PCBWay 3D Printing and Fabrication: www.pcbway.com/rapid-prototyping/ Support the Channel with Patreon: www.patreon.com/sdgelectronics Uni-T UTi690B Thermal Imager ~£160 at Banggood: uk.banggood.com/custlink/3vKJrhrL22
Learning so much more here than the application and the IC’s alone. It’s a treasure trove of information packaged in a perfectly explained video. THANK YOU !
Steve, brilliant, I’ve been messing about with constant current driver circuits to be as simple as possible, didn’t know about these little devices, reckon it will solve a load of problems, and allow me to then dim my light as per your suggestions, thank you….
well, your voltage drop will be still there, so you must rise your power supply to compensate, which means that 1st CC drivers will be hoter. to make drop voltage same, you must power thestring from both end in way X-mass lights are powered, plus on one side, minus on other
Indeed, I had planned for power from both ends, just not every 4 PCBs. We have a bit more margin now with supply rails to ensure consistent illumination but of course those closer to the power entry will dissipate more power.
Nice follow-up on this project. Since you have a good setup going, any chance of testing up to 1kHz and even higher, just to find out where these regulators really give up? There are a few use cases for superfast switching.
Power dissipation across the 1st driver will still be a lot higher, though. Couple thoughts: 1) If you double the supply voltage (or better yet, use a 48V supply) and double the led string (12 leds in series) you can get away with double the length before voltage drop becomes an issue. the pcb layout would have to change, of course. You could either double the pcb length or have a settable resistor that can make the pcb either a series or parallel configuration. (edit: I checked pcbway and doubling the length 280 x 90 doesn't actually double the price, so it's cheaper this way, too! $98 for 280x90 vs $69 for 140x90) This will improve your overall efficiency 2) doubling the copper weight, doubles the price on pcbway, so that's a no go. 3) If you supply only the positive supply from one end, and only the negative from the other end, you'll get approximately the same voltage across all the LEDs strings. Overall power dissipation is the same, though.
Thanks for sharing cool part! 👍 Indeed this may be very useful. No need to create discrete transistor current limiting supply.. As I did it long ago for lit up the amplifier front panel as an retro update.. At that situation I'm dealing with 50V-60V and transistor needs separate heatsink to drive two leds.. 😂 Linear regulator ICs wasn't an option due to high input voltage - they can't handle it by datasheet..
Great video as so often. Nice little current control devices, although obviously linear and therefore not suitable for dropping much voltage. However, I am not sure why you are using these devices and then PWM dimming. If using a PWM MOSFET, then a couple more components (current sense resistor and an opamp - or a specialised chip) would give you both functions (ie control of max current and a nice linear dimming curve at any freq you want) - and without the linear heating losses (may not be an issue at these low power levels). Also you could now remove / bridge out the 1 Ohm resistors?
It's because there's 70 segments in total, current matching of all the boards wouldn't be trivial. If it was only a few strips I'd likely use an adjustable constant current regulator.
@@sdgelectronics Ahhh. 70 x 150mA in parallel(!).. Yeah that's harder, you would need PWM current control for each one or at least each small group. Fair enough.
@@oschonrock FET, shunt and opamp is still a linear current limiter. But these little ones allow for effortless voltage controlled dimming from PSU side.
@@oschonrock "PWM MOSFET, then a couple more components (current sense resistor and an opamp" Either I am not reading well in English today or there exists some opamp+fet circuit that works using PWM principle that I have no clue about..
See also the ON Semi NSI450xx range. Although those don't seem to be available in as wide a range of current, and they're only rated for 45V. These two-pin constant-current drivers are quite neat, and I was considering using them for something a while back, but the downside is that they are quite expensive. So I went with a dual transistor arrangement instead.
Don't make the mistake of trying to make a discrete current mirror, I did that once and had to respin boards with a DC/DC converter and floating opamps instead, it was for a 4-20mA output circuit controlling air supply in a ~3000kW total power gas burner set and the current mirror would go crazy with a slight change in temperature
so presumably... the higher the pwm freqency, going up into several khz... then the driver will not have enough time to fully switch on and get up to the full rated 150ma current. which would then limit the maximum achievable brightness? (if am understanding this properly). which might be ok depending if there is enough usable upper range of duty cycle still being preserved. and also depending by how much the leds are 'too bright' / too much @100% for the given application. normally i would prefer a duty cycle about 20khz. not to be within any audible range. but that is way far higher than what these are speccd at. also a 100hz or 120hz sits exactly on the same frequency as the ac mains hum. which is usually already being filtered. so maybe that would make more sense here instead
It depends on the total voltage you have to drop. Having additional resistors creates additional voltage drop and reduce the voltage drop in the current regulators. This reduced the power it has to dissipate, because this is shared by the resistor (s).
Instead of using PWM, modifying feedback circuit of AC-DC PSU for voltage control would be not much more work than modifying it for different voltage. As a result you get a smooth dimming without long antennas radiating EMI (TBH, I care about flicker-free aspect more than this). Other than that, you seem to get 87,6% efficiency in LED strip alone, so ~2.6V drop from these drivers is big limiting factor. They are linear after all, not sure why I expect more from them.
PCBWay 3D Printing and Fabrication: www.pcbway.com/rapid-prototyping/
Support the Channel with Patreon: www.patreon.com/sdgelectronics
Uni-T UTi690B Thermal Imager ~£160 at Banggood: uk.banggood.com/custlink/3vKJrhrL22
Learning so much more here than the application and the IC’s alone. It’s a treasure trove of information packaged in a perfectly explained video. THANK YOU !
I never knew these devices existed, every day is a school day👍
Never met these CCD's before, they're an excellent way of achieving your required result. Another part to my arsenal. Thanks for sharing Steve.
Steve, brilliant, I’ve been messing about with constant current driver circuits to be as simple as possible, didn’t know about these little devices, reckon it will solve a load of problems, and allow me to then dim my light as per your suggestions, thank you….
well, your voltage drop will be still there, so you must rise your power supply to compensate, which means that 1st CC drivers will be hoter. to make drop voltage same, you must power thestring from both end in way X-mass lights are powered, plus on one side, minus on other
Indeed, I had planned for power from both ends, just not every 4 PCBs. We have a bit more margin now with supply rails to ensure consistent illumination but of course those closer to the power entry will dissipate more power.
Nice follow-up on this project. Since you have a good setup going, any chance of testing up to 1kHz and even higher, just to find out where these regulators really give up? There are a few use cases for superfast switching.
Yes please!
Power dissipation across the 1st driver will still be a lot higher, though.
Couple thoughts:
1) If you double the supply voltage (or better yet, use a 48V supply) and double the led string (12 leds in series) you can get away with double the length before voltage drop becomes an issue. the pcb layout would have to change, of course. You could either double the pcb length or have a settable resistor that can make the pcb either a series or parallel configuration. (edit: I checked pcbway and doubling the length 280 x 90 doesn't actually double the price, so it's cheaper this way, too! $98 for 280x90 vs $69 for 140x90) This will improve your overall efficiency
2) doubling the copper weight, doubles the price on pcbway, so that's a no go.
3) If you supply only the positive supply from one end, and only the negative from the other end, you'll get approximately the same voltage across all the LEDs strings. Overall power dissipation is the same, though.
Nice "bodge" .. simple.. elegant .. hurrayy to Henrik :)
Great device I have never heard of....cheers.
Thanks for sharing cool part! 👍
Indeed this may be very useful. No need to create discrete transistor current limiting supply.. As I did it long ago for lit up the amplifier front panel as an retro update.. At that situation I'm dealing with 50V-60V and transistor needs separate heatsink to drive two leds.. 😂 Linear regulator ICs wasn't an option due to high input voltage - they can't handle it by datasheet..
The AL5809 is available in SOD123 (0.36W) and PowerDI123 (0.68-1.24W), it is the latter that is close to the 1206 footprint isn't?
You can compensate for the non linear dimming easily. You just have to take into account the initial ramp up.
Great video as so often. Nice little current control devices, although obviously linear and therefore not suitable for dropping much voltage.
However, I am not sure why you are using these devices and then PWM dimming. If using a PWM MOSFET, then a couple more components (current sense resistor and an opamp - or a specialised chip) would give you both functions (ie control of max current and a nice linear dimming curve at any freq you want) - and without the linear heating losses (may not be an issue at these low power levels). Also you could now remove / bridge out the 1 Ohm resistors?
It's because there's 70 segments in total, current matching of all the boards wouldn't be trivial. If it was only a few strips I'd likely use an adjustable constant current regulator.
@@sdgelectronics Ahhh. 70 x 150mA in parallel(!).. Yeah that's harder, you would need PWM current control for each one or at least each small group. Fair enough.
@@oschonrock FET, shunt and opamp is still a linear current limiter. But these little ones allow for effortless voltage controlled dimming from PSU side.
@@Mr.Leeroy "PWM MOSFET" was shorthand for a "duty cycle circuit". But that's not the reason. See SDG Electronics reply above.
@@oschonrock
"PWM MOSFET, then a couple more components (current sense resistor and an opamp"
Either I am not reading well in English today or there exists some opamp+fet circuit that works using PWM principle that I have no clue about..
See also the ON Semi NSI450xx range. Although those don't seem to be available in as wide a range of current, and they're only rated for 45V.
These two-pin constant-current drivers are quite neat, and I was considering using them for something a while back, but the downside is that they are quite expensive. So I went with a dual transistor arrangement instead.
Arrow had these for $0.26 last week, so worth it for minimal rework. I'll test out some other solutions for LED control
Don't make the mistake of trying to make a discrete current mirror, I did that once and had to respin boards with a DC/DC converter and floating opamps instead, it was for a 4-20mA output circuit controlling air supply in a ~3000kW total power gas burner set and the current mirror would go crazy with a slight change in temperature
Did you check the temperatures when driving at the higher pwm frequencies?
It's generally consistent. The switching MOSFET will be the part that gets hotter
Interesting never knew that exsists
Reminds me of JFET. It is giant JFET.
Excellent! Gave me the idea to replace mi burned 50A1 regulator tube from a Zenith TO, in series with an adequate resistor. Thanks!
so presumably... the higher the pwm freqency, going up into several khz... then the driver will not have enough time to fully switch on and get up to the full rated 150ma current. which would then limit the maximum achievable brightness? (if am understanding this properly). which might be ok depending if there is enough usable upper range of duty cycle still being preserved. and also depending by how much the leds are 'too bright' / too much @100% for the given application.
normally i would prefer a duty cycle about 20khz. not to be within any audible range. but that is way far higher than what these are speccd at. also a 100hz or 120hz sits exactly on the same frequency as the ac mains hum. which is usually already being filtered. so maybe that would make more sense here instead
At higher brightness the pulse will be longer so tho give the regulator more time. And of course at 100% there are no pulses anymore. 😉
Correct, at some point, the LEDs will barely illuminate up to high duty cycles where suddenly they will go very bright.
Are you removing all resistors? Now with the current regulators they do not serve a purpose, right?
I'm leaving the 1R resistors. None of the UK distributors had a reel of 0R links in 1206 at normal prices
It depends on the total voltage you have to drop. Having additional resistors creates additional voltage drop and reduce the voltage drop in the current regulators. This reduced the power it has to dissipate, because this is shared by the resistor (s).
@@sdgelectronics 0R jumpers seems like a perfect job for CN suppliers (:
@@cmuller1441 in this case he had already raised the voltage, since the regulator itself drops approximately 2.5 V
@@sdgelectronics I forgot they were only 1R
Interesting CCS
Instead of using PWM, modifying feedback circuit of AC-DC PSU for voltage control would be not much more work than modifying it for different voltage.
As a result you get a smooth dimming without long antennas radiating EMI (TBH, I care about flicker-free aspect more than this).
Other than that, you seem to get 87,6% efficiency in LED strip alone, so ~2.6V drop from these drivers is big limiting factor. They are linear after all, not sure why I expect more from them.