithium batteries.  Once any cell reaches full, the charge circuit becomes a constant current source to trickle charge the battery.  The logic which controls this hangs from battery positive to input negative. So once in CC mode, it gets the full input voltage. With a straight 12 volt panel, the simple resistor-zener regulator works fine. Directly connect a 24 volt panel, and the resistor goes failsafe and lets the smoke out.  I've since fitted a transistor pre-regulator, but going to need a more elegant solution.

My other problem is hash on the output. I can't get any 2 meters to agree on what the output current is.  So I've added an extra LC filter, but haven't been able to test it yet.

Amanda

[Opera House]

Depressing  going around thinking that you will never fully charge a battery.  I don't see why setting a maximum battery voltage isn't a priority. 

"At best, it goes down to ~12% duty"

That would be a poorly designed regulator chip id that was true.  Every power supply would just keep climbing in voltage.  Read that again. Every regulator will shut off.  Some may limit minimum width when it is enabled.

[commanda]

Depressing  going around thinking that you will never fully charge a battery.

Not sure what sparked this comment, or even who it is directed at.  If it's for me, just let me say this is not the time nor the place for a lecture on the ins & outs of Lithium batteries.  Just let me say that they do fully charge.

"At best, it goes down to ~12% duty"

That would be a poorly designed regulator chip id that was true.  Every power supply would just keep climbing in voltage.  Read that again. Every regulator will shut off.  Some may limit minimum width when it is enabled.

Perhaps you should Google "Hysteretic control", or COT (constant on time).  Also note that many supplies have a minimum load requirement, often met by the cooling fan.

Getting back to Lithium batteries. Maybe I should do a major write-up. I've now got 3 separate setups, dating back over 2 years.  2KWHrs in my electric scooter, which is my daily commuter. 2KWHrs on my solar setup, which is hugely under-used at the moment, just runs the TV in the bedroom nightly.  And another 1/2KWHr in my pushbike, which will get ridden regularly, if this rain ever recedes.

Amanda

[bj]

  If you do decide to do a write-up Amanda, I for one would consider it a must read.

[Madscientist267]

Opera -

This is one of those times where I'm going to call you straight out in the open, and say that you don't know what you're talking about.

Amanda is using lithiums, I am using FLA.

Rest assured my testing battery is COMPLETELY full as well, having dumped yesterday for a few hours, the OTV after sitting overnight is 13.36V.

I would suggest downloading the datasheet for the MC34063, looking carefully at the schematic posted on page one, properly interpret what you see, and learn a little bit about what you're going to say before you say it.

This buck converter works absolute wonders, and was designed by one of the greater minds on this board. The only time I had problems was when I strayed a little too far away from her design.

I'm also going to re-affirm my original statement of 90-95% here, as there is almost NO heat anywhere in this unit. After I approached 4 amps yesterday going into the battery, the MOSFET finally began to get warm enough to warrant mounting the heatsink. No other signs of concerning heat anywhere else.

Steve

[madlabs]

Hi All,

I finally got around to reading this entire thread. When I saw that the MC34063 is available in PDIP, I got excited. I have messed around with some MPPT stuff, with mixed results.

Anyway, I have access to a CNC router to make PCBs with. I am wondering if there is a layout available, and if it is routable. There can't be any traces between .1 pitch pins for example, the router method can't go that fine. If there is one, or if someone wants to make one, I'll cut and send a couple of boards for ya.

A couple of questions:

Any reccomends for low ESR caps?
Is the IRFZ44 really ok? I see it mentioned, and I have a few, but I'm gonna be ordering some parts anyway.
Source for the inductor core? I can find one, but has someone already sourced this?
How much current is possible/realistic with this design? I have 660 watts already mounted that I need to control. I have another 440 watts that I have yet to mount.

Right now I'm doing a simple relay controller, and I need to milk MORE POWER!!

Jonatahn

[commanda]

Hi All,

I finally got around to reading this entire thread. When I saw that the MC34063 is available in PDIP, I got excited. I have messed around with some MPPT stuff, with mixed results.

