Solar power controller
- Thread starter boriz
- Start date
BeanieBots
Moderator
Boriz, I'd be very interested in how you get on with a SEPIC design.
Of all the SM topologies I've tried, SEPIC has always been the easiest to build but has always yealded the WORST results. (efficiency wise).
I put this down to its reliance on extremely low ESR for the caps which is much harder (IMHO) to achieve than it is to get high efficiency magnetics such as a low R transformer for a flyback converter.
Of all the SM topologies I've tried, SEPIC has always been the easiest to build but has always yealded the WORST results. (efficiency wise).
I put this down to its reliance on extremely low ESR for the caps which is much harder (IMHO) to achieve than it is to get high efficiency magnetics such as a low R transformer for a flyback converter.
Boriz, why do you want to do SEPIC? You seem to have gone from simple to tricky in no time at all.
I'm no expert and I suspect you aren't either but after reading the article by Maxim (who probably know a thing or two) it looks complicated with just a couple of benefits.
I just wonder why you want to do this for solar?
I'm no expert and I suspect you aren't either but after reading the article by Maxim (who probably know a thing or two) it looks complicated with just a couple of benefits.
I just wonder why you want to do this for solar?
As I have said, once I get the basic system installed and running, I will begin working on improving efficiency. I’ll need a baseline for comparison anyway.
By basic system, I mean…
- Panel flat on sloped south(ish) facing roof.
- Straight connection of panel to battery.
- Simple on/off overcharge and over discharge control.
- Flooded lead acid car battery.
After installing this, and some basic lighting for the load, and after I get a better idea of the power available, I will begin working on efficiency improvements like tracking and/or PWM or even MPPT charge control etc.
The SEPIC circuit is new to me, and fascinating. I’m just gonna play around with a little test circuit to get a feel for the technology and PICAXE control. Might even base some small LED drivers on it. Maybe one day I’ll build a full size practical MPPT charger, but certainly not for a while. Maybe I’ll find a more interesting alternative. Maybe I’ll get sidetracked by something else fascinating…
By basic system, I mean…
- Panel flat on sloped south(ish) facing roof.
- Straight connection of panel to battery.
- Simple on/off overcharge and over discharge control.
- Flooded lead acid car battery.
After installing this, and some basic lighting for the load, and after I get a better idea of the power available, I will begin working on efficiency improvements like tracking and/or PWM or even MPPT charge control etc.
The SEPIC circuit is new to me, and fascinating. I’m just gonna play around with a little test circuit to get a feel for the technology and PICAXE control. Might even base some small LED drivers on it. Maybe one day I’ll build a full size practical MPPT charger, but certainly not for a while. Maybe I’ll find a more interesting alternative. Maybe I’ll get sidetracked by something else fascinating…
Good on you Boriz. You've confused me (and maybe BeanieBots too?) 
#41: "Building a SEPIC charge pump test rig now "
#44: "As I have said, once I get the basic system installed and running," & "...Straight connection of panel to battery"
Mmmm... well SEPIC is a switched design, basically a Boost design with a series capacitor and an extra inductor. And most definitely NOT a "straight connection of panel to battery".
I can see the advantages in certain apps. But I can't see the point for solar where your PV Panel (assuming you have chosen properly) will always have a MPP voltage well above the battery voltage and where a standard Buck design can give excellent efficiency and intrinsically (with good design like mine) follow the MPP.
At the end of the day you need to suck out as much as possible from the PV - and forget fancy phrases which may impress the General Public.
[I get this feeling in me water that "MPPT" is a phrase glibly chucked around more and more... I think SEPIC may go the same way]
Anyway, I'm sure you know what you're doing, so I wish you luck.
#41: "Building a SEPIC charge pump test rig now "
#44: "As I have said, once I get the basic system installed and running," & "...Straight connection of panel to battery"
Mmmm... well SEPIC is a switched design, basically a Boost design with a series capacitor and an extra inductor. And most definitely NOT a "straight connection of panel to battery".
I can see the advantages in certain apps. But I can't see the point for solar where your PV Panel (assuming you have chosen properly) will always have a MPP voltage well above the battery voltage and where a standard Buck design can give excellent efficiency and intrinsically (with good design like mine) follow the MPP.
At the end of the day you need to suck out as much as possible from the PV - and forget fancy phrases which may impress the General Public.
[I get this feeling in me water that "MPPT" is a phrase glibly chucked around more and more... I think SEPIC may go the same way]
Anyway, I'm sure you know what you're doing, so I wish you luck.
BeanieBots
Moderator
Yep, I'm confused
I took "Building a SEPIC charge pump test rig now " to mean Boriz was going to use a SEPIC controller for MPPT. Obviously I was wrong and it meant he was going to build a rig to TEST the SEPIC converter he had ALREADY built
I think you're right Dippy re:-
[I get this feeling in me water that "MPPT" is a phrase glibly chucked around more and more... I think SEPIC may go the same way]
I also think that some people think MPPT is just switching the panel on/off in order to maintain a particular battery voltage.
