Help with modified buck converters...

Thread Starter

Dyslexicbloke

Joined Sep 4, 2010
566
Hi folk's,
I need to build a battery charger, fed from PV, and plan to manage the batteries using a uP
However it seemed wise to build a charger that had inherrent limits and then adjust it down when necessarry as opposed to having the uP in full control.
If this achieves nothing else it would allow me to operate manually if needs be.

It will need to deliver circa 50A on occasion, but should limit at that by default.
It will need to limit the maximum voltage to about 14.8
It should maintain a minimum PV voltage of about 16, possibly as much as 19, I will not actually know until it is running.

Assiming I use a buck converter, which I have, it starts out as a simple voltage source but finding such a high current device wasnt looking promissong, at least not with the budget I have.
I did some reading on paralell devices and also did some testing and found that the cheap V & I limmiting units all used the negative line to measure current. which clearly wasnt going to work keeping several units ballanced because the shunt wasnt in the positive output and there was no practiacal way to ensure that the negative currents are balanced.

However I did find that all the units I looked at used a simmilar control strategy when implimenting current limmiting... some circuit, it isnt importent what, raised a voltage as the current went up. This reference was then fed to the sens pin, enhancing the voltage that is applied by the voltage devider, via a diode. The net result is that until the forward bias of the diode is reached nothing happens and when it is the sens pin voltage rises causing the. buck chip to back off... Neet I thought.
After a bit of fiddling, and a little smoke, I got a simple circuit to work that allowed me to bend the buck board with an analouge, intigrated PWM,ii output from the IP.
I also found that many of these aditional 'limits' could be applied easily as the diode isolated each circuit keeping them simple.
Now I think about it is no different than a summing circuit with an opamp.

Anyway... I decided to try the simple buck boards in paralell. Obviously each needed a current limit as a minimum requirement. I added a current sensor, 5v supply +/_ 2v arround a 2.5v mid point. The sensor output go's via a current limiting resistor to a trimmir which in turn applies voltage to the buck sens pin via a smal diode. The limit is a little soft but in all honesty that just works to eliminate oscillation, or at least it seems to.
I could, and may, amplify the signal a little to get a harder tirn off but it will do for now.

After I had 4 simmilar units all modified with the high side current sensing and limiting circuit I hooked them all in paralell to see if it worked and it initially looked good. The at very low currents the ballance isnt great but still acceptable. At higher currents all the units top out nicely and show no obvious signs of oscillation or feedback, at least not that I am recognizing.

There is no uP in play just the stock voltage devider, athe new current limit and a second inverted voltage sensing circuit on the PV side.

Holdng the converters off untill the panel voltage is over 16, well above battery voltage, seems to prevent any reverse curent and apart from addressing some form of soft start, another little circuit and a diode I thought I was there.... Silly me!

Despite the fact tbat everything looked OK the convertrs are running very hot. Yes they are from afar with dodgy build quality but I bought 15A units anticipating being able to run them at 8 or 9A so that they wern't stresed at all.
Well that isnt working... I have 4 in paralell now and even st at 5A each they are warm enough to want forced cooling despite some chunky heat sinks.

Does anyone know what I am missing and or what to do about it?

In all honesty if I thought it was all stable but with way overstated specs I would simply add more modules but it seems more likely that I am stressing the kit in some way that I have failed to apriciate and if that is the case they will faill at some point however low I set them.

Thoughts, I dont have a clue.

Actually a couple of specific. questions, should I remove the output caps from each module and have a single separate bank, could too much capacitance be a problem. It would certainly have an effect on an AC circuit and as the effect increases with frequency I can see that it would be a problem at some point but I dont know how to work out when.
Also will a buck converter run hotter or cooler as the differential between input and output voltage increases.

Thanks for looking,
regards,
Al
 
Last edited:

Thread Starter

Dyslexicbloke

Joined Sep 4, 2010
566
I know bumping your own post is frowned upon but I am relay stuck, despite hours on the web looking for solutions.
This will be the first and last time I push this thread up the queue

I would welcome any input at all, even if it is only to tell me to go about this in some other way.

To be fair my first thoughts were to use PWM and but a high current relay but when looming at the potential issues with that I found that adding an inductor to handle current spikes and ripple, and as a result a diode to discharge it and prevent voltage spikes when the switch is open, I would have effectively been building a buck circuit anyway, which is what brought me here.

We are off grid so lack of solar has a huge effect of generator fuel and I need to find a solution as quickly as possible.

If you think you have a better plan than this PLEASE shout up and if you have any experiance with buck circuit design I would love to here from you, even if you are only pointing out how silly I am being

Thanks,
Al
 

ronv

Joined Nov 12, 2008
3,770
I know bumping your own post is frowned upon but I am relay stuck, despite hours on the web looking for solutions.
This will be the first and last time I push this thread up the queue

I would welcome any input at all, even if it is only to tell me to go about this in some other way.

To be fair my first thoughts were to use PWM and but a high current relay but when looming at the potential issues with that I found that adding an inductor to handle current spikes and ripple, and as a result a diode to discharge it and prevent voltage spikes when the switch is open, I would have effectively been building a buck circuit anyway, which is what brought me here.

We are off grid so lack of solar has a huge effect of generator fuel and I need to find a solution as quickly as possible.

If you think you have a better plan than this PLEASE shout up and if you have any experiance with buck circuit design I would love to here from you, even if you are only pointing out how silly I am being

Thanks,
Al
I don't think you will have much luck paralleling the smaller units. Do they have a current adjust pot? Maybe a link to their data sheet.
Post the specs on your panel and the battery(s).
 

Thread Starter

Dyslexicbloke

Joined Sep 4, 2010
566
No I adjust, I specifically bought them that way...
I realize that parallel units with no control other than voltage will never share the load but the units capable of controlling current, at least the 5 or 6 different ones I checked, all used a shunt on the negative rail so that wouldn't have helped anyway.

I have added current sensors, isolated ones, that measure and successfully regulate the output and as a result balancing the current isn't a problem.

The 4 I have hooked up now share the load well but if I take them above about 4.6 amps they run very hot!

Al
 

ronv

Joined Nov 12, 2008
3,770
No I adjust, I specifically bought them that way...
I realize that parallel units with no control other than voltage will never share the load but the units capable of controlling current, at least the 5 or 6 different ones I checked, all used a shunt on the negative rail so that wouldn't have helped anyway.

I have added current sensors, isolated ones, that measure and successfully regulate the output and as a result balancing the current isn't a problem.

The 4 I have hooked up now share the load well but if I take them above about 4.6 amps they run very hot!

Al
I don't think it matters that the sense resistor is in the return.
How hot is hot? Can you hold your finger on the heat sink?
Have you tried one by itself with a 5 or 10 amp load?
If they are fighting with each other a small resistor - say .1 ohms in series with each one might help.
How do you plan on controlling these?
 

Thread Starter

Dyslexicbloke

Joined Sep 4, 2010
566
I don't think it matters that the sense resistor is in the return.
The problem is that the components are not tight tolerances so the paralleled resistors will not share current accurately...
Obviously if the current is low enough a few % of imbalance wouldn't matter but as I want to manage the maximum current I need to know what it is for each device.
Current sensing and substituent limiting is working well.

How hot is hot? Can you hold your finger on the heat sink?
Right now at 4.5A about 45-50C but at 10A hot enough to cause a failure, queue the smoke!

Have you tried one by itself with a 5 or 10 amp load?

I have tried a single unit, the one that generated smoke, but although its output was isolated its input wast and as the entire system voltage was lower than the set output the others were probably fully on which would have hooked in all the capacitors, all be it via the switches in the buck chips. A properly isolated test is a good idea, I will try and do it over the weekend

If they are fighting with each other a small resistor - say. .1 ohms in series with each one might help.

