Author Topic: Commanda's design  (Read 43859 times)

0 Members and 1 Guest are viewing this topic.

Madscientist267

  • Hero Member
  • *****
  • Posts: 1181
  • Country: us
  • Uh oh. Now what have I done?
Commanda's design
« on: December 17, 2010, 06:59:52 PM »
As per just before the fork (heading toward push-pull forwarding converter) in the previous thread, I'm going to give Amanda's MC34063 based design a shot:

http://fieldlines.com/board/index.php/topic,144628.msg978165.html#msg978165

For the Push-Pull Forwarding Converter, see:

http://fieldlines.com/board/index.php/topic,144676.0.html

I'll be working the two threads independently, this one dealing with the encounters with the 34063 design. The other will be the KA3525 design.

Now then:

Amanda,

One question... just to make sure I'm following this correctly:

Effectively, a high input turns the pwm down, so the fet is on more.

Meaning:

Input meaning inverting comparator input (pin 5) I assume.

But high input turning the pwm down, which based on the OEM design, would have resulted in the regulator lowering the on-time of the duty cycle found at the output stage tranny in the chip, pins 1/2.

So if I'm reading your schemo correctly, you've flipped the duty cycle 'over' with the complementary pair so that (just for illustration), 30-on/70-off would become 70-on/30-off, correct?

Steve
« Last Edit: December 17, 2010, 07:26:20 PM by Madscientist267 »
The size of the project matters not.
How much magic smoke it contains does !

JW

  • Development Manager
  • Super Hero Member Plus
  • *****
  • Posts: 4052
  • Country: us
    • Flashsteam.com
Re: Commanda's design
« Reply #1 on: December 17, 2010, 07:06:44 PM »
Commanda is having some login problems, please bear with us, as we get this resolved.

JW

commanda

  • Hero Member
  • *****
  • Posts: 731
Re: Commanda's design
« Reply #2 on: December 17, 2010, 09:05:55 PM »
OK, back at my home computer. Work computer will have to wait till Monday.

Steve,

Yes, the pwm is inverted by using the chips output transistor to pull to ground.
The complementary pair is just to get the Fet's gate drive current back up.
If you connect the gate straight to pin 1, gate current to turn the fet on can only come from the pullup resistor.
This would be fine in a low frequency application, like motor speed control at a couple of KHz max.

For a buck converter running at 40-50 KHz, we need a lot more gate drive current.

With the inverted pwm, to get 70% on time of the fet, the chip has to run at 30% on time.
Remember this when choosing the timing cap.

I've got a prototype knocked up, just waiting for my chip to come in from Farnell.

Amanda

commanda

  • Hero Member
  • *****
  • Posts: 731
Re: Commanda's design
« Reply #3 on: December 18, 2010, 03:49:30 PM »
Update:
I have simulated this in LTspice, and it appears to work as expected.
I can post the files if anybody else wants to have a play.

Amanda

DamonHD

  • Administrator
  • Super Hero Member Plus
  • *****
  • Posts: 4125
  • Country: gb
    • Earth Notes
Re: Commanda's design
« Reply #4 on: December 18, 2010, 03:54:52 PM »
I'd be interested to have a quick look, but please don't bust a gut!

Rgds

Damon
Podcast: https://www.earth.org.uk/SECTION_podcast.html

@DamonHD@mastodon.social

commanda

  • Hero Member
  • *****
  • Posts: 731
Re: Commanda's design
« Reply #5 on: December 18, 2010, 04:56:47 PM »
Here's a zip file. Hope it works for you.

* MC34063.zip (5.62 kB - downloaded 548 times.)


