Author Topic: Driving (AC) mosfet switch directly from MPU using GDT  (Read 6123 times)

0 Members and 1 Guest are viewing this topic.

Offline beduinoTopic starter

  • Regular Contributor
  • *
  • Posts: 137
  • Country: 00
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #25 on: October 24, 2019, 08:08:25 pm »
This in turn suggests that Fsw should be about as many times slower as the gate drive is.  So, if we can't do 4A but we can do 14mA, we should run at a frequency 4/(0.014) times lower, or about 21kHz.  And again, lower would be preferable, to save on switching losses.
Now changed software and I've nice ~20kHz input to GDT, however bloody cheap YATO multimeter maybe is not capable notice those short  pulses ?  :wtf:

I think I'll try connect this AC mosfet switch do Saleae Logic analyzer with 470 Ohm pull up resistor and try to catch turn on/off  :-/O
 

Offline Simon

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #26 on: October 24, 2019, 08:10:02 pm »
a cheap meter will do 1kHz at best.
 

Offline beduinoTopic starter

  • Regular Contributor
  • *
  • Posts: 137
  • Country: 00
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #27 on: October 24, 2019, 08:16:27 pm »
light flickering at 100Hz disturbs your dreams?
Who knows how our brain works and what it does during sleep?
Even with eyes closed probably everyone is capable notice longer light pulses, but maybe we simply do not notice higher frequencies, but it might somehow affect neurons activity  ::)
 

Offline beduinoTopic starter

  • Regular Contributor
  • *
  • Posts: 137
  • Country: 00
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #28 on: October 24, 2019, 09:32:09 pm »
I'm not an expert in pulse transformers but the idea of a zener sounds like a good one.
It is strange, but when I've disconncted zener diode (I belive in) connected to AC mosfet switch gate (I had only one zener so far and it worked with 4VDC Vgs ) now it works fine with 1k load @ 4Vcc  :o



Even with 100k resistor on mosfets gate we have nice ~20kHz output on AC mosfet switch, but this is not supprise, since we have a few mA gate current for sure.

It is strange  why this zener diode, which I'll remove from this PCB to examine, makes GDT not working at all as gate driver?
Maybe we need two as suggested in antiparallel, to provide the same GDT secondary load in full period?  :-//

Anyway, now we drive AC mosfet switch directly from MPU using GDT, however it will be interesting to see what happends with this 80W 230VAC infrared light bulb, since at 4Vcc and 1k load resitor we had 2V on AC mosfet drains shown by this YATO multimeter, so probbaly it was fine, but this blood zener diode tried ruin my night today  >:D
Any ideas, why this zener diode sucks in this AC mosfet switch when we use GDT as gate driver @ eg. 20kHz?  :-\
 

Offline Simon

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #29 on: October 25, 2019, 07:19:43 am »
light flickering at 100Hz disturbs your dreams?
Who knows how our brain works and what it does during sleep?
Even with eyes closed probably everyone is capable notice longer light pulses, but maybe we simply do not notice higher frequencies, but it might somehow affect neurons activity  ::)


So your eyes can see the 100Hz but when you chop the still remaining 100Hz up at a higher frequency it sudenly dissaperes? I already explained, this is a fillament bulb. the time it take the filament to cool down and stop glowing is longor than 1 cycle at 100Hz. It is a pysical low pass filter.

As I already said this is all a waste of time. just rectify the mains and if you are hell bent on getting rid of the non existent flicker you are chasing around then you must use a smothing capacitor. Then chop the DC with whatever frequency you like. 1kHz is ample. more than that is plain stupid! Put an inductor in series with your bridge rectifier to try and do something towards restoring power factor.

Also remember that you plan to sleep in this thing. Given you obvious lack of experience I would suggest being extremely cautious about falling asleep with your contraption running. Unless you were having plans for an early death and automatic cremation.
 
The following users thanked this post: Yansi

Offline beduinoTopic starter

  • Regular Contributor
  • *
  • Posts: 137
  • Country: 00
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #30 on: October 25, 2019, 07:00:22 pm »
I'm not sure what numbers you used here.  Qg(tot) * Fsw does equal supply current for a gate driver IC.  The peak current is circa T2/t times higher though.
Thank you very much for so detailed calculations, which I've already tested previous night sleeping in capsule with heat coming from 80W 230VAC driven by gate transformer in a way where I use software pseudo-random generator with additional xor of AVR time counter for even better randomness to create gate drive pulses with avg ~20kHz frequency in average in a range something like 10kHz-30kHz  8)



So, It looks like those IRF840 mosfets in AC switch opens (turn on) very well at predicted switching frequency in this case, since after a few hours test during the night, no any significant temperature rise on mosfets- maybe slightly warm.