They don't look like they are fighting and I don't want to go down that route if I can help it. at high currents the losses are unacceptable and at lower currents they are manageable but still a significant overall los in solar output

How do you plan on controlling these?

With an Arduino eventually, I have tried this in the past and it seems to work well. I built a passive in integrator to smooth a PWM output and fed it, via a pot, current limiting resistor and a diode to the sense pin. I was able, reading an analogue and ramping the output, to mimic the effect of the normal voltage divider which gave me a pragmatically adjustable buck module.
I didn't try to build a control loop, the buck chip already has one, I simply set the output to some proportion of the input.

The plan is/was to build a circuit that had effective maximum limits, set with pots, but that could be overridden to reduce the duty cycle of the converter when circumstances demanded it.

The buck chip attempts to maintain the sense pin at 1.4V at least on these chips and if you decouple them with diodes vertualy any number of circuits can feed that pin.
If any drive it over 1.4V the converter starts ramping down until the voltage on the sense pin is within tolerance.

My current limits have a 5V rail with 0A being expressed as 2.5V. The 20A units I am using increase the output bu 0.1V/A This is fed to a pot to allow for threshold adjustment to be made and then on to the sense pin via a signal diode. when the current increases a point is reached where the diode is sufficiently biased forwards to turn on at which point the sense voltage starts to clime, backing of the converter and reducing the voltage and thus the current in the load, batteries in my case.

I appreciate your input it is always god to bounce ideas off somone

Regards
Al
 
Last edited:

ronv

Joined Nov 12, 2008
3,770
I don't think it matters that the sense resistor is in the return.
The problem is that the components are not tight tolerances so the paralleled resistors will not share current accurately...
Obviously if the current is low enough a few % of imbalance wouldn't matter but as I want to manage the maximum current I need to know what it is for each device.
Current sensing and substituent limiting is working well.

How hot is hot? Can you hold your finger on the heat sink?
Right now at 4.5A about 45-50C but at 10A hot enough to cause a failure, queue the smoke!

Have you tried one by itself with a 5 or 10 amp load?

I have tried a single unit, the one that generated smoke, but although its output was isolated its input wast and as the entire system voltage was lower than the set output the others were probably fully on which would have hooked in all the capacitors, all be it via the switches in the buck chips. A properly isolated test is a good idea, I will try and do it over the weekend

If they are fighting with each other a small resistor - say. .1 ohms in series with each one might help.

They don't look like they are fighting and I don't want to go down that route if I can help it. at high currents the losses are unacceptable and at lower currents they are manageable but still a significant overall los in solar output

How do you plan on controlling these?

With an Arduino eventually, I have tried this in the past and it seems to work well. I built a passive in integrator to smooth a PWM output and fed it, via a pot, current limiting resistor and a diode to the sense pin. I was able, reading an analogue and ramping the output, to mimic the effect of the normal voltage divider which gave me a pragmatically adjustable buck module.
I didn't try to build a control loop, the buck chip already has one, I simply set the output to some proportion of the input.

The plan is/was to build a circuit that had effective maximum limits, set with pots, but that could be overridden to reduce the duty cycle of the converter when circumstances demanded it.

The buck chip attempts to maintain the sense pin at 1.4V at least on these chips and if you decouple them with diodes vertualy any number of circuits can feed that pin.
If any drive it over 1.4V the converter starts ramping down until the voltage on the sense pin is within tolerance.

My current limits have a 5V rail with 0A being expressed as 2.5V. The 20A units I am using increase the output bu 0.1V/A This is fed to a pot to allow for threshold adjustment to be made and then on to the sense pin via a signal diode. when the current increases a point is reached where the diode is sufficiently biased forwards to turn on at which point the sense voltage starts to clime, backing of the converter and reducing the voltage and thus the current in the load, batteries in my case.

I appreciate your input it is always god to bounce ideas off somone

Regards
Al
I think your toast.
http://ijcsi.org/papers/IJCSI-9-6-3-487-494.pdf
There are ways to do it, but not with the little ebay guys.
I'm still thinking. :D
 

Thread Starter

Dyslexicbloke

Joined Sep 4, 2010
566
Actually that charge controler looks fairly good but it still lacks any method of refining the charging strategy...
It would be a good compromise at the price though.

Why do you think Toast?
The document on current sharing is interesting and in all honesty not that far from where I am.

I absolutly agree that cheap modules are a poor long term choice but they are a great learnin tool, provide you are happy to sacrifice a few!

Just to reitterate... current sharing is relativy poor with simply paralelled modules but I am not necessarraly seeing that as a problem because it is good when each module hits its limit, bear in mind I have implimented current monitoring as opposet to using the stock design found in cheap modules. My scheme is very simmilar to the one described in that document but only active at max current.
Given that cascading thr modules will give better efficiancy that was a design choice.

What I cant figure out is what could contribute to a hot chip, ther than high current.
I have been at work all day so havent had much time to think but I am wondering about reversu current in the modules when the PV voltage is low so I will build a simple input voltage monitor and ramp them down if PV falls below 18, that will ensure, whatever the battery statem that the output rail will always be at least 3.2v below the input.

Anyone know if bucks suffer when the input is the limmiting factor and prevents the output voltage / current being achieved.
Now I think about it I dont even know what the controller would do, essentally it would be outside its spec which is I guess why the datasheets dont seem to civer the eventuallity.

Anyway holding PVV above BATV, by limmiting current, will ensure that reverse curent cant occur even if the unsunct harmonic effects get silly.
By the way, ding this dynamically would costitute MPPT, future plan, which that interesting looking charge controllr couldnt do.

I may be wrong but aw far as I know the only way to adiquatly smooth a PWM based regulator, especilly when the input V is significantly higher than its output is to add an integrator of some sort.... going LC which is much more efficient than RC essentally make the circuit a buck anyway, hence my approach.

There are days when I think an eduaction beyond school might have been nice! but then when I was in school PLC,s lived in 19" racks and memory that didnt use ferriet cores was still new and exciting.

Thoughts?
 

ronv

Joined Nov 12, 2008
3,770
If you are thinking about building something this would be the way to do it I think.
http://www.linear.com/solutions/5464
It would be tricky to do, but not impossible and it would be buck and boost which would make maximum power point possible.
I'm having trouble visualizing what you are doing now. Maybe a schematic would help.
 

Thread Starter

Dyslexicbloke

Joined Sep 4, 2010
566
You are correct, simpler, well more monolythic, chip but there or there abouts.
I dont need boost, PV max power will be 16-19V depending on conditions and the batteries never need more than 15 even when equilising, less at float.

The chips are XL4016E1 on a simple buck board.
http://www.dx.com/p/adjustable-step-down-module-blue-dc-dc-12a-242230#.VwhMVm4o_MI

I didnt buy fro there but the info is good.

I have otner boards using this chip which impliment current limmiting by taking an opamp output to the sense pin via a diode.
Obviously as the devider is still in circuit the output voltage is controlled in the normal way untill the diode is passing current at which point the chip backs off, in fact as its is ontrplled by a loop internally as oposed to working like a linear regulator if the sens pin is pushed high even slightly the chip will shtt down asfr a short period.

My implimentation is to use a current segsing chip at 2.5V +/_ 0.1V/A which biases the sense pin via a devider, a trimmer in my case, and a diode. I am confident that this is not the issue.

I am at home tomorow and will try, if there is any sun, a single module to see it it runs hot in its basic configuration, and then with my limit circuit.

Do you have any tboughts RE output capacitance or insuficiant input current, and thus voltage, to maintain the output?

