Hey, it's an example. Even with your 3V difference, if you're operating at 1 ampere, that's a substantial power dissipation. If you're willing to adjust the supply, it's manageable.
DIY LED driver for reef lighting
- Thread starter der_wille_zur_macht
- Start date
Hey, it's an example. Even with your 3V difference, if you're operating at 1 ampere, that's a substantial power dissipation. If you're willing to adjust the supply, it's manageable.
tahiriqbal
Sialkot
Thanks, we all appreciate positive input towards this new technology and DIY projects. Thermal heat transfer is extremely important part when taking PCB style LED route. My engineer did a sample PCB on-board driver plus enough space for 5 x XP-G and 5 x XP-E LEDs.
If you look closely you can see very tiny hole sunk though the board and though very thin aluminium sheet which attached to the PCB itself. These holes were once again sunk with very thin copper plating for better heat transfer to the main heatsink. This would prevent PCB from overheating and maybe enhance the life of my LEDs and of course better thermal control. You would also need to place thermal tape between PCB and the main heatsink for safety reasons.
der_wille_zur_macht
Team RC
Read on - we eventually figured that out.
I wouldn't run the CAT4101 (or any other linear reg) in this application unless I could ensure the total Vf was as close as possible (half a volt in this case) to supply voltage, through either careful selection of LEDs and count of LEDs, or trimming the power supply.If I had a situation where I only wanted to drive a small number of LEDs (say, three) from a 24v supply, I would definitely go for a switcher; probably the NCP3066.
I finally finished reading the whole thread.... Phew. The driver I'm considering building meets those or most of those requirements but it looks like it would cost about twice what DWZM's drivers ended up costing.
I am looking at the National Semiconductor LM3409HV. It definitely meets #s 1, 3, 4 and the PWM part of #6. I'm not sure what you mean by #2, or I'm having trouble distinguishing it from #3.
#5 I would like to implement with an associated ATtiny25 for each driver (part of that greater cost). And while there is analog dim control (#6b), it is 0 - 1.24V, not 0 - 10V, so one would need a level shifter or translator.
The LM3409HV goes up to 75V. It does not pass the string current through the part.
The disadvantage of not passing the string current through the part is that you must have an external power transistor. That's an additional part. Plus, as a true Buck regulator, one needs an inductor and hefty diode as well. Then choosing to support the full (or almost full) capabilities of the LM3409HV, the ratings on those parts must be quite high, typically 100V.
The LM3409's datasheet mentions up to 5A LED current, but I've targeted 2A for now, which is beyond what most folks want to use anyway. I also have a scheme for a set of three (or four) DIP switches which will select maximum currents ranging from 350mA to 2A in eight steps. Plus analog dimming within the selected range.
The other idea I have is to put an ATtiny25 on each driver to provide PWM control and temperature sensing and use an I2C bus to communicate with them. I think four drivers per board is about right. But one should be able to daisy chain the I2C bus from board to board, and thus have individual control of every driver. I've begun working on a list of I2C commands for the drivers/ATtiny and a rough flowchart for the control program.
I did a preliminary BOM and calculate the cost of parts per driver at $8 - $16 per driver (depends on quantity bought) and the cost of board and board specific parts (four drivers per board) at $6 - $10. So at the cheapest group buy, it would still be $42 per four drivers, but they would be very nice drivers.
I would enjoy discussing these preliminary ideas more, but want to check first if they belong in this thread, before I get too carried away. This thread was started as a driver thread, but it was oriented towards an economical driver. I have in mind a driver which will cost a little less than a Buck Puck, but be extremely versatile, resilient and feature laden. So while it is a driver, it is a somewhat different goal.
kcress
New member
Interesting.
What's meant by #2 is feed back from the string's current - A resistor is included in the string, at the bottom usually. A voltage is developed across the resistor that is directly related to the current running thru the string. This provides for accurate current control.
Having an external FET is not bad! It allows the unit to control more current, otherwise the controller chip gets too hot.
"Current limit" is some final say in what you can command the driver to reach. This is also important so that if a LED shorts the current doesn't increase across the string, or you send something dumb to the driver from your controller.
Sometimes the resistor value chosen will automatically set the maximum current. That's a solid dependable method too.
Your ATs can certainly monitor the temps and throttle back the current if the cooling isn't hacking it for some reason.
The only thing you haven't covered is that this controller is a BUCK controller.. so where are you going to get 75Vdc? That is not common to come by.
Perhaps you should keep looking and go for a boost controller. That will make a higher voltage allowing you to amortize your controller costs over very many more LEDs.
What's meant by #2 is feed back from the string's current - A resistor is included in the string, at the bottom usually. A voltage is developed across the resistor that is directly related to the current running thru the string. This provides for accurate current control.
Having an external FET is not bad! It allows the unit to control more current, otherwise the controller chip gets too hot.
"Current limit" is some final say in what you can command the driver to reach. This is also important so that if a LED shorts the current doesn't increase across the string, or you send something dumb to the driver from your controller.
Sometimes the resistor value chosen will automatically set the maximum current. That's a solid dependable method too.
Your ATs can certainly monitor the temps and throttle back the current if the cooling isn't hacking it for some reason.
The only thing you haven't covered is that this controller is a BUCK controller.. so where are you going to get 75Vdc? That is not common to come by.
Perhaps you should keep looking and go for a boost controller. That will make a higher voltage allowing you to amortize your controller costs over very many more LEDs.
That is how the 3409 works. The resistor is actually directly upstream of the power transistor. The inductor and LED string are after the power transistor in the current flow.
It has pluses and minuses. The advantages are as you say, plus one can choose the parameters to suit one's desires. The disadvantage is simply additional components and complexity. However, that additional complexity is a bit of an illusion.
I almost chose the LM3406 for my design (1.5A internal power transistor) simply because it does not require an external transistor. However, when I started looking at the various "extras" one should probably add to protect the 3406 from various failure modes, the two or three extra diodes plus handful of resistors looked like more trouble than just having an extra transistor. As long as the 3409 is paired with a transistor that can handle the full Vin, it is protected from things like disconnecting the load.
So the basic design of the 3409 is more complex because it needs an external transistor. But it saves another handful of components in real world design because it is more naturally resistant to failure modes.
Are you referring to number 3 here, "sets max allowable current"? The sense resistor does this. I suppose that if the sense resistor failed shorted though then one would probably get the full Vin across the LED string. I don't see how any design could protect against that though in a general fashion.
For the LM3409 the maximum current is Vcst/Rsense, where Vcst =~.25V. The Vcst can be adjusted downward with the Iadj pin. The maximum value for Vcst (according to the datasheet) is .261V which is about a 4% variation off the nominal. So the maximum current set with the sense resistor shouldn't be able to wander more than about 4% without the sense resistor failing shorted.
The 3409 also has a built-in thermal shut down, but it may be at a higher temperature than one wishes. My thought is that one of the I2C commands will be to set the ATtiny25 to alert the master controller if a temperature threshold is exceeded, or perhaps simply to perform a shutdown. Or perhaps two commands, one to set each mode and limit. The ATtiny has a built-in temperature sensor tied to its ADC, which is very cool. There's a ton of functionality in that tiny 8 pin chip.
I will probably use it with 24V supplies. The *maximum* is 75V. One can still use it at lower voltages. I need to go through the equations and make sure that the components I have on my preliminary BOM yield sensible results in all the boundary operating modes. E.g. Vin = 6V, 1 LED; Vin = 75V, 1 LED; Vin = 75V, 18 - 19 LEDs. And perhaps some middle values as well such as, Vin = 24V, 3 LED or 6 LEDs. And Vin = 48V, 3, 6, 9 or 12 LEDs.
But this would give others (assuming anyone else is interested) the ability to use it with higher capacity supplies. If no one else is interested then I'll economize the parts by targeting them at my environment. But the savings from a moderate group buy, would exceed the savings of limiting the components' capacities.
I've looked at Boost. Partly because I have a dozen 12V 47A supplies on hand from old Compaq servers.
I do not like two things about boost. Most of the available components seem fairly limited in the output current they'll deliver, rarely getting up to 1.5A with any margin. Second, I do not like the inability to operate at voltages below Vin. I want one to be able to attach any number of LEDs (well, 2+) without concern that the driver will kill the LEDs. And have the ability to analog dim down very close to zero.
The Meanwells apparently have a minimum operating voltage of about 21V and more than one person has burnt out smaller strings of LEDs on the 48V version. I wonder if the 48V meanwells are actually a 24V supply with a boost circuit following the voltage regulator.
Now, in reality, 1A is probably plenty. But as I wrote, I'm aiming for versatile and heavy duty. It may be foolish. But this is where I'm starting.
It might be more reasonable to just aim at 24V and 1A as the maximum ratings. The history of my investigations has kind of led me to the maximum versatility version, though.
I like the 3409 topology. The non-HV version of the 3409 has a 42V limit. But 48V supplies are not unheard of. And the 3409HV is not that much more expensive than the non-HV. In larger quantities they're practically the same price. But once one has the 3409HV in the design, now the chip limit is 75V. If one chooses the other components for 24V or 48V operation, that's okay for personal use, but if others come along on a group buy they might get confused and think that 75V is the limit because of the 3409HV spec.
So why not go for a full out 75V (even though most folks will go 24V and a few will go 48V) 2A limit. Then with the switch settings to set the maximum current from 350mA to 2A it will be very easy for anyone to choose (roughly) their initial operating point and then adust that with a pot. on Iadj and/or the PWM through the ATtiny25.
I'm also considering connecting the potentiometer to an ADC pin on the ATtiny25 and a PWM with low pass filter to the Iadj pin of the 3409. Then the ATtiny25's EEPROM can be programmed with a flag which tells it to either use the value at the ADC pin to set Iadj, or to use a programmed value which was sent over the I2C. That way one could set the analog levels manually, or by controller depending on the setting of the flag.
But that's wandering over into a discussion of the command set for the ATtiny25 controller. Which is a discussion I'd like to have, but again, it's one I'm wondering if it belongs in this thread. I would love to see a group effort to create a sort of standard command set (via I2C) for the ATtiny25. Then I can whip up a program to implement the commands and we'll have a general control system for any PWM capable LED driver, not just the one I'm proposing here. It would be extremely versatile because up to 128 drivers could be controlled on a single I2C bus (yes, I know about capacitance limitation, *and* bus extenders).
So instead of a hodge podge of port expanders that are currently being used, we could have a standard client control system for LED drivers which any host control system (with an I2C interface) could easily access. The cost would be one ATtiny25 per driver, but the things are cheap, especially if one buys 100 at a time.
Last edited:
TheFishMan65
New member
and another don'tdo wht I did
and another don'tdo wht I did
Actually it is not terrible. I followed Kkcress's (I think) and ordered a nice power supply with power factor correction and a voltage adjust trim pot. My mistake was ordering off ebay - it was exactly as described. They just did not mention that the trim pot had been glued :sad1
probably loctite) so it is really fixed at 23.6 volts.
So if you order a used power supply make sure the trim pots are not glued.
Some loctite is breakable and I am working on figuring out how to break the seal. Worst case I guess I can replace the pot.
and another don'tdo wht I did
Actually it is not terrible. I followed Kkcress's (I think) and ordered a nice power supply with power factor correction and a voltage adjust trim pot. My mistake was ordering off ebay - it was exactly as described. They just did not mention that the trim pot had been glued :sad1
probably loctite) so it is really fixed at 23.6 volts.So if you order a used power supply make sure the trim pots are not glued.
Some loctite is breakable and I am working on figuring out how to break the seal. Worst case I guess I can replace the pot.
Taqpol
New member
Are those the ebay power supplies that DWZM posted on my thread? If so, I bought two of those and didn't check to see if the trim pots were glued...Actually it is not terrible. I followed Kkcress's (I think) and ordered a nice power supply with power factor correction and a voltage adjust trim pot. My mistake was ordering off ebay - it was exactly as described. They just did not mention that the trim pot had been glued :sad1
So if you order a used power supply make sure the trim pots are not glued.
Some loctite is breakable and I am working on figuring out how to break the seal. Worst case I guess I can replace the pot.
TheFishMan65
New member
I know where 7 of the 10 went now So far 4 are glued one is not and 2 are up in the air.
I tried to force one and pressed to hard with a screw driver. Oops it looks like at least one is getting a new potentiometer.
So far 4 are glued one is not and 2 are up in the air. I tried to force one and pressed to hard with a screw driver. Oops it looks like at least one is getting a new potentiometer.
WOW
I just read through this whole thread and am pretty excited but unfortunately still confused.
It sounds as if the CAT4101 would be the way I would want to go. I am pretty ignorant when it comes to electronics, but I can follow recipes and instructions, so I think I could probably assemble some of these.
I am a little shy about trying to set up the - what are they called Gerber files - that the pcb fab places want/need in order to produce your pcb's.
So, I am wondering if anyone out there has any of the 3 driver per board pcb's that DWZM designed that they are willing to sell or if anyone has any of the gerber files that seeed can use so I can have some made.
I do have a couple of concerns.
1. I am wanting to build something to cover about a 24 x 18 footprint, and am thinking about 12 cool whites and possibly up to 20 RB - most likely all XR-E's, unless someone has some decent optic choices for XP-G's. At any rate, I think 12 whites on 1 string should be fine with a 24v supply, but what about the RB's? If I went with either 2 strings of 10 or 1 string of 12 and 1 with 8, how would I handle adjusting the power supply voltage to properly handle the different sized strings? Would I need to use 2 power supplies?
2. For dimming I am a bit lost on the pwm. I was thinking that I wanted to use 0-10v dimming and was hoping that I could use my ACIII to do this, but have not done enough research, any input on this?
Thanks.
I just read through this whole thread and am pretty excited but unfortunately still confused.
It sounds as if the CAT4101 would be the way I would want to go. I am pretty ignorant when it comes to electronics, but I can follow recipes and instructions, so I think I could probably assemble some of these.
I am a little shy about trying to set up the - what are they called Gerber files - that the pcb fab places want/need in order to produce your pcb's.
So, I am wondering if anyone out there has any of the 3 driver per board pcb's that DWZM designed that they are willing to sell or if anyone has any of the gerber files that seeed can use so I can have some made.
I do have a couple of concerns.
1. I am wanting to build something to cover about a 24 x 18 footprint, and am thinking about 12 cool whites and possibly up to 20 RB - most likely all XR-E's, unless someone has some decent optic choices for XP-G's. At any rate, I think 12 whites on 1 string should be fine with a 24v supply, but what about the RB's? If I went with either 2 strings of 10 or 1 string of 12 and 1 with 8, how would I handle adjusting the power supply voltage to properly handle the different sized strings? Would I need to use 2 power supplies?
2. For dimming I am a bit lost on the pwm. I was thinking that I wanted to use 0-10v dimming and was hoping that I could use my ACIII to do this, but have not done enough research, any input on this?
Thanks.
TheFishMan65
New member
dwl,
At 24 volts you can only get 6 LEDs per string. 6 LEDs x 3.5 forward voltage = 21 volts. Each string should have about the same voltage drop or you may have heat problems.
Sorry not familiar enough with the ACIII to help.
At 24 volts you can only get 6 LEDs per string. 6 LEDs x 3.5 forward voltage = 21 volts. Each string should have about the same voltage drop or you may have heat problems.
Sorry not familiar enough with the ACIII to help.
I knew that I could only drive 6 leds per string at 24 volts... too much reading too quickly and not enough digestion I guess.
OK, so with 12 cool whites that is easy, 2 strings. For 20 RB though, the same question still applies, I would need to split this up with a different number of leds on at least 1 string, but lets assume I just went with 4 strings of 5 to get my 20. That would be something like 17.5 volts per string rather than the 21 volts that the strings of white would be.
I assume that I cannot adjust the power supply to ideal voltages for woth the whites and the blues, so I would need seperate power supplies, is that correct?
As for the ACIII I do not know what to do there either since the module that might work would most likely cost more than an arduino...
this is like knocking down dominoes...
OK, so with 12 cool whites that is easy, 2 strings. For 20 RB though, the same question still applies, I would need to split this up with a different number of leds on at least 1 string, but lets assume I just went with 4 strings of 5 to get my 20. That would be something like 17.5 volts per string rather than the 21 volts that the strings of white would be.
I assume that I cannot adjust the power supply to ideal voltages for woth the whites and the blues, so I would need seperate power supplies, is that correct?
As for the ACIII I do not know what to do there either since the module that might work would most likely cost more than an arduino...
this is like knocking down dominoes...
tahiriqbal
Sialkot
Why not :thumbsup:, the layout I have posted out was just to have a play around with PCB driver board. The new and improved layout will have 10 x XP-G (white)and 20 x XP-E (royal blue) on 140mm x 240mm PCB driver board. I am only leaving 140mm x 140mm space for 30 LEDs which I know many of you may think insane!! Once I finished my DIY project, I would be more than happy to publish any relevant information related to my DIY project which is still two months away.
stugray
Premium Member
tahiriqbal,
Very Good!
I am interested to see how these LEDs work with standard PCB vs. MCPCBs.
I honestly feel that not using a MCPCB for the base of the LED (like Stars) is a bad idea, but perhaps using the LEDs at less than Max current and adding a lot of thru-holes might bring the die temp within spec.
Only time & experimentation will tell....
Stu
Very Good!
I am interested to see how these LEDs work with standard PCB vs. MCPCBs.
I honestly feel that not using a MCPCB for the base of the LED (like Stars) is a bad idea, but perhaps using the LEDs at less than Max current and adding a lot of thru-holes might bring the die temp within spec.
Only time & experimentation will tell....
Stu
TheFishMan65
New member
dwl,
2 power supplies or some power resistors in series to drop some of the power. It also depends on the driver, but even buck pucks don't seem to like the heat. So I plan to drop as much power as I can anywhere but in the driver.
However if you mix white XP-G (5) with royal blue XR-E (6) you might have close enough voltage match.
There are probably some other options that someone will chime in with soon
2 power supplies or some power resistors in series to drop some of the power. It also depends on the driver, but even buck pucks don't seem to like the heat. So I plan to drop as much power as I can anywhere but in the driver.
However if you mix white XP-G (5) with royal blue XR-E (6) you might have close enough voltage match.
There are probably some other options that someone will chime in with soon
tahiriqbal
Sialkot
Agree, I guess we only learn from trail and error and only time will tell if it works or not!! On the positive side my new PCB, I have decided to go with extra holes behind the each LED which currently amount to 10. Maybe three rows of 10 holes with good copper sink which should help to reduce heat buildup and not to damage the PCB itself. You can also see two temperature sensors on my current PCB which would constantly monitor the LEDs and PCB temperature and increase or decrease fan flow accordingly. Fans will also take power from the same PCB with PWM signal and if any fan fails for some reason the fixture won't lit up and error message will be displayed on my touch screen controller.
TheFishMan65
New member
I am not using XP-G becuase of a lack of optics. Maybe someone else will chime in.
Hello all!! I have read all the page but i have some questions for you guys. If i want to setup 120 leds what is the best driver for my setup??? I will have around 60b and 60w. I want to have 4 sections each sections is 30 leds because i want to be able to dim left to right for sunrise and sunset. I want the less and a compact driver with less power supply possible like max 4 power supply. Thank you!!