/Program Files/LTC/LTspiceIV/lib/sub/mc34063.lib
/Program Files/LTC/LTspiceIV/lib/sym/mc34063.asy
(anywhere_you_like) /MC34063_mppt.asc

Amanda

DamonHD

  • Administrator
  • Super Hero Member Plus
  • *****
  • Posts: 4125
  • Country: gb
    • Earth Notes
Re: Commanda's design
« Reply #6 on: December 18, 2010, 05:17:49 PM »
Do you have ... ahem ... a PNG or GIF of the schematic?  %-O

McCad won't open any of those, at least not in cheapskate mode...

Rgds

Damon
Podcast: https://www.earth.org.uk/SECTION_podcast.html

@DamonHD@mastodon.social

commanda

  • Hero Member
  • *****
  • Posts: 731
Re: Commanda's design
« Reply #7 on: December 18, 2010, 06:16:29 PM »
You specifically need LTspice to run the simulations.
It works on Linux under wine, so should be the same on Mac.
Also, it's a free download.

But I grabbed some screenshots as png.

1685-0

Schematic is basically what I posted previously in the other thread, with a few values added to make it work.
And a simulated solar panel and battery load.

1686-1

This is the waveforms.
Top grey trace is the input voltage.  (17.656V)
Green trace is charge current into the battery. (6.1668A)
Red trace is the fet gate drive.  (varies from 29KHz to 39KHz)
Pink trace is the timing capacitor voltage.
Blue trace is the input current. (4.3465A)

By comparing the pink & red traces, you can see how the chip works.
Constant off time, variable on time. You can see the pwm varies between 1 off:2 on  to 1 off:3 on.
So 66% to 75%.

Power in = 17.656 * 4.3465 = 76.74 watts
Power out = 12 * 6.1668 =74.00 watts
= 96.4% efficient.
Whether you can actually achieve that in real life is a good question. Should be able to get close though.

Amanda

DamonHD

  • Administrator
  • Super Hero Member Plus
  • *****
  • Posts: 4125
  • Country: gb
    • Earth Notes
Re: Commanda's design
« Reply #8 on: December 19, 2010, 03:52:43 AM »
Thanks for that.  Now may have to break out my Horowitz and Hill to remind myself how that bipolar driver stage works with the emitters together rather than the collectors: presumably low gain but very low impedance output for driving the FET gate capacitance?

Rgds

Damon
Podcast: https://www.earth.org.uk/SECTION_podcast.html

@DamonHD@mastodon.social

commanda

  • Hero Member
  • *****
  • Posts: 731
Re: Commanda's design
« Reply #9 on: December 19, 2010, 03:03:08 PM »
Damon.
The bipolar driver stage is fairly standard, and found in all good texts on smps design.
Your term "low gain" is misleading. These have high current gain.

Amanda

Opera House

  • Sr. Member
  • ****
  • Posts: 261
Re: Commanda's design
« Reply #10 on: December 20, 2010, 02:48:17 PM »
I'd make R5 a lot smaller.  In most cases a high current or multiple FETs would have a lot of capacitance and I can't see a 4700 ohm resistor giving that much drive even with the booster transistors.

Madscientist267

  • Hero Member
  • *****
  • Posts: 1181
  • Country: us
  • Uh oh. Now what have I done?
Re: Commanda's design
« Reply #11 on: December 20, 2010, 06:37:10 PM »
VERY nice... I'm definitely going to try it... Be interesting to see what you get out of real world.

Still intend on building it live right?

Steve
The size of the project matters not.
How much magic smoke it contains does !

commanda

  • Hero Member
  • *****
  • Posts: 731
Re: Commanda's design
« Reply #12 on: December 20, 2010, 08:17:41 PM »

Still intend on building it live right?

Steve

Just had my very rough veroboard prototype running.
Bench supply through 2 ohm resistor to the input. 12 volt 40AHr sla (already fully charged, unfortunately) as a load.
Pumping 2 amps into the battery. Appears to work.

Component values used in prototype:
C1  4700/35 low esr
C2  100pF
C3  1000/25 low esr
R1  2K2
R5  100R
220/16 Q1 collector to gnd
M1  IRF540
L1 33uH  18 turns 1.2mm wire on a 28 x 14 x 11 toroid HY2 material
R2 & R3  39K + 5K pot + 2K2 to gnd
V2 7812 regulator
Q1 & Q2  BC547 BC557

Still got to knock up a heatsink, then apply it in anger. Probably not feasible till xmas eve or xmas day due to earning-a-living commitments.

Amanda

commanda

  • Hero Member
  • *****
  • Posts: 731
Re: Commanda's design
« Reply #13 on: December 20, 2010, 11:02:18 PM »
Just one Gotcha!

Needs a diode in series with the output. 

I used an ideal diode (LTC4357) which I had prepared earlier.

Madscientist267

  • Hero Member
  • *****
  • Posts: 1181
  • Country: us
  • Uh oh. Now what have I done?
Re: Commanda's design
« Reply #14 on: December 20, 2010, 11:38:33 PM »
Quote
then apply it in anger

???

Runs hotter than you expected?

Might be switching transition issues... sims are nice but they don't always account for everything. :( Depends on who wrote the models.

Still sounds decent... what's the actual current at the input? Efficiency, etc...

Steve
The size of the project matters not.
How much magic smoke it contains does !

commanda

  • Hero Member
  • *****
  • Posts: 731
Re: Commanda's design
« Reply #15 on: December 21, 2010, 12:50:27 AM »
Actually it runs quite cool. Most heat was generated in R5, which I've now changed for 1K.
Don't expect the switching to be brilliant, considering the crappy layout thrown together on veroboard without a whole lot of thought.
Discharged the battery load a bit, and had it pumping 5 amps comfortably.

Checked with an IR thermometer. Nothing rose more than a few degrees.

Haven't specifically checked efficiency, but considering it isn't making much heat, seems pretty good.

I've started doing a proper pcb layout. I'll include the full mppt control on that, so I can do more development work on that side of things.

Amanda

Madscientist267

  • Hero Member
  • *****
  • Posts: 1181
  • Country: us
  • Uh oh. Now what have I done?
Re: Commanda's design
« Reply #16 on: December 21, 2010, 10:05:08 AM »
Sounds like it's going to work well... so where did the 'apply it in anger' part come in? Still confused about that...

Steve
The size of the project matters not.
How much magic smoke it contains does !

commanda

  • Hero Member
  • *****
  • Posts: 731
Re: Commanda's design
« Reply #17 on: December 21, 2010, 04:04:06 PM »
Sounds like it's going to work well... so where did the 'apply it in anger' part come in? Still confused about that...

Steve


Hook it up to a solar panel and use it for real, vs running it on the test bench.
Just a colourful turn of phrase.

Amanda

Madscientist267

  • Hero Member
  • *****
  • Posts: 1181
  • Country: us
  • Uh oh. Now what have I done?
Re: Commanda's design
« Reply #18 on: December 21, 2010, 04:11:39 PM »
Ahh gotcha... I'm close to finishing take IV of the forwarder... We'll have to compare notes. :)

Steve
The size of the project matters not.
How much magic smoke it contains does !

commanda

  • Hero Member
  • *****
  • Posts: 731
Re: Commanda's design
« Reply #19 on: December 23, 2010, 09:16:15 PM »
So, it's xmas eve and we've all knocked off work early.

It's generally overcast, but in a couple of short periods of clear sunshine, I can confirm my prototype works as expected.
Tweaked the pot to get max amps, and can confirm there is a peak, with the current decreasing either side of that.
With the panel volts at 15 1/2, the battery at 13.4, amps into the battery got to a little over 7 amps during a brief period of sunshine.
With 4 square inches of aluminium heatsink, heat rise is minimal. So efficiency is good. Need a good run of pure sunshine to be more definitive.

For the next iteration, I've got some really good low rds fets, and really low forward voltage schottky's.  Be even better.

Next, I'm going to throw another panel on the roof, which I believe may give me a higher Vp.