Anyway, I have access to a CNC router to make PCBs with. I am wondering if there is a layout available, and if it is routable. There can't be any traces between .1 pitch pins for example, the router method can't go that fine. If there is one, or if someone wants to make one, I'll cut and send a couple of boards for ya.

A couple of questions:

Any reccomends for low ESR caps?
Is the IRFZ44 really ok? I see it mentioned, and I have a few, but I'm gonna be ordering some parts anyway.
Source for the inductor core? I can find one, but has someone already sourced this?
How much current is possible/realistic with this design? I have 660 watts already mounted that I need to control. I have another 440 watts that I have yet to mount.

Right now I'm doing a simple relay controller, and I need to milk MORE POWER!!

Jonatahn

Jonathan,

May I suggest you slow down just a bit. I haven't released any pcb files for this, as it depends too much on exactly what components are going to be used, and how it is physically housed.
Steve may have a toroid working, but I was getting too much temperature rise with what I had available, at the power levels I am working at.  I am now using a gapped ferrite E-core, not a toroid.
And speaking of power levels, I wouldn't be taking this beyond 100 watts or so.  Suits me just fine, as my solar panels are a motley collection picked up second hand, so I want one converter per panel, basically.
At increased power levels, the gremlins become harder to tame.  To go much beyond where this is at, would first require going to a synchronous buck converter, and then to an interleaved polyphase design.

By all means, have a play. Steve's is apparently working fine on matrix board, and my original prototype was working on veroboard. So it is certainly possible. An oscilloscope and inductance meter would be very high on my list of "must-haves".

Amanda

[Madscientist267]

Amanda -

Mine may not be seeing the heat issues for two reasons... First, lower power. Peaking out around 4A output or so. Second, I have a feeling the toroid I am using is much larger physically than it probably needs to be.

I got away with just 1 in hand on this one, but if I were to go smaller, I would definitely use more. I almost rewound it with more anyway, but got a little giddy toward the end and laziness took over. Guess it worked out to just be a 'magical' combination. 

Also, I think your 'confused meter' situation is probably due to output capacitance. I really laid them on thick, using like 8 caps in parallel in the 'final' design. 4 of these are 22uF 'lytics, the others are something like 1000 and 470.  At the input I also have 2 hefty caps as well. Because the completion of the conversion has to run all the way out to a battery, I'm thinking low ESR on both the input and output is paramount.

The meters are probably reacting differently to the RMS of the ripple, Hence disagreement.

I know when I first looked at your board, the first thing that came to mind was the lack of caps; and that was before you made mention of the meter issues. 

Considering that you're pushing twice the power through that I am, I would wager a bet that this is the problem.

FWIW

[commanda]

Just one of my many failings. I never put enough caps on anything.
Re-vamping the pcb layout now, will stuff in more capacitors.

Also, consider some protection. Those long leads coming from the panels, and going to the batteries, act like antenna's.
A decent lightning strike within several Km's will induce unbelievable voltages.  Series inductance and parallel MOV's will go a long way towards mitigating this.
Incidentally, this is part of the reason why I like to run the controller chip off a 3 terminal regulator. Regulator fails short cct, and protects the rest of the electronics.

Amanda

[Madscientist267]

LOL I hadn't thought about that...

I put a small inductor in series with the output, on the positive rail just before the last cap (470uF). It's not much of anything; about a dozen turns of #16 or 18 on a 1 x 3/16 inch ferrite rod (not sure what that is in metric exactly). The thinking there was different however, really only thought about radiating switching harmonics etc. Probably should have had one on the input as well if even only for the same reason.

I wonder if in addition to coils and MOVs, has anyone considered some kind of common mode rejection for the wires coming from the panels? Something like UTP comes to mind, but not sure if the concept is limited to that or not... Or is this just thinking too far outside the box (ie running them side-by-together in the same cable already provides all the common mode rejection to be had)

An interesting side trip:

I remember reading something about a guy who was doing something to his system one evening during a storm (he was inside IIRC). He was killed when lightning struck closeby enough that the light from the bolt energized his panels briefly, and he had ahold of the wrong wires and that was that.