BTW: I have no idea why peak current might be circa T2/t times higher - I'll study also another manuals howto estimate this current needed to drive mosfets at given frequencies, but for the moment it looks like logic analyzer output from AC mosfet test and real experiments with ~100W 230VAC light bulb shows that your calculations were really helpfull to make this thing working as expected  :-+

WARNING: Off course I didn't touched those mosfets on  230VAC circuit side by hand while  connected to mains power grid  contact !!! I never do this even of datasheets claims sometimes some triacks and other stuff heatsink tap is insulated !!!

 

Offline beduinoTopic starter

  • Regular Contributor
  • *
  • Posts: 137
  • Country: 00
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #31 on: October 25, 2019, 07:09:13 pm »
As I already said this is all a waste of time.
Nope, it works as expected now, thanks to very helpfull detailed analysis of gate current requirements at given frequency  made by T3sl4co1l  :-+

I will try debunk this flickering free at 100Hz pure mains claims, while powering infrared light bulb later - I need setup this experiment with phototransistor  :-/O
 

Offline beduinoTopic starter

  • Regular Contributor
  • *
  • Posts: 137
  • Country: 00
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #32 on: October 25, 2019, 08:19:03 pm »
Also remember that you plan to sleep in this thing.
Nope, this thing I mean sleep capsule covers bed, but 230VAC 80W-100W is outside at least 0.5m and infrared light comes inside by using aluminium 110mm pipe which reflects light inside, so 230VAC light bulb is outside sleeping place and is not more danger than  table night lamp not far away from bed  8)



AC mosfet switch with control electronics is a few meters away in another room right now.

I've made photo right now how this infrared light pump looks like in the dark, but it might be patent pending, so sorry no image in daylight  >:D

Now, since we have AC mosfet switch working @ 20kHz +/-10kHz and powered light bulb from mains 230VAC 50Hz for the moment It could be interesing to see if there are any disturbances in output infrared light other than 100Hz coming from mains AC power, since I've implemented random switching frequency of AC mosfest switch and sample waveform below - already posted above earlier:



« Last Edit: October 25, 2019, 08:24:38 pm by beduino »
 

Offline Simon

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #33 on: October 26, 2019, 06:16:35 pm »
oh now it's a random frequency that is above the frequency that is already above the low pass frequency of the bulb :palm: I'm sure your patent will be wonderful.
 

Offline beduinoTopic starter

  • Regular Contributor
  • *
  • Posts: 137
  • Country: 00
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #34 on: October 26, 2019, 06:32:31 pm »
Given you obvious lack of experience I would suggest being extremely cautious about falling asleep with your contraption running.
This thing below was running previous night for about 8 hours sleep and I can confirm amazing dreams comes   >:D

Patent for good sleep  :D



Anyway, what do you think about the way I disipate heat in this AC mosfets switch from mosfets drain taps directly to wires using 3mm hole connectors in a way shown above.
It is at least 0.5mm2 copper wire and at least 2m long from this pcb to mains contact and more than 3m to infrared light bulb ~100W?
This means, that at 230VAC mains 50Hz we have I~0.434A  with IRF840 max RDSON R~1 \$\Omega\$, but ~@ 20kHz PWM we have 50% on/off so, it looks like we can expect power loses something like: 2*(I^2*R)*50% ~0.2W total while two mosfets in series, but only ~0.1W per mosfet?  :o

After this 8 hours test today those mosfets were maybe only slightly warmer than room teperature, so do we need any heatsinks, while it looks like in this case, copper wires via those soldered 3mm connectors mounted directly under mosfet drain taps shouldn't quite well disipate any heat into >0.5mm2 copper wires >5m long on both sides of this AC mosfet switch?  ::)

« Last Edit: October 26, 2019, 06:34:03 pm by beduino »
 

Offline Simon

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #35 on: October 26, 2019, 07:46:42 pm »
heat won't magically race up the wire. The first short length may help but after the first few 10's of mm it's not helping anymore.

