Maximizing Heat Dissipation in High-Power SMPS Design

[Baluncore]

Attached here is a useful reference for use when designing SMPS transformers.

 

[Baluncore]

Here are more. Everyone has a different approach. Get all these files, then browse them until they start to make sense.

slup123 “Magnetics Design for Switching Power Supplies”; http://www.ti.com/lit/ml/slup123/slup123.pdf

slup171 "Magnetic Field Evaluation in Transformers and Inductors". http://www.ti.com/lit/ml/slup171/slup171.pdf

Attached is; slup205 "Transformer and Inductor Design for Optimum Circuit Performance"

 

[Salvador]

OK, now I have a lot of material will take me a few days to get through it all with a cool head.
meanwhile i can say that he first smps (two of them being on my board) is running with a "3c85" rated ferrite core, I am not sure i'll try to look up maybe judging by this ferrite index we can approximately know how many turns would be required for a given frequency since we know the ferrite brand.

also what makes me wondr is that I was using a 82v zener for the feedback, since i though that i could have more voltage on he rails since the amplifier can take it.
yet still I am only doing +-70 volts DC output at max. so maybe i should just get a 66 volt zener and see what happens , the octocoupler should work then and maybe youre right , maybe the pulse width isn't regulating.it probably isnt.
i will have more oscilloscope tests next week then i will also try to measure the duty cycle with different loads.
it has to be the transformer that limits the current , the output diodes used are MUR1560, the driving circit is working fine.
maybe the fault lies in the transformer design , still a mystery why the FCH devices failed

 

[Baluncore]

yet still I am only doing +-70 volts DC output at max.

If the output is not regulated then do not mess with the feedback circuit.
You should instead lower the test load and identify the reason why the supply has such poor power output that it will not regulate.

maybe judging by this ferrite index we can approximately know how many turns would be required for a given frequency

That core is for a wound transformer, NOT an inductor. Frequency is determined by the components on the PWM clock.
The important thing now with the transformer is the ampere*turns which could be causing current limiting by saturation of the core.
The inductance of the transformer determines magnetisation current.
You need as many turns as possible without core saturation at maximum output current, so as to keep the magnetising current as low as possible.

Keep reading the references until they start to make sense and you can put numbers in the equations.

 

[mheslep]

You must be careful testing that circuit. I use a 1:1 mains isolation transformer and Earth the common reference when working inside live switching power supplies.

Yes, that's step two or three for me. Before getting anywhere close to the mains voltage, I'll run the primary off of a low voltage AC bench supply and the control circuitry off a bench DC supply, with the feedback loop jiggered accordingly. Later with with the mains isolation transformer in place, first use a well below design rating fuse, and so on. Little by little.

 

[Salvador]

ok here's what i think i should do , take a 0.1 resistor and place it in series with primary winding then attach scope probes across the resistor and see the waveform under the condition when the secondary voltage sags abut 10 to 20 volts under load. this should show me whether the core is saturating or not, even though i read that in half bridge the core should't saturate under load if it hadnt saturated at idle already, kinda makes sense because as you add load to the secondary it becomes harder for the primary to keep the same magnetizing current as it did with no secondary load isn't this true?

also i think i should insert the 66v zener instead of the 82 volt one and then drive the smps under the same load and monitor the signal on the gates of the mosfets with the scope to see if the waveform chages under load (whether PWN works or not)

also if my calculations are correct the supply did 110 volts instead of 140 idle when it had a 20 ohm resistor as its secondary load , a simple calculation shows that V/R = 110/20 =5.5 amps , now take 110 volts times 5.5 amps turns out to be just a little over 600w of power delivered to those resistors at that time and load, which isn't exactly small and maybe isn't my goal either , so right now it seems that either core saturation or non working PWN could be the cause for the drop , as if i would maintain 140 volts with a 20 ohm load , the power would be 980 watts , and quite frankly that would be way too good for two IRFP 460 driving a homemade half bridge.
so maybe its not that bad after all.
ok I will continue on reading the materials provided and other information and also measure and test my supply to see where the falt may lie, also the destruction of the FCH devices is still an open case.
any information from you folks is much appreciated and also I want to say thank you for the help already given on this topic.

 

[Salvador]

also , i think I will increase or at first for experimental purposes use a reostat, in place of the resistor between sg3525 pins 7 and 5, now there is a value of 33 ohms as shown in the original schematic , but with this low value the deadtime mayb be too short and given that i have more turns than originally shwn on primary may have been the cause for the FCH untimely death.
i also saw on scope when i probed the gate signals that there is close to no deadtime.my plan is to increase deadtime with this resistor and then after tests increase frequency with the decreasing of the other resistor for which i already have a reostat inserted and then will measure the voltage response to applied load.

 

[mheslep]

What kind of scope probe do you intend to connect to the primary?

 

[mheslep]

Jim Williams of Linear Tech has the definitive smps design guide, imo.
http://www.linear.com/docs/4120
See appendix C in particular for a safe method of testing out your magnetics and switches for saturation prior to assembling the entire design.

 

[Salvador]

I see that in the west you are extremely concerned about safety , I mean it's not like I or anyone else I know measures the potentials with his fingers or probes a live switching circuit while washing his body in a bathtub full of water and a nice long earthing cable attached to it.
and while on this topic , quite frankly I have seen safety regulations being abandoned much more by professionals than amateurs like me , because when i get to a place where once in a while I am not certain what might be waiting at the other side i use a safety tool instead of my hand to check things or sometimes when i have worked with live AC at the wall installation just those rubber gloves that insulate from electricity also stepping on a rubber pad combined with the gloves does the trick quite nicely.as for the question you asked , I will be using as I said alow value resistor in series with the primary so that i can attach an oscilloscope and see the waveform , i mena i could do it otherways but then i have to use some resistors in other way because its an old but simple soviet made scope and the max you can see in it is about 50 volts +-.
the man with whom I am working with has repaired tons of tv's back in the day using that and other scopes so we will just probe the primary to see what going on with the waveform.
also the deadtime , i have to make it longer at first to put less stress on the mosfets and to save them from a potential current shooting through.
also will see whether the length of each pulse has to do with driving the core into saturation which maybe limiting the power output.