Also, now to connect up the true mppt tracking control circuit.  Stay tuned.

Amanda

Madscientist267

  • Hero Member
  • *****
  • Posts: 1181
  • Country: us
  • Uh oh. Now what have I done?
Re: Commanda's design
« Reply #20 on: December 23, 2010, 11:37:30 PM »
Cool.. thanks for taking this on... I realize I kinda 'threw' you into it... I apologize for that... but between us, we'll come up with a good move... :)

Steve
The size of the project matters not.
How much magic smoke it contains does !

commanda

  • Hero Member
  • *****
  • Posts: 731
Re: Commanda's design
« Reply #21 on: December 24, 2010, 12:37:58 AM »
Cool.. thanks for taking this on... I realize I kinda 'threw' you into it... I apologize for that... but between us, we'll come up with a good move... :)

Steve


You didn't 'throw' me into anything. I've been messing with this on & off for a few years. Lately, I've been concentrating on getting a working DC-DC converter, which I believe I now have. Steep learning curve these things.  You may have spurred me on to greater heights, however.

The second half of the equation, the tracking mppt, I've had the basic design for a number of years. Have had it running on a wind turbine as proof-of-concept.
(Well, the generator head in a drill press, at least).
As an aside, I've also figured out how to turn the rpm into a cubic function control voltage.

Have just been roughing out a concept diagram to explain it, which I'll post in the very near future.
Might even post a photo of the prototype. Just to prove you can make these things work on veroboard.

Amanda

Madscientist267

  • Hero Member
  • *****
  • Posts: 1181
  • Country: us
  • Uh oh. Now what have I done?
Re: Commanda's design
« Reply #22 on: December 29, 2010, 03:21:55 PM »
I think I'm going to re-steer to this design for a little while...

Any more data from strong sunlight with actual panels yet (ie I/V in and I/V out)?

I'm bailing on the other design until I get some kind of eureka moment... I'm tired of replacing MOSFETs in the damn thing.

I did figure out a way to eliminate the 78XX regulator from both designs [EDIT - the other design; I just realized that yours is floating ground output :(] - power the chip from the battery side! DUH  ::) Dunno why I didn't see it earlier.

That being said, the only thing I'd need to modify on your design is the values of the feedback resistors to scale it up to the ~37V input, right?

Steve
« Last Edit: December 29, 2010, 03:36:14 PM by Madscientist267 »
The size of the project matters not.
How much magic smoke it contains does !

commanda

  • Hero Member
  • *****
  • Posts: 731
Re: Commanda's design
« Reply #23 on: December 29, 2010, 04:53:07 PM »
Steve,

I don't believe the forward converter design will ever achieve enough efficiency to make it worthwhile pursuing.

I've had mine putting 7 amps at 14 volts into the battery, and running cool doing it.
Don't have enough meters on hand to simultaneously measure power in & power out.

I'm powering the circuit from the input side for a good reason. Once the sun goes down, it all stops. So no phantom load on the battery.
Be warned, the MC34063 will not go to zero percent pwm.

I will also be going to 5 volts to power the thing, because of the voltage requirements of the current sensor that feeds the mppt.
Once I get it sorted, and post the circuit, you'll understand better what I'm on about.
I brought the oscilloscope home from work yesterday, and the weather looks like a few days of sunshine.  Hopefully, some progress can be made.

And yes, provided you size the inductor appropriately, changing the voltage divider resistors should be all that's required for your input voltage.

Amanda

Madscientist267

  • Hero Member
  • *****
  • Posts: 1181
  • Country: us
  • Uh oh. Now what have I done?
Re: Commanda's design
« Reply #24 on: December 30, 2010, 09:41:47 AM »
Went out last night and grabbed a myriad of parts to give this thing a shot...

