TS100 soldering iron MOSFET repair and circuit reliability improvement

[Gyro]

Well after a couple of years of duty as my main iron, running at 24V, I finally managed to blow the MOSFET in my TS100. It was my own stupid fault, I was trying to unsolder a component leg from a solid copper (not FR4) sheet heatsink.  Instead of taking a moment to change to the heavy C4 tip, I was trying to use the smaller BC2 tip while leaning on the temperature boost button for an extended period.

Normally, a MOSFET going short like this results in a glowing tip, but luckily my 'modified to 24V' HP laptop adapter decided that it didn't like 100% duty cycle on the element and went into overload cycling, so the tip didn't get past a very dull red.

Time to look for a replacement mosfet then. The original was a Chinese (HM Semi) HM4953, equivalent for the Si4953 Dual P-channel 30V mosfet. R_DS(on) is listed as 53mR @ -10V_GS. A quick browse through the Vishay Siliconix website parametric search revealed the Si4925, which has an R_DS(on) of 25mR @ -10V_GS in the same SO8 package.  An ideal beefier replacement candidate.

Looking at the TS100 schematic (1st image), the mosfet is badly drawn, it's not immediately apparent that it is a dual channel part with completely separate Drain, Source and Gate pins. The diodes shown are the parasitic ones. What also becomes apparent is that they've wired the two mosfets in inverse series, with Q1B effectively being a redundant ideal rectifier (if turned off, the parasitic diode is still forward biassed).

I assume they must have put the mosfets in series for PCB layout reasons, however it means that the on-resistance of the package is effectively doubled to 100mR, no wonder it got hot! As the Si4925 replacement has half the on-resistance, a single mosfet channel would have a quarter of the power dissipation of the original package. It would also have the same gate capacitance as the pair of mosfets in the HM4953. So my initial decision was just to bypass Q1B and use just one mosfet in the replacement.

The second image shows the pad layout with Q1A and Q1B identified, together with pad idents. You can see that I had already isolated the relevant Q1B pads with kapton tape and added a wire to the +ve centre pin of the power connector.

The third image shows the Si4925 fitted in single mosfet configuration. The gate of Q1B is solder bridged to the source to stop it floating.

At this point I started thinking that maybe using both mosfets in parallel might be a better solution. Granted, the gate capacitance would be doubled, but looking at the schematic, the gate drive is rather sluggish anyway (2k5 source resistance). Having both mosfets in parallel should significantly improve the SOA, and result in half the on-resistance for any given gate voltage. This is how I've left it, as shown in the fourth image.

It's debatable whether the single or parallel mosfet configuration is best. I should really have compared switching waveforms of the two. In either case most of the heat conduction to the PCB will be through the Q1A Drain pads, area (the PCB Drain pads for Q1B being isolated by the kapton). I tried to compare the package temperatures with my thermal camera in both configurations. |t is partly obscured by the CPU PCB when assembled, but after prolonged operation at maximum tip temperature I couldn't see any difference between the two, and the package didn't exceed 40'C with the case open. I couldn't detect any difference in the stability of the F/W PID stability either, the tip temperature is nice and stable.

Note: The marked up schematic in the first image shows the parallel mosfet configuration.

[Gyro]

Just an update on the repair...

A couple of days ago I again had an issue with the iron. This time was different though - the 24V mains adapter went into overcurrent shutdown and display went back to the idle screen, rather than the whole thing of going into continuous overcurrent shutdown-restart cycling with the tip starting to glow.

After a few minutes thought, I realised that this time the MOSFET must still be operational. Hmm, maybe my modified mains adapter's current limit had suddenly dropped. As a rather foolhardy test, I connected it to my modified Lidl 20V Lithium drill battery pack. It blew the fuse in there too when I pushed the heat button (my add-on socket is protected by a T2A fuse).

Long story short, I did what I should have done in the first place and checked the tip. I found that it had developed an intermittent element to barrel short (the barrel is grounded to the -ve on the earthed adapter). When cold it read fine, but after putting in the iron and firing up I immediately heard a tick and after removing and measuring again the direct short (+ve connection to barrel) manifested.

Whether the original MOSFET failure was down to the tip developing an intermittent short, or whether the intermittent short was down to being heated to dull red heat, I will never know (I suspect the latter), but I now have good confidence that my new MOSFET selection and circuit arrangement will repeatedly withstand a shorted element with both of my power sources.

Looking back through my ebay emails, I found that this BC2 tip has been in regular use since 2017 ( ), not just a couple of years, as I had thought. It might have lasted much longer if the original mosfet hadn't failed, the plating is still fine.