I refer to my comment you quoted in your post: you don't know what you are doing. most of your loses are not resistive while the mosfet is on but during the switching that happens at twice your drive frequency. The MOSFET does not instantly go from full on to yull off or viceversa. The transition will be "resistive" and will dissipate more heat than your 50% on. As I said 1kHz is ample for what you are doing but you know everything so why bother telling you. So long as no one else dies when you burn your house down I am happy to let Darwins law take it's course.
 

Offline beduinoTopic starter

  • Regular Contributor
  • *
  • Posts: 137
  • Country: 00
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #36 on: October 26, 2019, 08:29:21 pm »
The MOSFET does not instantly go from full on to yull off or viceversa. The transition will be "resistive" and will dissipate more heat than your 50% on.
Even when it will be 100% not 50% given only to estimate this is still less than 1/4W per each mosfet in TO220 package connected to copper wires directly from drain tap, so etimated amount of additional copper in wires is ~0.018kg which means it is huge heatsink, while this thing will usualy work in room temperature below 25*C  :-DMM

Experiments shows that this AC mosfet switch doesn't need any additional heat sink in this case where we have huge amount of copper in wires connected directly to mosfet taps - this is beauty of custom design, where thanks to taking into account other things than only to be competitive with low cost on the market and look for as many customers as we can, which means another constrains to lower down price including shipping, while this design do not need to hit so huge market and thanks to using mains wires in a smart way as a heat sink, it can easy fit requirements of custom design.

Probbaly you believe in too much in market electronics products, which sometimes fit into some government regulations, but sometimes you can find CE mark on things not safe at all, so sometimes it is better to know what is inside and how it works, because of it can save your ass and money when fails - easy to repair  ;)
 

Offline beduinoTopic starter

  • Regular Contributor
  • *
  • Posts: 137
  • Country: 00
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #37 on: October 26, 2019, 08:35:09 pm »
heat won't magically race up the wire.
No magic there  ;)
Mosfets in "bulky" TO220 package connected directly via 3mm holes tap drain to soldered copper wires >0.5mm2 lets assume 4m long we have more than 0.0178kg of copper so less than 1/4W will flow througth 2m long copper wires in room 25*C temperatures... No  :-DMM  neither  :-BROKE needed to imagine that heat will disapear and heat will be transfered to room air  :o
 

Offline Simon

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #38 on: October 26, 2019, 09:04:46 pm »
heat won't magically race up the wire.
No magic there  ;)
Mosfets in "bulky" TO220 package connected directly via 3mm holes tap drain to soldered copper wires >0.5mm2 lets assume 4m long we have more than 0.0178kg of copper so less than 1/4W will flow througth 2m long copper wires in room 25*C temperatures... No  :-DMM  neither  :-BROKE needed to imagine that heat will disapear and heat will be transfered to room air  :o


I REPEAT, YOU DON'T HAVE A CLUE WHAT YOU ARE TALKING ABOUT! YOU ARE TALKING BOLLOCKS WHILST PLAYING WITH THE MAINS. THAT IS A RECEIPE FOR BAD THINGS. THIS THREAD IS A WASTE OF TIME  BECAUSE YOU WON'T LISTEN. PLEASE MAKE SURE THAT NO ONE ELSE IS INVOLVED WITH YOUR EXPERIMENTS BECAUSE THE ONLY PERSON YOU SHOULD END UP KILLING IS YOURSELF.

Yes of course your rihgt, copper offers no resistance to heat and all of that plastic insulation is as conductive to heat as copper. You carry on with your delusions and don't listen to anyone! Or are you trolling us?
 

Offline beduinoTopic starter

  • Regular Contributor
  • *
  • Posts: 137
  • Country: 00
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #39 on: October 26, 2019, 11:11:44 pm »
Yes of course your rihgt, copper offers no resistance to heat and all of that plastic insulation is as conductive to heat as copper.
Plastic insulation doesn't have to be as conductive to heat as copper, since quick estimation of area around 4m long 0.5mm2 copper wire is A: 0.010 [m^2] which is equivalent to 10cm x 10cm sheet of thin insulation, so now imagine what can happen when we heat this quite big area with lets say 0.5W from one side having 25*C room air on other side?   :palm:

Problem is you didn't provide any calculations to proove your claims  :-DMM
I listen to people like T3sl4co1l , who provided very usefull estimation of gate current needed, so big letters with warning not needed in your posts.