 

[Baluncore]

I see that in the west you are extremely concerned about safety ,

Are you suggesting that those in the East are not, or the West should not be extremely concerned about safety?

PF is seriously concerned both for everyone's health and the liable for any injury or death to anyone. PF does not have to carry that liability, the thread could simply be locked. I know that you will continue to experiment even if this thread is locked. Personally I consider it safer to remain in contact and keep warning you than to lock the thread. Others may read this thread who are neither as experienced nor as careful as you. They definitely need the warnings. You are not the only reader of this thread.

When probing the active part of a SMPS you really should be using a 1:1 isolation transformer and earthing the common reference that would otherwise be connected to the live mains voltage.

You really do need to acknowledge the concern expressed and agree to take the advice seriously, or you may find this thread being locked like so many threads like this have been locked in the past.

 

[Salvador]

by the " West" i was reffering more to a kind of political/economic /cultural part of the worls rather than specific places , I'm sure you took it as that.well there's a lot of talking involved in this and this is not the place for it , all I can say is where I come from (the former USSR) nowadays Eastern europe (badly misrepresented in the world media) we just do things (yes sometimes with much less care) differently.Yes the things you mentioned are correct no doubt but since I don't have an isolation transformer , but i do have non conducting floor (non conducting with respect to domestic voltage levels) automatic circuit breakers , a fuse before the smps and working with tools that are made of non conducting materials , I personally am safe. the advice to put a light bulb in series with the ac mains entering the smps is on the other hand a really good advice as to save the components from blowing in case something goes wrong and saving extra money and work.

I think it boils down to the difference in understanding who is to blame if someone does something foolish, in the " western society" you feel and maybe even are physically responsible under law to bear responsibility for someone who had an accident if one proves that he did it because someone told him on the internet , in our side of the world no one would ever be held responsible for that.More like each man for himself ,
in no way I am implying that simply because of this me or anyone else should give false safety instructions or no safety instructions at all, I am just saying that not always the laws of physics and the safe handling with them go together with what PF or any other human made website or organization considers safe or not safe.
I have felt many times this limits the amount or especially depth of information one could go into sharing all kinds of know how and tricks but you see I can't because the moderators of this forum think I'm beyond law or telling others unsafe practice even though that is far from truth.
My personal opinion is such that everyone is responsible only for his own actions and not those of someone else unless he is physically or in other ways pushing that person to do the things he wants. If someone in the world decides to do something crazy he will do it anyways no matter what you tell on some internet forums.Ok for the sake of this thread let's forget about this philosophical issue and Baluncore , youve been a help so far , could you please share some opinion on the questions posed in my previous two replies? like for example the pwm part of the half bridge etc.
personally i think i would be better of reducing the duty cycle length and increasing frequency , as to give the transistors more deadtime , since the core saturates probably quite fast anyways so the longer duty cycle is useless beyind saturation yet the higher frequency could push more useful power through the core.

 

[Baluncore]

personally i think i would be better of reducing the duty cycle length and increasing frequency , as to give the transistors more deadtime , since the core saturates probably quite fast anyways so the longer duty cycle is useless beyind saturation yet the higher frequency could push more useful power through the core.

I am not going to answer your endless streaming thoughts on modifications to a tested SMPS design if you are not prepared to focus on the real issues. So long as you avoid those critical issues, you are lost. You are reacting emotionally to the circuit. I expect you will mess it up even more by going off on further “reality avoidance tangents”.

It is clear that you have not resolved a fundamental transformer problem. Timing is irrelevant if the transformer saturates with the required output current. That will quickly destroy mosfets. Your PWM FB loop is not regulating, maybe it is the transformer, or maybe you just used cheap output capacitors with high ESR. We have no idea because you have not posted a single photo of the layout. If you vary component values for any reason you will almost certainly fail to meet the specifications of what was once an optimised design.

You cannot expect to successfully prototype a SMPS without an isolation transformer. If you need to, you can make an isolation transformer out of two identical microwave oven power transformers. Knock out the current shunts, throw out the secondaries, put both identical primary windings on the same core, you then have a 1:1 isolation transformer rated well above 1kW.

 

[Salvador]

first things first.By putting a reostat for both deadtime and frequency control of the SG3525 I am in no way 'messing' anything up, rather it could be considered as an extra option to stabilize or improve the circuit considering its put together using many leftover parts and maybe not in the best engineering sense.

Not the first time people start "hatin" on me here on PF.probably because i always ask side questions and talk much , yes I do have a weird way of exploring things including the ones i like but eventually I get there either with or without the help of others or established learning methods.

Ok just for a little relief , here comes some pictures.
http://foto3.inbox.lv/girtsliep

in this link there should be two folders , one is a folder with some pics of a few of my home made and repaired amplifiers, the other one should be called 'smps'
a few things I want to note, when viewing the mainboard circuit trace side , don't worry before this application i made it originally for other intents so the only traces used here are the mains input rectifier and filter caps , everything else is either on other boards or uses wires.
then above the filter caps there is the small board which has the SG3525 and both IR2110 on it., from that board there extends a bunch of wires no longer than 5cm to the gates of the mosfets.
everything else well is what it seems, two ferrite e core transformers , aluminum heatsink for the high frequency rectifier diodes etc.

 

[mheslep]

so that i can attach an oscilloscope and see the waveform

I asked the question because, if you use the scope probe ground on one side of that inline test resistor on the primary, without an isolation transformer, you may well short the high voltage primary to Earth ground through the scope. That won't go well for the scope lead or your primary circuit.

 

[Salvador]

ok mheslep, I see your point ,. but the scopes my friend has have no grounding at all in the socket , nor are they otherwise grounded, nor the scope chassis has any ground connection, . most of our sockets don't have any grounding , just the phase and the neutral , even those who have the separate ground connection are useless because the older cables had only two wires in most home AC installations.
the only places were a grounding wire extended to the local ground spot, were the ones installed in bathrooms or kitchens for washing machines , water heaters etc.Also nothing bad can happen to the scope otherwise because the voltage will be only that which is dropped across the resistor hence very small.

P.S. a theme for a whole other topic but a little mentioning, separate grounding in places like bathrooms in apartment buildings in cities is kinda trivial , I once measured the resistance of the cold water pipe in my 4 story appartment building (most of it metal some of it plastic) I made the measurment such that one probe was attached in the bathroom the othe in the kitchen , the pipe in the kitchen extends all down the house through the basement up at another place into the bathroom , I was surprised i found about 2 ohms of resistance and 1.2 of them came from my copper wire extension which went for about 3 meters.
so technically in cities were miles of pipes are all connected together they basically act like a big very low resistance grounding wire.I'm pretty sure one couldn't make a better lower resistance ground path than the one that naturally occurs in such a pipe.
So if my electrical water heater developed a ground fault at some point it would be hard to tell whether the ac would be cut by my (nonexistent) ground fault circuit breaker or by the disconnection of the typical automatic circuit breaker(there is one) in series with the heater due to the sudden large current running through the attached pipe down into the city's pipeline.
in the end of the day i assume it's all about which one of these devices are faster in cutting the AC supply.faster in terms of the splits of a second.

 

[Salvador]

I have a question specifically for Baluncore, while reading your given documents i came upon a place were it is said quote

In practice, the flux density change, ΔB, is
limited either by core saturation or by core loss,
thereby limiting the volt-seconds per turn that
can be applied to a specific core cross-section
area. To fully utilize any core, the design should
result in ΔB close to the saturation or core loss
limit, whichever governs, by adjusting either the
core area, the number of primary turns, or the
ON time.

so this is basically saying the same thing i said earlier about having a reostat to control the duty cycle manually, or to see whether the feedback is working , even though I already have the close maximum duty cycle as i saw in the scope while measuring gates.
anyhow i can only be certain after i probe the primary and see whether its saturating or not , spekaing of which would probing the secondary also work for determining core saturation ?

And since I didint get a reply on this question I would like to restate it, when the core saturates , does the saturation decreases with a heavy load on the secondary or not?
my reasoning here is such that in order to drive a core into saturation one needs energy applied for a given amount of time to the primary but under heavy secondary loads most of that energy goes into the secondary load so why would a core saturate under heavy load even if it did that at light or no load on the secondary? reading further i think i found the answer

The flux density in the core (which links both
windings) is determined solely by volt
seconds per turn applied to the primary
(Faraday’s Law), independent of load
current.

thank you.

 

[Baluncore]

ok mheslep, I see your point ,. but the scopes my friend has have no grounding at all in the socket , nor are they otherwise grounded, nor the scope chassis has any ground connection, . most of our sockets don't have any grounding , just the phase and the neutral , even those who have the separate ground connection are useless because the older cables had only two wires in most home AC installations.

The Coroners verdict will be that "the deliberate use of an oscilloscope with a live common chassis" was tantamount to the murder of those who responded to your suicide. I need not say more.

so this is basically saying the same thing i said earlier about having a reostat to control the duty cycle manually, or to see whether the feedback is working ,

NO. Live controls are dangerous to you and the circuit. You can adjust it to the point where it destroys itself while you are trying to optimise output. It is one thing creeping up on an optimum operating point, but you must also be able to get there and back safely when applying, removing or momentarily interrupting power to the circuit at any time. In aerodynamics that region of operation is called "coffin corner".

You must first design everything by the numbers. The data and equations are there in the references. A SMPS is too complex a system to rely on luck.

Contrary to your belief, the celebration of a “Wild West” approach to engineering represents engineering incompetence. Not only is your engineering reasoning unsound, but so is your critical thinking about your approach to engineering. If you cannot analyse your approach to engineering then you cannot expect to analyse the technical engineering issues.

 

[Averagesupernova]

Also nothing bad can happen to the scope otherwise because the voltage will be only that which is dropped across the resistor hence very small.

Ok, the above comment to me shows this thread is on the verge of needing to be shut down. There is no place in the primary circuit you can put a series resistor and have one side of it already grounded simply due to the nature of the design of the circuit. Sticking a scope probe ground on either side of a series resistor in the primary circuit guarantees disaster. One way to 'see' the voltage across a resistor of this sort using a scope is to have a 2 channel scope that you can invert one channel and use the summing function of the scope. The difference in voltage from one side of the resistor to the other will then show up in the scope display.
-
If you want to test for proper feedback vary the line voltage up and down slightly with no load. If a power supply can't regulate with no load you cannot possibly expect it to regulate with a load. My impression throughout this whole thread is that you are not troubleshooting this in a manner that yields any useful information. A lot of shoot-in-the-dark here.

