The 2nd circuit you showed was the one i used during the IC shortage after the pandemic. Worked really well for the motor controller i was tasked with designing.
thanks for these valuable practical electronic lessons, it was beneficial, please make more of such videos in which every circuit is validated with real test
Thank you for the lesson!! I'm new to electronics and I was searching for a solution on high side switching for a week but I couldn't find such a good explanation. I'd like to see a video explaining the UDN2981 or M54563P ic's!
Thanks you for the detailed explanation. I just want to add a comment that MOSFET turn on and turn off speed is the time the transistor changes state. In most of the video, you are referring to the shifter delay as turn or and off time but this is just the shifter delay and not the turn on/off time. While important to prevent shoot through and ringing, the delay time is not as critical as the time the mosfet spends not fully on or off.Power dissipation is current x voltage and partially open time is where most of power loss is generated. One of the most important characteristics of the mosfet driver is how fast it can switch the transistor. In that respect the slope of the open/close curve is more important than the shifter delay. If you measure it in your experiment you will probably find out that the IC does that best and will result in least power loss.
very well explained, even though i am in mechanical field and learning electronics now. i can understand your video at a good pace. Thank you very much.
I was always under the impression that the energy dissipation in the Mosfet was due to the speed of the actual switching of the fet and had nothing to do with the delays being on or off. Remember if there is a delay of what ever time say 500ns the fet is still dissipating no energy because it is either full on or full off. It is only at the time that it is actually switching from full on to full off (or vias versa) that the power is dissipated. In your tests the actually switching times are 200ns (using IR2104 circuit) and that is what should be used for the power dissipation calculation. Basically the time of the slope of the mosfet switching is what should be used for the power dissipation. And this is based on the speed of charging the gate capacitance of the mosfet. In the advanced transistor driver and the IR2104 a 10 ohm resistor was used and similar actual switching times of the moset resulted. In the basic version the low current of the 1K resistor slowly charged the gate capacitance resulting in the slow actual turn on of the mosfet. Am I wrong on my assumptions?
Your videos explaining circuit operation are some of the best available, free or paid. If you’re excepting requests, will your touch on the topic of the “Slayer Exciter” circuit? It’s used to drive Tesla Coils in lieu of a spark gap circuit. I ask, because in studying data sheets of N-channel Mosfets, I found that the slayer circuit closely resembles a test circuit in the back of many datasheets (an unclamped inductive load) I’m still new to field/hobby of electronics. I’m pushing 50 and not without experience. I’ve built several Tesla Coils that use a single MOFET Slayer Exciter that were functional. I’ve seen a few simulations of the circuit but none with the in-depth explanations that make your videos stand out from the rest. Thanks for your time and the videos!!!
Thank you! I know the Slayer Exciter, and it is not a metal band from the 80's. I have no experience but I expect you can use the low side MOSFET driver push pull part to make your slayer exiter more efficient. I did see that for Tesla coils it is key to drive it pulsed, e.g. with 5-10Hz PWM with low on cycle (5% or so) to get nicer sparks and prevent burning your MOSFET. You can use the arduino for that (or a 555....) So yes I am interested and if I find the time I will build one myself 🙂
Thanks for explaining the working and differences of the circuits. Also the simulations were quite interesting and helped to better understand the circuits. I also liked your Kicad video😊
Photovoltaic mode optocouplers floated on the source will also work. LEDs will exhibit PV effect if illuminated by same colour LED, so it's fairly easy to "dead bug" such an arrangement for experimentation.
Very interesting! I was hoping to see a comparison with the simpler P channel high side switch circuit, particularly since that one can be driven at zero Hz. I'll make that my suggestion for a future video :)
Nice video, good explanations. But a comment on the use of "current sink" (BJT with emitter resistor) to drive the boot-strap circuits. The **minimum** voltage achievable at the collector of the level-shifter BC547 (call this Q1) is about 4.5V, due to the output voltage from the Arduino, and the Vce (sat) of Q1. Add to this the Vbe of the pull-down PNP in the buffered boot-strap, means that the mimum voltage at the gate of the MOSFET will be about 5.2V. In most cases this should be sufficiently low to ensure the MOSFET stays OFF, but not always. Is there any particular reason you did not just connect the emitter of Q1 directly to GND, and insert a suitable resistor in series with its collector if current or power dissipation became a concern?
I wish I found this a year ago, but being able to switch high side with pwm using cheap components is super useful. Still struggling to understand the transistors, so I'll be reviewing that literature...
Hello! Interesting video! I made experiments with discrete (only transistors) class D amplifiers. In my opinion most good design of a gate driver is the " UM10155 Philips application note" . There are Q12, Q10 is a high side gate driver. The transistor Q8 turns on/off this driver.You can replace it with usual BJT transistor , but the maximum CB voltage can be greater that your power positive bus (in this video -24V). To decrease current consumption it may be a low power MOSFET, for example 2n7000.
The "UM1055" app note circuit provides no provision against C5+C6 overdischarge so insufficient gate drive of the top mosfet, e.g. when the top transistor is requested to be ON for way too long (in case of the amplifier, when the input signal overdrives it towards clipping on the top rail - so pretty much normal operation unless the input has properly set and even supply tracking limiter). It is even worse as the Q8 draws current from the C5+C6, so is actively discharging them, so the critical maximum ON time is not that long (barely covering the normal switching frequency).
@@annaplojharova1400 So , we need simple forward one-transistor isolated converter to supply high side - if you want to turn on high side during long time.Or we can use a usual half-bridge gate driver IC and supply high side (without bootstrap diode) with mentioned one-transistor forward self-oscillating isolated DC/DC converter.
@@andrewandrosow4797 You can use the simple bootstrap, but the predriver has to provide protections that ensure the bootstrap capacitor gets recharged (e.g. by forcing an off pulse in the class D amplifier or the buck converter examples) and/or the transistor gets shut down properly if it can not handle the current with the available gate drive (in fact this could be done by a simple 2-transistor, few resistors and a capacitor addon circuit monitoring the ON state Vds, if we want to stay with "simple" discretes). Or use one of the many available gate driver ICs, which do have at least the bootstrapped supply monitoring build in (that is enough, when either the current is monitored elsewhere, and a direct short circuit is not an issue, like e.g. in an active speaker box amplifier). Here I see the gate driver IC just as the simplest practical solution, offering at least the minimum protection (essentially against input undervoltage for the buck converter, or the signal overdrive for the amplifier). There are also ICs that provide the overload protection (either the large Vdson, or current sense, or both).
