Drive signal issues with half bridge

[zeus_threat]

Hi Microsim thanks for the reply sim shows too much offset with 741 in single supply mode LM393 gives very good sim results am completing testing on the overtemperature circuit

 

[zeus_threat]

Overtemperature circuit completed on breadboad and it works nicely. I need to check what the transients look like on scope when its switching into protection.

I have a question regarding over and under voltage protection. Do you normally sense the "voltage window" after ac mains rectification or at the smps rectified output? Am confused with all i see on the web.

The more i see the amount of protection growing the more i ask myself if all those sensors: overcurrent, over + undervoltage, overtemperature would not make things easier feeding all this into a pic and use only 1 output to switch off the oscillator. I think that could be an interesting project to post here, a universal SMPS protector. Soon maybe for the next smps as i complete this one

 

[zeus_threat]

Hi just an update the pcb layout are still going on. I will post first version here as its completed. Any help on that overvoltage/undervoltage protection is welcome thanks

 

[zeus_threat]

Controller Card

Hi here's what the controller card looks like and working on the main power pcb also

 

[zeus_threat]

Power PCB

Power PCb ver 1 completed Its almost identical to the prototype layout i need to rethink if everything is positioned properly before going further

 

[zeus_threat]

Just an update going on pcb version 1.1 the distance between the GDT primary and the controller card is too large and there are power lines crossing below. I am working on relocating the items to solve this issue


While this pcb is on its way gradually i have stocked some and need to purchase some other of the following items, straight from ebay and shipped internationally for a 800W smps that will be next smps project:

Core:http://cgi.ebay.com/ETD44-ETD-Ferri...910?pt=LH_DefaultDomain_0&hash=item3a66873bf6

Wire(REquired no strand will be used):http://cgi.ebay.com/31-AWG-GAUGE-MA...938?pt=LH_DefaultDomain_0&hash=item335f005732

Controller:http://cgi.ebay.com/10PCS-G3525ADW-...176?pt=LH_DefaultDomain_0&hash=item3a687ee680
OR
http://cgi.ebay.com/10-pcs-IR2153-H...549?pt=LH_DefaultDomain_0&hash=item2eb34e4435


Totem pole transistors: http://cgi.ebay.com/30-x-Transistor...458?pt=LH_DefaultDomain_0&hash=item27b122f7e2
http://cgi.ebay.com/30-x-Transistor...356?pt=LH_DefaultDomain_0&hash=item2a0056c2a4

Output rectifiers:http://cgi.ebay.com/20PCS-MUR860-MU...=260815968501&ps=63&clkid=2107557213619257932

Mosfet: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=250804860159&ssPageName=STRK:MEWAX:IT

 

[zeus_threat]

Currently rebuilding my exposure box to use led instead of old uv fluorescent tube. Once done i'll complete the adjustment on the pcb and post it here

 

[zeus_threat]

The exposure box is being rebuilt. I've noticed how bad those fluorescent tubes can get as they heat up a dark layer starts to build up and reduces exposure surface. This shouldn't happend with the UV leds.

 

[IGBT]

The exposure box is being rebuilt. I've noticed how bad those fluorescent tubes can get as they heat up a dark layer starts to build up and reduces exposure surface. This shouldn't happend with the UV leds.

UV is better

 

[giololucas]

Hi zeus.
I want to congratulate you to your efforts in your smps , I read the complete topic.:UP:

Could you help me?
I´m doing some try outs with the smps in attached.

MicrosiM told me to ask you about some troubles I´m facing x-
-Did you faced big voltage drop on secondary?
-Big ripple in secondary, like 20V!

Xformer nominal voltage is -+ 45V; real world no load 48V, with 7,5amps load 35V.

Thank you.

 

[zeus_threat]

Hi giololucas i've looked at your design. honestly i got a voltage drop from +-35 to +-32V when loading the SMPS from light to medium-high. you are mentionning 2 problems
1) the heavy voltage drop could be something inherent with IR2153 that has quite a high deadtime built in but it coud also be someting else
2) 20V ripple is huge with my basic capacitors didn't even yield such high ripple. It didn't even bother to measure it as i could not even see it when loading the SMPS.
Personally i wouldn't put a controller board facing mosfets like that just put your scope probes in front and you will see how much EMI they radiate when they conduct the board is also likely to heat up severely with heat radiated from the mosfet + heatsink on heavy loading.
Could you show GS and DS waveforms at safe level? Make sure you take necessary precautions to isolate everything and do this at a safe voltage level <40V
My guess but am not sure is that the mosfets are radiating back energy to IR2153 OP pins causing cross conduction. In turn protection triggers and latches the OP. Your caps energy get sucked out by the load since the mosfet stopped conducting causing the heavy ripple you see and it repeats like a cycle.
Look at Ludo and Microsim 1KW SMPS they are based on IR2153 its properly laid out