 

[Salvador]

ok I have usually quite a lot of enthusiasm but even with all of that I have to say I am done in this thread.Instead of answering a few simple questions about core saturation etc I read only about that damn scope thing.I do realize it's not among the safest of practices.And whne i come to think of it maybe there is some other way to probe the primary involving a capacitor but not in series.
the man who has the scope is turning 61 this year he works as an electrician and repaired crt's back in the day for a living, also keep in mind that this scope came from the factory with no grounding wire and such scopes don't have their input test ground also directly connected to their chassis.otherwise when making any kind of measurment you would almost always spoil it because your body capacitance affects the readings and everyone knows this since i assume we all have touched the measuring wires with hands at some point to see the waveform on the screen being distorted.

i'm not an expert on smps's but I also don't like or expect to have a " wild west' approach.Enthusiasm and wild west in not exactly the same thing.
My second question was totally disregarded in the light of my first one, I don't consider that fair.PF becomes more and more useless by the day since now I don't even see the old staff coming by to comment except for a few , and then there are a bunch of people with pretty good feedback and even some titles (aint going to call names even though i would love to) which just cruise by and every thread that somehow is dangerous to their mind or where they spot some possible inconsistency with the rules they just report and sometimes give their "golden" advice in a few lines of comment which are sometimes on the border of being intellectually disrespectful, I mean maybe they should just consider their candidacy for the local police since police is meant to watch the law but a forum for physics is meant to learn and explore.
and after all you have to realize that no matter jow much you enjoy pointing out various legal issues about giving some safety advice which let's be honest you don't really care about someone at the other side of the world that much might as well just answered his question with explaining the possible drawbacks of the situation he is in.

if i'll get any positive results I will come here maybe to write a few short lines (even though i dotn see much use of it)

as for what averagesupernova said , yes indeed I would tend to agree that it cannot regulate itself if even at small loads it would " walk" a little.
I actually measured this when i was driving a light load , the mains voltage drifted some 5 volts and my output with a constant resistive load also did some 2.5 volts lower and then back again.
i've made some nice resistive loads of various resistances and after a bit of adjusting and testing will see what goes on.and what this thing is eventually capable of.

 

[Averagesupernova]

When it is difficult to test for something that is suspected of happening in your circuit don't you think it would be wise to approach it from a different angle? You admit that it is poorly regulated with no load. So, reduce the line voltage until you are fairly sure there is no core saturation. Then look elsewhere for the cause of poor regulation. You may find a problem that solves everything and core saturation is no longer a consideration. If you have to reduce the line voltage to the point that the 5 volt regulator goes out of regulation then use a battery ahead of the 5 volt regulator until you have the problem solved.
-
As far as some of the advice here, don't complain about it. Unless you just have not followed a single piece of advice given here, how can you know everything would have gone along just fine had you just gone about it without any feedback from this forum? How do you know how much stuff you would have blown to pieces by now? I have a family member who complains about the direction their life went due to advice given by their parents. I have always questioned where their life would have gone without the guidance. It can always get worse.

 

[Salvador]

well , yeah but to reduce the line voltage gradually i would need an autotransformer and i don't have one, maybe i could find an old one somewhere but it surely would be just as dangerous to use as the probing the primary method.quite frankly I don't bet much on saturation , i tested a few hours ago using a a 1600w rated mains iron for clothes which has a resistance of 34 ohms and then for extra fun i put the now probably famous 230w halogen bulb in parallel. :D since the bulb changes its resistance as it heats up the exact precise resistance under load is not sure but it was definitely less then those 34 ohms of the iron alone. the voltage dropped to about 125 to 128 volts from the average 140, with light loads. quick maths tells me I was averaging about 500 watts of power out and the iron got so hot that it started to change color.
heres what I noticed even i didint have a scope this evening , i decreased the duty cycle a bit and the cold start load was 11 ohms , last time i did this without modifying the circuit the IRFP's blew up.

heres what i believe , this smps even with no faults aint going to regulae to the level where mains suddenly dropping 10 volts due to whatever reason leaves no trace on the output under medium to high load.how can i know this ?when i did my load tests today i changed not only the duty cycle i also changed the frequency (manually with reostats) , and i found a sort of best place , going higher in frequency resultad in no voltage increase and also going for more deadtime started to yield lower voltage, so i left the deatime where it didnt affect my readings but was bigger than in the schematic to make an easier life for my mosfets and the frequency simply were the voltage sagged less.also the feedback isn't working as of yet because i haven't got the right voltage zeners i will get them soon.but i kinda think that aint going to change much.
i would follow BAluncores advice to rebuild the transformer but i first want to see whether its really the problem.
please don't consider this rude or whatever but i will talk with the guy and well find a way to probe the transformer , not that big of a deal, simply because seeing the waveform would tell best what's going on.after all the smps is fine enough for what i built it for originally, which is an audio amplifier board which could take an average of no more than 400w rms i think ,

as for the advice here I mean I am always thankful for one but as long as its not just constant pondering about how stupid everything is and how foolish I am and the others questions are just then left and ignored and the responder simply takes off , as has happened all to many times here on PF.like he thinks its not worthwhile to say anything anymore simply because someone has taken a less safe approach to something according to his viewpoint.
I think it's always a guessing game to tell whether someone is where he is because of what others have said to him or what they haven't said.
as far as my life goes I started building circuits and doing yes back in the day unsafe practices with electricity when i was a kid.long before any PF was around. as a kid i loved electricity but didint have the respect for it as I have now so back then i just touched live wires with my fingers simply to see if the laws of physics are true whenever they said about not getting zapped while standing on an isolating floor etc etc.also got some bad shocks a few times , even to the point were i fell down because my legs were paralyzed for a while so had to use my hands to get myself up the stairs.
so whenever you guys tell me be careful with electricity or floating a scope with live chassis , trust me I know darn well and I keep a very close check on the safety of myself when I am doing anything like that because i have a good personal experience with things going wrong.
if anything of all the folks who know me that are electricians they actually think I am overly cautious whenever i do anything with electricity , i sometimes even switch off power when changing a lightbulb in my room.
i dotn want to sound like some annoying mr.knows it all, I'm just explaining were i stand.