Al
 

ronv

Joined Nov 12, 2008
3,770
You are correct, simpler, well more monolythic, chip but there or there abouts.
I dont need boost, PV max power will be 16-19V depending on conditions and the batteries never need more than 15 even when equilising, less at float.
I understand, but it is useful when the sun is not bright and the panel voltage is below the battery voltage.

The chips are XL4016E1 on a simple buck board.
http://www.dx.com/p/adjustable-step-down-module-blue-dc-dc-12a-242230#.VwhMVm4o_MI

I didnt buy fro there but the info is good.

I have otner boards using this chip which impliment current limmiting by taking an opamp output to the sense pin via a diode.
Obviously as the devider is still in circuit the output voltage is controlled in the normal way untill the diode is passing current at which point the chip backs off, in fact as its is ontrplled by a loop internally as oposed to working like a linear regulator if the sens pin is pushed high even slightly the chip will shtt down asfr a short period.

My implimentation is to use a current segsing chip at 2.5V +/_ 0.1V/A which biases the sense pin via a devider, a trimmer in my case, and a diode. I am confident that this is not the issue.

I am at home tomorow and will try, if there is any sun, a single module to see it it runs hot in its basic configuration, and then with my limit circuit.
Not sure about this.
Do you have any tboughts RE output capacitance or insuficiant input current, and thus voltage, to maintain the output?

Al
I think the fet in the module will just turn all the way on if the input voltage is low. Assuming there is a diode or something in the panel to keep it from taking voltage from the battery it should be ok I think.
Of course one will quit before the others. Not sure how that works out. okay, I think.[/quote]
 

Thread Starter

Dyslexicbloke

Joined Sep 4, 2010
566
Synchronised ... Yes I have seen that and I apricoate that if things are done that way there is probably a good reason, understanding it howevr is a different matter.

Low sun... I have played with this in the past putting panels in seriese in attempt to get usable power from very low light levels and it doesnt work. PV has a relativly high impedence when the light levels are low. In fact most large panels have forward biased diodes that can bypass shadex sections of the panel which, given that it works, means a poorly illuminated cell must drop at least 0.6V

Fully on... Thats what I thought but it looks as though ther is always about 1.4V differential, assuming that there is some illumination.
Reverse current flow is possible so this volt drop cant be the forward drop of a diode so I cant explain it right now.
My PV does have reverse protection diodes but I plan to remove thes e once I have an activly controlled system, IE converters isolated woth relays when the PV is below its cut in voltage.

Your comments are helpful, thanks, even though tbey are not definative answers they are helping me reevaluate and in some cases revisit my assumptions.

Cheers,
Al
 

Thread Starter

Dyslexicbloke

Joined Sep 4, 2010
566
Well I think I found the issue with overheating....
The board was advertised as using an XL4016 and being capable of 15A...
When I first checked I found that a genuine XL4016 is rated at 12A which is a little irritating and begs the question as to why the board has a 15A fuse...

HOWEVER...

I just looked a little carefully and found that an XL4016E1, which is what is actually fitted, is an unbranded variant and only rated at 8A. This explains why the magic smoke escaped when I pushed one to 10A.
It maly may also explain why the buck is audable under load when it is supposed to ne 300kHz
I had assumed I was hearing harmonics resulting from the paralelling and minor differences in each converters freequency but I am left wondering now if they are simply running at 20k

Mixed news I suppose.... On a positive note it means that the heat is explained and the paralelling probably isnt contributing but if I run each module at 5 or 6A I am going to need more of them which I suspext will come with its own issues not to mention increesed cost, al be it modest.

I think I acn feel a rant comming on... I bought from a UK supplier, as opposed to direct from China, assuming that the part had been validated as performing as advertised but it clearly hasnt been.
I am not happy and about to pass on the none-joy

Hey at least it looks like the concept ist too flawed. The bank of 4 I have connected now are all in current mimiting mode with the PV at 20 odd volts and doing a combined current of about 18.5A - warm but not exessivly hot and all about equilibrium

Al
 
Top