 

[zeus_threat]

Hi IGBT just saw your post well the fluorescent tubes i have used were UV insect killer lamps they worked well but heated up a lot during use causing those dark areas to appear UV leds are definitely better am still working on them with respect to distance to exposure area and proper layout. Id di also try standard CFL lamps they work fine as well but you need to be more patient and do some corection after exposure and developement

 

[giololucas]

Thank you for reply!

Hi giololucas i've looked at your design. honestly i got a voltage drop from +-35 to +-32V when loading the SMPS from light to medium-high. you are mentionning 2 problems
1) the heavy voltage drop could be something inherent with IR2153 that has quite a high deadtime built in but it coud also be someting else
2) 20V ripple is huge with my basic capacitors didn't even yield such high ripple. It didn't even bother to measure it as i could not even see it when loading the SMPS.
Personally i wouldn't put a controller board facing mosfets like that just put your scope probes in front and you will see how much EMI they radiate when they conduct the board is also likely to heat up severely with heat radiated from the mosfet + heatsink on heavy loading.
Could you show GS and DS waveforms at safe level? Make sure you take necessary precautions to isolate everything and do this at a safe voltage level <40V
My guess but am not sure is that the mosfets are radiating back energy to IR2153 OP pins causing cross conduction. In turn protection triggers and latches the OP. Your caps energy get sucked out by the load since the mosfet stopped conducting causing the heavy ripple you see and it repeats like a cycle.
Look at Ludo and Microsim 1KW SMPS they are based on IR2153 its properly laid out

Hi zeus!

Voltage drop is better now.
1- Output inductors removed (tks microsim).
2- In Brazil we have a local supplier of ferrite (see attach). This local material could not be used in 1100 ~ 1200 Gauss, value I considerated for primary turns calculation. My transformer must be used at least 1700 Gauss. So I slowed down the switching frequency and higher flux helps voltage drop to sustain at 6 volts for 8 amps of load (48,5V no load to 41,5V with 350W load). NOTE:I applied 350W in each rail in separated times.
3- Installed four 270pf ressonant caps in paralel secondary diodes.

About big ripple
I provide a isolation for oscilloscope from the ac mains. Big ripple gone. Now it has a ripple of 2V maximum.
Ripple has 2 points of observation:
1- Switching ripple = less than 1 Volt.
2- Ondulation from mains cap reflected in secondary voltage = less than 2 Volts.

I don´t have pics of GS and DS loaded. sorry (see attach).
If you have some time, I can take pictures of waveforms to discuss.

What do you think? values are good ? bad?
Should I put interference suppressor (inductors) in mains power?
Waiting for your comments.
Thank You for your tips:UP:

 

[zeus_threat]

Hi your waveforms look good for GS with SMPS unloaded But you will have to do more measurements GS + DS unloaded and loaded it will tell you if everything is correct. Also if possible try to measure the current flowing through the primatry loaded and unloaded it can be done with a current probe in case you have one or a current sense trasnfo and be careful to do this with gradually increasing voltage. Don't try it straight at full voltage even with isolation. The current wave form should be like a triangular waveform for the inductor and like a square wave with an upward slope for the mosfet like in this document: http://www.coremaster.com/appnotes/an108.pdf on page 2

A is the mosfet current B is the transformer voltage C is the transformer current am sure you must already know all that I think what you showed in previous attachment was the primary transformer wave form right? It looks basic simple but the proper functioning of any SMPS lies in those wave shapes. Another advice if you can't load both SMPS secondaries at once with 350W load both simultaneously with a smaller load.

Now another point its strange that all those ripple drop by x10 by just isolating the SMPS, were your scope and SMPS both grounded when you made the output voltage measurement? Ground loops maybe. As far as i know all smps measurements are made with isolation either at SMPS or using differential probes.