[vidarr]

I am in this same position now. After changing the mosfet, how has it held up?

[Gyro]

Yes, I'm pleased to report that the mosfet change has been completely reliable (on a sample size of one anyway). The TS100 has become my 'daily driver' and I run it at the maximum 24V supply, and occasionally on the 6S Lithium 20V Drill battery pack.

I have stayed with the parallel mosfets configuration, as detailed in the marked up schematic and final image in my OP. This reduces the SO8 package dissipation to a tiny fraction of the original. I haven't seen any adverse effects from the increased gate capacitance.

As I mentioned in my follow-up, I found my original tip intermittent short circuit after the overheating episode. I still don't know whether this was a chicken or egg situation, but it would be a good idea to measure the resistance of your element - It should read around 8 ohms, with open circuit (or at least several Megohms) from element to tip).

I hope this helps.

[vidarr]

In my case, I am not sure what happened. I turned it on the other day and the iron turned on for a second, then shut off completely and the PSU started blinking the light. I put a brand new tip in and the tip went straight to glowing red in seconds.

I am confused about what you did. Your last picture is how your iron is now? Or did you make another change after your "update" comment?

Your idea to use the Si4953 is good and I will go with that, but lacking the electronics knowledge that you have, I think maybe I should solder the new mosfet in the stock design. Am I missing a chance to make a big improvement to the iron if I go with stock?

Thank you for taking the time for this!

VK

[Gyro]

In my case, I am not sure what happened. I turned it on the other day and the iron turned on for a second, then shut off completely and the PSU started blinking the light. I put a brand new tip in and the tip went straight to glowing red in seconds.

From the sequence that you describe, it sounds as if the first tip went short-circuit, this must have caused the mosfet to fail and the PSU to go into overcurrent mode. As this tip was shorted, you would not see it glow (it would not have enough resistance to heat up). When you put in the new (good) tip, the already failed mosfet caused it to glow. That's my best guess anyway.

I have only ever purchased genuine Miniware tips, I avoid the 'compatible' ones of unknown quality. I don't know if it helps, but I only had that one failure with the glowing tip and the exact circumstances are still unclear to me.

I am confused about what you did. Your last picture is how your iron is now? Or did you make another change after your "update" comment?

Your idea to use the Si4953 is good and I will go with that, but lacking the electronics knowledge that you have, I think maybe I should solder the new mosfet in the stock design. Am I missing a chance to make a big improvement to the iron if I go with stock?

Yes, my iron is configured as the last picture (and marked up schematic) in the first post. No further changes. Also, yes, you could simply fit the Si4953 in place of the original part. This will provide roughly half the heat dissipation in the mosfet package and twice the current capability. Hopefully this might be sufficient to survive the maximum current of your PSU into a short circuit, I don't know, but it is certainly an improvement (and you could also buy a spare).

My analysis was more about correcting a design weakness in the TS100 circuit. It uses the two mosfets in the SO8 pkg in series, when they could just as easily have put them in parallel - which would have doubled the current handling capability, and reduced the mosfet package dissipation by a factor of four (even before changing to the Si4953).

Using the Si4953, and in parallel configuration, doubles the current handling by another two times (now four times the original) and reduces the package dissipation by a factor of sixteen. I have (accidentally) proved that this is sufficient to survive a short circuit with my PSU and blow the 2A fuse on my battery pack without destroying the mosfet.

If you do decide to copy my parallel mosfet modification, then don't forget to isolate the two +Vin pads (pins 5 and 6) with Kapton tape and straighten these two legs so that they do risk puncturing the tape (pin 4, is not isolated for the parallel mod). The wire that I used is probably thicker than necessary. The best solution for you depends on your soldering skill and level of comfort - I wouldn't want to think of you killing your iron by performing a mod that you are not comfortable with.

Thank you for taking the time for this!

You're welcome. It's nice that somebody has found this thread useful.

[vidarr]

Originally, I was fixated on your pics, but then I just realized something when I reread your post: There are TWO MOSFETS on the board. How do I know which one is bad? Do I replace both of them with the Si4953?

edit: The second MOSFET has no label on it.

Thank You so much!

[Gyro]

The two mosfets are contained in a single package, the HM4953. The no label package you marked is an RT7272B switching regulator which powers the analogue parts, microcontroller and display, you must not replace this.

Just one correction. If you check back in my original post you will see that the higher current part that I used is the Si4925, not the Si4953. Sorry, I didn't spot this error in your Reply #4. The Si4953 is a direct substitute for the HM4953, with no improvement in specification.


As you seem uncertain, I would probably suggest that you simply replace the dead HM4953 with an Si4925. This will give you increased robustness without the risk of damage due to accidental errors in copying the full modification. I can't say if this would survive another shorted element with your PSU but it has a better chance and you can always stock a spare in case you should ever need it.

[vidarr]

Yes, I am going to do what you suggested and just replace the original with Si4925. The flux and solder where I am in Brazil is terrible quality, which makes soldering small things like this a nightmare ---especially being a novice.

If I had not found your post here, this iron would have been in a garbage bin. So, your help is so much appreciated.

[vidarr]

My replacement parts arrived a couple weeks ago from China. The shipping cost more than the parts.

Another question I have is about the soldering tips. It is clear to me that the first tip shorted and fried the chip. I put a 2nd tip in and it went glowing red in seconds, which is how I knew the problem was more than the tip. Is this tip ruined now? Or did it just get really hot with no damage? Is there a way to tell?

The chips came in a strip of ten, so there are extras, if the tip fries another chip. Any advice is much appreciated.

Thank You!

[Gyro]

I'm glad your parts arrived. I know part sourcing must be difficult for you (from your comments about flu and solder), hopefully the parts from China are genuine, or at least a reasonable quality substitute.

As for the second tip, it is very hard to say. The element should read 7 - 8ohms between its contacts, and of course no continuity between the element contacts and the barrel. My original tip only developed a short between the element and barrel once it started to heat up though and otherwise measured ok. If your second tip only glowed briefly, then there is still a reasonable chance that it survived, but no guarantee. The best thing would be to insert a 2A fuse in series with the DC supply to the iron as some level of protection when testing. Worst case you may need to change the mosfet a second time (but hopefully the higher current rating of the new one will save it).

[ledtester]

Using a current limiting power supply is another way to protect the TS100 in case the MOSFET is bad.

At around 9V the iron shouldn't draw more than 1 A. (Below 9V and the firmware might report "DC LOW" or something like that.) Also, at a lower voltage your iron will heat up slower giving you more time to react if it overshoots the set temperature.

Here is a video of my TS100 being powered at 9V. Note that I installed a version of the Ralim firmware a couple of years ago. You can tell when the PID control kicks in (and that the MOSFET is working) by watching the ammeter needle.

https://youtu.be/bBedKOVtnaU

[vidarr]

The device came with a 12V power supply and it is all I used.

"The best thing would be to insert a 2A fuse in series with the DC supply"

"Using a current limiting power supply is another way to protect the TS100 in case the MOSFET is bad."

So, will this 2A fuse still be necessary if I am using the original 12V supply? Is it a "current limiting power supply"? I have a fuse and supplies to easily hook it up.

Thank you guys! 

[Gyro]

The original supply will have some sort of overcurrent protection (to protect itself) - you saw it with the light blinking, but the current level was too high to prevent the tip from glowing.

If you don't have a bench PSU with current limit knob and a suitable 2.5x5.5mm plug to connect it to the TS100, then yes, a 2A fuse is the best bet. Worst case (if it's not possible to hook a fuse up), monitor the temperature display closely when you first power it up - I can't remember what happens if the mosfet is shorted, but I think it immediately displays a very high temperature.

[atax112]

Hello!

Registered just to reply...

I just blew my ts100 mosfet...doing some hack on a PIR light, turned the socked off but when I touched the phase connection on the PCB(I wanted to remove a test lead) I blew the whole flat dark and fried the mosfet and soldering tip...uuugh.  stupid

Anyway, found this gem of a write up and already ordering Si4925 (as you corrected yourself) off aliexpress.

Will try to save the iron, but it doesnt look good...

Wish me luck and thanks again!

[Gyro]

Hi atax, welcome to the forum.

Well that's an impressive way to remove the legs from a mosfet packeage! I'm glad you are chastened rather than injured.

I've been pondering the sequence of events...

- I suspect that you were working with the tip not grounded, but your power supply is a decent one with grounded output.
- When the tip touched the phase connection, the insulation on the element broke down, at the positive end because the fault current has clearly passed down the +ve line through the mosfets.
- From there, it probably found a path through the +ve line of the power supply to ground, either that or it found a path through Gate-Source breakdown of the mosfets and through the TS100 circuitry to the grounded PSU -ve.

As you say, the tip will have been destroyed, and obviously the mosfet package. What concerns me is what other damage has occured. The main worry is that the buck regulator U2, which feeds the logic supply from the main +ve rail may have broken down internally, passing some portion of mains to the microcontroller, display etc. There are also non mains rated capacitors and transistors in the mosfet gate feed and regulator input that might have gone short. There is a 24V protection Zener (D3) which might have clamped the surge, but probably have taken out D4 and R30 in the process.

I suggest that you proceed as follows..

- Check that there isn't a short circuit reading at D3 and then try to power the TS100 from a 5V supply via the microUSB connector. If the display doesn't show something then U2 and/or the logic probably died. Check for 3.9V output at U2 and 3.3V at the output of U5.
- If you get past that step, check the functionality of your power supply - does it still put out the expected voltage?
- If both of the above succeed, then carefully check all the passives, (and Q2) between Vin and U2 and also the Po signal path from CN2.

The mising pads on the mosfet footprint won't be a problem if you go for the parallel mosfet configuration in the OP, but I'm afraid you have rather a lot to check first. Of course, you will need to replace the tip before trying a heating test.

Yes, Good luck.

[atax112]

Hi atax, welcome to the forum.

Well that's an impressive way to remove the legs from a mosfet packeage! I'm glad you are chastened rather than injured.

I've been pondering the sequence of events...

- I suspect that you were working with the tip not grounded, but your power supply is a decent one with grounded output.
- When the tip touched the phase connection, the insulation on the element broke down, at the positive end because the fault current has clearly passed down the +ve line through the mosfets.
- From there, it probably found a path through the +ve line of the power supply to ground, either that or it found a path through Gate-Source breakdown of the mosfets and through the TS100 circuitry to the grounded PSU -ve.

As you say, the tip will have been destroyed, and obviously the mosfet package. What concerns me is what other damage has occured. The main worry is that the buck regulator U2, which feeds the logic supply from the main +ve rail may have broken down internally, passing some portion of mains to the microcontroller, display etc. There are also non mains rated capacitors and transistors in the mosfet gate feed and regulator input that might have gone short. There is a 24V protection Zener (D3) which might have clamped the surge, but probably have taken out D4 and R30 in the process.

I suggest that you proceed as follows..

- Check that there isn't a short circuit reading at D3 and then try to power the TS100 from a 5V supply via the microUSB connector. If the display doesn't show something then U2 and/or the logic probably died. Check for 3.9V output at U2 and 3.3V at the output of U5.
- If you get past that step, check the functionality of your power supply - does it still put out the expected voltage?
- If both of the above succeed, then carefully check all the passives, (and Q2) between Vin and U2 and also the Po signal path from CN2.

The mising pads on the mosfet footprint won't be a problem if you go for the parallel mosfet configuration in the OP, but I'm afraid you have rather a lot to check first. Of course, you will need to replace the tip before trying a heating test.

Yes, Good luck.

Hey, thanks for the quick reaction!

As usual, i wasnt patient, but found some 4925 at local store(small wonder in itself).... let me explain

Here are some things I didnt tell before:

After the whole incident(iron wasnt grounded, the point of it occured to me after), I flipped breakers on, got back to my "workhole", small place I set my tools up in our flat.
Checked the area, checked the lamp, on to the TS100...it turned on, didnt recognize the tip, thats when I found out. a miracle in itself

So ok, its on, fried tip, no biggie, lets swap it for another...oh ok, wont heat above 24C...so somethings off.

I picked it apart, immediately seen pieces fall out and carbon residue all around, started cleaning and saw the mosfet(had no idea initially what the IC is) blew up...

Google got me to your post, so glad it did and the course of action was clear.

Knowing our local options here from experience, I got on aliexpress and ordered replacement mosfets, both kinds and 2 oleds as well, since I burned it a bit during IC removal(turns out to be ok, just the upper layer got heat)

Kicking myself about what happened I went out to 1/2 of our local parts shops trying my luck...and lucky I got, 4 Si4925 in stock, got two, just in case.

Off of that bit of good karma I got home, did some work, started lunch...taped the PCB, did prep work.

Managed to find my Lidl, Parkside 3AA battery 4.5W soldering iron I modded to be able to run it off my LAB PSU...

Did the mosfet replacement, knowing the iron worked after the incident, did the mod like in OP last pic.

Et Voila! It works!

What a ride... have to do some real work with it yet, but Im confident its alive again. Hopefully it will stay that way, at least until I have another  moment

Got the whole set from Novel Life (or really close) store from Aliexpress, silicone psu cable and so on, so its not like the generic chinesium stuff I guess.

Cheers

[Gyro]

Wow, you did get lucky. I had very little hope that you would get away with just the mosfet. Nice save. 

I doubt you'll ever make the same mistake again!