I just hope we could all be more of a community and less like a cop whose just doing his job for the sake of it.
I'm sorry for my part if I have said something which have made someone else think less of me.

 

[Averagesupernova]

Have you probed around on the Vref, +in and -in pins of the 3525 while changing loads? Do you know what should be happening here? Again, be careful here. The device may be run from a 7812 but that is floating well above conduit ground and is potential for shock.
-

If you have to reduce the line voltage to the point that the 5 volt regulator goes out of regulation then use a battery ahead of the 5 volt regulator until you have the problem solved.

My bad ^^^^^^^^^^^^. The 5 volt regulator is was referring to is actually the 12 volt regulator.

 

[Salvador]

the device is not run from a 7812, its run from a small transformer , rectifier , and two 12 volt zeners in series., in other words from a regulated 24 volt supply.
but no i haven't probed around there , actually a good idea , ok the way you asked that is kind of intriguing so as much as I know once a mosfet is driven with enough gate current to get to its threshold (open) state in the necessary time , that gate current should change on the load the mosfet has to drive ?
so my logic is that on the supply pins of the 3525 the voltage should be constant under varying loads at the output or whatever.?

 

[Salvador]

Ok here's a quick sidenote , yet an important one.i tried to launch the second smps with my other two leftover FCH104N60, this time i put a 100w globe lighbulb in series with the AC incoming mains as a current limiter.attached a small load at the secondary just to see if anything comes through together with a voltmeter.
now guess what , as I switch the power on the lightbulb in the ac mains barely glows for a little while might have been the split second then I hear a crak noise and the light bulb lights fully.nothing happens at the secondary at all. i switch the power off, make a measurment and guess what both FCH mosfets dead short on all legs no matter which way one measures.I mean for Christ's sake they are 600v 37 A continious at room temp (MAX) rated mosfets , even if i switched them both on and left them there , simply like a lightswitch they should whitstand the current and voltage I am having here , its only 325v DC about 20 amps max , the wall socket can't even supply more than that.
and I am having a light bulb in series which limits the current to maximum about 1.5 amps at 230 volts ,
and the IRFP'S work just fine even when i tortured them yesterday when the transfomer core started to hiss and whine a little.

either I am so unaware of some alline technology here or something doesn't add up in my book.even though i did not get a clear answer from the datasheet i read that these mosfets are easier to drive for the IR2110 than the IRFP'S due to lower gate charge , so technically the IR2110 isn't to blame for not being capable of driving them
i spent like 33 euros on them , waste of money.

 

[mheslep]

the device is not run from a 7812, its run from a small transformer , rectifier , and two 12 volt zeners in series., in other words from a regulated 24 volt supply.

Your schematic indicates a 7812 off a 15V (?) transformer.

 

[Baluncore]

as I switch the power on the lightbulb in the ac mains barely glows for a little while might have been the split second then I hear a crak noise and the light bulb lights fully.

The design was for kilowatts, but you were only testing 100 watt. Obviously, the energy stored in the full capacitance was capable of destroying the MOSFETS. When doing initial testing you should have been using much less storage capacitance after the line rectifier. You were warned.

the device is not run from a 7812, its run from a small transformer , rectifier , and two 12 volt zeners in series., in other words from a regulated 24 volt supply.

So you have no 7812 = +12V regulated floating supply, but instead use a 24 volt zenner clamp? Do you have a limiting resistor before the zenners ? Does that explain why the half-bridge driver may be unable to correctly drive the MOSFET gates?

How many other things are different from the original design?

 

[mheslep]

Another good application note here.
http://cds.linear.com/docs/en/application-note/an19fc.pdf

The PWM controller is another make but the math is all applicable, the discussion is accessible. Especially note the discussion on how fast current rises in an inadequate magnetic core, on the order of amps per microsecond, and here the switch period is up to 20us.

 

[Averagesupernova]

Hmmmm.

How many other things are different from the original design?

Yep, it just keeps getting better.

 

[Salvador]

, thanks to all who are still interested even though I see its turning more into an amusement thing rather than a thread here.
well you may think I am dumb but it's not that bad.just because i didint use something mentioned in the schematic doesn't mean its not working , I had a small transformer , and a few zeners so i thought i will use them and not buy the regulator.the transformer supplies 28v DC after rectifier and smoothing, i thought ok will give me some headroom.
i used two 12 zeners in series with a 10 ohm 2w resistor. if my math is about right this gives me an output of about 24v and close to 500 mA.
after all the chips can take up to 35v for the sg3525 and 25 for the IR2110, and since i have two IR2110 each driving a half bridge the extra voltage doesn't do bad.
now why would this be a problem here? as I said I probed the gates , the square wave is nice with no sparks and spikes.its about 16v in level on average.
i did probe the power supply for the IC's under load. it drops from the 24volts which i get with no IC's inserted to a steady 20 volts.the difference between driving a light load and a heavy load was only 0.5 to max 1 volt drop in the IC power rail.
again given that the FCH devices have less gate capacitance I fail to see how this could be a problem. Now ofcourse you can go the simple way and just blame a bad design from an amateur but up to the point when I am writing this i really can't see why they failed.
and no one also mentioned the fact that they are rated almoust twice the IRFP's power in amps and a 100 volts more in volts and they fail so easily, I can't understand that.
i guess i should have tried using one of them simply as a light switch , just turn it fully on and leave it there and see what happens.

you may say the gate voltage was too high but the gate voltage to both these devices are standard +-20 volts so if one failed the other should too.
just please don't tell me that the IRFP is a foolsafe thing :Dsadly the time it took for them to fail is so short i can't physically manage to probe the gates with a scope but my guess here is that either the gates are driven too weakly and something goes wrong even though i find it weird as already mentioned a million times the capacitance is smaller and should be lighter to drive or yes the other scenario call me names all you want but maybe there is something wrong with the mosfets, as in the gate shoots through like breaks down.and then it shorts out on all three pins.a rather rare but not uncommon failure for a mosfet.I mean i tested them on two transformers each a bit different, also with current limut etc , i doubt anything in the circuit except for the gate drive could have caused teir mysterious falure.especially since the high current and voltage ratings I am pretty sure by now there is something wrong with the gates or with how they were driven , but i can't still figure out what , I will measure the gate voltage next time on scope to very precise levels at the IRFP gates , maybe just maybe because the FCH had less gate charge the IR2110 were driving the gates to higher voltage than the IRFP's and that could have caused the gate to break down and short itself with the mosfet's current channel.

also this begs the question which will probably be left unanswered in the light of some funny and " smart' jokes about my circuit :D but does the IR2110 drives the gates with the voltage level it is being supplied or does it has a built in drive mechanism which adjusts the voltage even though i don't see any circuitry on the ic's periphery which would be doing that, so could it be that if the IR2110 gets supplied by 22 or at most say 23 volts DC it would also drive the mosfet with that voltage level?

 

[mheslep]

If the FETs are repeatedly failing, instantly, then, either 1) their breakdown voltage is being exceeded, or 2) the current ratings are being exceeded while gated on. Period. If the failure occurs over slightly longer than instantly then include 3) exceeding the heat dissipation of the package. There is nothing "mysterious" about the failures, other than determining which of the various things you think are so are actually not. Given the conservative nature of IC ratings they are probably being exceeded by a high margin. So resolve to methodically investigate which.

For instance, what's the highest possible ring voltage against the FET, without the snubber? At FET shutoff, V=(1/2)Ldi/dt. If, say, L=200uH, di = 10A and dt is 0.1uS, Vpeak=1000. To investigate current problems, disconnect the AC and run the primary off, say, 18VDC from a couple of nice and safe 9VDC batteries ( scaling your feedback network accordingly.) Try some of those industry references supplied above for help.

Also, if you manage to get this switcher working, your layout and assembly shown in your picture is as likely to render it a big unstable 50 khz radio and spike noise feed into your AC mains as it is a power supply. Because of the current parameters in the example above, i.e. 100A/usec, one can not easily throw a SMPS assembly together on the bench as is possible with other electronic projects, i.e. long wires and no ground planes on the PCB (both of which are the case in your photo).

 

[Baluncore]

could it be that if the IR2110 gets supplied by 22 or at most say 23 volts DC it would also drive the mosfet with that voltage level?

That is the case.
Recommended Operating Conditions. IR2110. VCC Low side fixed supply voltage 10V to 20V
Absolute Maximum Ratings. IR2110. VCC Low side fixed supply voltage 25V

You were warned. In post #40.
If you want your components to survive testing.
1. Get a 100W filament globe in series with the AC supply.
2. Reduce the 4x 680uF to something like 2x 100uF during early low power testing. That will reduce the current spike on control failure.
3. Use a 40W filament globe as the load during early testing.

 

[Salvador]

Ok here is the summary so far from what I have seen , tested , thought and read from your asnwers.
even though the smps looks like a mess it's surprisingly stable , i mena i haven't probed the AC mains side to see any bad feedback thrown into the grid but as long as the secondary side dc and primary side gates are concerned everythings nice and square or nice and flat.

as for the mosfet destruction problem , I think I know the problem , too high IR2110 supply voltage. Can't be anything else beacuse if it were to do with the Drain source voltage being too high or having spikes or the transformer inducing some spikes upon switching then the IRFP's would have long failed because they are less powerful.but they havent.
Due to the IRFP's having three times the capacitance on their gates as the FCH ones the IR2110 has a harder time driving the IRFP460 and so its supply voltage sags to an acceptable value , but when i had the FCH104N60 it's gate charge is lower and the IR2110 probably were driving the gate with something about 21 to maybe 22 volts based on what I know about the capabillities of my small transformer zener supply.
the most logic explanation here and also the most likely one is that somehow the FCH devices were driven with too high gate voltage due to their lower capacitance rating and hence them being easier to drive by the PSU.
also the second time i tried the FCH ones i was having a light bulb in series with the AC mains so current was little and probably didin't cause the damage but voltage was still the same so the IR2110 still got the same voltage and that explains the same results obtained.also when i did the second test with the series AC lightbulb the current could not destroy the mosfet simply by going D-S because the mosfets capability is rated at 37 amps and I was no way near that even wtihout the globe.should have been better to simply use the all in one linear voltage controller. or just get a transformer whose secondary is closer to what i need because messing with these small transformer to rewind them is a pain in the ***.
or should have changed my zeners to regulate to 20 volts with a 15 ohm resistor.

well i guess a mosfet's gate is a really sensitive thing , but now to prove my theory right i should order two more FCH104n60 or anything close to that and reduce my gate voltage levels.would love to the bad side is that they have to be ordered, while the IRFP's are here and three times cheaper and i have gotten good enough output power with them so far.I'll take my tikme to think about this.