@@annaplojharova1400 I had made in past the class D amplifier using this application note.There also was simplest protection against short-circuit. The amplifier works well. I am totally agree - an IC gate driver much reliable and robust that discrette driver.Mosfets much resistant against avalanche breakdown when voltage becomes high.
In my experience, for the discrete bootstrap push-pull it’s best to use the common base (or better yet, common gate) mode for the level-shift transistor. I tend to use the jellybean 2N7002 for this purpose. No extra caps & resistors needed, much shorter rise & fall time, and the logic levels non-inverted. It’s especially useful when your microcontroller/logic chip/comparator/whatever has an open drain/collector output. Penalty: it has to sink both the pull-up resistor’s and the PNP base current, which may be too much for some low-power chips. At voltages higher than 24V it becomes necessary to replace the pull-up resistor with a current source, which further improves the performance.
Interesting, do you mean a circuit as mentioned in this discussion (2nd schematic, scroll down). Never tried that! electronics.stackexchange.com/questions/296879/logic-level-converter-using-transistors
@@smartpowerelectronics8779 Pretty much yes, but as I’ve said, I tend to use a low-power MOSFET for this, just for the convenience (no need for a base current limiting resistor). Gate is connected directly to Vdd (+5V or +3.3V, whatever the chip is powered by), and source to the output of the chip. The typical Vgs ON of these FETs is abt. 1.5-2V, so it needs to be either a rail-to-rail output or open drain. Might not work with the old TTL logic. HIGH level at the output of the chip turns the power MOSFET ON, so there’s no inversion like with the common-emitter level shifter.
Thank you for an interesting presentation sir. For me it's not a problem to have another voltage source in a system, but I see many examples in a real life (a hybrid solar controller for instance) when people actively use a transformer gate driver. Then they can have one powerful low voltage source and control a MOSFET bridge. But I see no other advantage with the transformer. Because it's hard to manage a dead time in this case if a 1:1:1 transformer is used. And ferrite cores are quite expensive (pretty comparable with a drive IC price) Could you touch that in one of your further videos pls?
I have never encountered the 2nd circuit in your demo. So I learned a new thing! Also, I wonder if the discrepancy in the turn on time measured vs simulated could be due to the way you are generating the 12v. Would be interesting to see if the 12v rail is bouncing during turn on transitions. Its not just the gate that needs to be charged. Also the base current of the high side transistor will pull the 12v down slightly. I will probably make this circuit myself to investigate since it is entirely new to me.
Great to hear you learned something william4749. For practical use I do recommend an IC, much easier to build and it also has some protection against low voltage etc. ....but circuit2 works OK
Wow great tutorial. Can I ask about the #3 level shift circuit if the R-load/output is paralleled to high capacitance capacitor can it still charge the bootstrp capacitor during off state of HO switch while the output capacitor still have some charge on it if VS need to be in GND level voltage when High side is off to charge bootcap.
Question is if you use higher voltage like 400 volts DC at the high side drain then need more than 400 volts at the gate source to drive the load at the source and the limit voltage for gate source is not More than 20 volts DC how does this bootstrap works for high voltage at the source you may want to make a video for that.
@@doost6233 the circuits can work with 400V dc, you will need to replace the “level shifting “ bc547 transistor with a high voltage mosfet. The IC circuit works for 400 volt.
They are different indeed. delay will not cause any switching losses because the MOSFET is just waiting. The speed is very important because during the slopes/transitions you have current and voltage at the same time = power loss...
I have a question as I'm not an expert in power electronics. Could you make a separate charge pump circuit with its own oscillator and switching transistor that's permanently on and is solely devoted to generating bootstrapping voltage for the high side MOSFET? I reckon it would allow you to keep it switched as long as you like since putting the charge pump directly besides the MOSFET only generates the bootstrapping voltage for a brief moment
To my surprise, I found from digikey some P-Channel mosfets that don't cost that much but are similar to the N-Channel RDS(on) levels. E.g. G040P04T has less than 3.7mΩ and is only $1.82. The G050PO3T has less than 5mΩ costing $1.17.
I use Simetrix/SIMPLIS (www.simetrix.co.uk/) the free version, however I would recommend you to use LTSpice which has much more users and tutorials online. I change to LTSpice now. For nice looking schematics, I use KiCAD.
Not bad. It would help greatly though if you could explain the defensive design techniques for preventing false triggering on the high side due to fast dv/dt at the output e.g. when driving a (half-)bridge.
Thank you p0lyglot, you must be a professional in this field. The IR2104 has built-in logic and non-overlap time for that. For the discrete solution you will need to build in the dead time and logic in your logic signals. Half bridges are tricky for beginners, especially prevention of capacitive mode and hard switching are complex to manage.
The false triggering thing isn't a dead-time issue, but a matter of physical design, parasitics and low-side turn-on speed. It's not a shortcoming of this high-side video but if you do a half-bridge or full-bridge video, it'd be worth including some additional considerations for high-side driver design. The problem is that when the low side comes on, the high-side source voltage goes down very fast. Any stray C between high side gate and a power supply rail mean that the high-side driver has trouble "following" the FET (keeping a fixed low Vgs) so suddenly you have a bump in Vgs, the high side turns on and all the smoke gets out.
If you search the datasheet of the IRFZ44 you can find the Thermal resistance is "60 K/W". You can then calculate the temperature rise: T= Tamb + Powerloss*Rthermal , in this case 25C + 0.85*60 = 76 C. A TO220 without a heatsink is 60K/W a small BC547 in TO92 is 200 K/W 🙂
Thanks for the video. I am having trouble understanding some of the things you said at around 4:00. Why does the "pull down current" have to be more than the current in the 1k resistor? Maybe it is a stupid question but I don't get it...