LC pi filters or double pi filters are there to prevent interference in and out of the circuit. you can see many designs on the net even app notes not using them but it needs to be implemented that interference will either end up in your smps or in one of your home appliance. If you are on your way to adding that filter, a varistor for added surge protection won't be something excessive in that circuit. EMI and surge are the two basic thigns you need to filter before rectifying AC mains to send to your smps

Another thing as well, most proper implementations of IR2153 use a totem pole or mosfet driver at their output, have you checked if their output current is enough to charge Cgs of the 840? In some cases i have read of these buffers protecting the IC from failing when the mosfets blow up but that needs to be verified but nothing won't replace a proper OC/OV/UV/OT set of protections. Not to get your design complicated but that fuse you have used won't prevent your mosfet and ic from blowing up in case something goes wrong.

 

[giololucas]

Hi your waveforms look good for GS with SMPS unloaded But you will have to do more measurements GS + DS unloaded and loaded it will tell you if everything is correct. Also if possible try to measure the current flowing through the primatry loaded and unloaded it can be done with a current probe in case you have one or a current sense trasnfo and be careful to do this with gradually increasing voltage. Don't try it straight at full voltage even with isolation. The current wave form should be like a triangular waveform for the inductor and like a square wave with an upward slope for the mosfet like in this document: http://www.coremaster.com/appnotes/an108.pdf on page 2

A is the mosfet current B is the transformer voltage C is the transformer current am sure you must already know all that I think what you showed in previous attachment was the primary transformer wave form right? It looks basic simple but the proper functioning of any SMPS lies in those wave shapes. Another advice if you can't load both SMPS secondaries at once with 350W load both simultaneously with a smaller load.

Now another point its strange that all those ripple drop by x10 by just isolating the SMPS, were your scope and SMPS both grounded when you made the output voltage measurement? Ground loops maybe. As far as i know all smps measurements are made with isolation either at SMPS or using differential probes.

LC pi filters or double pi filters are there to prevent interference in and out of the circuit. you can see many designs on the net even app notes not using them but it needs to be implemented that interference will either end up in your smps or in one of your home appliance. If you are on your way to adding that filter, a varistor for added surge protection won't be something excessive in that circuit. EMI and surge are the two basic thigns you need to filter before rectifying AC mains to send to your smps

Another thing as well, most proper implementations of IR2153 use a totem pole or mosfet driver at their output, have you checked if their output current is enough to charge Cgs of the 840? In some cases i have read of these buffers protecting the IC from failing when the mosfets blow up but that needs to be verified but nothing won't replace a proper OC/OV/UV/OT set of protections. Not to get your design complicated but that fuse you have used won't prevent your mosfet and ic from blowing up in case something goes wrong.

Hi zeus.
Thanks for fast response.

I don´t have a current probe. I will try current sense transformer :UP: . This is an important issue. I made current measurement with a 10 ohm resistor and scope using ohms law, looks crap, but was the first and fast idea to did it.
When I connected load on half secondary was to analise how variation will be in the unloaded side, because in past I had some troubles with this. Another point is that I dont have a ligth load for a while, sorry.

Rigth, that was a just a simple primary unloaded waveform.

This ripple was a nigthmare!!! Now I´m free from this crap! When ripple was big my scope was not grounded to smps, only cable of probe was conected (GND and signal).

Yes I did checked the gate charge of 840s direct in IR2153, ok. My next prototipe will be with drivers, they are inexpensive and could save the 2153, even though I had blow just 1 IC and 4 mosfets until now..

For sure you are completely right, set of protections is the best choice, my simple fuse is only to prevent big damages in electrical wiring.

For my next prototype, I don´t know if I start a new thread, since the design will not be different from many desingns posted here.

Thank you

 

[zeus_threat]

Hi giololucas glad to hear you got everything working and under control. Good luck for your next prototype

 

[giololucas]

Everything going good!

Hi giololucas glad to hear you got everything working and under control. Good luck for your next prototype

I did more tests and SMPS survived to:
-600W load (88V x 7 amps)
-Total ripple 4 volts maximum
-Voltage drop 8 volts per rail. Looks ok, because its a heavy load for this mosfets)
-Big heat 65ºC in the 840s (only 3 minutes to reach it !!! .
I did some tests with an amplifier I have build some years ago and result was pretty good, stable, nice sound and no noises.

I´m working in the second version slowly. More caps in main stage (4000uF), better switching transistors (ifrp460), better wiring in transformer and bigger heatsinks will be the principal improvements.
I think it will reach 800W minimum with stability. This is very good for what I´m looking for.
When I reach results I will post everything here, like I did first time prototype.
Thank you.

 

[MicrosiM]

Very good


Glad to hear that!