Primarily needed the PNPs and some mag wire. Radio shack is getting further and further away from being helpful to us tinkerers...  :'(

The only thing they have left in terms of mag wire is a 3 pack; #20 was the biggest they had in anything. Ended up buying 2 packs just so I'm stocked up for future projects.

That being said, basing what you have listed in your parts, I am thinking ~1 turn / volt, and bunching them, say, 3 strands twisted together and ~36 turns or so for the toroid? Sound about right?

I think the core I'm using is roughly the same size as yours, maybe just a touch smaller. Is there a trick to doubling up the donuts so that they are stacked? I've seen it done every once and a while but I'm not sure of the logistics and efficiency issues of doing it that way... Compensate with more turns? Less?

Steve
The size of the project matters not.
How much magic smoke it contains does !

Madscientist267

  • Hero Member
  • *****
  • Posts: 1181
  • Country: us
  • Uh oh. Now what have I done?
Re: Commanda's design
« Reply #25 on: December 30, 2010, 11:12:30 AM »
Heh - may have done one better, although the # of turns may be a bit excessive, i'll still give it a shot.

I had problems deciding what to use as a 'universal' size comparison, and after ditching a quarter, I came up with the IEC connector that (i believe) is in use on computer power supplies worldwide...



Can't get a 100% accurate count on the turns because there is some layering, but a rough count gives ~115 turns of #18(?) wire...

The core is blue, and it is 32mm OD, 18mm ID, 10mm Thick.

Here's a closeup of the top:



Steve

The size of the project matters not.
How much magic smoke it contains does !

commanda

  • Hero Member
  • *****
  • Posts: 731
Re: Commanda's design
« Reply #26 on: December 30, 2010, 02:36:37 PM »
Steve,

Here in Australia, the electronic tinkerer's supply house used to be Dick Smith electronics.  It was bought out by Woolworths (the supermarket people),  Tandy/Radio Shack was merged in, and now they sell stereos & TV's.

Fortunately, we still have Jaycar, who do a roaring trade with no opposition.
I mostly get my stuff now by mail order, from Jaycar's wholesale division, Soanar.
Failing that, it's Farnell or Radio Spares.  All of these will courier deliver next day.

You really need to build that inductance meter add-on for your multimeter. Or purchase a digital LCR from ebay.
http://cgi.ebay.com.au/ws/eBayISAPI.dll?ViewItem&item=220710626879

To work out what size inductor you need;
http://schmidt-walter.eit.h-da.de/smps_e/smps_e.html#Abw

That inductor in the picture, my best guesstimate would be 1.5 mH.

And remember, inductance increases as the square of the turns. If you double the turns, you get 4 times the inductance.

If I remember correctly, you were looking at 4 amps out. One of those toroids should do it. I wouldn't bother trying to stack toroids, unless the core starts suffering excessive heat rise.

Amanda

Madscientist267

  • Hero Member
  • *****
  • Posts: 1181
  • Country: us
  • Uh oh. Now what have I done?
Re: Commanda's design
« Reply #27 on: December 30, 2010, 03:30:17 PM »
Broke down and did it...

http://www.amazon.com/Sinometer-MS8222H-Digital-Multimeter-Meter/dp/B000MFIDC0

I hate inductors...  ;D

I wonder what would have happened if the 1940s had gone a little different... Hitler could have just sent his messages in wound inductors and NOBODY would have ever figured it out hahahaha

So until this thing arrives, whats the worst case scenario likely to be with the pictured inductor?

More zapped MOSFETs?  :-\

Steve
The size of the project matters not.
How much magic smoke it contains does !

Madscientist267

  • Hero Member
  • *****
  • Posts: 1181
  • Country: us
  • Uh oh. Now what have I done?
Re: Commanda's design
« Reply #28 on: December 30, 2010, 06:28:32 PM »
Ok.. dug through the java proggy... pretty cool... but leaves me with some questions that I can't seem to find the answer to.

How much of a role does the inductor's current delta play in efficiency?