There is no such thing as a stupid question, every question may make you smarter. Let me try to explain: 1 - the 1 kΩ resistor connected to the gate "Pulls" the gate to +12V of the capacitor 2 - this way the MOSFET G0S has ~ 12V volt and the MOSFET is ON 3 - If you switch on the BC547 a current will flow from the 12B via the 1 k, and the BC547 this current "pulls the gate down" if (I visualize a schematic like elastic bands connecting the parts ) 4 - If the current is too small (say 0.0001mA) nothing will happen, if the current is big (say 1A) the gate will be pulled down very fast. In this case the current is 20mA, 5 how to know if 20mA is enough? well: 12V/1kΩ is 12mA, it must be more than that for it to "win" the pulling, so 13mA would be just ok. To be sure we set it at 20mA. (also no too much to keep your transistor from breaking )
Mosfets that switch any real power need gate protection, back to back zener diodes, plus an ultrafast clipping diode on output (drian to source) . Many other improvements but for now I will qualify my statements with this,,,I designed Mosfet EDM powersuppies for decades, and although the methods shown will function, they by no means represent best circuit design. The problem is caused by the Miller effect. My supplies switch up to 400 amps at just under 1 mhz or less. These will get you started but switching real power needs more components.
At these currents, it’s best to use some kind of resonant topology (normally ZVS). Rather than battling the FET parasitic capacitance by brute-forcing the shortest rise & fall times possible, you can take advantage of it. :) It’s only applicable if you’re driving an inductive load of course, but at these frequencies any load becomes somewhat inductive. Re: the bootstrap push-pull circuit, replacing the pull-up resistor with a current source and using a common base or common gate transistor instead of common emitter/source would be a big improvement.
Thank you Steven! Here is the link: drive.google.com/file/d/1_vD16bBGvdzPY8TEF4GOULbfYH5T9_QT/view?usp=sharing I also added in the video comments. Please let me know it it works!
Can a purpose-built driver IC (such as the last one) keep the load on with DC (0 hz) ? No switching involved - just to use an N-channel as a basic switch.
Hi Sir, I have tried Power Mosfet driver IR2101 for high side Buck Converter with Vcc between 12 to 15V. But Mosfet IRF3205 is not going to off state. Hence the required output voltage is not observed across the load. Please suggest any additions to the circuit. Thank You.
Thank you infoolyugun! A level shifter can drive a MOSFET with a "floating" source. So the IR2104 is a level shifter. It drivers the MOSFET AND handles that the MOSFET source is flying at some high voltage. A MOSFET driver only drives the MOSFET at ground level, see this video: ua-cam.com/video/wv4uAz8FV2w/v-deo.html The IR2104 has a MOSFET driver at the LO (we did not use it in the video) and a Levelshifter for the HO.
The 2n222 bjt is very fast I don’t have any issues with the first boostrap and you can add a speed driver of pnp 5401 bjt to increase speed in series it’s great speed
@@annaplojharova1400 Even a small mention like "You can replace this BJT with an opto if you want isolation" and show a simple circuit would make this video better without any drawback.
@@XiaZ These "simple basic circuits" are already quite bad from safety perspective (microcontroller fault -> output ON; microcontroller supply failure -> output ON; in both cases even not completely, so mosfet meltdown ensured). With an optocoupler attempt the same issue becomes way more complex (more power supplies to handle, parasitic capacitive coupling between LED and the phototransistor within the optocoupler, dealing with the typical optocoupler parameter instability in general,...). By the way optocouplers are normally not used with fast switching, it goes usually via transformers or balanced capacitive coupling (when the insulation barrier is within ICs).
Good discussion guys! I did consider the opto-solution but that would make the video even longer because it comes with many caveats (speed being one of them....) I do agree that could have mentioned it.
How can I use this bootstrap push-pull to drive H-bridge load? For example: 1. DC motor 2. Inverter That is to drive both the HO & LO. Can you modify the circuit in such a way that it would not invert the HO side while it can invert the low side ?
You can use it for an inverter. You can not use it for a DC Motor: all circuits cannot handle DC output in thje HIGH state (LO high) because the bootstrap capacitor will slowly lose its charge. So you have to switch "low" once i a while to top up it's charge. Solutions for motor drive: 1) stay at 99.5% (1kHz PWM) or 0.5% 2) a charge pump, this adds extra complexity - would not recommend 3) Motor drive IC like L293
Is it possible to measure the efficiency of the gate driver circuit by considering only the driver output and input? is the efficiency going to be very low? And why used RMS value for power calculation instead of taking the average value? Thank you..
Yes you can, for the simulationin the Simetrix tools, efficiency calculator. I did not measure the efficiency, however it can be done with some multi-meters. Regarding the RMS value...I checked a bit more online and you are right I should use mean...... never too old to learn thank you for pointing that out!
Can I use these circuits to swap batteries? I would like to keep a second battery full until the fire battery is exhausted. After that I want to turn on the power from second battery and jettison the empty battery. But this means that after battery swap the second battery would power everything including the control circuit.
@@smartpowerelectronics8779 The one from the link is not really good for me. The current will be like 80Amps continuous and I have no space for a heatsink. So I am still looking for some low Rdson n-mosfet that will be under 1mOhm.
Normally less than 1MHz is not very critical. I suggest you make a ground plane, use the bottom layer and fill it with as much copper as you can. Search fill function in KiCad and search ground plane in general with Google
Nice video, but both of these discrete circuits are examples what should never be used just like that. And the slow switching of the first one has nothing to do with any of them. There are two major problems: No protection against mosfet destruction (by insufficient bootstrap voltage causing incomplete turn on because of abnormal load, no output overload) and mainly everything but fail safe (if processor fails, it keeps the mosfet partially on, without any control). And to make this circuit OKish at least a bit, we are talking about roughly 10 extra components, even for the first "slow" predriver. To me it does not make much sense, when that IC is available...
The IC version is for sure the easiest and very robust. However understanding the basic principles is key to make a good design with integrated circuits. By the a different version of circuit 1 has been produced in the millions for a 20W, 300V half bridge running in ZVS at 50kHz and proved robust. When I saw the circuit the 1st time I was actually shocked just like you... but it was cheap, did the job and reduced the electromagnetic noise so much that more savings were found in reducing EMI filtering.
Hello, I try the simulation circuit #3, but Simetrix returns errors : There are errors in the circuit. Simulation aborted IR2104.sub (2): C1 Com In 10p Rpar=1e6 *** ERROR *** Unknown parameter 'Rpar' IR2104.sub (13): C3 HO VS 22p Rpar=100k *** ERROR *** Unknown parameter 'Rpar' IR2104.sub (17): C4 Com _SD 10p Rpar=1e6 *** ERROR *** Unknown parameter 'Rpar' IR2104.sub (28): C6 LO Com 22p Rpar=100k *** ERROR *** Unknown parameter 'Rpar' *** ERROR *** Cannot find definition for model '50n' *** ERROR *** Cannot find definition for model '.35' *** ERROR *** Cannot find definition for model '20n' *** ERROR *** Error in model 'IR2104.SH': VH parameter is below minimum value of 0 *** ERROR *** Error in model 'IR2104.SL': VH parameter is below minimum value of 0 *** ERROR *** Cannot find definition for model '25n' It seems to become Rpar and all A descriptions in the IR2104 model
Did you succeed adding the IR2104 model? There is a link in the video with model and a short description how to add it. If you are stuck with SimeTrix it is hard to find a solution, because of this I now use LTSpice, which has much more online information, tutorials and is really free to use. The schematics and interface of SimeTrix are much better though ;-)
thanks for reply. Yes, I have droped the IR2104 Sub file in the shell and after I can edit the model in local/global or use it from the user library. Is there anything to do juste after added model like special setups or modifications in model in local mode ? So I try to understand the problem but like you said, it is very difficult to find informations juste for exemple on A Spice command, witch represente probably an amp ou an amplication function. However, the IR2104 spice model seems to wheel organised...
Why is everyone on YT all of a sudden does high side bootstrap drivers? Cause I've seen like 3 already. Also discrete drivers + arduino noobies = bad idea. You also need to mention that MOSFET needs to be constantly switching for the circuit to work properly. So beginners better stick to IC drivers that have internal protection features.
Hi Zeeebrenn, yes that part is tricky, but the key point of a bootstrap supply. If you look at circuit #1 (0:07 in the video, make a screenshot) *If the MOSFET is off, the 100n cap is charged from 12V via the top 4148 and the load to about 12V right? *OK, so we have 12V on the capacitor, that will remain there if nothing discharges it... *Now if we let go of the gate (BC547 off) the 100n has 12V and will charge the G-S of the MOSFET, the G-S is only 2-3nF so the 100nF keeps at 12V! *If the MOSFET is on...D-S will be like a short circuit, so the bottom of the 100n is now connected to +24V...therefore the top of the 100n will be at almost 36 Volt! Please draw or print the schematic and use a pen to draw the current s and voltages, hope you understand it! 🙂
Yes thanks, I understood it by realising the mosfet D-S just acts as a short when it turns on, resulting in 24V on the source, which is then added to the cap's 12v. I've seen these bootstrapping used in amplifiers and didn't understand until now. I don't however understand your explanation of the gate diode. You mentioned that there might be cases where the gate voltage can be -24V, but I don't see how. Could this be simulated with a capactive load or something? @@smartpowerelectronics8779
Well that used to be my opinion too, for instance if you simulate a astable multivibrato r with 1 npn's, it does not even start because the parts are 100% balanced mathematically..(you add a 100G resistor anywhere and it works ) ..But if you know what you do and accept the limitations of simulation it can be helpful. And I fully agree with you, build & testing is gold and more fun to do! :-)
Thank you for pointing the traps of bootstraps in HV circuits.
Haha that is quite a tongue twister there...so you do not get trapped in the traps of bootstraps?
this is no 1 video on all youtube that explained the MOSFET gate driver very well , ASAP you will be top 10 youtube channel
Wow thanks man
The 2nd circuit you showed was the one i used during the IC shortage after the pandemic. Worked really well for the motor controller i was tasked with designing.
haha cool, yes these component shortages...you could still buy old stuff like transistors and NE555's.
thanks for these valuable practical electronic lessons, it was beneficial, please make more of such videos in which every circuit is validated with real test
Thank you for the lesson!! I'm new to electronics and I was searching for a solution on high side switching for a week but I couldn't find such a good explanation. I'd like to see a video explaining the UDN2981 or M54563P ic's!
Thanks you for the detailed explanation. I just want to add a comment that MOSFET turn on and turn off speed is the time the transistor changes state. In most of the video, you are referring to the shifter delay as turn or and off time but this is just the shifter delay and not the turn on/off time. While important to prevent shoot through and ringing, the delay time is not as critical as the time the mosfet spends not fully on or off.Power dissipation is current x voltage and partially open time is where most of power loss is generated. One of the most important characteristics of the mosfet driver is how fast it can switch the transistor. In that respect the slope of the open/close curve is more important than the shifter delay. If you measure it in your experiment you will probably find out that the IC does that best and will result in least power loss.
very well explained, even though i am in mechanical field and learning electronics now. i can understand your video at a good pace. Thank you very much.
I was always under the impression that the energy dissipation in the Mosfet was due to the speed of the actual switching of the fet and had nothing to do with the delays being on or off. Remember if there is a delay of what ever time say 500ns the fet is still dissipating no energy because it is either full on or full off. It is only at the time that it is actually switching from full on to full off (or vias versa) that the power is dissipated. In your tests the actually switching times are 200ns (using IR2104 circuit) and that is what should be used for the power dissipation calculation. Basically the time of the slope of the mosfet switching is what should be used for the power dissipation. And this is based on the speed of charging the gate capacitance of the mosfet. In the advanced transistor driver and the IR2104 a 10 ohm resistor was used and similar actual switching times of the moset resulted. In the basic version the low current of the 1K resistor slowly charged the gate capacitance resulting in the slow actual turn on of the mosfet. Am I wrong on my assumptions?
You you are right, during the delay the MOSFET is just waiting, so no switching losses.
Your videos explaining circuit operation are some of the best available, free or paid.
If you’re excepting requests, will your touch on the topic of the “Slayer Exciter” circuit? It’s used to drive Tesla Coils in lieu of a spark gap circuit.
I ask, because in studying data sheets of N-channel Mosfets, I found that the slayer circuit closely resembles a test circuit in the back of many datasheets (an unclamped inductive load)
I’m still new to field/hobby of electronics. I’m pushing 50 and not without experience. I’ve built several Tesla Coils that use a single MOFET Slayer Exciter that were functional. I’ve seen a few simulations of the circuit but none with the in-depth explanations that make your videos stand out from the rest. Thanks for your time and the videos!!!
Thank you!
I know the Slayer Exciter, and it is not a metal band from the 80's. I have no experience but I expect you can use the low side MOSFET driver push pull part to make your slayer exiter more efficient. I did see that for Tesla coils it is key to drive it pulsed, e.g. with 5-10Hz PWM with low on cycle (5% or so) to get nicer sparks and prevent burning your MOSFET. You can use the arduino for that (or a 555....) So yes I am interested and if I find the time I will build one myself 🙂
Did you check this video, seems like a robust design to me. ua-cam.com/video/xImmwL5MR0o/v-deo.htmlsi=BMNipvCvJpJh1Tgh
thank you for a very good course with all the details and examples to be tried out GREAT
Thank you Wliterow!
Thanks for explaining the working and differences of the circuits. Also the simulations were quite interesting and helped to better understand the circuits. I also liked your Kicad video😊
Great channel! Looking forward to more usefull circuit! I like your approach, explain, simulate and reallife. Keep it up!
Thanks Rick_er2481! More will come!
Photovoltaic mode optocouplers floated on the source will also work.
LEDs will exhibit PV effect if illuminated by same colour LED, so it's fairly easy to "dead bug" such an arrangement for experimentation.
Optocouplers are a solution, I do wonder what the propagation delay of the opto is, maybe you will need some special fast types.
Very interesting!
I was hoping to see a comparison with the simpler P channel high side switch circuit, particularly since that one can be driven at zero Hz.
I'll make that my suggestion for a future video :)
Nice video, good explanations. But a comment on the use of "current sink" (BJT with emitter resistor) to drive the boot-strap circuits. The **minimum** voltage achievable at the collector of the level-shifter BC547 (call this Q1) is about 4.5V, due to the output voltage from the Arduino, and the Vce (sat) of Q1. Add to this the Vbe of the pull-down PNP in the buffered boot-strap, means that the mimum voltage at the gate of the MOSFET will be about 5.2V. In most cases this should be sufficiently low to ensure the MOSFET stays OFF, but not always. Is there any particular reason you did not just connect the emitter of Q1 directly to GND, and insert a suitable resistor in series with its collector if current or power dissipation became a concern?
I wish I found this a year ago, but being able to switch high side with pwm using cheap components is super useful. Still struggling to understand the transistors, so I'll be reviewing that literature...
Great to hear that, the discrete cheap components can really show how this circuit works, success with studying transistor operation !
High-quality tutorial! Love to see more
Thanks man! Bit busy lately, ~ July I will make some more videos 🙂
Very useful content ❤
Hello! Interesting video! I made experiments with discrete (only transistors) class D amplifiers. In my opinion most good design of a gate driver is the " UM10155 Philips application note" . There are Q12, Q10 is a high side gate driver. The transistor Q8 turns on/off this driver.You can replace it with usual BJT transistor , but the maximum CB voltage can be greater that your power positive bus (in this video -24V). To decrease current consumption it may be a low power MOSFET, for example 2n7000.
Thank you, that is a great appnote, found it:
educypedia.karadimov.info/library/pwm%20vegfok%20philips%20UM10155_1.pdf
The "UM1055" app note circuit provides no provision against C5+C6 overdischarge so insufficient gate drive of the top mosfet, e.g. when the top transistor is requested to be ON for way too long (in case of the amplifier, when the input signal overdrives it towards clipping on the top rail - so pretty much normal operation unless the input has properly set and even supply tracking limiter). It is even worse as the Q8 draws current from the C5+C6, so is actively discharging them, so the critical maximum ON time is not that long (barely covering the normal switching frequency).
@@annaplojharova1400 So , we need simple forward one-transistor isolated converter to supply high side - if you want to turn on high side during long time.Or we can use a usual half-bridge gate driver IC and supply high side (without bootstrap diode) with mentioned one-transistor forward self-oscillating isolated DC/DC converter.
@@andrewandrosow4797 You can use the simple bootstrap, but the predriver has to provide protections that ensure the bootstrap capacitor gets recharged (e.g. by forcing an off pulse in the class D amplifier or the buck converter examples) and/or the transistor gets shut down properly if it can not handle the current with the available gate drive (in fact this could be done by a simple 2-transistor, few resistors and a capacitor addon circuit monitoring the ON state Vds, if we want to stay with "simple" discretes). Or use one of the many available gate driver ICs, which do have at least the bootstrapped supply monitoring build in (that is enough, when either the current is monitored elsewhere, and a direct short circuit is not an issue, like e.g. in an active speaker box amplifier). Here I see the gate driver IC just as the simplest practical solution, offering at least the minimum protection (essentially against input undervoltage for the buck converter, or the signal overdrive for the amplifier). There are also ICs that provide the overload protection (either the large Vdson, or current sense, or both).
@@annaplojharova1400 I had made in past the class D amplifier using this application note.There also was simplest protection against short-circuit. The amplifier works well. I am totally agree - an IC gate driver much reliable and robust that discrette driver.Mosfets much resistant against avalanche breakdown when voltage becomes high.
In my experience, for the discrete bootstrap push-pull it’s best to use the common base (or better yet, common gate) mode for the level-shift transistor. I tend to use the jellybean 2N7002 for this purpose. No extra caps & resistors needed, much shorter rise & fall time, and the logic levels non-inverted. It’s especially useful when your microcontroller/logic chip/comparator/whatever has an open drain/collector output. Penalty: it has to sink both the pull-up resistor’s and the PNP base current, which may be too much for some low-power chips. At voltages higher than 24V it becomes necessary to replace the pull-up resistor with a current source, which further improves the performance.
Interesting, do you mean a circuit as mentioned in this discussion (2nd schematic, scroll down). Never tried that! electronics.stackexchange.com/questions/296879/logic-level-converter-using-transistors
@@smartpowerelectronics8779 Pretty much yes, but as I’ve said, I tend to use a low-power MOSFET for this, just for the convenience (no need for a base current limiting resistor). Gate is connected directly to Vdd (+5V or +3.3V, whatever the chip is powered by), and source to the output of the chip. The typical Vgs ON of these FETs is abt. 1.5-2V, so it needs to be either a rail-to-rail output or open drain. Might not work with the old TTL logic.
HIGH level at the output of the chip turns the power MOSFET ON, so there’s no inversion like with the common-emitter level shifter.
Thank you for an interesting presentation sir. For me it's not a problem to have another voltage source in a system, but I see many examples in a real life (a hybrid solar controller for instance) when people actively use a transformer gate driver. Then they can have one powerful low voltage source and control a MOSFET bridge. But I see no other advantage with the transformer. Because it's hard to manage a dead time in this case if a 1:1:1 transformer is used. And ferrite cores are quite expensive (pretty comparable with a drive IC price)
Could you touch that in one of your further videos pls?
Thanks for this lesson.
My pleasure Carlosmf6954!
I have never encountered the 2nd circuit in your demo. So I learned a new thing! Also, I wonder if the discrepancy in the turn on time measured vs simulated could be due to the way you are generating the 12v. Would be interesting to see if the 12v rail is bouncing during turn on transitions. Its not just the gate that needs to be charged. Also the base current of the high side transistor will pull the 12v down slightly. I will probably make this circuit myself to investigate since it is entirely new to me.
Great to hear you learned something william4749. For practical use I do recommend an IC, much easier to build and it also has some protection against low voltage etc. ....but circuit2 works OK
Wow great tutorial. Can I ask about the #3 level shift circuit if the R-load/output is paralleled to high capacitance capacitor can it still charge the bootstrp capacitor during off state of HO switch while the output capacitor still have some charge on it if VS need to be in GND level voltage when High side is off to charge bootcap.
These are incredibly helpful videos
thank you!
Question is if you use higher voltage like 400 volts DC at the high side drain then need more than 400 volts at the gate source to drive the load at the source and the limit voltage for gate source is not More than 20 volts DC how does this bootstrap works for high voltage at the source you may want to make a video for that.
@@doost6233 the circuits can work with 400V dc, you will need to replace the “level shifting “ bc547 transistor with a high voltage mosfet. The IC circuit works for 400 volt.
Great videos but did you confuse turn off/on delay with turn off speed ?
They are different indeed. delay will not cause any switching losses because the MOSFET is just waiting. The speed is very important because during the slopes/transitions you have current and voltage at the same time = power loss...
Thanks a lot for your extremely helpful and informative videos! Highly appreciated! May I ask what kind of spice simulator you're using?
I use Simetrix, the free version.
@@smartpowerelectronics8779 Thank you!🙌
I have a question as I'm not an expert in power electronics. Could you make a separate charge pump circuit with its own oscillator and switching transistor that's permanently on and is solely devoted to generating bootstrapping voltage for the high side MOSFET? I reckon it would allow you to keep it switched as long as you like since putting the charge pump directly besides the MOSFET only generates the bootstrapping voltage for a brief moment
Informative Content!
To my surprise, I found from digikey some P-Channel mosfets that don't cost that much but are similar to the N-Channel RDS(on) levels. E.g. G040P04T has less than 3.7mΩ and is only $1.82. The G050PO3T has less than 5mΩ costing $1.17.
THANK YOU FOR THIS BUT PLEASE CAN YOU TELL THE NAME OF THE SIMULATION SOFTWARE YOU WORKED WITH? THANKS
I use Simetrix/SIMPLIS (www.simetrix.co.uk/) the free version, however I would recommend you to use LTSpice which has much more users and tutorials online. I change to LTSpice now. For nice looking schematics, I use KiCAD.
i m very much thankful to you .
thanks for sharing such nice video.
Not bad. It would help greatly though if you could explain the defensive design techniques for preventing false triggering on the high side due to fast dv/dt at the output e.g. when driving a (half-)bridge.
Thank you p0lyglot, you must be a professional in this field. The IR2104 has built-in logic and non-overlap time for that. For the discrete solution you will need to build in the dead time and logic in your logic signals. Half bridges are tricky for beginners, especially prevention of capacitive mode and hard switching are complex to manage.
The false triggering thing isn't a dead-time issue, but a matter of physical design, parasitics and low-side turn-on speed. It's not a shortcoming of this high-side video but if you do a half-bridge or full-bridge video, it'd be worth including some additional considerations for high-side driver design.
The problem is that when the low side comes on, the high-side source voltage goes down very fast. Any stray C between high side gate and a power supply rail mean that the high-side driver has trouble "following" the FET (keeping a fixed low Vgs) so suddenly you have a bump in Vgs, the high side turns on and all the smoke gets out.
The other methods for driving high-side mosfets are: transformer driver, opto-isolated driver and dc:dc convertor driver.
Isolated methods are much more expensive than non-isolated.
You are correct, transformer drives have been still popular though, especially for bipolar transistor half bridges
Very well explained ... Hope the ghouls keep away from your designs !
sure hoe so! Thanks mumbaiverve2307!
Thank you for your explanation, I was wondering how do you get 76 degree, given 850mWatt at 60 kelvin per watt?
If you search the datasheet of the IRFZ44 you can find the Thermal resistance is "60 K/W".
You can then calculate the temperature rise: T= Tamb + Powerloss*Rthermal , in this case 25C + 0.85*60 = 76 C.
A TO220 without a heatsink is 60K/W a small BC547 in TO92 is 200 K/W 🙂
I think with 31khz the pwm resolution is only 9 bits, it's 0 to 511. Very good video btw
Thanks for the video. I am having trouble understanding some of the things you said at around 4:00. Why does the "pull down current" have to be more than the current in the 1k resistor? Maybe it is a stupid question but I don't get it...
There is no such thing as a stupid question, every question may make you smarter.
Let me try to explain:
1 - the 1 kΩ resistor connected to the gate "Pulls" the gate to +12V of the capacitor
2 - this way the MOSFET G0S has ~ 12V volt and the MOSFET is ON
3 - If you switch on the BC547 a current will flow from the 12B via the 1 k, and the BC547 this current "pulls the gate down" if (I visualize a schematic like elastic bands connecting the parts )
4 - If the current is too small (say 0.0001mA) nothing will happen, if the current is big (say 1A) the gate will be pulled down very fast. In this case the current is 20mA,
5 how to know if 20mA is enough? well: 12V/1kΩ is 12mA, it must be more than that for it to "win" the pulling, so 13mA would be just ok. To be sure we set it at 20mA. (also no too much to keep your transistor from breaking )
Mosfets that switch any real power need gate protection, back to back zener diodes, plus an ultrafast clipping diode on output (drian to source) . Many other improvements but for now I will qualify my statements with this,,,I designed Mosfet EDM powersuppies for decades, and although the methods shown will function, they by no means represent best circuit design. The problem is caused by the Miller effect. My supplies switch up to 400 amps at just under
1 mhz or less. These will get you started but switching real power needs more components.
At these currents, it’s best to use some kind of resonant topology (normally ZVS). Rather than battling the FET parasitic capacitance by brute-forcing the shortest rise & fall times possible, you can take advantage of it. :) It’s only applicable if you’re driving an inductive load of course, but at these frequencies any load becomes somewhat inductive.
Re: the bootstrap push-pull circuit, replacing the pull-up resistor with a current source and using a common base or common gate transistor instead of common emitter/source would be a big improvement.
Very interesting and useful video, any chance of a link to download the Elements schematics to save the work of recreating them? - thanks.
Thank you Steven! Here is the link: drive.google.com/file/d/1_vD16bBGvdzPY8TEF4GOULbfYH5T9_QT/view?usp=sharing
I also added in the video comments. Please let me know it it works!
Thank you - yes the simulations work and will save me much time @@smartpowerelectronics8779
Can a purpose-built driver IC (such as the last one) keep the load on with DC (0 hz) ? No switching involved - just to use an N-channel as a basic switch.
What about a drive circuit that works at DC? (Say to shut down voltage on the load when a fault condition is detected)?
Hi Sir,
I have tried Power Mosfet driver IR2101 for high side Buck Converter with Vcc between 12 to 15V. But Mosfet IRF3205 is not going to off state. Hence the required output voltage is not observed across the load. Please suggest any additions to the circuit.
Thank You.
What software you use to simulate process?
it's in the corner. openhantek6022
Appears to be some version of SIMetrix/SIMPLIS Elements.
No that is my oscillosope! The simulation software is Simetrix, free version
yes you are correct, I added the simulation models in the comments today
Hi 🤩🍇
thnx 4 nice vids Mr. T
is level shifter same as mosfet driver?
thnx 4 answere and we wish u nice day 🌈🙋♀
Thank you infoolyugun!
A level shifter can drive a MOSFET with a "floating" source. So the IR2104 is a level shifter.
It drivers the MOSFET AND handles that the MOSFET source is flying at some high voltage.
A MOSFET driver only drives the MOSFET at ground level, see this video: ua-cam.com/video/wv4uAz8FV2w/v-deo.html
The IR2104 has a MOSFET driver at the LO (we did not use it in the video) and a Levelshifter for the HO.
undustund thank You 🤟@@smartpowerelectronics8779
The 2n222 bjt is very fast I don’t have any issues with the first boostrap and you can add a speed driver of pnp 5401 bjt to increase speed in series it’s great speed
Yes the 2N2222 is faster and has more current capability than the BC547!
I still believe that a lowish VGSmax small signal N-channel MOSFET would work best for the inverting level shifter transistor.
Yes, you are right, and it will be very fast.
For my demonstration I try to use the most common parts possible so the circuit is easy to build.
Should include isolated version too.
That is out of the scope for this video...
@@annaplojharova1400 Even a small mention like "You can replace this BJT with an opto if you want isolation" and show a simple circuit would make this video better without any drawback.
@@XiaZ These "simple basic circuits" are already quite bad from safety perspective (microcontroller fault -> output ON; microcontroller supply failure -> output ON; in both cases even not completely, so mosfet meltdown ensured). With an optocoupler attempt the same issue becomes way more complex (more power supplies to handle, parasitic capacitive coupling between LED and the phototransistor within the optocoupler, dealing with the typical optocoupler parameter instability in general,...).
By the way optocouplers are normally not used with fast switching, it goes usually via transformers or balanced capacitive coupling (when the insulation barrier is within ICs).
@@annaplojharova1400 That's why I said you need to address it. You just proved my original point.
Good discussion guys! I did consider the opto-solution but that would make the video even longer because it comes with many caveats (speed being one of them....)
I do agree that could have mentioned it.
What if 12v and 24v are connected and can I use this circuit as a step down converter or half bridge circuit?
what is the software you used? i like it
Simetrix, also called simplis, free version
Nice
How can I use this bootstrap push-pull to drive H-bridge load?
For example:
1. DC motor
2. Inverter
That is to drive both the HO & LO.
Can you modify the circuit in such a way that it would not invert the HO side while it can invert the low side ?
You can use it for an inverter.
You can not use it for a DC Motor: all circuits cannot handle DC output in thje HIGH state (LO high) because the bootstrap capacitor will slowly lose its charge. So you have to switch "low" once i a while to top up it's charge.
Solutions for motor drive:
1) stay at 99.5% (1kHz PWM) or 0.5%
2) a charge pump, this adds extra complexity - would not recommend
3) Motor drive IC like L293
What’s the name of circuit simulator have you used? (I see the word “Elements” in the window title bar but I can’t find it on the web). Thanks
www.simetrix.co.uk/ the demo version
Is it possible to measure the efficiency of the gate driver circuit by considering only the driver output and input? is the efficiency going to be very low? And why used RMS value for power calculation instead of taking the average value? Thank you..
Yes you can, for the simulationin the Simetrix tools, efficiency calculator. I did not measure the efficiency, however it can be done with some multi-meters.
Regarding the RMS value...I checked a bit more online and you are right I should use mean...... never too old to learn thank you for pointing that out!
Can I use these circuits to swap batteries? I would like to keep a second battery full until the fire battery is exhausted. After that I want to turn on the power from second battery and jettison the empty battery. But this means that after battery swap the second battery would power everything including the control circuit.
Probably not the best circuit for that, I would search another circuit online , maybe something like this: www.wellpcb.com/battery-backup-circuit.html
@@smartpowerelectronics8779 The one from the link is not really good for me. The current will be like 80Amps continuous and I have no space for a heatsink. So I am still looking for some low Rdson n-mosfet that will be under 1mOhm.
Can this work with 14v drain voltage
That would be about the minimum, expect it works ok
which simulation software you are using?
Simetrix
Free version
what should i do if i want to use circuit at 650 khz ?
Normally less than 1MHz is not very critical. I suggest you make a ground plane, use the bottom layer and fill it with as much copper as you can. Search fill function in KiCad and search ground plane in general with Google
@@smartpowerelectronics8779 not works. Pmos IRF9520 didnt switch off. i want to make either half or full bridge driver for inductive load, at 700 KHz
You are using Simetrix
Correct, free version. I now added the simulation files download link in the comments
Nice video, but both of these discrete circuits are examples what should never be used just like that. And the slow switching of the first one has nothing to do with any of them.
There are two major problems:
No protection against mosfet destruction (by insufficient bootstrap voltage causing incomplete turn on because of abnormal load, no output overload) and mainly everything but fail safe (if processor fails, it keeps the mosfet partially on, without any control).
And to make this circuit OKish at least a bit, we are talking about roughly 10 extra components, even for the first "slow" predriver. To me it does not make much sense, when that IC is available...
The IC version is for sure the easiest and very robust. However understanding the basic principles is key to make a good design with integrated circuits.
By the a different version of circuit 1 has been produced in the millions for a 20W, 300V half bridge running in ZVS at 50kHz and proved robust. When I saw the circuit the 1st time I was actually shocked just like you... but it was cheap, did the job and reduced the electromagnetic noise so much that more savings were found in reducing EMI filtering.
Hello, I try the simulation circuit #3, but Simetrix returns errors :
There are errors in the circuit. Simulation aborted
IR2104.sub (2): C1 Com In 10p Rpar=1e6
*** ERROR *** Unknown parameter 'Rpar'
IR2104.sub (13): C3 HO VS 22p Rpar=100k
*** ERROR *** Unknown parameter 'Rpar'
IR2104.sub (17): C4 Com _SD 10p Rpar=1e6
*** ERROR *** Unknown parameter 'Rpar'
IR2104.sub (28): C6 LO Com 22p Rpar=100k
*** ERROR *** Unknown parameter 'Rpar'
*** ERROR *** Cannot find definition for model '50n'
*** ERROR *** Cannot find definition for model '.35'
*** ERROR *** Cannot find definition for model '20n'
*** ERROR *** Error in model 'IR2104.SH': VH parameter is below minimum value of 0
*** ERROR *** Error in model 'IR2104.SL': VH parameter is below minimum value of 0
*** ERROR *** Cannot find definition for model '25n'
It seems to become Rpar and all A descriptions in the IR2104 model
Did you succeed adding the IR2104 model? There is a link in the video with model and a short description how to add it.
If you are stuck with SimeTrix it is hard to find a solution, because of this I now use LTSpice, which has much more online information, tutorials and is really free to use. The schematics and interface of SimeTrix are much better though ;-)
thanks for reply. Yes, I have droped the IR2104 Sub file in the shell and after I can edit the model in local/global or use it from the user library.
Is there anything to do juste after added model like special setups or modifications in model in local mode ?
So I try to understand the problem but like you said, it is very difficult to find informations juste for exemple on A Spice command, witch represente probably an amp ou an amplication function. However, the IR2104 spice model seems to wheel organised...
Why is everyone on YT all of a sudden does high side bootstrap drivers?
Cause I've seen like 3 already.
Also discrete drivers + arduino noobies = bad idea.
You also need to mention that MOSFET needs to be constantly switching for the circuit to work properly.
So beginners better stick to IC drivers that have internal protection features.
Good point, that is also why I recommend an IC.
Using a chip with signal delay and comparing it to drivers that don't have this delay is not the sign of a professional.🤦♂
Your explanation is not helpful. "The 12V will be carried on top of the 24V". Why, how does that work?
Hi Zeeebrenn, yes that part is tricky, but the key point of a bootstrap supply.
If you look at circuit #1 (0:07 in the video, make a screenshot)
*If the MOSFET is off, the 100n cap is charged from 12V via the top 4148 and the load to about 12V right?
*OK, so we have 12V on the capacitor, that will remain there if nothing discharges it...
*Now if we let go of the gate (BC547 off) the 100n has 12V and will charge the G-S of the MOSFET, the G-S is only 2-3nF so the 100nF keeps at 12V!
*If the MOSFET is on...D-S will be like a short circuit, so the bottom of the 100n is now connected to +24V...therefore the top of the 100n will be at almost 36 Volt!
Please draw or print the schematic and use a pen to draw the current s and voltages, hope you understand it! 🙂
Yes thanks, I understood it by realising the mosfet D-S just acts as a short when it turns on, resulting in 24V on the source, which is then added to the cap's 12v. I've seen these bootstrapping used in amplifiers and didn't understand until now. I don't however understand your explanation of the gate diode. You mentioned that there might be cases where the gate voltage can be -24V, but I don't see how. Could this be simulated with a capactive load or something? @@smartpowerelectronics8779
Simulation is stupid and misleading
Well that used to be my opinion too, for instance if you simulate a astable multivibrato r with 1 npn's, it does not even start because the parts are 100% balanced mathematically..(you add a 100G resistor anywhere and it works ) ..But if you know what you do and accept the limitations of simulation it can be helpful. And I fully agree with you, build & testing is gold and more fun to do! :-)