I took "Building a SEPIC charge pump test rig now " to mean Boriz was going to use a SEPIC controller for MPPT. Obviously I was wrong and it meant he was going to build a rig to TEST the SEPIC converter he had ALREADY built
I think you're right Dippy re:-
[I get this feeling in me water that "MPPT" is a phrase glibly chucked around more and more... I think SEPIC may go the same way]
I also think that some people think MPPT is just switching the panel on/off in order to maintain a particular battery voltage.
Yes, but:-
"The intelligent charging controller uses SEPIC and microprocessor technology in order to provide MPPT functionality - therefore maximising the power transfer efficiency"
- sounds so much better ; even if it's just an Ebay relay ticking on and off
The neighbours will be SO impressed! Their NAND gates will be twitching with excitement.
"The intelligent charging controller uses SEPIC and microprocessor technology in order to provide MPPT functionality - therefore maximising the power transfer efficiency"
- sounds so much better ; even if it's just an Ebay relay ticking on and off
The neighbours will be SO impressed! Their NAND gates will be twitching with excitement.
BeanieBots
Moderator
Sounds so much better to who? Another marketing department?
Single-Ended Primary Inductance Converter (SEPIC) and Maximum Power Point Tracking (MPPT) are new to me. But I’m gonna learn.
@BB
“I put this down to its reliance on extremely low ESR for the caps which is much harder” –Paralleling normal electrolytics is a good way to get high capacitance and low ESR. EG: Three 100uF electrolytics in parallel have 3* the capacity and 1/3 of the ESR.
@Dippy
Sorry for my bad English. It’s my native (and only) language, so I failed my exams with flying colours. Actually I’m a bit stupid, but I mean well. I have not received delivery of my solar panel yet and have no idea about it’s actual power output (in this location) or how to best manage it.
By ‘test rig’ I mean a breadboard or stripboard SEPIC circuit, with a swap-out facility for the main components. I’ll be winding the inductors myself. I want to get a feel for how it works and how different control schemes and component values effect performance. EG: How does frequency and duty-cycle relate to input load and output power. Maths is not my strong point, so I have to experiment and see some curves to understand. Prolly gonna use two series PP3s to supply it with 18v, which according to the 75W panel specs is it’s MPP voltage. But I don’t even know yet if it can reach that voltage on cloudy days. Could I use an LDR to move the controllers peak-efficiency-point around according to conditions. Could I get useful power even when the panel voltage has fallen to say 6v using a SEPIC type circuit? Maybe two separate SEPIC circuits optimised for different input ranges with auto switching?
SEPIC is just one of several types of DC-DC power converter, but have to start someplace. It seems to have good efficiency, low complexity, low component count and the PDFs. show a complete solar charge controller with good up-scaling info.
I have already created a spice simulation of the 10W circuit described in the PDFs. Works well ‘straight out of the box’, I’ll post some images presently.
To clarify. I WILL NOT be installing a MPPT controller for quit a while. Probably many months. But once my basic solar power system is in operation, and I begin the process of improving efficiency, an MPPT controller will be considered as an essential add-on. They are too expensive to buy, so I’ll have to build one. Here and now, I’m just trying to learn how. Thanks for all your help.
@BB
“I put this down to its reliance on extremely low ESR for the caps which is much harder” –Paralleling normal electrolytics is a good way to get high capacitance and low ESR. EG: Three 100uF electrolytics in parallel have 3* the capacity and 1/3 of the ESR.
@Dippy
Sorry for my bad English. It’s my native (and only) language, so I failed my exams with flying colours. Actually I’m a bit stupid, but I mean well. I have not received delivery of my solar panel yet and have no idea about it’s actual power output (in this location) or how to best manage it.
By ‘test rig’ I mean a breadboard or stripboard SEPIC circuit, with a swap-out facility for the main components. I’ll be winding the inductors myself. I want to get a feel for how it works and how different control schemes and component values effect performance. EG: How does frequency and duty-cycle relate to input load and output power. Maths is not my strong point, so I have to experiment and see some curves to understand. Prolly gonna use two series PP3s to supply it with 18v, which according to the 75W panel specs is it’s MPP voltage. But I don’t even know yet if it can reach that voltage on cloudy days. Could I use an LDR to move the controllers peak-efficiency-point around according to conditions. Could I get useful power even when the panel voltage has fallen to say 6v using a SEPIC type circuit? Maybe two separate SEPIC circuits optimised for different input ranges with auto switching?
SEPIC is just one of several types of DC-DC power converter, but have to start someplace. It seems to have good efficiency, low complexity, low component count and the PDFs. show a complete solar charge controller with good up-scaling info.
I have already created a spice simulation of the 10W circuit described in the PDFs. Works well ‘straight out of the box’, I’ll post some images presently.
To clarify. I WILL NOT be installing a MPPT controller for quit a while. Probably many months. But once my basic solar power system is in operation, and I begin the process of improving efficiency, an MPPT controller will be considered as an essential add-on. They are too expensive to buy, so I’ll have to build one. Here and now, I’m just trying to learn how. Thanks for all your help.
Oh I'm more confused than ever. I couldn't get any real info from that site, sorry.
And no, if your (nom.12V) solar panel is chucking out 6V open circuit (or high impedance load) then it is giving out zilch. Nothing useful.
No you can't step it up and magically produce a decent charge.
The bit I can't underestand is: why do you want to use the SEPIC technique?
I may have considered it in a step-up app especially where the Vin may creep beyond your intended Vout. But as your (well illuminated) PV Panel will spend most of its time providing Vout>> Vbatt then I can't see the point.
Here's a link for more maths written by proper people:-
http://www.maxim-ic.com/appnotes.cfm/appnote_number/1051/
I'd appreciate an explanation of your thinking behind this and your intended benefits over a simple buck which can poduce >90% efficiency easily. I am seriously worried I am missing something significant
And no, if your (nom.12V) solar panel is chucking out 6V open circuit (or high impedance load) then it is giving out zilch. Nothing useful.
No you can't step it up and magically produce a decent charge.
The bit I can't underestand is: why do you want to use the SEPIC technique?
I may have considered it in a step-up app especially where the Vin may creep beyond your intended Vout. But as your (well illuminated) PV Panel will spend most of its time providing Vout>> Vbatt then I can't see the point.
Here's a link for more maths written by proper people:-
http://www.maxim-ic.com/appnotes.cfm/appnote_number/1051/
I'd appreciate an explanation of your thinking behind this and your intended benefits over a simple buck which can poduce >90% efficiency easily. I am seriously worried I am missing something significant
BeanieBots
Moderator
Boriz, thanks for the ESR tip.
Experimentation is a great way to learn. Especially with switchers. You will soon discover how useless spice is when it comes to modelling DC-DC converters unless you include models of your layout and EMI patterns.
Only building a few and doing thorough tests will make it sink in.
Don't forget, building it on breadboard WON'T give the same results as a properly layed out board. Winding your own magnetics is a great idea. If you are going to try to get them small by going up in frequency, don't forget to try bi and tri-filer windings. It can really help with keeping down R due to skin effect. Far less "PCB real estate" than multiple low ESR caps.
You WILL need to brush up on your maths if you ever want to get ultra high efficiency within the next few years. Snubber networks and compensation feedback can't be experimented without a LOT of time. It would be quicker to learn and understand the principles than it would be to grope around in the dark trying things out. You'll also need a very good 'scope.
The first thing to get rid of is the diode. This requires getting your timing right or you will get shoot-through which does not need to be more than 100nS or more to lose everything you gain. Not to mention popping the FETs if it's much bigger. Oh yes, get the snubber wrong and the FETs go in uS.
There's a lot of info available on the NET to help you in THEORY, but you simply can't beat practice to find out what it's like for real.
I wish you all the best.
"Could I use an LDR to move the controllers peak-efficiency-point around according to conditions"
That's NOT the way to do it.
Monitor the power in/out and adjust accordingly. That's the whole point of MPPT. (I'm not sure you've got this yet!)
"a swap-out facility for the main components"
After you've built one, let us know how that idea went.
(don't forget to model the "extra cap/inductor lengths in your spice model).
"How does frequency and duty-cycle relate to input load and output power"
It doesn't. But then you knew that from your spice model didn't you
Please let us know how it all goes. Switchers can easily become a lifetimes work.
EDIT:-
If you find Dippy's link a little too much. Here is the BASIC version.
http://www.powerdesigners.com/InfoWeb/design_center/articles/DC-DC/converter.shtm
Like Dippy, I don't understand your desire for SEPIC. However, if you want to learn about switchers, a FULL understanding of the SEPIC topology will go a long way.
(you'll also understand my comment about low ESR caps vs low R inductors)
Experimentation is a great way to learn. Especially with switchers. You will soon discover how useless spice is when it comes to modelling DC-DC converters unless you include models of your layout and EMI patterns.
Only building a few and doing thorough tests will make it sink in.
Don't forget, building it on breadboard WON'T give the same results as a properly layed out board. Winding your own magnetics is a great idea. If you are going to try to get them small by going up in frequency, don't forget to try bi and tri-filer windings. It can really help with keeping down R due to skin effect. Far less "PCB real estate" than multiple low ESR caps.
You WILL need to brush up on your maths if you ever want to get ultra high efficiency within the next few years. Snubber networks and compensation feedback can't be experimented without a LOT of time. It would be quicker to learn and understand the principles than it would be to grope around in the dark trying things out. You'll also need a very good 'scope.
The first thing to get rid of is the diode. This requires getting your timing right or you will get shoot-through which does not need to be more than 100nS or more to lose everything you gain. Not to mention popping the FETs if it's much bigger. Oh yes, get the snubber wrong and the FETs go in uS.
There's a lot of info available on the NET to help you in THEORY, but you simply can't beat practice to find out what it's like for real.
I wish you all the best.
"Could I use an LDR to move the controllers peak-efficiency-point around according to conditions"
That's NOT the way to do it.
Monitor the power in/out and adjust accordingly. That's the whole point of MPPT. (I'm not sure you've got this yet!)
"a swap-out facility for the main components"
After you've built one, let us know how that idea went.
(don't forget to model the "extra cap/inductor lengths in your spice model).
"How does frequency and duty-cycle relate to input load and output power"
It doesn't. But then you knew that from your spice model didn't you
Please let us know how it all goes. Switchers can easily become a lifetimes work.
EDIT:-
If you find Dippy's link a little too much. Here is the BASIC version.
http://www.powerdesigners.com/InfoWeb/design_center/articles/DC-DC/converter.shtm
Like Dippy, I don't understand your desire for SEPIC. However, if you want to learn about switchers, a FULL understanding of the SEPIC topology will go a long way.
(you'll also understand my comment about low ESR caps vs low R inductors)
Last edited:
Thanks.
@Dippy
“…why do you want to use the SEPIC technique?...”
I DON’T SPECIFCALLY.
I just want to study high efficiency solar power conversion and SEPIC is the first one that came along (with schematics descriptions and scaling mod suggestions as seen in the PDFs). If you can direct me to a better one to start on, please do.
@Dippy
“…why do you want to use the SEPIC technique?...”
I DON’T SPECIFCALLY.
I just want to study high efficiency solar power conversion and SEPIC is the first one that came along (with schematics descriptions and scaling mod suggestions as seen in the PDFs). If you can direct me to a better one to start on, please do.
BeanieBots
Moderator
Buck converter would be the obvious choice. As already mentioned by Dippy.
Much simpler, easier to build, easier to get right, easier to understand, more forgiving and above all, more appropriate to the job in hand.
Much simpler, easier to build, easier to get right, easier to understand, more forgiving and above all, more appropriate to the job in hand.
BeanieBots
Moderator
I've still got some non-polarized electrolytic caps for building SEPICs with.
They're as rare as rocking horse dung. Another reason not to try that topology.
Only good for about 1/2A but won't explode like a polarised one would when reversed. 10uF/50v.
They're as rare as rocking horse dung. Another reason not to try that topology.
Only good for about 1/2A but won't explode like a polarised one would when reversed. 10uF/50v.
I can see the initial attraction for SEPIC - it boosts or bucks and gives an output regardless of whether the input is higher or lower than the output.
In practice though, you don't really need a boost. A 36 cell panel will give about 18V open circuit and if you are charging a 13.8V battery, you can convert some volts into extra watts. As the light level falls, due to cloud etc, you get less volts - but if you look at the curve of volts vs amps, the volts stay pretty constant until there is not much light. Then the volts fall off very rapidly. So in practice, if you had a solar panel that was producing 10V open circuit, and you wanted to boost that to 13.8V, you would find there are hardly any amps left at 10V.
So the boost bit of the SEPIC doesn't add much.
A buck is much simpler. But still very complex at these sorts of power levels.
BeanieBots and Dippy have some very sage advice above. I'll second the comments.
For switchers and MPPT, I'd buy a commercial unit.
For battery voltage monitoring and charging and preventing over discharge - sure, that would be a great picaxe project. Even getting a picaxe to display the % capacity of a battery would be a real challenge - and a very useful project. But it isn't as simple as measuring the volts.
In practice though, you don't really need a boost. A 36 cell panel will give about 18V open circuit and if you are charging a 13.8V battery, you can convert some volts into extra watts. As the light level falls, due to cloud etc, you get less volts - but if you look at the curve of volts vs amps, the volts stay pretty constant until there is not much light. Then the volts fall off very rapidly. So in practice, if you had a solar panel that was producing 10V open circuit, and you wanted to boost that to 13.8V, you would find there are hardly any amps left at 10V.
So the boost bit of the SEPIC doesn't add much.
A buck is much simpler. But still very complex at these sorts of power levels.
BeanieBots and Dippy have some very sage advice above. I'll second the comments.
For switchers and MPPT, I'd buy a commercial unit.
For battery voltage monitoring and charging and preventing over discharge - sure, that would be a great picaxe project. Even getting a picaxe to display the % capacity of a battery would be a real challenge - and a very useful project. But it isn't as simple as measuring the volts.
Dr_A: Wise advise, & the PV thoughts echo my findings. Charging benefits with PVs in series versus parallel relate to the battery voltage & lighting conditions. Hence (in overcast light) 2 series 12V 10W PVs may still push 100mA into a 12V SLA, & even though each is only delivering maybe 10V they'd have 20V total. However if in parallel, they'd not deliver anything, since their 10V output would not overcome the ~12V SLA voltage. In general (under bright sun where the PVs output V exceeds the battery V), parallel charging may be superior- if only because any shading of a series panel (unless it used bypass diodes) would seriously cut the supplied current. Panels in parallel allow one to go wonky, & still have the system deliver (pro rata).
Re voltages Versus charge: It's indeed tricky,as temperature & battery age are issues- the battery may be "all hat & no cattle" as the Texan's say - but lead acid terminal voltages CAN be a fair measure of charge level, & thus it's possible capacity. Although of course open to endless debate, Charge level versus Voltage tables abound- many garages have them up.
Re voltages Versus charge: It's indeed tricky,as temperature & battery age are issues- the battery may be "all hat & no cattle" as the Texan's say - but lead acid terminal voltages CAN be a fair measure of charge level, & thus it's possible capacity. Although of course open to endless debate, Charge level versus Voltage tables abound- many garages have them up.
Last edited:
Thank you! Something as seemingly simple as a solar panel sure is complicated.
boriz, you are going to have to design a controller that determines which way the panels need to be connected at any one time, and I'll expect 1 percent of your profits
Edit: make that 1/2 percent, and everybody else that helped gets to split the other 1/2 lol...
boriz, you are going to have to design a controller that determines which way the panels need to be connected at any one time, and I'll expect 1 percent of your profits
Edit: make that 1/2 percent, and everybody else that helped gets to split the other 1/2 lol...
Post 58 "Well...I'll ask something and show my ignorance. Would two panels give more power connected in series than the same two in parallel? Seems like you would get usable voltage even in low light."
212 not an ignorant question at all.
Indeed, in low light, put them in series. Then you still get a charge current. In normal light, put them in parallel. If you use a latching relay to switch between each state the relay uses no power. If you then have a picaxe controlling the switching, you can set up some hysteresis so the relay doesn't flip back and forward.
This is a MPPT of sorts. It still doesn't optimise charge at high noon but then again, that is probably when the extra power isn't needed so much anyway.
One can envisage extending the concept further with more panels.
And it may end up a lot simpler than inductor switching regs. Possibly even more efficient at certain light levels as there are no switching reg losses.
212 not an ignorant question at all.
Indeed, in low light, put them in series. Then you still get a charge current. In normal light, put them in parallel. If you use a latching relay to switch between each state the relay uses no power. If you then have a picaxe controlling the switching, you can set up some hysteresis so the relay doesn't flip back and forward.
This is a MPPT of sorts. It still doesn't optimise charge at high noon but then again, that is probably when the extra power isn't needed so much anyway.
One can envisage extending the concept further with more panels.
And it may end up a lot simpler than inductor switching regs. Possibly even more efficient at certain light levels as there are no switching reg losses.
Hmm. Series connected in low light, parallel other times. Sounds good, but I can only afford one panel and it's bought!.
Here’s a capture of the circuit I’m playing with in the simulator (Circuit maker 2000 SP1):
V1 is the solar panel, R1 is added series resistance to make it a bit more realistic, same at the other end, V3,R3. R2 is added so that I can ,measure drain current. The transistor is just one that I have a model for with high current handling and >60v Vdrain.
Without V3,R2, the output voltage on C3 climbs quite high. But I can’t find a frequency/duty combo that gives higher current through R2 than R1.
Here’s a capture of the circuit I’m playing with in the simulator (Circuit maker 2000 SP1):
V1 is the solar panel, R1 is added series resistance to make it a bit more realistic, same at the other end, V3,R3. R2 is added so that I can ,measure drain current. The transistor is just one that I have a model for with high current handling and >60v Vdrain.
Without V3,R2, the output voltage on C3 climbs quite high. But I can’t find a frequency/duty combo that gives higher current through R2 than R1.
premelec
Senior Member
Your configuration looks a bit unusual for a down convertor - which would likely be the most efficient thing to do where you are trying for maximum power transfer... that configuration would simply have a voltage switched _into_ a series inductor with a backward diode to V- from the inductors feed point and the inductor out to the battery - then you look for maximum current into the battery as you vary the switching pulse... [I hope I've got this right as it's in my head and not on paper... ].
].BeanieBots
Moderator
Boriz, I think Dr_Acula has given the best advice so far. At this power level, buy one.
"But I can’t find a frequency/duty combo that gives higher current through R2 than R1."
So what! Besides, what do you mean by "current" wrt R2?
Average or peak? Do the sums, what do think the peak might be?
What do you think will happen if your switching frequency is too low?
What would happen if L1 saturated? (clue, have a spare Q1 ready)
What is the peak and average current in C2?
Where are you going to get C2 from? (manufacturer & part number please)?
TIP:
If you are going to make that circuit into a MPPT, put R1 & R3 on the other side of the load/battery. It will make life a LOT easier unless you are happy making differential amplifiers. They will also need heatsinking. IMHO they are both MUCH too high in value. They account for about 15% loss on their own!
If on the other hand, R1 & R3 are only there for the purposes of simulation, then where are your cap/inductor/diode 'simulated' resistances? For simulation, you also need to include track resistance.
I would expect about 60% efficiency from what you have drawn when it runs at 75W (assuming C2 lasts long enough to take any measurements). Hence, you are better off without it. Even with PERFECT components, it will run at about 85%. (R1,R3 & D1 alone).
"But I can’t find a frequency/duty combo that gives higher current through R2 than R1."
Just occured to me that maybe you meant R1 & R3.
You probably won't. It's too inefficient. Try increasing V1 and decreasing V3 just to prove the point. Although I would have thought the simulator might have implied that it would.
Let's say 80% efficient. (about right for your simplified model)
75/18=4.17A. 75(*80%)/13=4.6A
At 60% (more realistic)
75/18=4.17A 75(*60%)/13=3.46A
Maybe the I^2 losses in R2 (Q1 Ron) are higher than I thought.
Anyway, I've had my say. I've pointed you to the maths.
You now need to TRY IT and draw your own conclusion.
"But I can’t find a frequency/duty combo that gives higher current through R2 than R1."
So what! Besides, what do you mean by "current" wrt R2?
Average or peak? Do the sums, what do think the peak might be?
What do you think will happen if your switching frequency is too low?
What would happen if L1 saturated? (clue, have a spare Q1 ready)
What is the peak and average current in C2?
Where are you going to get C2 from? (manufacturer & part number please)?
TIP:
If you are going to make that circuit into a MPPT, put R1 & R3 on the other side of the load/battery. It will make life a LOT easier unless you are happy making differential amplifiers. They will also need heatsinking. IMHO they are both MUCH too high in value. They account for about 15% loss on their own!
If on the other hand, R1 & R3 are only there for the purposes of simulation, then where are your cap/inductor/diode 'simulated' resistances? For simulation, you also need to include track resistance.
I would expect about 60% efficiency from what you have drawn when it runs at 75W (assuming C2 lasts long enough to take any measurements). Hence, you are better off without it. Even with PERFECT components, it will run at about 85%. (R1,R3 & D1 alone).
"But I can’t find a frequency/duty combo that gives higher current through R2 than R1."
Just occured to me that maybe you meant R1 & R3.
You probably won't. It's too inefficient. Try increasing V1 and decreasing V3 just to prove the point. Although I would have thought the simulator might have implied that it would.
Let's say 80% efficient. (about right for your simplified model)
75/18=4.17A. 75(*80%)/13=4.6A
At 60% (more realistic)
75/18=4.17A 75(*60%)/13=3.46A
Maybe the I^2 losses in R2 (Q1 Ron) are higher than I thought.
Anyway, I've had my say. I've pointed you to the maths.
You now need to TRY IT and draw your own conclusion.
This may be of interest.
Page 3 shows how to Spice Up your SEPIC.
http://www.powersystemsdesign.com/design_tips_nov06.pdf
and there are some good points made.
I hope you get something working. It will be educational even if you can't follow the maths.
(It's been educational for me so far and has certainly re-inforced my dull preconception that Buck is better for this app. Seriously though, the maths and function have been interesting).
If you are hoping to get significant boost from a drooping solar panel then you may see why BB & I have questioned this after you get hold of your PV. You'll get to a stage where it takes more current to do he switching than you get from the PV
Yonks ago I think(?) BB & I discussed a switched charge transfer method to squeeze the pips; can't exactly remember. If we didn't then forget I just said that.
I'm sure you'll be converted to simple Buck circuits soon
I'll leave you to it and the best of luck.
Page 3 shows how to Spice Up your SEPIC.
http://www.powersystemsdesign.com/design_tips_nov06.pdf
and there are some good points made.
I hope you get something working. It will be educational even if you can't follow the maths.
(It's been educational for me so far and has certainly re-inforced my dull preconception that Buck is better for this app. Seriously though, the maths and function have been interesting).
If you are hoping to get significant boost from a drooping solar panel then you may see why BB & I have questioned this after you get hold of your PV. You'll get to a stage where it takes more current to do he switching than you get from the PV
Yonks ago I think(?) BB & I discussed a switched charge transfer method to squeeze the pips; can't exactly remember. If we didn't then forget I just said that.
I'm sure you'll be converted to simple Buck circuits soon
I'll leave you to it and the best of luck.
BeanieBots
Moderator
Dippy, I also came across that series of articles while looking for information to point Boriz towards. Reading through, they actually look quite good. However, I TOTALLY dismissed them without reading because EVERY article he has written, has a picture of himself in the header
That level of vanity and engineering simple don't belong together.
A bit like the power meter in the other thread. Cutting edge esthetics, shame about the engineering
Boriz, despite the outrageously tacky presentation, Ray Ridley has actually written some good tips. Worth reading if you can get past the pictures!
Looks like a nice boy
That level of vanity and engineering simple don't belong together.
A bit like the power meter in the other thread. Cutting edge esthetics, shame about the engineering
Boriz, despite the outrageously tacky presentation, Ray Ridley has actually written some good tips. Worth reading if you can get past the pictures!
Looks like a nice boy
Hello Sailor!
So many articles and reports have these ponsy pictures at the front. Makes me gag too.
To me it just says "Look at me. I had a haircut and hired this suit to look clever."
I think dodgey professors of medicine started it off. Certainly all the articles I have on viagra have a picture of a bloke with a big permanent smile. I was up all night reading them.
So many articles and reports have these ponsy pictures at the front. Makes me gag too.
To me it just says "Look at me. I had a haircut and hired this suit to look clever."
I think dodgey professors of medicine started it off. Certainly all the articles I have on viagra have a picture of a bloke with a big permanent smile. I was up all night reading them.
BeanieBots
Moderator
Perhaps a tad too much information there Dippy. Don't want to give newbies nightmares do we?
Pretty pictures of groomed men have their place, but please, not in power circuit design articles!
Pretty pictures of groomed men have their place, but please, not in power circuit design articles!
Ballpark figures (and I've lost the envelope I drew them on), I've got a 10W panel charging a 7Ah SLA and running a picaxe circuit drawing 10mA 24/7.
From the first post, I gather you are using a 75W panel charging a 100Ah batt. So ballpark you have the ratios roughly right there. So, again (on the back of a new envelope), that gives maybe 75-100mA continuous draw. Or higher if you want to run things intermittently.
That of course is allowing for cloudy days and the need to run all the time. If you can accept it failing after a week of clouds, and you have mains backup, then you can draw a lot more current.
Having got lost in the intricacies of switch mode supplies before, and having spend months on them and not got very far, there is another factor to consider. Say one takes 3 months to build a circuit that gives x% extra power. All well and good, but it just wasted 3 months of sunlight that the panel didn't capture.
I'd build this in stages. Start by connecting the panel to the battery and disconnect the panel with a relay when it is charged. Then improve it with a MPPT or whatever. But maybe datalog it with a picaxe - then you can work out how much extra power it captured. Then build other more complex things.
It is a general build philosophy I've used in a number of circuits (borrowed from power stations). Build a simple circuit with a three way switch. Panel to battery. 3 positions are on/off/auto. Test some currents on a cloudy day late in the afternoon. Compare with the 75W nominal rating. Once you have it working on a simple switch, build something more complex to go in the "auto" circuit. But if the SEPIC circuit blows its caps, you can still go back to the manual. Each "auto" circuit can have its own 3 position switch. So as it grows, you can add more and more automation - right up to PC control. But if bits fail, you can go back to manual solutions too.
Just an idea, anyway.
Out of interest, how many leds round the house are you driving?
From the first post, I gather you are using a 75W panel charging a 100Ah batt. So ballpark you have the ratios roughly right there. So, again (on the back of a new envelope), that gives maybe 75-100mA continuous draw. Or higher if you want to run things intermittently.
That of course is allowing for cloudy days and the need to run all the time. If you can accept it failing after a week of clouds, and you have mains backup, then you can draw a lot more current.
Having got lost in the intricacies of switch mode supplies before, and having spend months on them and not got very far, there is another factor to consider. Say one takes 3 months to build a circuit that gives x% extra power. All well and good, but it just wasted 3 months of sunlight that the panel didn't capture.
I'd build this in stages. Start by connecting the panel to the battery and disconnect the panel with a relay when it is charged. Then improve it with a MPPT or whatever. But maybe datalog it with a picaxe - then you can work out how much extra power it captured. Then build other more complex things.
It is a general build philosophy I've used in a number of circuits (borrowed from power stations). Build a simple circuit with a three way switch. Panel to battery. 3 positions are on/off/auto. Test some currents on a cloudy day late in the afternoon. Compare with the 75W nominal rating. Once you have it working on a simple switch, build something more complex to go in the "auto" circuit. But if the SEPIC circuit blows its caps, you can still go back to the manual. Each "auto" circuit can have its own 3 position switch. So as it grows, you can add more and more automation - right up to PC control. But if bits fail, you can go back to manual solutions too.
Just an idea, anyway.
Out of interest, how many leds round the house are you driving?
Michael 2727
Senior Member
Solar Charts info, plot your own
If you get really bored try these -
http://solardat.uoregon.edu/SunChartProgram.html
The .pdf charts plotted are around 90kb.
Look-up Latitude and Longitude - Other Countries-
http://www.bcca.org/misc/qiblih/latlong_oc.html#AUSTRALIA
UTC/GMT Conversion-
http://www.dxing.com/utcgmt.htm
If you get really bored try these -
http://solardat.uoregon.edu/SunChartProgram.html
The .pdf charts plotted are around 90kb.
Look-up Latitude and Longitude - Other Countries-
http://www.bcca.org/misc/qiblih/latlong_oc.html#AUSTRALIA
UTC/GMT Conversion-
http://www.dxing.com/utcgmt.htm
If the aim is to collect as much energy as possible, SEPIC/buck/boost/tracking may still not be worthwhile. The flawed logic (I'm guilty of this too) is to say - I have a 100W panel and it is night half the time (on average) so I've got 50W - so store that in a battery with 90% efficiency and there is 45W to run devices etc.
The problem is that some days that 100W panel may produce only 1W. On those days, can you live with no power. Ah, you say, you can draw power off the battery. But that assumes the battery is a deep cycle battery. Car batteries don't like being discharged more than 20% or so, and if you use bigger batteries the cost of batteries and replacing them every 7 years becomes higher than the panels.
So, for continuous power, you end up having to grossly overrate the panel. That means that most of the time you are throwing away power - either dumping it into a resistor or simply disconnecting the panel.
And if you are throwing away power most of the time, there is no point in trying to get a little bit more energy with all the extra expense of MPPT and trackers.
There are two exceptions to this that I can think of, where the aim is to collect as much energy as possible and use 100% of the energy, and where MPPT and tracking are very much worthwhile. The first is grid intertie systems, where the solar power goes straight into the grid and there are no batteries at all. The second is solar water pumping, where the aim is to fill a big tank and it doesn't matter if pumping doesn't happen on some days. (in this scenario, the tank acts as the battery).
I've got solar panels all round my place running various things. But their purpose is not to collect energy per se, their main purpose is to avoid stringing wires areound the garden. For collecting energy (eg to offset power bills, save the planet etc), get a grid intertie. Or even better, go solar hot water.
The problem is that some days that 100W panel may produce only 1W. On those days, can you live with no power. Ah, you say, you can draw power off the battery. But that assumes the battery is a deep cycle battery. Car batteries don't like being discharged more than 20% or so, and if you use bigger batteries the cost of batteries and replacing them every 7 years becomes higher than the panels.
So, for continuous power, you end up having to grossly overrate the panel. That means that most of the time you are throwing away power - either dumping it into a resistor or simply disconnecting the panel.
And if you are throwing away power most of the time, there is no point in trying to get a little bit more energy with all the extra expense of MPPT and trackers.
There are two exceptions to this that I can think of, where the aim is to collect as much energy as possible and use 100% of the energy, and where MPPT and tracking are very much worthwhile. The first is grid intertie systems, where the solar power goes straight into the grid and there are no batteries at all. The second is solar water pumping, where the aim is to fill a big tank and it doesn't matter if pumping doesn't happen on some days. (in this scenario, the tank acts as the battery).
I've got solar panels all round my place running various things. But their purpose is not to collect energy per se, their main purpose is to avoid stringing wires areound the garden. For collecting energy (eg to offset power bills, save the planet etc), get a grid intertie. Or even better, go solar hot water.
BeanieBots
Moderator
Some good points Doc.
Also, a 100Ahr battery will have an internal leakage current in the order of 50mA. If you fit a desulphator, that will use another 30mA or more.
Also, a 100Ahr battery will have an internal leakage current in the order of 50mA. If you fit a desulphator, that will use another 30mA or more.
Thanks all. Good info.
@Dippy
“Out of interest, how many leds round the house are you driving?”
None at the moment. I haven’t even purchased any. I haven’t even started looking for any in earnest. I’ll do that after the panel is on the roof and I have some wiring in place. I have in my spares box a few 22000mcd 5mm white LEDs. I’ll prolly start with those and work up.
@BB
Sorry. I did indeed mean R3, not R2.
Looking at the maths, it seems I may have bitten off more than I can chew. Even the ‘basic’ introduction to switching power converters nearly made my head pop. Oh, for the old days, when a simple series resistor was sufficient charge control.
I know. I’ll just use a simple series power resistor and have it submerged in a cup of water. Saves boiling the kettle
@Dippy
“Out of interest, how many leds round the house are you driving?”
None at the moment. I haven’t even purchased any. I haven’t even started looking for any in earnest. I’ll do that after the panel is on the roof and I have some wiring in place. I have in my spares box a few 22000mcd 5mm white LEDs. I’ll prolly start with those and work up.
@BB
Sorry. I did indeed mean R3, not R2.
Looking at the maths, it seems I may have bitten off more than I can chew. Even the ‘basic’ introduction to switching power converters nearly made my head pop. Oh, for the old days, when a simple series resistor was sufficient charge control.
I know. I’ll just use a simple series power resistor and have it submerged in a cup of water. Saves boiling the kettle
I'd certainly recommend you now put time & energy into checking LED lighting types/ mounts/ shades etc at this stage. Aside from DIY, some recent cluster types (rated as low as 1W), are absolutely gorgeous -with a dozen of these you could light up the house. 3W types will almost burn holes in the wall. You may NOT even need significant solar input if the end lighting placement is well thought out. Compare LHS reading conditions below (using a bright LED torch) with the RHS 1W LED cluster lamp.
Attachments
I have been playing with simulating switching circuits using CircuitMaker sp1.
I have started simple. A LED driver.
For the purposes of this simulation, I have used a blue LED (coz the sim does not have any white) with a forward drop of 3.5v and rated max current of 40mA. The diode is a switching type (10nS). The transistor is of a type I have in my spares box and is driven with 0.5% duty at 5KHz.
Circuit 1:
Circuit 2:
In the first circuit, the LED is fed forward current on both the charge and discharge cycles of the inductor, but the extra diode dissipates some power. It also adds capacitance and the switching delay must count for some loss. I would have thought this would be the less efficient of the two.
In the second circuit the LED is fed forward current only during the inductor discharge cycle. But all the power stored in the field should be available.
Why is Circuit 1 (~75%) is more efficient than Circuit 2 (~65%)?
(Efficiency determined by average watts in verses average watts in the LED)
I have started simple. A LED driver.
For the purposes of this simulation, I have used a blue LED (coz the sim does not have any white) with a forward drop of 3.5v and rated max current of 40mA. The diode is a switching type (10nS). The transistor is of a type I have in my spares box and is driven with 0.5% duty at 5KHz.
Circuit 1:
Circuit 2:
In the first circuit, the LED is fed forward current on both the charge and discharge cycles of the inductor, but the extra diode dissipates some power. It also adds capacitance and the switching delay must count for some loss. I would have thought this would be the less efficient of the two.
In the second circuit the LED is fed forward current only during the inductor discharge cycle. But all the power stored in the field should be available.
Why is Circuit 1 (~75%) is more efficient than Circuit 2 (~65%)?
(Efficiency determined by average watts in verses average watts in the LED)