The larger the inductor, the flatter the delta, and from what I can gather, this is good up to a point, where the DC resistance of the inductor starts to become detrimental. Flatter delta results in less stress of the output components (ie output filter cap etc), and reduces the ripple current at the output. So the bigger inductor won't be a problem until I really start to try and pull serious juice from it. Am I close?

It also seems better to try and keep it in continuous mode because of something about the current spikes in discontinuous mode? I kinda got a little lost on that, but it makes sense looking at the graphs in the java applet. Something tells me though that staying in continuous mode won't be much of a problem with this application, only when the battery gets disconnected by the following uber-modified version of a Ghurd controller, and at that point, no current is flowing at the output anyway, so who cares, right?

In the 'proposal' department, the program tries to keep a semi healthy (~25%?) delta on the inductor. What's the reason for this? Balance between DC resistance losses in the inductor (tending toward a 'shorter' coil?) and ripple current at the output?

With a higher frequency, it states that a smaller inductor will cause greater switching losses, reducing efficiency. But to me, this is counterintuitive to hysteresis/eddy etc losses in the core of a larger inductor.

There seems to be some conflicting information, and I stopped when I saw a line in wikipedia referring to something like 'the design is always a tradeoff based on the intended load'... I get it, I think, but damn...

Did I mention I hate inductors? LOL

Steve
« Last Edit: December 30, 2010, 06:49:50 PM by Madscientist267 »
The size of the project matters not.
How much magic smoke it contains does !

commanda

  • Hero Member
  • *****
  • Posts: 731
Re: Commanda's design
« Reply #29 on: December 30, 2010, 08:45:12 PM »
Quote
the design is always a tradeoff based on the intended '...

This is true of all design, not just electronics design, and especially not just switchmode converters.

For the homemade converter, 20 - 50 Khz seems to be an acceptable tradeoff.

A higher freq (edit: not sure what happened there, It just went & posted in mid sentence)

A higher frequency results in higher losses in the switching transistor.

Some back of the envelope calculations.
Buck Converter
Vin_min = 28.0V    Vin_max = 40.0V    Vin = 40.0V
Vout = 13.0V    Iout = 5.0A    f = 30.0kHz
L = 151.4uH   ΔILfor Vin_max = 2.0A   

Assuming your toroids are similar to mine (your early photo of the green & blue toroid is a dead ringer)
AL  = 100 nH/N2

so 150uH  = 39 turns

20 awg; suggest using at least 2 in hand.

Amanda
« Last Edit: December 30, 2010, 08:55:35 PM by commanda »

commanda

  • Hero Member
  • *****
  • Posts: 731
Re: Commanda's design
« Reply #30 on: December 30, 2010, 09:58:21 PM »


This is the conceptual block diagram of the solar mppt implemented without an mcu.

We start with a dc-dc converter, with voltage feedback taken from the input.
This will hold the panels at a fixed voltage.
There is a schematic of this implemented with an MC34063 earlier in this thread.

Functionally, the mppt uses the Perturb & Observe (P & O) hill climbing algorithm.

Starting with the square wave oscillator, through R4, to the voltage feedback node.
This is the perturb signal. This causes the pwm to increase & decrease slightly at a half second rate.

As the pwm goes up & down slightly, we measure the current going to the battery.
This is differentiated, to give us a signal which tells us if the current is increasing or decreasing.

We then compare this signal to our square wave signal, using an Xor gate.
There are 2 possible states.
1: Increasing and decreasing the pwm causes an increase & decrease in the output current (in phase)
2: Increasing and decreasing the pwm causes a decrease & increase in the output current (out of phase)

The output of the Xor gate is integrated, and used as the average control signal to the pwm.

So, when it's in phase, it drives the average control signal higher.
When it's out of phase, it drives the average control signal lower.

The outcome is that it hunts back & forth across the maximum power point.

In a practical implementation, to reduce component count, the differentiator & integrator are a dual op-amp (LM358)
and the square wave oscillator & Xor gate is a CD4046 phase locked loop.

