Author Topic: sg3525 gets mosfets heated  (Read 5574 times)

0 Members and 1 Guest are viewing this topic.

Offline dave_j_fanTopic starter

  • Regular Contributor
  • *
  • Posts: 169
  • Country: in
sg3525 gets mosfets heated
« on: December 28, 2019, 05:42:44 pm »
dear all
 i am trying Sg3525 inverter.
issue is Mosfets getting superhot in 2-3min


original maker showed no use of heatsinks..so i have not added heatsink

just running 60w tablefan

pcb ofthe originalmaker is like this u cab see the connections
very.clearly.

please advice how to reduce heating.

i swapped diodes 4007 and mosfets z44

running 60wtable fan

transformer  15amp 12v
« Last Edit: December 28, 2019, 06:00:33 pm by dave_j_fan »
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: sg3525 gets mosfets heated
« Reply #1 on: December 28, 2019, 09:12:45 pm »
Schematic?  :-//

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

Offline dave_j_fanTopic starter

  • Regular Contributor
  • *
  • Posts: 169
  • Country: in
Re: sg3525 gets mosfets heated
« Reply #2 on: December 29, 2019, 01:31:39 am »
attached pcb
has good clarity.


i will draw a schem.if not clear.
 

Offline oPossum

  • Super Contributor
  • ***
  • Posts: 1472
  • Country: us
  • Very dangerous - may attack at any time
Re: sg3525 gets mosfets heated
« Reply #3 on: December 29, 2019, 04:11:21 am »
High side gate drive is probably horrible. Slow opto to turn it on, and a 2k resistor to slowly turn it off. A proper bootstraped gate driver IC would be a big improvement.
 

Offline dave_j_fanTopic starter

  • Regular Contributor
  • *
  • Posts: 169
  • Country: in
Re: sg3525 gets mosfets heated
« Reply #4 on: December 29, 2019, 09:59:21 am »
any alternate circuits.. avoiding ic due to cost increase
 

Offline dave_j_fanTopic starter

  • Regular Contributor
  • *
  • Posts: 169
  • Country: in
Re: sg3525 gets mosfets heated
« Reply #5 on: December 29, 2019, 10:42:33 am »
898516-0

here is diagram

update: some where i found that  it is better to drive one pair of mosfets with boot strap(probably dcdc booster of low current)so that they get atleast 18v on gates
« Last Edit: December 29, 2019, 11:12:27 am by dave_j_fan »
 

Online Zero999

  • Super Contributor
  • ***
  • Posts: 20358
  • Country: gb
  • 0999
Re: sg3525 gets mosfets heated
« Reply #6 on: December 29, 2019, 11:57:06 am »
(Attachment Link)

here is diagram

update: some where i found that  it is better to drive one pair of mosfets with boot strap(probably dcdc booster of low current)so that they get atleast 18v on gates
What do you expect, the high side MOSFETs are source followers, so will never turn fully on.

Do away with the h-bridge and use a centre tapped transformer with the +V going to the tap and a MOSFET going to each side of the primary.

Why not use a 12V fan to start with? Then no inverter is required.
 

Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11714
  • Country: my
  • reassessing directives...
Re: sg3525 gets mosfets heated
« Reply #7 on: December 29, 2019, 11:58:26 am »
i suspect Q1 and Q4 must be P channel mosfet. Q2 and Q3 must be N channel mosfet which is correct from your drawing.
« Last Edit: December 29, 2019, 12:01:28 pm by Mechatrommer »
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Online Zero999

  • Super Contributor
  • ***
  • Posts: 20358
  • Country: gb
  • 0999
Re: sg3525 gets mosfets heated
« Reply #8 on: December 29, 2019, 12:00:29 pm »
i suspect Q2 and Q3 must be N channel mosfet. Q1 and Q4 must be P channel mosfet which is correct from your drawing.
No, they're clearly drawn as N-channel and if they were P-channel, the circuit wouldn't work at all, since they would be off, when the N-channel devices were on.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: sg3525 gets mosfets heated
« Reply #9 on: December 29, 2019, 12:02:05 pm »
Oh yeah, those optos take like 100us to turn off, plus gate capacitance.  Those poor transistors are toast.

Also the capacitance at the output probably isn't doing what you thought it was doing.  At best / most likely, it's just making things heat up more, but exactly what happens depends wildly on what the transformer is.

Try a better driver circuit, like, sorta like this but a simpler version will do: https://www.seventransistorlabs.com/Images/CMBuck_Output.png I don't seem to have a better example to hand.  The bottom-left stuff is level shifting, including differential signaling.  You could change out the that with an always-on current source or sink, and only one switching channel (the inverting channel -- it matters!) from the level shifter section.  You might be able to get rid of the emitter followers, but in any case do take care that turn-off is faster than turn-on so as to avoid shoot-through.

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

Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11714
  • Country: my
  • reassessing directives...
Re: sg3525 gets mosfets heated
« Reply #10 on: December 29, 2019, 12:02:16 pm »
i suspect Q2 and Q3 must be N channel mosfet. Q1 and Q4 must be P channel mosfet which is correct from your drawing.
No, they're clearly drawn as N-channel and if they were P-channel, the circuit wouldn't work at all, since they would be off, when the N-channel devices were on.
yeah i got it mixed up, earlier post corrected.
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline dave_j_fanTopic starter

  • Regular Contributor
  • *
  • Posts: 169
  • Country: in
Re: sg3525 gets mosfets heated
« Reply #11 on: December 29, 2019, 12:31:58 pm »
i suspect Q1 and Q4 must be P channel mosfet. Q2 and Q3 must be N channel mosfet which is correct from your drawing.
no the 4 mosfets are n channels as per design.
the diodes helping for boot straping.

but i have used 4007 in place FR107.may be slow diodes not ableto boost voltage
 

Offline dave_j_fanTopic starter

  • Regular Contributor
  • *
  • Posts: 169
  • Country: in
Re: sg3525 gets mosfets heated
« Reply #12 on: December 29, 2019, 12:52:11 pm »
after searching a bit i got a high side driver with 4148
and also just 547 npn transistors .

i am bit confused how do i use it
 

Online Zero999

  • Super Contributor
  • ***
  • Posts: 20358
  • Country: gb
  • 0999
Re: sg3525 gets mosfets heated
« Reply #13 on: December 29, 2019, 02:12:07 pm »
i suspect Q1 and Q4 must be P channel mosfet. Q2 and Q3 must be N channel mosfet which is correct from your drawing.
no the 4 mosfets are n channels as per design.
the diodes helping for boot straping.

but i have used 4007 in place FR107.may be slow diodes not ableto boost voltage
Oh, I missed the bootstrapping part. Still you've got the slow switching speed of the opto-couplers and the 2k2 pull-down resistors.
It's possible to make a bootstrapped high-side MOSFET driver from discrete parts. Two transistors need to be used: one to pull-up and another to pull-down.
LTSpice simulation

https://www.eevblog.com/forum/projects/high(ish)-frequency-bootstrap-circuit-for-high-side-switching/msg2678304/#msg2678304

My advice is not to bother. Use a centre tapped transformer. Or better, scrap the inverter and use a 12V fan in the first place. It's always best to avoid voltage conversion whenever possible,
 

Offline dave_j_fanTopic starter

  • Regular Contributor
  • *
  • Posts: 169
  • Country: in
Re: sg3525 gets mosfets heated
« Reply #14 on: December 30, 2019, 04:33:12 am »
thank u .first of all, "issue is fixed "after adding high speed diode.

radically deviating from inverter is not intended.
the unit should be deliverable to poor via volutary organization.
ct transformers have lower efficiency.i got10 one coil transformers  made here before putting the proto.So i willl use this with them.

dc fan no ways a solution.many households have Ac fans only can't force to buy dc.they must be able to use ac led bulbs.in  my country dc led bulbs have very poor quality.

some changes i will do
-adding feed back,i have some circuits at hand
-will try with dual transistor highside and compare performance.
need to add overload shutoff ..low battery off and a smps charger




 

Online Zero999

  • Super Contributor
  • ***
  • Posts: 20358
  • Country: gb
  • 0999
Re: sg3525 gets mosfets heated
« Reply #15 on: December 30, 2019, 09:09:42 am »
Oh, I remember now. You want to make an inverter on the cheap, so it can be used in poor countries.

This is not the way to make the cheapest inverter possible.

Big, iron cored transformers are relatively expensive. The cheapest way is to use a small, high frequency transformer to build a DC:DC converter to boost the 12V to the peak voltage of the mains, so 340V, assuming 240V out and add the H-bridge to convert the 340VDC to 230VAC on the secondary side. If you really want to make it cheap, then you can skip the isolation and voltage regulation. I accept it may seem more complex, but the it works out cheaper, because the high frequency transformer in the DC:DC converter is a fraction of the size of a mains frequency transformer and as the power level is increased further, the cost saving becomes ever greater.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: sg3525 gets mosfets heated
« Reply #16 on: December 30, 2019, 11:57:29 am »
Iron core transformers may not be all that unavailable.  Keep in mind 3rd world countries get all our junk.  Which includes tonnes of wall warts that are too heavy for our tastes anymore.

