Having significant switching currents in long lengths of wire might cause oscillation and EMI problems. A gate resistor, like 3-10 ohms, will save your design. The gate pull-down resistor of 100k is quite high, I might use within 20k in this design .. The gate resistor also saves the driver transistors from exceeding their max when the gate cap starts charging
I amended the schematic slightly before ordering PCBs. I added the pull-downs before the gate driver rather than at the gate and did indeed add a position to add gate resistors. Thanks for the comment!
If you are going to use long cables I would also recommend a small simple RC snubber on the MOSFET output, at turn-off you will get a lot of ringing without one.
Well timed, I just sent a panel of 8x LED strips off to PCB way. My board was laid out almost exactly like yours; next step is exactly this, so thanks for the parts selection. Looking forward for the next one.
Great video Steve, nice to see your process of choosing which parts you're going to use and the thinking behind the choices. I love this style of video. Thanks for sharing.
It's worth mentioning that for similar sized FETs, devices with a higher max Vds rating tend to come with higher Rds on figures, so it's worth considering the minimum voltage rating you can get away with. Generally I avoid PWM as a method of power delivery unless it's constrained to a small area on a PCB, having significant switching currents in long lengths of wire sounds like an EMI problem.
In this case, the load is a constant-current LED driver (see previous videos). If you'd create an analog voltage, you will have a huge difference in intensity caused by the cable drop, which was the reason to use the constant-current drivers in the first place. Low-frequency PWM like this is also used a lot in automotive. They often just use use unshielded cables and the EMI requirements are quite strict, so I think this will work just fine.
@@tomvleeuwen The load is current limited in the on period of the PWM cycle, but the current is zero in the off period. This constantly changing current is the fundamental problem as it has significant energy at high frequencies, and the parasitic inductances and capacitances cause ringing. EMI standards are relative to the environment, a car is a very noisy environment already with alternators and spark plug discharges, the concern is whether one system will interfere with another. In a home environment there are different EMI concerns, for example PWM controlled lights or even some DC LED drivers with high AC ripple cause interference with oscilloscope and spectrum measurements, and cause noise on magnetic pickups and high impedance cabling, even shielding only helps to an extent. Your standards may vary for your application.
Yeah, my concern with that driver that it has been drawn as an emitter follower. So that simplification suggests that there will be no "level-shifting".
Nice video, it’s always good to see the thinking behind the design. The times I’ve looked at 150A power fets only to find that if you want to put it on a normal size heatsink you can’t pull anything close to its max, they seem to have strange ideas of real world properties of metals used for heatsinks.
I destroyed many in my youth with high-power experiments; even with all that thermal mass, you can burn out the die before the package or heatsink has started warming up!
@@sdgelectronics that's part of the fun working out all the thermal resistances. In the hope that the junction temperature won't go above the melting point of silcon 🙂
Be careful with Marketplace items on Digikey. Especially from Rochester, the MOQ is frequently much greater than one. I tend to always check "in-stock", "hide Marketplace items" and then pick part status "active" for any Digikey search.
very great video thank you. learned a lot here to better understand these considerations for mosfet. and how to better read those graphs in a typical mosfet datasheet... before had been feeling just a bit confused. also the gate drivers... never knew they existed! seems pretty handy. thanks you so much, loved it!
I hope you get no audiable issues. (with smps they are not granted). I always tend to avoid 3kHz switching like plague, because ears are the most sensitive there. 3 interlaced pwm at 1kHz = 3kHz. Of course, you can always adjust the frequency to say 250kHz to make any issues much less of a problem. To be safe, I use a design criteria to keep everything much below 3kHz or well above audiable frequencies.
Check the gates with an oscilloscope for oscillations ( "ringing" ). Not using gate resistors might cause some trouble ( increased EMI emissions at least ). Even small, 10 ohm gate resistor can reduce ( dampen ) any gate driving circuit oscillations.
I would have the mosfets by quantity available and then filter by voltage and rdson and the other specs ... in theory those available in higher quantities are more common and easier to source from multiple stores... you picked that GOFORD brand mosfet in crappy package (power dissipation wise) ... For example AON2260 and AON2290 seem much better and they're 50 cents each and there's 200k+ of them in stock... and if it's about price, TME has AOD4184A or AOD4186 (max 40v) both to252 packages at 15-20 euro cents, if you order 20...