 

[Baluncore]

The simple resistor–zener regulator really must go. It is very inefficient in that power supply application. It would require an emitter follower to improve efficiency.
Much better to replace it with a 7812 / LM340-12, you might find one in the output of an old desktop computer power supply. The 7812 can handle a maximum input of 35V, so the transformer secondary must be below 24VAC.

There is no advantage in switching MOSFET gates through more voltage than needed as it takes more time to move more charge with no reduction in Rds. A power MOSFET has thousands of gates in parallel. Only one has to fail at 20V to destroy the device.

 

[mheslep]

A comment on cost concerns of a SMPS project like this.

If the primary object is to DIY and thus save money over a complete supply purchase, then that goal is pretty much impossible, here and in eastern Europe, unless all the parts are somehow already on the bench. One can simply not buy and ship all the piece parts of a kW supply, even with zero value on labor and the necessary test equipment, for less than the cost of an off the shelf supply. I see $40/1000W, new, including cooling fan, chassis with safety and FCC approvals, connectors. Used, less still. Otherwise, an effective goal here must then be self-education.

 

[Salvador]

It's basically the fun of building something yourself , the money is not that much of a concern here.Also I really don have most of the parts on the desk , the only ones i have to buy are the power mosfets and driving IC's.
I agree the linear regulator would be better, I just checked my local store (no shipping time) and they have in stock the UA7815, I think I will just go for that , regulated 15 volts @1.5 amps max.the output after bridge and capacitors from my small transfomer is 28v DC so the regulator should be fine.
One question though, ok the zener regulator isn't that effective but since this is a linera regulator wouldn't it produce quite some heat considering it would have to drop bit more than 10 volts on an average of 500mA?
or maybe it has some advanced little circuitry inside and that's not the case.Because usually when I imagine a classical linear regulator I see a half conducting BJT which then is only suited to low current applications due to the excessive heating.

as for the mosfets to take in place , I've been doing some search and I found basically these , IRFP32N50K, is in stock at my local store and costs about 7 euros., but then from the RS catalog I can get SiHG32N50D-GE3 for bit less than 3 euros a piece , I think it's a good price and the datasheets tells me the parameters are good.
surely the fairchild devices may have been better but they also cost more and for this particular application the power I'm getting even with the 460's is almsout enough.
Or maybe the Vishay products are less quality ? Also I don't quite understand the reason for pricing , as from the datasheets i see that amon similar rating mosfets some cost for example in the 7-10 euro range and some others just about 3 to 4 per device. Why is that why would some of otherwise similar rating devices cost more than half less?

Also I see some good high current but lower voltage rated devices.Now here's a question , for example if I used a 50a 200v device for both mosfets in the half bridge , as long as the deadime is kept in check and no spikes occur , the devices should work because the voltage any of the device has to whitstand is only half the rectified mains so about 162 volts in my case.I do realize the device would have a higher chance of failing under some unwanted events like spikes from the primary or else but what do you think?

 

[Baluncore]

One question though, ok the zener regulator isn't that effective but since this is a linera regulator wouldn't it produce quite some heat considering it would have to drop bit more than 10 volts on an average of 500mA?

What is this? The zenner or the 7812? Where does that average 500mA come from? Only current pulses are needed when MOSFETs are switching. Resistor with Zener must carry current maximum needed over the full cycle. It will waste energy whenever current is below peak of cycle, = very poor for power pulses.
Why use a 15V regulator when 12V is more than enough. Higher gate voltages take longer to switch, so overlap of conduction is more likely. Your refusal to follow the original design suggests you must be an anarchist. Come the revolution, things will be different. Not better, just different.

Or maybe the Vishay products are less quality ?

Buy on specification, not a guess of quality. The quality of your circuit construction is nowhere near as high as quality of Vishay products.

Also I see some good high current but lower voltage rated devices.Now here's a question , for example if I used a 50a 200v device for both mosfets in the half bridge , as long as the deadime is kept in check and no spikes occur , the devices should work because the voltage any of the device has to whitstand is only half the rectified mains so about 162 volts in my case.I do realize the device would have a higher chance of failing under some unwanted events like spikes from the primary or else but what do you think?

When one mosfet is on the other has full voltage of -162V to +162V = 324V total.

 

[Averagesupernova]

One question though, ok the zener regulator isn't that effective but since this is a linera regulator wouldn't it produce quite some heat considering it would have to drop bit more than 10 volts on an average of 500mA?
or maybe it has some advanced little circuitry inside and that's not the case.

A zener regulator with a series resistor is less efficient than a 7815 since the series resistor with the zener is ALWAYS conducting. A 7815 will not operate in this manner. The 7815 only wastes power when there is current passing through it.

Because usually when I imagine a classical linear regulator I see a half conducting BJT which then is only suited to low current applications due to the excessive heating.

There are plenty of linear power supplies out there that are high current supplies. 50 amps is not uncommon.

 

[Salvador]

fun thing tho about the anarchists , i see you like political philosophies, me too, I could call you some interesting names but i doubt that would be productive for this thread, no offence.and after all, I've followed the desing in everything except this thing , sadly i didint rethink it through about the voltage exceeding the mosfets gate capabillity, might have saved the fairchild's.After all I am not that rich, and the local metal recyclers are having enough copper already so my two cents are not necessary.

I see the thing right i forgot that once the high mosfets turns on the only thing between the +ve and ground is the low side mosfet so it then has to be able to whitstand the full ractified mains.

by this i was reffering to the 7812.I think of buying the 7815 because 15 volts is in the acceptable voltage level for a mosfets gate and mainly because the 7812 is not in my store but the 7815 is.after all I've been driving my mosfets with 20 volts and they have stood except for the fairchilds which took a bit more than those 20 and lost the game, 15 vmax should be good.i will see if i can get the 12 volt version but if not ill go with the 15v one.

as for the quality of my circuit, it's not like i can't do nice and high quality stuff it's just that the board I am using here i was actully making it for the very first smps i had years ago which used a simple IR2153 driving a pair of mosfets.I was just too lazy to go all over the board making stuff again and since i had the amplifier boards already and so i just built some periphery stuff and made this smps.
if you have any interest whatsoever I will post some pics of the box my friend welded for me from aluminum which i got at a warehouse.the idea is simple to make the box smaller and requiring less space it's made from aluminum and that also is the heatsink , if thre smps would have been not the first one of such kind i would have made its heatsink also together with the chassis.oh and by the way I see you are a bit eccentric yourself or atleast that's the impression one would get from you r avatar picture which I assume is painted by Salvador Dali, maybe some other not sure , after all Dali is my personal favourite much of his work resonates within me, also the reason why I picked this username. :D Sadly haven't yet had enough will to push myself to put an avatar that would go togeher with this username.

 

[Averagesupernova]

by this i was reffering to the 7812.I think of buying the 7815 because 15 volts is in the acceptable voltage level for a mosfets gate and mainly because the 7812 is not in my store but the 7815 is.after all I've been driving my mosfets with 20 volts and they have stood except for the fairchilds which took a bit more than those 20 and lost the game, 15 vmax should be good.i will see if i can get the 12 volt version but if not ill go with the 15v one.

Why not look into a 317 adjustable regulator if your source does not have the 7812? Even a 7805 can be used as an adjustable regulator. Changing a design may be acceptable when you consider all of the consequences.

 

[Baluncore]

oh and by the way I see you are a bit eccentric yourself or atleast that's the impression one would get from you r avatar picture which I assume is painted by Salvador Dali, maybe some other not sure ,

My photo of a picture in Tiroler Volkskunstmuseum, unknown 17thC European artist, there are several copies and later versions by different people.
Title is usually “Bird of self knowledge” or “Do not take yourself by the nose”. Google images “bird of self knowledge”.

 

[Salvador]

as to what average supernoca said, since i already modified the circuit and was driving the gates with almoust their maximum and sometimes over that, I think the 15v regulator could barely be called " considering the consequences"
7805? but the last two digits indicate it's a 5v regulator?

thanks Baluncore for the info on that very interesting avatar of yours.Will check out his works.

 

[Baluncore]

LM340-xy = 78xy = positive xy voltage regulator.
If you put a 7V zenner in the reference lead of a 7805 it becomes a 12V regulator.

Depending on the MOSFET design, 5V is usually insufficient to guarantee gates are fully conducting. 12V is sufficient. 15V is unnecessary and delays switching.

Will check out his works.

Let me and the Museum know if you work out who the artist was. She was well ahead of her time.

 

[Salvador]

:D:D, yeah right I said it so that it sounds like the unknown author is someone I know :D
you say she? so they think it was a woman?

 

[Baluncore]

You assumed male, but the best way of being written out of history was to be a woman.
Yes, I know it is sexist, but I contend that an unknown artist is therefore more probably a woman than a man.

 

[Averagesupernova]

7805? but the last two digits indicate it's a 5v regulator?

You can also wire it like an adjustable regulator. We know that the regulator will do what it needs to in order to get 5 volts between the ground pin and the output pin. So if we make a voltage divider between the output and ground and tie the ground PIN to the node between the resistors of our voltage divider the output voltage will be determined by the ratio of the resistors in the divider.

 

[Salvador]

oh interesting so technically , if I make a voltage divider in which the middle point is say at 50 volts measuring to the actual circuit ground then putting the ground pin of the 7805 to that divider's middle point the output from the regulator would be 55 volts?
in other words the regulator only regulates between the two points it is being attached to, no matter that the point which the regulator uses as the reference might actaully be way above the actual common circuit ground or lowest circuit point?

that's quite handy , even though when using the regulator with a resistor voltage divider to achieve higher regulated voltage the current would then be limited to that of the voltage divider correct?
by the way I got my 7815 regulators. will try them soon.

 

[Averagesupernova]

The current in the divider has nothing to do with the current that the regulator can source. This part has not changed. The divider is in parallel with the load. I don't like to take a regulator very far away from the design spec when using that trick. The heat sink tab of a 78xx is connected to the ground pin so this also needs to be kept in mind.
-
Just a quick google: https://www.google.com/#q=using+a+7805+as+an+8+volt+regulator

 

[Baluncore]

The 78xx series requires a good solid reference. Once you mess with the ground reference terminal weird things start to happen. For example, a LED in the ref will lift the output by a few volts, dependent on the colour, but in multi-rail supplies it may obstruct the start-up sequence and the output of a regulator may be pulled to reverse polarity.

If you use a 7805 as a 55V reg then all seems OK for steady state, but do not forget that the maximum ref to input voltage is only about 35V. That configuration cannot be started or stopped without risking device failure. An output short circuit will destroy an otherwise protected regulator. By the time you have provided enough protection it has become uneconomic.

Protection circuits are inherently unreliable because they are not always exercised and so cannot be easily tested.

 

[Salvador]

I asked more about the 55 volts as a thought experiment rather than an actual idea of doing something like that since these days one can get all kinds of semiconductors without the need to make something from scratch