BTW: I use maxima software under Linux  instead of Window$ calculator, so probably that is why no smoke so far in many projects  :popcorn:

Attached detailed calculation of estimation area around a few meter copper wire. Show us your calculations to proove your claims  :o
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #40 on: October 27, 2019, 01:56:41 am »
The heat spreading length for wire of that size is about an order of magnitude lower than you imagine.  The wires contribute negligible dissipation area.  The transistor assembly is dominant, and can dissipate a couple watts in free air.

Evidently you aren't dissipating much power, so it doesn't matter much.

Quote
BTW: I use maxima software under Linux  instead of Window$ calculator, so probably that is why no smoke so far in many projects  :popcorn:

As the saying goes, it is a poor craftsman who blames his tools, or who trusts them blindly.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline Simon

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #41 on: October 27, 2019, 07:38:37 am »
Yes of course your rihgt, copper offers no resistance to heat and all of that plastic insulation is as conductive to heat as copper.
Plastic insulation doesn't have to be as conductive to heat as copper, since quick estimation of area around 4m long 0.5mm2 copper wire is A: 0.010 [m^2] which is equivalent to 10cm x 10cm sheet of thin insulation, so now imagine what can happen when we heat this quite big area with lets say 0.5W from one side having 25*C room air on other side?   :palm:

Problem is you didn't provide any calculations to proove your claims  :-DMM

Oh you want me to do your job for you? personally I am thinking if you die in a fire that involves you alone now that will be better as at the rate you are going clearly at some point you will do something really stupid and injure or kill other people!

Quote
I listen to people like T3sl4co1l , who provided very usefull estimation of gate current needed, so big letters with warning not needed in your posts.

I too deeply value Tim's contrimutions when he answers because he is very knowledgeable but you are expecting other people to do your design for you or at least come up with complete and utter proof as to why you are wrong. But I won't bother. If you kill yourself by being stupid and not taking the hint so be it.

Quote

BTW: I use maxima software under Linux  instead of Window$ calculator, so probably that is why no smoke so far in many projects  :popcorn:


So you admit being a troll, good, good, I will feel less bad when i ban you.

Quote
Attached detailed calculation of estimation area around a few meter copper wire. Show us your calculations to proove your claims  :o

And what have you proved? that you can calculate geometry, well done you. Do you know what thermal resistance is? If the thermal resistance in 10mm of wire is already too high to let the heat flow to the rest of the wire all of that wire was wasted. I was astounded when I discovered what a "poor" conductor an aluminium heasink was and that putting a power resistor at one end did not really heat up the other end. If you are intelligent you would understand what I am saying.

i could of course set up a test with my thermal imaging camera and show you but why should i waste my time helping you? You will just ignore anything i say that does not resonate with you. So you go ahead and do as you please. i just hope you don't injur anyone else. This project may not hurt you but with your attitude clearly in the future you are likely to do something really stupid and possibly hurt other people.

 

Offline beduinoTopic starter

  • Regular Contributor
  • *
  • Posts: 137
  • Country: 00
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #42 on: October 27, 2019, 07:56:21 am »
i could of course set up a test with my thermal imaging camera and show you but why should i waste my time helping you?
So do something to proove your claims  :palm: 

I've working circuit in reality not simulator and it performs as expected  :popcorn:
 

Offline Simon

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #43 on: October 27, 2019, 08:00:33 am »
I never said you circuit would not work. I was trying to explain to you that you are not using the wire as a heatsink, not that much anyway. Obviously there is so little heat that your setup works but you have no idea about what is happening, it's pure luck. This one may not kill you, the next could.

I'm done here.
 

Offline beduinoTopic starter

  • Regular Contributor
  • *
  • Posts: 137
  • Country: 00
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #44 on: October 27, 2019, 11:02:51 pm »
Obviously there is so little heat that your setup works but you have no idea about what is happening, it's pure luck.
There is no luck, when you turn on/off mosfet using GDT :o
Everything depends on application - in this case it doesn't matter when something goes wrong and for example mosfet will be shorted and conduct 100% instead of in a controled way, because of it works as a smart switch, but even when we provide full light bulb power not such a big deal - slightly uncomfortable temperature rise inside sleeping capsule, but it can be easy detected and for example fan forced to higher RPM or turn on alarm.

Yes, I've in my sleeping room alarm clock, but instead of clock it has light sensor, so it will be integrated with sleeping capsule in a way, that you wake up when it is daylight outside-no need to set alarm clock time anymore  :-+
 

Offline beduinoTopic starter

  • Regular Contributor
  • *
  • Posts: 137
  • Country: 00
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #45 on: October 28, 2019, 06:55:06 pm »
The wires contribute negligible dissipation area.  The transistor assembly is dominant, and can dissipate a couple watts in free air.
Mosfet assembly, since mounted ~0.5mm above PCB while drain tap sits on wire connector (3mm inside hole and more than 7mm outer ring) also helps dissipate heat with additional copper layer on opposite side.

However, it is clear when you soldering connectors with copper wire inside a lot of heat is transfered into wire since you have to be carefull to keep connector during soldering in a position to avoid solder flow into the wire.

In this case, where we have not to much heat to dissipate this heat transfer into wire can be not significant.

I think the only way to debunk theory that those wires with connectors on mosfet drains can not be used as heatsinks is to use thermal camera  :-//

Anyway, estimated amount of copper in wires connected to those two mosfet drains in my AC mosfet switch is around 10g copper in wire per each side which is significant thermal mas in comparision to TO220 package taps coper volume, so I'm confident without any doubts that it must be significant heat transfer into those wires  :o
Some IRF840 sellers says it weight 1.5g total vs ~10g copper in wires not included connectors.

It is a pity Simon didn't made experiment using thermal camera if he thinks I'm wrong.
Maybe it could be interesting topic for Dave fundamentals fridays everybody likes?  :bullshit:
 
« Last Edit: October 28, 2019, 07:00:55 pm by beduino »
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #46 on: October 29, 2019, 01:22:37 am »
Thermal cam just shows heat distribution.  Doesn't show heat flows.  Soldering wires requires a lot of thermal mass, but not nearly so much steady-state heat (not that you'd ever be cooking a soldering joint for minutes at a time, anyway!).

I could equally well turn around your argument by asking how far down the wire you have to hold it, when soldering -- does the heat conduct along the whole length, or just a dm or two? ;)

Which, again, because of thermal mass, is a similarly flawed argument; the heat will indeed conduct further down in the steady state.  But not by much.

In any case, if a thermal cam is used, we can at least estimate the heat dissipation by measuring the surface area of the item in question, assuming an emissivity (mostly irrelevant to heat dissipation at low temperatures, but obviously quite important to the camera) and convectivity (heat dissipation into the air, the dominant sink), and combining everything to get a total power figure.  Hopefully this meshes with the actual electrical power dissipation...

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline beduinoTopic starter

  • Regular Contributor
  • *
  • Posts: 137
  • Country: 00
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #47 on: October 29, 2019, 04:00:48 pm »
I could equally well turn around your argument by asking how far down the wire you have to hold it, when soldering -- does the heat conduct along the whole length, or just a dm or two? ;)
Soldering 3mm connectors takes a few seconds while clamped over copper wire, so heat conduction along the whole length of a few meters link is not possible at such high speed, but ac mosfet switch in my case creates heat for hours continuously which means the closer the mosfet drains taps, wire temperature will be higher, than at some point depending on disipated heat power closer to room temperature and heat disipated into free air.

Additionally, as I've already read in other posts, those a few meter long wires on each side of ac mosfet switch connected directly to mosfet taps probbaly should help (thanks to its  inductance) limit (di/dt)  ::)

 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Driving (AC) mosfet switch directly from MPU using GDT
« Reply #48 on: October 29, 2019, 06:18:09 pm »
Heh... heh heh...

A few meters won't limit mains dI/dt in any meaningful way, but mains itself is usually modeled as ballpark 100uH, including the inductance of all the wiring up to the pole/pad transformer (and a bit behind that as well, but because medium voltage distribution is done at a pretty modest impedance, it doesn't account for much inductance; the local wiring dominates).

But that same limit in dI/dt is also what blows up your transistors.  It stores energy at turn-on, and has nowhere to go at turn-off.

In short, if you had enough switching speed or inductance to matter, you'd know it very quickly. ;)

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: beduino


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf