noelex
Well-known member
Good LED drivers have efficiencies of about 90-95%. They are available as pre-built modules,
Without a driver you can have good brightness, or a long life, take your pick.
http://www.cutter.com.au/proddetail.php?prod=cut131
DIY LED Bulb Project PBO April 2011 P88
- Thread starter nigelmercier
- Start date
Good LED drivers have efficiencies of about 90-95%. They are available as pre-built modules,
Without a driver you can have good brightness, or a long life, take your pick.
http://www.cutter.com.au/proddetail.php?prod=cut131
Rigger Mortice
N/A
I agree with you about the brightness/life compromise without some form of regulation but you really are a trouble maker! SHMBO is not happy with a kitchen table which has been covered with regulators, switch mode supplies and unregulated LED arrays for a couple of weeks. She is now going to have to wait until I get hold of one of your BuckPucks to play with. 
Interestingly drawings 5 & 6 on the data sheet show just how I was suggesting arrays be built up.
Interestingly drawings 5 & 6 on the data sheet show just how I was suggesting arrays be built up.
William_H
Well-known member
Resistors for LEDs
OK it is all about the math and the different voltages you supply with.
Resistors for current limiting can be excellent and simple but are better for higher voltage supplies.
Take 5 LEDs in series assume 2.25 v drop in each LED. (This can vary) With a 14v supply as in battery on charge you need to drop (5X 2.25) 11.25v from 14v is 2.75 volts> assume a current for the LEDs of 25ma. R+ E/I so resistance needed is 110 ohms.
Now when the voltage drops to 12v battery not on charge. current becomes .75 volts across 110 Ohms is about 6.5 ma. So LEDs will give only quarter brightness /power.
However if you sacrifice efficiency by using 4 LEDs in series you need 200 ohm series resistance for 25ma at 14v. At 12v input LED current is 15ma. or 60% of original power. However with one less bulb you have 20% less brightness regardless of input voltage.
Now using a L200C regulator you will find from the specs that min voltage drop through the regulator is 2.5 volts and you may need another .5 volts dropped in the current sensing resistor. So at best you have 11v available for LEDs at 14v input. While at 12v input you have just 9volts available for LEDs.
Using the above example 9v will only drive 4 LEDs in series. (perhaps less). So you will have no better efficiency with a linear regulator. However you will be able to set it up for max brightness/ power at lowest expected battery voltage (11v would be a good number) and have automatic power limiting if battery voltage is higher. (with more power wasted in regulator)
Now as light efficiency is so good compared to incandescent you might be happy to lose some efficiency with less bulbs in series so resistive current limiting can be attractive for its simplicity. I would use a separate resistor for each string because if one string becomes open then using a common resistor the remaining strings will get too much current. Likewise using a current regulator but you may be willing to risk because of cost of additional regulators.
You can also see that LEDs are good on dry cell batteries as voltage is more stable so max power /light on good battery that then declines as cell discharges.
Or as said use a switching current controller for best efficiency but risking radio noise interference. These will usually have all LEDs in series by producing a high voltage.
good luck olewill
OK it is all about the math and the different voltages you supply with.
Resistors for current limiting can be excellent and simple but are better for higher voltage supplies.
Take 5 LEDs in series assume 2.25 v drop in each LED. (This can vary) With a 14v supply as in battery on charge you need to drop (5X 2.25) 11.25v from 14v is 2.75 volts> assume a current for the LEDs of 25ma. R+ E/I so resistance needed is 110 ohms.
Now when the voltage drops to 12v battery not on charge. current becomes .75 volts across 110 Ohms is about 6.5 ma. So LEDs will give only quarter brightness /power.
However if you sacrifice efficiency by using 4 LEDs in series you need 200 ohm series resistance for 25ma at 14v. At 12v input LED current is 15ma. or 60% of original power. However with one less bulb you have 20% less brightness regardless of input voltage.
Now using a L200C regulator you will find from the specs that min voltage drop through the regulator is 2.5 volts and you may need another .5 volts dropped in the current sensing resistor. So at best you have 11v available for LEDs at 14v input. While at 12v input you have just 9volts available for LEDs.
Using the above example 9v will only drive 4 LEDs in series. (perhaps less). So you will have no better efficiency with a linear regulator. However you will be able to set it up for max brightness/ power at lowest expected battery voltage (11v would be a good number) and have automatic power limiting if battery voltage is higher. (with more power wasted in regulator)
Now as light efficiency is so good compared to incandescent you might be happy to lose some efficiency with less bulbs in series so resistive current limiting can be attractive for its simplicity. I would use a separate resistor for each string because if one string becomes open then using a common resistor the remaining strings will get too much current. Likewise using a current regulator but you may be willing to risk because of cost of additional regulators.
You can also see that LEDs are good on dry cell batteries as voltage is more stable so max power /light on good battery that then declines as cell discharges.
Or as said use a switching current controller for best efficiency but risking radio noise interference. These will usually have all LEDs in series by producing a high voltage.
good luck olewill
pteron
Member
You (or your kids) almost certainly have a mobile device that uses a white LED backlight driven by a controller that I either designed or managed the design of.
The *only* way to be efficient and control the light output is to use a switching regulator.
White LEDs have massively varying forward voltage drops, and unless you buy them by the tens of millions you'll find that Nichia et al are not interested in binning them by voltage for you. So your circuit has to cope with the total drop over your 3 diodes varying from 9V (3 x 3V) to 10.8V (3 x 3.6V) or more. In addition the voltage drop varies with temperature, so let's say our voltage varies from 8.5V to 11.5V at 20mA current drive.
Your input voltage could be 11.8V to 14V or more depending on the state of charge of your battery.
To set the current correctly at worst case, (11.8V input, 11.5V forward drop), you need a resistor (or a linear regulator of any kind which looks like a resistor for this calculation) of
V = IR
(11.8 - 11.5)/20e-3 = 15 Ohms
At the other extreme we get
(14 - 8.5)/15 = 366 mA i.e. we've just blown our diodes up.
LEDs are *current* driven devices, you control them by setting the current and ensuring that your driver has just enough headroom to supply the needed volts.
In our case above, you design a buck converter that measures the current through the diode and sets the voltage to be just above the voltage needed. Typically you need about 200mV to measure the current, so:
11.8V input, 11.5V needed across the diodes, you drop the 11.8V to 11.7V in the buck converter and the current source sets the current to 20mA. Efficiency overall is:
270mW input, 230mW in the diodes, 4mW in the current source, so 84% efficient.
14V input, 8.5V needed? The buck drops the voltage to 8.7V at about 85 to 90 % efficiency. The current source then sets the current to 20mA. Efficiency overall is:
204mW input, 170mW in the diodes, 4mW in the current source, so 82% efficient.
Compare this to the resistor case (assuming we fix the resistor, making it 275Ohms so the diodes don't blow):
At 14V input: 280mW in, 170mW in the diodes, 110mW in the resistor, 44% efficient.
At 11.8V input your 275 Ohm resistor has dropped the current to 1mA so you have no light.
So the switcher is almost twice as efficient and has a constant brightness over voltage and temperature variation. The resistor approach causes massive variations in brightness and is less efficient.
That's why your mobile device does not use a resistor to set the LED current.
If you want more light, you add parallel strings and control the current in each one or you boost the voltage and use a single string of more LEDs with one current controller.
If anyone tries to sell you an LED bulb that does not have a purpose designed switching regulator in it, you are giving up consistency of light output or efficiency or both.
The *only* way to be efficient and control the light output is to use a switching regulator.
White LEDs have massively varying forward voltage drops, and unless you buy them by the tens of millions you'll find that Nichia et al are not interested in binning them by voltage for you. So your circuit has to cope with the total drop over your 3 diodes varying from 9V (3 x 3V) to 10.8V (3 x 3.6V) or more. In addition the voltage drop varies with temperature, so let's say our voltage varies from 8.5V to 11.5V at 20mA current drive.
Your input voltage could be 11.8V to 14V or more depending on the state of charge of your battery.
To set the current correctly at worst case, (11.8V input, 11.5V forward drop), you need a resistor (or a linear regulator of any kind which looks like a resistor for this calculation) of
V = IR
(11.8 - 11.5)/20e-3 = 15 Ohms
At the other extreme we get
(14 - 8.5)/15 = 366 mA i.e. we've just blown our diodes up.
LEDs are *current* driven devices, you control them by setting the current and ensuring that your driver has just enough headroom to supply the needed volts.
In our case above, you design a buck converter that measures the current through the diode and sets the voltage to be just above the voltage needed. Typically you need about 200mV to measure the current, so:
11.8V input, 11.5V needed across the diodes, you drop the 11.8V to 11.7V in the buck converter and the current source sets the current to 20mA. Efficiency overall is:
270mW input, 230mW in the diodes, 4mW in the current source, so 84% efficient.
14V input, 8.5V needed? The buck drops the voltage to 8.7V at about 85 to 90 % efficiency. The current source then sets the current to 20mA. Efficiency overall is:
204mW input, 170mW in the diodes, 4mW in the current source, so 82% efficient.
Compare this to the resistor case (assuming we fix the resistor, making it 275Ohms so the diodes don't blow):
At 14V input: 280mW in, 170mW in the diodes, 110mW in the resistor, 44% efficient.
At 11.8V input your 275 Ohm resistor has dropped the current to 1mA so you have no light.
So the switcher is almost twice as efficient and has a constant brightness over voltage and temperature variation. The resistor approach causes massive variations in brightness and is less efficient.
That's why your mobile device does not use a resistor to set the LED current.
If you want more light, you add parallel strings and control the current in each one or you boost the voltage and use a single string of more LEDs with one current controller.
If anyone tries to sell you an LED bulb that does not have a purpose designed switching regulator in it, you are giving up consistency of light output or efficiency or both.
nigelmercier
RIP
- Joined
- 20 Jun 2007
- Messages
- 16,234
- Location
- Live in Kent, boat in Canary Islands
I've searched for months for a simple to use buck converter that will drive a long string of LEDs, say 20 or so. The only ones I have found need a zillion other components to make them work, and have far more features than I need (such as dimming).
Any pointers?
pteron
Member
None that spring to mind - I have the luxury of access to TSMC!
You want a boost converter controller with an external FET driver and a sense voltage of about 200mV. You'd need a couple of caps, the FET, a coil, diode and the resistor. Try the usual suspects - Maxim, Linear, TI, STM.
noelex
Well-known member
Any serious LED experimentation should be conducted on the dinning room table. Reserve the kitchen table for final assembly and testing.I agree with you about the brightness/life compromise without some form of regulation but you really are a trouble maker! SHMBO is not happy with a kitchen table which has been covered with regulators, switch mode supplies and unregulated LED arrays for a couple of weeks. She is now going to have to wait until I get hold of one of your BuckPucks to play with.
Interestingly drawings 5 & 6 on the data sheet show just how I was suggesting arrays be built up.
Graham_Wright
Well-known member
It's a long time since I dabbled with analogue electronics but the word "varistor" has just emerged from the mire that used to be a useful brain.
I have noticed with leds used in vehicles and sign lighting, that there can be a very disturbing flicker in peripheral vision presumably from switching circuits. Why is the frequency so low?
I have noticed with leds used in vehicles and sign lighting, that there can be a very disturbing flicker in peripheral vision presumably from switching circuits. Why is the frequency so low?