Constant current regulation with boost converter, PWM and 555 timer.

[AlgoJerViA]

I'm working on this schematic based on parts I have available. I intend to replace R1 with a 100 Watt LED. Can that be done in multisim or do I have to make arrays of LEDs? I have replaced the normal diodes and potentiometers with Q2 and Q3 but thats the part I have not copied from the Internet but it seems to work fine in the simulation. This allows me to control the PWM with a voltage level instead of a manual potentiometer. The next step and the goal of this schematics is to sense the current somehow, possibly with the current transformer I1 and regulate the PWM so that the current holds constant at 3 amps. Is this possible at all or am I on the wrong track altogether?

//AlgoJerViA

 

[(*steve*)]

For a constant current supply you place a low value resistor in series with the load and regulate the voltage across this "current sense" resistor.

Your schematic does not appear to show this.

 

[Fish4Fun]

Hi AlgoJerVia! Welcome to EP!

What *Steve* said, but a couple of other things too....

First, a 100W LED driven by a buck-converter and a 12V source suggests the Vf of the LED should be << 9V and this in-turn implies the current will be >10A....not really a chore for a buck converter....

Next, if you are planning on using something like this: http://www.ebay.com/itm/1X100W-LED-...ing_Parts_and_Accessories&hash=item19f3db59fd

Please Note that the Vf = 32V to 35V.... And If = ~3.0A....which would be really difficult to achieve with a 12V source, a555 timer and a single N-Channel mosfet configured as a high-side switch....perhaps not impossible to achieve...but really really difficult at the very least.

I would suggest you simply BUY an LED driver designed to work with the specific LED you have selected....I can't see any way you could build one for for less than 10x what you could simply buy one for.....Certainly aren't going to build a suitable driver with the parts in your schematic.....

Good Luck!

Fish

 

[KrisBlueNZ]

First, a 100W LED driven by a buck-converter and a 12V source suggests the Vf of the LED should be << 9V and this in-turn implies the current will be >10A....not really a chore for a buck converter....

It's a boost converter.

Please Note that the Vf = 32V to 35V.... And If = ~3.0A....which would be really difficult to achieve with a 12V source, a555 timer and a single N-Channel mosfet configured as a high-side switch....perhaps not impossible to achieve...but really really difficult at the very least.

I agree, I don't think a 555 is a good control device. Google for designs based on the UC3842/3/4/5 current mode controller. It's normally used in isolated flyback applications but a boost converter is almost the same as a flyback converter without isolation.

I would suggest you simply BUY an LED driver designed to work with the specific LED you have selected....I can't see any way you could build one for for less than 10x what you could simply buy one for.....

True dat.

 

[AlgoJerViA]

Hi all. Thanks for your replies.
What are the problems you see with this configuration? When I tried this out a month or so ago I had no problem running the led att something like + 25 volt, I didn't want to go any higher since I didn't have any current regulation and worried about burning the LED. The only problem I run into after a couple of minutes or so was that the breadboard started melting under the mosfet but the heatsink was not very hot.

This project is just for fun and started because I need something to do and want to learn in the process. I also appreciate the idea of reusing part and the general "hacking" philosophy. If I one day decides I need to use the LED for something important where quality is desired I will probably buy a driver.

However, as a complete side note then... If this works works it is rather the opposite around, this design is about 10x cheaper than any premade driver.
2N2222, 2N2907, NE555P, C1, C2 and C6 0.16€ (already bought this in small quantity around 50 of each)
IRF520 0,41€ (but I took it from the arduino starter kit, can also get other suitable from trash)
L1 and D1 are recycled but I can't see why they would cost more than a few euro cent either.
Then some resistors and a lab pcb and I think we are talking about something like 1 or 2€ top.
And yes the current transformer costs about 1€ but the only reason I thought that would be a good idea is just that I had i laying around.

This is the alternative I can find http://www.aliexpress.com/item/LED-...odule-car-laptop-mobile-power/1382703483.html at 9€.

 

[Arouse1973]

Isn't a current transformer an A.C device?
Adam

 

[Fish4Fun]

AlgoJerViA,

If the purpose of the project is learning and the venue is "hobby", then BY ALL MEANS lets work this out...if you just need a driver the link you sent for $11 (not sure what the cost to your country would be...) sounds like a "deal" to me...There are other drivers available from China for less, but as this is an educational project, even if you end up @ $100 building one...IF you learn a lot in the process then it would be money WELL SPENT! Besides, I view "Hobby Accounting" as having very different rules than "business accounting"....."Hobby Accounting" is much closer to Government Spending....but I won't diverge into that here...

OK, As Kris pointed out the configuration you are working on is "Boost-Buck" with a High-Side Switch.....In order for Q1 to "turn on" in this configuration you will need the gate to be at least 10V above the rail voltage....there are several ways to achieve this....but the complexity likely out-weighs the benefits, so it will be easier for you to simply re-configure your design with Q1 as a "low side" switch....As previously pointed out you will need to incorporate current feed-back as that is the primary feature of an LED driver....You might also consider the IC that Kris linked or a uController to replace the 555 timer.....While the 555 has been used for a million different purposes, it is RARELY the "best choice" for any of them....

Do some Googling and figure out how you want to proceed then post back with questions or thoughts!

Good Luck!

FIsh

 

[AlgoJerViA]

Isn't a current transformer an A.C device?
Adam

Haha now when you say it, sounds very logical.

"Hobby Accounting" is much closer to Government Spending....

You get my point.

OK, As Kris pointed out the configuration you are working on is "Boost-Buck" with a High-Side Switch.....In order for Q1 to "turn on" in this configuration you will need the gate to be at least 10V above the rail voltage....there are several ways to achieve this....but the complexity likely out-weighs the benefits, so it will be easier for you to simply re-configure your design with Q1 as a "low side" switch....As previously pointed out you will need to incorporate current feed-back as that is the primary feature of an LED driver....You might also consider the IC that Kris linked or a uController to replace the 555 timer.....While the 555 has been used for a million different purposes, it is RARELY the "best choice" for any of them.....

I had understood it as Q1 is fully on when the gate to source potential is 10v and that it will acts like a closed switch then? However putting Q1 as a low side switch sounds easy enough (rather like an suitable challenge) . But obviously I need to read more about the difference.

I think the question I really had about current feedback but failed to formulate before is if a bipolar transistor (2N2222) is a good amplifier for the signal from the shunt resistor.

More than one people has told me that the 555 timer is a bad choice but they say it's because the supposed low frequency. However I think the 555 timers I have is of cmos typen since I have no problem generating a square wave in the megahertz range and when I point that out I have not get any further explanation why they would be a bad choice. I have a Arduino Nano I could use but it seems a bit over kill since I don't see the problem using the 555 timer.

//AlgoJerViA

 

[KrisBlueNZ]

More than one people has told me that the 555 timer is a bad choice but they say it's because the supposed low frequency. However I think the 555 timers I have is of cmos typen since I have no problem generating a square wave in the megahertz range and when I point that out I have not get any further explanation why they would be a bad choice.

The CMOS versions of the 555 have much lower output drive capability. You need a high current driver to charge and discharge the MOSFET's gate-source capacitance quickly. If you don't switch the MOSFET quickly, it will overheat because it will stay in its linear region too long. That is not the only reason why the 555 is not a suitable controller! It is a timer/oscillator not a PWM controller! It is really not suitable!

 

[Arouse1973]

Hi AlgoJerVia! Welcome to EP!

What *Steve* said, but a couple of other things too....

First, a 100W LED driven by a buck-converter and a 12V source suggests the Vf of the LED should be << 9V and this in-turn implies the current will be >10A....not really a chore for a buck converter....

Next, if you are planning on using something like this: http://www.ebay.com/itm/1X100W-LED-...ing_Parts_and_Accessories&hash=item19f3db59fd

Please Note that the Vf = 32V to 35V.... And If = ~3.0A....which would be really difficult to achieve with a 12V source, a555 timer and a single N-Channel mosfet configured as a high-side switch....perhaps not impossible to achieve...but really really difficult at the very least.

I would suggest you simply BUY an LED driver designed to work with the specific LED you have selected....I can't see any way you could build one for for less than 10x what you could simply buy one for.....Certainly aren't going to build a suitable driver with the parts in your schematic.....

Good Luck!

Fish

What is Vf 32 to 35 Volts and high side switch. Are you saying he should use a high side switch?
Cheers
Adam

 

[AlgoJerViA]

OK, As Kris pointed out the configuration you are working on is "Boost-Buck" with a High-Side Switch....

Now I have googled and I must conclude that this thread got messed up. I claim that this is a pure boost converter (no buck) and that the mosfet is a low-side switch. You Fish misread the schematic and turned the polarity of the diode so that you did see it as a buck converter. Kris then corrects you and says it is a boost converter, since this makes sense and I guess you did not care to study the schematics in detail you of course trusted yourself and assumed it was a boost-buck converter. Kris then agrees with you that the 555 timer is a bad choice for this task and in the text he quotes is your claim that the mosfet is high-side, he does probably not agree specifically with that part of the quote. This is of course just my best guess of what's going on and it has been a very helpful learning process to try to be brave enough to question the master.

The CMOS versions of the 555 have much lower output drive capability. You need a high current driver to charge and discharge the MOSFET's gate-source capacitance quickly. If you don't switch the MOSFET quickly, it will overheat because it will stay in its linear region too long. That is not the only reason why the 555 is not a suitable controller! It is a timer/oscillator not a PWM controller! It is really not suitable!

I don't disbelieve you that the 555 is not suitable, this is just a hack I trying to do for fun. It is not supposed to be perfect. However after some googling I have decided to take your advice and buy some UC384x but it will take several weeks before I get them. I'm still interested in trying to push this project as far as possible in the meantime. Now when you point it out I remembered I figured that out the last time I worked on this circuit about a month ago. I was then driving the LED at around 24v and didn't want to go any higher since I had no current regulation. The mosfet got hot enough to melt the breadboard. As far as I understand that specific problem could be solved by using a gate driver and that it could be easily constructed with just a pair of normal transistors.
I guess another bad thing with the 555 as a PWM is quite a large frequency drift when changing the duty cycle that is not good, then also some functions like under/over voltage protection is of course missing. Would you care to add to the list?

 

[KrisBlueNZ]

Now I have googled and I must conclude that this thread got messed up. I claim that this is a pure boost converter (no buck) and that the mosfet is a low-side switch. You Fish misread the schematic and turned the polarity of the diode so that you did see it as a buck converter. Kris then corrects you and says it is a boost converter, since this makes sense and I guess you did not care to study the schematics in detail you of course trusted yourself and assumed it was a boost-buck converter. Kris then agrees with you that the 555 timer is a bad choice for this task and in the text he quotes is your claim that the mosfet is high-side, he does probably not agree specifically with that part of the quote. This is of course just my best guess of what's going on...

Yes, that's what I think. I didn't agree with Fish about the MOSFET being a high-side switch. It clearly isn't, and the converter is a standard boost converter.

I don't disbelieve you that the 555 is not suitable, this is just a hack I trying to do for fun. It is not supposed to be perfect.

And you're not the first person here to want to use a 555 to control a high-power converter. But in the fight between "it's not supposed to be perfect" and "the 555 is not suitable and there are much better choices", I know which one wins for me. And I'm glad you have decided to follow this advice.

However after some googling I have decided to take your advice and buy some UC384x but it will take several weeks before I get them. I'm still interested in trying to push this project as far as possible in the meantime. Now when you point it out I remembered I figured that out the last time I worked on this circuit about a month ago. I was then driving the LED at around 24v and didn't want to go any higher since I had no current regulation. The mosfet got hot enough to melt the breadboard. As far as I understand that specific problem could be solved by using a gate driver and that it could be easily constructed with just a pair of normal transistors.

A gate driver will help the MOSFET switch quickly, and this will help it run cooler if it's currently switching too slowly. But that's not the only reason why the MOSFET could be running hot.

Overvoltage at the drain causing breakdown is another likely scenario, as is uncontrolled saturation in the inductor. Saturation can be avoided with proper inductor design and the use of the UC384x. Drain overvoltage and breakdown can be avoided by using a MOSFET with a higher V_DS(max) rating, and if it's due to leakage inductance in the transformer, a snubber circuit.

I guess another bad thing with the 555 as a PWM is quite a large frequency drift when changing the duty cycle that is not good, then also some functions like under/over voltage protection is of course missing. Would you care to add to the list?

Yes. The 555 is designed for making LEDs flash, and speakers go beep. It can generate an AC signal, but that's only one of many functions you need for a switching converter. Compare the block diagram of the 555 against the block diagram of the UC384x. The only similarity you'll find is that they both have eight pins!