I've built that before, which is also basically a commercial unit (which you might rightly ask, isn't very expensive either so why not just buy one and have it shipped in? -- but therein lies the problem, what if you can't even do that?).



Shame I didn't write down the schematic, I don't remember what all it does.  Bottom-center: TL494 control board (with opto feedback, I think), bottom left: push-pull forward converter (think there's a dual gate driver on the board).  Current mode control.  Also a UC3843 flyback converter for HV side supply; this is a fully isolated design.  Transformer is good for about 200W, and runs at 100-some kHz I think.  FWB output and choke filter.  Uses SiC schottky because I can (they run very cool; you'll have to be mindful of heat dissipation using FR307s or whatever).

Power (170V DC) crosses to the inverter board on the right, a TL494 at ~120Hz and fixed (~50%) duty, and, I forget what I did for current limiting but there is active current limiting at the output.  Chokes on the breadboard are for output filtering, and I later added some film caps to filter things further.

Output: "modified sine" (50% pulsed bipolar) waveform, correct peak and RMS but dV/dt obviously wrong.  Current limiting gives excellent performance into capacitive loads and short circuits (I didn't test shorts, come to think of it, but it should be okay for at least a little while..).

But as you can see, it takes a lot of parts to do this, and a heck of a lot less to use an iron core transformer.

OP: would it not also be attractive to use BJTs instead of MOSFETs?  A discrete circuit could be made, without needing anything more than C1815 and A945 (or any other GP BJTs), some power transistors, and which drives a 12V winding H-bridge style, 100% square wave, without current limiting.  (And could add current limiting for probably less than a dozen additional components.)

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

Online Zero999

  • Super Contributor
  • ***
  • Posts: 20358
  • Country: gb
  • 0999
Re: sg3525 gets mosfets heated
« Reply #17 on: December 30, 2019, 10:47:51 pm »
Iron core transformers may not be all that unavailable.  Keep in mind 3rd world countries get all our junk.
That's a good point, although the schematic in the original post shows a 60W (the apparent power will probably be higher) fan which is too big for a wall wart transformer, unless lots are used in series/parallel which would be very bulky and inefficient.

I wonder if it's possible to overclock a wall wart transformer by using it a higher frequency? A 120V unit could be run at double the primary voltage and 120Hz, to give 240V, thus double the power rating. Although it will be difficult to find the correct ratio to get 340V out with 12V in and a wall wart rated to even 30VA is still rare.

How about using salvaged cores from old PC switched mode power supplies? It shouldn't be too difficult to design something which would work with a rewound PC PSU transformer.

Another thing to bear in mind is the load. A modified sine wave is unsuitable for powering refrigerators and cheap LED lamps which use capacitive voltage droppers.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: sg3525 gets mosfets heated
« Reply #18 on: December 31, 2019, 01:28:17 am »
Just an example, anything will do of course.  Hm, I wonder what offhand would use a lot of transformers that size, it's enough VAs for a medium TV or small desktop PC, but neither of those have used iron transformers, almost forever...  Audio amps of course, but always at higher voltages.

In any case, OP seems to be confident in their supply.  Yes, rewinding is an option, cheap labor and recyclable materials (and a lax demand for safety compliance) would allow lots of mix and match.

Seems like 12-20V 50-300W SMPS (power packs, laptop chargers, etc.) are hella common these days (and have been for a decade or more), for which a solution like I suggested in the other thread would be fine,
https://www.eevblog.com/forum/beginners/mc34063-high-voltage-dc-dc-boost-converter/msg2850352/#msg2850352

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

Online Zero999

  • Super Contributor
  • ***
  • Posts: 20358
  • Country: gb
  • 0999
Re: sg3525 gets mosfets heated
« Reply #19 on: December 31, 2019, 12:13:34 pm »
That seems reasonable. I suppose if you really want to cheap out, skip the SG3525 and go for a self-oscillating topology.

I wonder how well a small mains frequency transformer would work for driving the MOSFET h-bridge on the secondary side? Obviously it would need to be rewound to give the desired gate drive voltage and the mains PWM circuit buffered with snubber diodes to work properly.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf