MT3608 Fails and burns

[@GnuReligion]

Years ago, I bought a reel segment of these MT3608 chips, and made a some hilarious spider-like circuits, in effort to keep the tiny SOT23-6 package cool. The device probably shunts a lot of current from the inductor as it operates, making it suitable only for low load applications. There are some switchers that have a big rectangular pad underneath, to help dissipate heat. This one is 1.2Mhz. A slower switcher (with a larger ring inductor), that controls a big honking TO-220 trenchfet outside the package may be a better rig to deliver significant power.

[@retireeelectronics2649]

The sheet states pin 2 (GND) is important heat sink. Very tiny heat sink even when coupled to the other side. Also the sheet indicates soft start and thermal shutdown at 155C. None of the modules worked after failing, even the ones that didn't smoke.

[@barnowl6807]

The sudden failure with a step change in current is a classic example of inductor saturation. Lacking the test equipment to verify this, (high bandwidth 'scope, high frequency current probe, etc.) you could load to just below smoke point and then carefully heat the inductor. The saturation flux density will change rapidly and smoke will appear.

[@retireeelectronics2649]

That is a nice idea. At this rate I have only 5 more tries (-:

[@richardpereyra5183]

Could the inductor be saturated while it is working by bringing a magnet close to it instead of applying heat?

[@mikropower01]

10:56 - I did not expect such a fire and smoke. I had created my own H-bridge with MosFETs and one was smoking and then I could see the chip inside white glowing and a little bit of smoke came out. It was bad that the P-Channel MosFET was damaged, but the light was nice.

[@retireeelectronics2649]

I didn't expect it to keep burning, might try burning a few parts and comparing it to high quality parts and see if there is much difference

[@mikropower01]

@@retireeelectronics2649 - I use this Step-Up converter to change the LiIon-LED lamp from 2 cells (7.2V - 8.4V) in line to only 4 cell, but all parallel, so that I have only 3.6V - 4.0V to charge, but I have to increase the voltage to 6 Volt for the LED-module. Now I have a constant brightness. I set this module for the lamp so that the power is constant 4 Watt.

[@rcinfla9017]

Issue is likely too physically small boost inductor saturating. As inductor approaches saturation due to high peak current its inductance plummets causing switching current by I.C. to shoot up very high, blowing MT3608. With 22 uH, it is operating in constant conduction mode. (large inductance for switching frequency). The larger inductance makes the inductor saturate at lower peak current.

[@retireeelectronics2649]

Have an assortment of SMD inductors and might try varying the value. Suspect I will smoke more of these before I am done.

[@LawpickingLocksmith]

Day one IDF training: This is how smoke looks like!

[@retireeelectronics2649]

(-:

[@DustinWatts]

As you said a the beginning those weren't real MT3806 IC's. The real ones actually work fine like in de datasheet. But the input and output capacitors are really important to prevent spikes and failures. I wonder what would happen with real MT3806's and bigger in/out capacitors. 4.7uF is too small. 22uF or even 47uF would do better. Don't know if you have them, but you could replace the in/out caps with higher ones and see if the behaviour changes. Anyway, fun to watch. Cheers!

[@retireeelectronics2649]

I have some 22uF SMD so will try that a bit later and see what happens. $0.38 per puff, I can live with that. Don't have any real MT3806's so, can't try replacing those yet. Have a good day

[@DustinWatts]

@@retireeelectronics2649 Yeah they are pretty cheap. I can understand you don't care if you blow one up :)

[@fabioeleuterio9382]

I use a lot of these SU/SD converters but always use like half the amps capacity they're designed for as precaution. So far so good...

[@retireeelectronics2649]

It was interesting how it stayed cool when I ramped up the current draw, then went smoky real fast. I agree, that using these in the lower current area is a much better idea.

[@110v4]

This module is very nice actually. Just be sure not to exceed 1A of input current for long term project. Interestingly they can also be modified to buck-boost converter using dual winding inductor I also try to modify it into 3.7V to 100-240V DC converter. It works but not very efficient. Maybe because the inductor that I use is not suitable to be used in MHz range (Its difficult to find dual winding inductor with high enough inductance that can operate at such high frequency) I also find that this module didn't have proper soft start feature

[@retireeelectronics2649]

The specs claim soft start but my bet is there are both the real chip and fake ones out there. It would be nice if price was an indicator but probably not. At the low current output, so far it stays cool. Going to do a bit more testing but don't like that it caught fire and continued to burn @18:50. I have seen a few circuits blow, but then it is all over, not this time, just flames. Good module to play with. Just looked at your video, nice modification.

[@johncoops6897]

Keep in mind that the rated power on all Boost Converters is based on INPUT power (and current). It's a common misconception that the max rating applies to the output side. You cannot expect a Boost Converter rated at 2A to output 18V at 1.5A... of course it will fail. Eg. 10:33 - because you are pulling 2.2A through a 2A rated device. I'd guess that the clone SDB628 doesn't actually have the thermal protection like the MT3608. The datasheet of SBD628 is actually a photoshopped version of the MT3608 one 😂

[@retireeelectronics2649]

In this case for the Maximum rating they probably used someone else's data sheet. The efficiency curve for 13Vin and 18V out graph shows 1.7 amps. They were a bit optimistic for their product since it failed much earlier.

[@johncoops6897]

@retireeelectronics2649 - yeah, the copy of datasheet is obviously photoshopped - they didn't even match the font when they replaced "MT3608" with "SBD628" 😮

[@retireeelectronics2649]

@@johncoops6897 I thought that was funny with the mismatch Font (-:

[@norbertsmallegange6331]

How to use regulatus perform under Dynamic Load so if you would make a Load from a transistor with a Square Wave opening and Closing the transistor making more and less current overshoot /sag?

[@retireeelectronics2649]

Don't see any reason why you can't make a variable load using pulse width modulation into a MOSFET with a resistive load in series. Would be a nice project.

[@deterdamel7380]

I ordered 10 of them on AE for 2.54€. Out of the box one failed at beginning (sorry no smoke). It seem that they interfere in my application with the 433MHz receiver a reducing the sensibility a bit. I'm not checked the capacitors yet.

[@retireeelectronics2649]

I was wondering how noisy these might be. The ones I have don't have magnetic shielding on the inductor. Some I see on AE do have better looking inductors and I wonder if that quality is carried through the rest of the parts.

[@charlesurrea1451]

I take it you haven't invested in one of those FLIR cameras? When Caterpillar came out with their phone that had one built in I had to snap one up. I was tired of burning up my digital thermometer (finger) looking for shorts on boards.

[@GnuReligion]

I'm proud to still possess all my fingers, and at least some feeling in most of them. There have been some close calls.

[@retireeelectronics2649]

Have a few items on my list before a FLIR camera, anyway the price keeps coming down or resolution up. My wife wonders how I can handle hot items so easily, I just say "I solder"

[@elvinhaak]

@@retireeelectronics2649 Ÿes, well, I have some problems in my nerves too since long but after soldering and desoldering for a while, not only do I not feel the temperatures but also almost or no pain before I smell my skin getting white. After a 'soldering-holiday', first days is quite painfull and then I am more likely to use some tools to hold. For getting more acurate temperature-measurements, I also like to use a K-thermocouple on a cheap multimeter like 'we' all had in the days. For just getting a bit like ' heat camera', there are small sensors to combine with a smartphone or computer. I want to test those first before I will invest in a real flir. I have a small one from Ali that is very low in resolution and gives some indication about where it is getting hot so I can measure on those points with the thermocouple.

[@daCount0]

@@retireeelectronics2649 There are good cameras out there for about 100 $ which fits on the phone USB-C. Not that big deal anymore.

[@retireeelectronics2649]

The ones I looked at need to download software from their company sites. That is not going to happen on my primary phone. I have been looking for a dirt cheap expendable phone and if I find one, then it becomes an option.

[@elvinhaak]

Is that the specs of the boards, or of the chip that is on the board? I would not have expected they would go over an Amp for sure without extra supporting parts on these voltages. Maybe with other capacitors inplace? Maybe worth another test with the original or more capacity and then just around one amp. Maybe measure with a scope on it and see when they go into oscillation-ramp which is probably why the chip failed soo rapidly. I guess it is ramping up in frequency or base and go into 'diode mode' without doing any switching, makes a short and then create smoke and more heat. Hope you have good ventilation... ;-)

[@retireeelectronics2649]

Specs are for the chip and in most websites the output current 2A is listed. I started with 10 modules and now down to 5 (-:. Most likely the B628 does not have the same specs as the MT3608 which it is being substituted. Plan on more smoke testing, I mean I mean current testing (-:

[@ColinMcCormack]

Would the specified capacitor help the spike?

[@retireeelectronics2649]

Not sure the the right capacitor size would help. An inrush resistor might be more in order.

[@TheEmbeddedHobbyist]

I expect that they just looked at the words in the data sheet but did not read them. There is a range of inductor values and a statement that it should have a low core loss at 1.2 MHz and a low DCR. I bet they just took a cheap indicator out of the parts bin as they are not going to go to the expense of going right. Not a very good data sheet as most of switching ic's datasheets go into a lot of detail about inductor selection and pwm period etc. But as you say you get what you paid for, if your going to design electronics circuits why not just include the switching psu as well. The you know it will do what you want it to.

[@jyvben1520]

some circuits need several voltages

[@retireeelectronics2649]

At roughly $0.38 each can't expect a lot. But the fact it burned and kept burning is very poor.

[@TheEmbeddedHobbyist]

@@jyvben1520 that's why you design the psu chip to provide the voltage it needs for each part. these one design does all will never work. and the components shown on the datasheet are the minimum needed to get it to work. not to work over every condition you throw at it.

[@TheEmbeddedHobbyist]

@@retireeelectronics2649 well if the inductor is not tot the required spec then switching currents will most likely be out of spec for the chip to handle.

[@retireeelectronics2649]

Will try another inductor in the future but at the rate I am cooking these I have only 5 more chances (-:

[@ekiskaliburnirvana9047]

i bought 5 of them recently, they all burned while adjusting voltage without load at around 17 volts. Not my fault because i use these modules for years. Anyway i bought a few from another supplier and they are all fine up to 25 volts. So time to time these are faulty. I think of a wrongly used low voltage output capacitor maybe 16 volts instead of 35 volts. Capacitance drops and peak currents burn the chip

[@retireeelectronics2649]

Probably comes down to fake vs real chips. An would not know until it is plugged it.

[@SymbolTech21]

Don't set the house on fire 🔥🔥🔥

[@retireeelectronics2649]

I always have a fire extinguisher sitting near the door. Where I worked last we all got a fire fighting course, 5 pounder vs gasoline fire, neat course. They stressed, everything in the building is replaceable, hence get out and never let the fire get between you and the exit.

[@BLOCKsignallingUK]

Had a lot of failures with these. Now use XL6007.

[@retireeelectronics2649]

The project I have for these will be 200 to 500ma at 18V plus some 12V at similar mAs. I figured these were overrated, but was hoping for 18V at 750mA ish as a safety margin. Big concern I have now is it burned to easy.

[@gazzacroy]

i've had a few things like that over the passed and i never trust them over about 500ma for a long term project. and a lot of them use fake chips :( cool theo :)

[@retireeelectronics2649]

Failing was bad enough, but to keep burning was very poor. Probably not top quality for the price. Have a good day.

[@gazzacroy]

@@retireeelectronics2649 100% agree. in some ways these could be a safety issue could end up burning your bloody house down :( not good really lol.

[@micbanand]

those fake cant handle the frekvens. and it is, absolute not heping it, when caps are to small. caps is very importen. try with current messure on input below 85% and caps above the recommended 22uf :)

[@retireeelectronics2649]

The 22uF will be the next test, but at best these will only be useable at under 500mA. Main concern is "are these a fire hazard". Have a good day

[@GoodWill-s8j]

Is this Chinese stuff? I find it very sus 💀

[@retireeelectronics2649]

Yes and was a super good deal (-: (-: $0.38 each. At some point I might try getting real chips and see if swapping onto the board would work.