@@sdgelectronics It's not only about oscillations. Gate capacitance acts like a dead short when driver changes it's state. Current that flows then is much higher than the driver's max Ion/Ioff. Makes it getting warmer, in some cases might bring it's VCC too low, causing more oscillations, and in extreme cases damage the chip. Gate resistors are especially important when the MOSFET is driven straight from a MCU, which usually can't source/sink more than 20...40mA.
I'm wee bit confused by your 60 x 0.15A calculation; Each 11x140mm board draws 150mA (regulated by the AL5809, so for 60x are you powering 60x boards? The series "string" of 6x LEDs only sees 150mA right? I'm not sure how it would see 6x150mA with the CC regulator in place? **One possibility is having 10x boards in parallel, that's 60 LEDs with each 6-segment getting 150mA; Still each board is regulated so the Isupply would be 150x10 = 1500mA, only 1.5A. No where close to 9A. Hmm...
@@jaro6985 Single board has a CC AL5809 - this limits max current to 6 LEDs to 150mA. This means the entire PCB can only draw 150mA (shared across 6x LEDs). Since it's series each LED gets 150mA - and the Vsupply is just 150mA too. So yeah, 60x PCBs = 360 LEDs.
@@bsodmike All the PCBs are in parallel to keep the overall voltage below the 50 DC limit for installation in the bathroom, so it's 60 PCBs, all in parallel, drawing 24V*0.15A
you selected "half bridge" in digikey, thats why that driver didnt show up. half bridge drivers are for ... half bridge mosfet configurations obviously. you should have selected "low side" instead.
Calling the ZXGD3009... a gate driver is a joke, ...obviously! ...not to mention the lack of voltage level shifting, ...obviously! ...get a proper gate driver please! :(
PCBWay 3D Printing and Fabrication: www.pcbway.com/rapid-prototyping/
Support the Channel with Patreon: www.patreon.com/sdgelectronics
Having significant switching currents in long lengths of wire might cause oscillation and EMI problems. A gate resistor, like 3-10 ohms, will save your design. The gate pull-down resistor of 100k is quite high, I might use within 20k in this design .. The gate resistor also saves the driver transistors from exceeding their max when the gate cap starts charging
I amended the schematic slightly before ordering PCBs. I added the pull-downs before the gate driver rather than at the gate and did indeed add a position to add gate resistors. Thanks for the comment!
@@sdgelectronics Well done, pads do not have to be populated and save you PCB revisions.
If you are going to use long cables I would also recommend a small simple RC snubber on the MOSFET output, at turn-off you will get a lot of ringing without one.
Well timed, I just sent a panel of 8x LED strips off to PCB way. My board was laid out almost exactly like yours; next step is exactly this, so thanks for the parts selection. Looking forward for the next one.
Thank you for the lesson on component selection, Steve!
Great video Steve, nice to see your process of choosing which parts you're going to use and the thinking behind the choices. I love this style of video. Thanks for sharing.
Thanks for the informative video and for every ones great comments!
Thoroughly enjoyed this design walk thru....cheers.
It's worth mentioning that for similar sized FETs, devices with a higher max Vds rating tend to come with higher Rds on figures, so it's worth considering the minimum voltage rating you can get away with. Generally I avoid PWM as a method of power delivery unless it's constrained to a small area on a PCB, having significant switching currents in long lengths of wire sounds like an EMI problem.
In this case, the load is a constant-current LED driver (see previous videos). If you'd create an analog voltage, you will have a huge difference in intensity caused by the cable drop, which was the reason to use the constant-current drivers in the first place.
Low-frequency PWM like this is also used a lot in automotive. They often just use use unshielded cables and the EMI requirements are quite strict, so I think this will work just fine.
@@tomvleeuwen The load is current limited in the on period of the PWM cycle, but the current is zero in the off period. This constantly changing current is the fundamental problem as it has significant energy at high frequencies, and the parasitic inductances and capacitances cause ringing. EMI standards are relative to the environment, a car is a very noisy environment already with alternators and spark plug discharges, the concern is whether one system will interfere with another. In a home environment there are different EMI concerns, for example PWM controlled lights or even some DC LED drivers with high AC ripple cause interference with oscilloscope and spectrum measurements, and cause noise on magnetic pickups and high impedance cabling, even shielding only helps to an extent. Your standards may vary for your application.
Yeah, my concern with that driver that it has been drawn as an emitter follower. So that simplification suggests that there will be no "level-shifting".
Yes, this is possibly an error on my part
Nice video, it’s always good to see the thinking behind the design. The times I’ve looked at 150A power fets only to find that if you want to put it on a normal size heatsink you can’t pull anything close to its max, they seem to have strange ideas of real world properties of metals used for heatsinks.
I destroyed many in my youth with high-power experiments; even with all that thermal mass, you can burn out the die before the package or heatsink has started warming up!
@@sdgelectronics that's part of the fun working out all the thermal resistances. In the hope that the junction temperature won't go above the melting point of silcon 🙂
Be careful with Marketplace items on Digikey. Especially from Rochester, the MOQ is frequently much greater than one.
I tend to always check "in-stock", "hide Marketplace items" and then pick part status "active" for any Digikey search.
very great video thank you. learned a lot here to better understand these considerations for mosfet. and how to better read those graphs in a typical mosfet datasheet... before had been feeling just a bit confused. also the gate drivers... never knew they existed! seems pretty handy. thanks you so much, loved it!
I hope you get no audiable issues. (with smps they are not granted). I always tend to avoid 3kHz switching like plague, because ears are the most sensitive there. 3 interlaced pwm at 1kHz = 3kHz. Of course, you can always adjust the frequency to say 250kHz to make any issues much less of a problem. To be safe, I use a design criteria to keep everything much below 3kHz or well above audiable frequencies.
Fortunately, the power supply is in the attic, so even if it makes noise, it'll not be heard in the house.
Check the gates with an oscilloscope for oscillations ( "ringing" ). Not using gate resistors might cause some trouble ( increased EMI emissions at least ). Even small, 10 ohm gate resistor can reduce ( dampen ) any gate driving circuit oscillations.
I would have the mosfets by quantity available and then filter by voltage and rdson and the other specs ... in theory those available in higher quantities are more common and easier to source from multiple stores... you picked that GOFORD brand mosfet in crappy package (power dissipation wise) ... For example AON2260 and AON2290 seem much better and they're 50 cents each and there's 200k+ of them in stock... and if it's about price, TME has AOD4184A or AOD4186 (max 40v) both to252 packages at 15-20 euro cents, if you order 20...
It looked like you were being quite cost conscious and then suddenly $5 jumps to $45 for the gold finish :)
Yea, a 9 x increase in price was descried as a small increase :D
Thank you, that is very interesting! Shouldn’t we protect the gate of the MOSFETs with a zener diode (although it might add a bit of capacitance) ?
No, the fet gate can handle +/- 20V
Isn't the gate resistors required anyway, even if one wants the maximum current?
Not on a MOSFET. You can use a gate resistor to help with ringing, though.
@@sdgelectronics It's not only about oscillations. Gate capacitance acts like a dead short when driver changes it's state. Current that flows then is much higher than the driver's max Ion/Ioff. Makes it getting warmer, in some cases might bring it's VCC too low, causing more oscillations, and in extreme cases damage the chip. Gate resistors are especially important when the MOSFET is driven straight from a MCU, which usually can't source/sink more than 20...40mA.
@@sdgelectronics you will always get ringing with a capable driver, so it is essential.
So calc its value at least for Imax of the driver.
I'm wee bit confused by your 60 x 0.15A calculation; Each 11x140mm board draws 150mA (regulated by the AL5809, so for 60x are you powering 60x boards? The series "string" of 6x LEDs only sees 150mA right? I'm not sure how it would see 6x150mA with the CC regulator in place?
**One possibility is having 10x boards in parallel, that's 60 LEDs with each 6-segment getting 150mA; Still each board is regulated so the Isupply would be 150x10 = 1500mA, only 1.5A. No where close to 9A. Hmm...
60 boards * 0.15A is 9A. Its not 60 LEDs its 360 right.
@@jaro6985 Single board has a CC AL5809 - this limits max current to 6 LEDs to 150mA. This means the entire PCB can only draw 150mA (shared across 6x LEDs). Since it's series each LED gets 150mA - and the Vsupply is just 150mA too. So yeah, 60x PCBs = 360 LEDs.
@@bsodmike All the PCBs are in parallel to keep the overall voltage below the 50 DC limit for installation in the bathroom, so it's 60 PCBs, all in parallel, drawing 24V*0.15A
@@sdgelectronics thanks. That makes a whole lot of sense!! Would be great if you could share shots of the installation. It must be pretty large!
you selected "half bridge" in digikey, thats why that driver didnt show up.
half bridge drivers are for ... half bridge mosfet configurations obviously.
you should have selected "low side" instead.
Calling the ZXGD3009... a gate driver is a joke, ...obviously!
...not to mention the lack of voltage level shifting, ...obviously!
...get a proper gate driver please! :(