At least that was the story, they said it wasn't a direct hit or even a streamer, because nothing electronic was damaged, and there was no 'entry wound' anywhere on the house.

This being said, I kinda wonder after the fact if it might have been antenna effect rather than a pulse from the photons...

Anyways, just reminded me of it.

Steve

[madlabs]

Ok, I'll *try* to slow down. Would be a first, but I can *try*.

My panels are rated for 3.3 amps, it's a 12V system. So if I could put 3 of them per controller, I'd be happy. That is above your power level, but not by a huge amount. I don't mind making prototypes that end up getting changed. Or blown up. I have a couple of 'scopes, analog and digital, and an inductance meter. I have some cores and have wound many inductors and transformers. I found driving a FET right on proto(vero) board to be tough at higher powers (> 5 amps). Since the CNC is close by, I don't mind making ones that will be superceeded, just to get in the game. The thing for my is that while I can build and troubleshoot OK, I don't know enough to design the circuit myself. But DIY MPPT is a holy grail for me. I want to be able to fix it if it craps out.

So, are you interested in making a routable layout? If so, the offer of boards stands. If the layout is routable, it's also home etchable. That would make it available to the DIY person.

Jonathan

May I suggest you slow down just a bit. I haven't released any pcb files for this, as it depends too much on exactly what components are going to be used, and how it is physically housed.
Steve may have a toroid working, but I was getting too much temperature rise with what I had available, at the power levels I am working at.  I am now using a gapped ferrite E-core, not a toroid.
And speaking of power levels, I wouldn't be taking this beyond 100 watts or so.  Suits me just fine, as my solar panels are a motley collection picked up second hand, so I want one converter per panel, basically.
At increased power levels, the gremlins become harder to tame.  To go much beyond where this is at, would first require going to a synchronous buck converter, and then to an interleaved polyphase design.

By all means, have a play. Steve's is apparently working fine on matrix board, and my original prototype was working on veroboard. So it is certainly possible. An oscilloscope and inductance meter would be very high on my list of "must-haves".

Amanda

[Madscientist267]

Well, if you're up for experimentation, I'm good with it. But Amanda would have to release the schematic to you, I can't (ok won't) do that 'for' her, it's her design.

I have completely clear skies today, and am experiencing technical difficulties. I'm not seeing quite the output I had the other day, and can't quite put my finger on the reason why... 

As it sits, right now I'd be better off just running the panels in parallel. Wish I knew why...

More to come when I figure out the problem.

Steve

[Madscientist267]

Well, I have come to the conclusion that there is only one explaination for the efficiency shift.

I reverted the two mods I made and it made no difference whatsoever. Kept one of the reverts however, due to parasitic duty cycle shifting while trying to adjust the trimmer. I don't own a plastic tweaker, and didn't feel like making one, so it's 10K from input rail to ground, slid almost all the way to ground to get my 1.25V. Shrug.

Another thing I tried when I was dinking around with the board is a small cap across pin 4 to ground. All this resulted in was a very noticeable hiss of toroid chaos regardless of setting, accompanied by a 15 or 20% loss in efficency. Undone.

The only conclusion I come to is that the cells in the panels are running a bit warmer in the continuous full sun than when I got my 15 second 'miracle' the other day. It is about 40 degrees F out, but this is the only thing left in the chain that I can figure has a play in it. I gather this since I never really saw a whole hell of a lot more than 1.5A in the panels. I don't know what they were simultaneously, but I was able to finally tweak it in and get about 3.5A of output from the buck...

I've pulled the plug on the test, and just letting it 'do its thing' now, monitoring only battery voltage from time to time until all the sunlight I stole out of the battery and meekly shot back at the universe via a 60W light bulb gets replaced.

Interesting thing, speaking of which, is that for the most part, there was only a slight overall discharge with the buck dialed in, and bulb bound. The problem with this is that the lamp is 60W, which works out to 5A at 12V (ignoring losses and trivial battery deviations). How the &@*% does 3.5 translate to 5? Is this one of those obnoxious algebraic rings or something?  

I'm thinking something like real world numbers that look something like this (based on that whole geometric shape under a paper plate on a billiard table type thing):

Buck to battery:

12.3V @ 3.5A = 43.05W

Battery to inverter:
12.3V @ 3.65A = 44.895W

What SHOULD be:

12.3V @ 4.87A = 60W (again ignoring losses)

Include a 15% hit in the inverter:

12.3V @ 5.6A = 69W

Anybody got a good one for that?

Steve

[commanda]

Ok, I'll *try* to slow down. Would be a first, but I can *try*.

My panels are rated for 3.3 amps, it's a 12V system. So if I could put 3 of them per controller, I'd be happy. That is above your power level, but not by a huge amount. I don't mind making prototypes that end up getting changed. Or blown up. I have a couple of 'scopes, analog and digital, and an inductance meter. I have some cores and have wound many inductors and transformers. I found driving a FET right on proto(vero) board to be tough at higher powers (> 5 amps). Since the CNC is close by, I don't mind making ones that will be superceeded, just to get in the game. The thing for my is that while I can build and troubleshoot OK, I don't know enough to design the circuit myself. But DIY MPPT is a holy grail for me. I want to be able to fix it if it craps out.

So, are you interested in making a routable layout? If so, the offer of boards stands. If the layout is routable, it's also home etchable. That would make it available to the DIY person.

Jonathan

Jonathan,

I suggest you start with a prototype on vero or matrix board, mainly so you can get a handle on the inductor. Magnetics is not my specialty, and I only succeed by educated trial and error.
Driving the fet is not that difficult, this circuit works well, just keep the totem pole driver close to the fet, and a supply bypass cap close to that.
I'l try and tidy up the basic circuit and get it posted in the next day or so.  Meanwhile, there is a screen grab from the simulator, might be in the previous thread?

What does your cnc router require in the way of input files?

Amanda

[Madscientist267]

Aye eye to that...

It's a serious challenge bringing it all together on a successful level. I'm on number... oh hell lost count, and still am not entirely sure what my little efficiency drop came from in a 2-3 day period.

The most difficult aspects in building this thing:

    The proximity of key components physically on the board (MOSFET, input caps, output caps, totem pole)

    Getting ESR as low as possible

    And last but DEFINITELY not least, the %&$#ing coil.

The hardest part about the whole project. The inductor makes this whole thing a Pandora's box from hell. 


Other things that come to mind...

    Heavy traces for the current path, particularly the output side amongst the output caps, since you will be dealing with higher current there than on the input side. Don't skimp on the input either, but it's not AS important as the output.

    Caps, caps, and more caps.

    Keep a FUSE on it during development, and after if you're going to go above 50W or so. You'll be dealing with some 'fake panelling' for some of your early test runs, and current can get away from you QUICKLY!

    In addition to a fuse, use BALLAST RESISTORS! I have a string of 3 8R 20W Dale ceramics that I use for this. They will save you a lot of time and trouble in things like replacement MOSFETs, and also provide you a controlled way of estimating your real world (actual panels) gains. They are not perfect replacements for panels, because the 'curve' is linear rather than a wonky hyperbola, but you'll get close enough to say whether its ready for the panels or not.

Get all your guess work out on the bench, because unless you have a completely clear day with at least a few hour stretch, you'll learn all about frustration as even the thinnest clouds come and go, screwing with your head. The 'fake panels' won't do that, and any discrepancies will have to be with the circuit itself.

Be prepared to let the smoke (and fire if you do it just right) out of a half dozen MOSFETs before you get a handle on things. Granted, I didn't have as much trouble with the buck as I did trying to get the forwarder going (dead count was like 20 before I gave up), but I still had some problems along the way.

One day I'll master the forwarder too, just because I refuse to concede that a bunch of parts on a board can out-wit a human. 

TTFN

Steve