Without this block diagram & explanation, the real circuit diagram wouldn't make a whole lot of sense.

Needless to say, the real circuit diagram will follow in due course.

Amanda

Madscientist267

  • Hero Member
  • *****
  • Posts: 1181
  • Country: us
  • Uh oh. Now what have I done?
Re: Commanda's design
« Reply #31 on: January 01, 2011, 10:21:01 PM »
Dunno what I'm doing wrong, but I can't seem to get even just the buck working at all this time.

Made everything pretty close (as possible) to the schematic for the buck itself, ran across a few snags early on, but got past those. A shorted zener kept the 7805 from getting any input, and I bumped something after fixing that which made the prereg tranny explode. Shrug.

Lowered the input voltage temporarily so as to not directly blow the 7805, and got back to getting 5V out of it.

Only subbed parts that made sense (or otherwise weren't specified, ie for the prereg I used a 2N2222).

I'm getting power to everywhere I'm supposed to (I think) but there's excessive heat coming from the 7805 and the 34063 (not enough to worry about particularly, but it just seems a bit much, considering yours runs cool... ?

I don't have any of the parts to do the mppt sampling, so I just built the buck itself to check for basic functionality.

Supply voltage to the whole thing right now is ~30V via a 10R 10W wirewound ceramic to limit current.

I can get the MOSFET to turn on if I ground the bases of the totem, and it turns back off if I disconnect them. So that seems ok.

An odd consideration I noticed was that with the MOSFET held on, there's no considerable drop on the input as I would have expected (due to the effective short due to the toroid being held across the input) but there is a drop if a load is placed on the output terminals.

Is the MOSFET drain connected in the wrong place?

Steve
The size of the project matters not.
How much magic smoke it contains does !

commanda

  • Hero Member
  • *****
  • Posts: 731
Re: Commanda's design
« Reply #32 on: January 01, 2011, 11:06:49 PM »
Dunno what I'm doing wrong, but I can't seem to get even just the buck working at all this time.

Made everything pretty close (as possible) to the schematic for the buck itself, ran across a few snags early on, but got past those. A shorted zener kept the 7805 from getting any input, and I bumped something after fixing that which made the prereg tranny explode. Shrug.

Lowered the input voltage temporarily so as to not directly blow the 7805, and got back to getting 5V out of it.

Only subbed parts that made sense (or otherwise weren't specified, ie for the prereg I used a 2N2222).

I'm getting power to everywhere I'm supposed to (I think) but there's excessive heat coming from the 7805 and the 34063 (not enough to worry about particularly, but it just seems a bit much, considering yours runs cool... ?

I don't have any of the parts to do the mppt sampling, so I just built the buck itself to check for basic functionality.

Supply voltage to the whole thing right now is ~30V via a 10R 10W wirewound ceramic to limit current.

I can get the MOSFET to turn on if I ground the bases of the totem, and it turns back off if I disconnect them. So that seems ok.

An odd consideration I noticed was that with the MOSFET held on, there's no considerable drop on the input as I would have expected (due to the effective short due to the toroid being held across the input) but there is a drop if a load is placed on the output terminals.

Is the MOSFET drain connected in the wrong place?

Steve


I doubt a 2N2222 is going to last long. A TO220 package, like a TIP31, was what I had in mind.

The mosfet source goes to input negative. Drain goes to output negative.

Output goes to your battery, which is your load.  When the mosfet turns on, current flows from input positive, through the inductor, through the load, through the mosfet, and back to input negative.

When the mosfet switches off, the collapsing magnetic field in the inductor causes current to flow from the inductor, through the load, back through D1 to the inductor.

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

If you ground the bases of the totem pole driver, the mosfet should switch OFF.
You do have an N-channel fet don't you?

Pin 5 should be at 1.25 volts.

And the IC should not be getting hot.

Amanda