Hi all,
My shiny new oscilloscope is due here by Friday, so after thinking it over, I am going to "slow the line" for a couple of days to allow it to get here and give me an opportunity to identify and fix the MOSFET failure issue. The failure rate due to this issue is too damn high, and I believe it is a solvable problem. I want to solve it now, before I ship any more heaters.
I now have a good idea of both of the root cause and also how to fix it, but I need the scope to tell me exactly what's happening, and also if my fix solves the issue.
To Science - huzzah!
I currently have a mini-batch of ~20 heaters that are almost done, as well as a few repairs, and instead of shipping them immediately I'm going to hold onto them for a few more days, until I've had a chance to look at the heater with the new scope.
I expect to start shipping again on Monday, hopefully with a fix in place, or at least a much better understanding of the issue(s).
Thanks for your patience while I work on getting to the bottom of this issue.
Cheers,
Well said , have been waiting my turn patiently wishing I'd ordered earlier , but am more confident after reading this thread that not only will I receive a quality product but the Cs is also top notchThank you @mr_cfromcali ! While I go through a rollercoaster of emotions while waiting my turn, I empathize with you and how much frustration you are dealing with. Keep at it.
I am very much interested in learning more about your oscilloscopic investigation and results. Best of luck!
Is your shadow cap loose? Maybe a refit is in order? (Easy fix)
Thank you @mr_cfromcali ! While I go through a rollercoaster of emotions while waiting my turn, I empathize with you and how much frustration you are dealing with. Keep at it.
I am very much interested in learning more about your oscilloscopic investigation and results. Best of luck!
Well said , have been waiting my turn patiently wishing I'd ordered earlier , but am more confident after reading this thread that not only will I receive a quality product but the Cs is also top notch
Thanks for the support. In retrospect, I should have jumped on this issue harder in mid-February, when it first appeared, but hindsight is 20/20, etc. and I've learned a lot through my missteps and false starts.
I have done a fair amount of reading on this recently, as I realized a simple parts swap wasn't going to solve it. Electronics is a hobby for me (as I'm sure any EE browsing this thread quickly realizes, ), and my previous projects have mostly been tube (valve) amplifiers for guitars. Consequently, there's a lot I don't know about MOSFETs, which are anathema in the tube amplifier world. Others do know MOSFETs, though, and it turns out that MOSFET-killing voltage spikes are well understood phenomena, and there are some straightforward strategies to mitigate them. TIL.
But all of these solutions require a decent oscilloscope for diagnosis and confirmation, and my previous 'scope wasn't up to the task.
The latest update from UPS has the new one arriving tomorrow. I'll definitely give this thread an update as soon as I have some test results - could be Friday (Pacific time), or it may be later if I have trouble getting the 'scope set up.
I don't think this problem will be hard to spot once I can trace the energy path of the circuit on the screen. I'm expecting to see a BIG voltage spike when the tactile switch is released and the circuit opens, as the unspent inductive energy is rapidly converted into heat and electrical energy*. I think that's where the damage occurs, and as I said, other people much smarter than me have already worked out proper ways of dealing with issues like these. I just need to see what's occurring in all of its noisy glory, so I can figure out which approach is the best to use to tame it.
That's what I know at the moment. More to come.
Glad mine is still working great but also glad to know I bought it from you in case there is a problem. I don’t think mine has a glass spacer between the switch and the pressure switch. Looks like a domed piece of plastic semi melted lol, could that be the reason it runs real hot with a ti tip(14mm coil), it goes in too deep? Or does it matter?
Are you having problems with the “14mm” or all sizes ???
This truly is a positive development. Thank you for providing immediate feedback with the oscilloscope. Looking good man!I would say that all sizes are subject to the issue, but the 14mm is the most at risk for MOSFET failure due to the way inductive energy spikes are generated. Of the sizes I offer, that size has the fastest response time and creates the most heat, while also drawing the most power. It uses the most energy, and it uses it rapidly, so it follows that it is more likely to suffer from a fault caused by too much residual energy. Which leads me to this...
Good news, folks: I now understand at least one significant cause of these MOSFET failures (with proof), and it looks like I can address it pretty easily (also w/proof).
I'll write up something longer tomorrow or Sunday, but this should do for now.
This is a screen shot of the oscilloscope's view of one heater MOSFET's source pin (in purple), and one power MOSFET's source pin (in yellow) when the tactile switch opens (i.e., your vapcap has clicked and you withdraw it). These are their vulnerable nodes and the ones where failure likely occurs:
The significant values: In this case, the max voltage on the heater MOSFET source pin was 34.8V in the above run, but I had other instances on the bench this evening where it went over 50V; the max spec for this pin is 40V. It can tolerate some over-voltage, but not a huge amount, and you'd like to stay further below that upper limit. Sorry I missed this until now.
This is the same setup, same measurements, but with the addition a "flyback" diode to the heater circuit, installed in reverse on one of the heater MOSFET source pins. This diode acts like a release valve and allows that spike of voltage caused by the remaining inductance energy to bleed back into the battery (which can easily absorb it) instead of frying the MOSFET:
Max voltage is down to 26.8V, and the pulse is dampened. The post-switch closing ripple in the power supply MOSFET, in yellow, is now completely gone, too.
I have some more testing to do, but so far, so good.
Just wanted to share some good news on this. Cheers
What is the half-life of the flyback diode, or does it not work like this? In other words, will the repeated redirection of the voltage spike to the battery eventually wear this diode down (hope this makes sense)?
mr c, u rock....i cant wait. Awesome!I would say that all sizes are subject to the issue, but the 14mm is the most at risk for MOSFET failure due to the way inductive energy spikes are generated. Of the sizes I offer, that size has the fastest response time and creates the most heat, while also drawing the most power. It uses the most energy, and it uses it rapidly, so it follows that it is more likely to suffer from a fault caused by too much residual energy. Which leads me to this...
Good news, folks: I now understand at least one significant cause of these MOSFET failures (with proof), and it looks like I can address it pretty easily (also w/proof).
I'll write up something longer tomorrow or Sunday, but this should do for now.
This is a screen shot of the oscilloscope's view of one heater MOSFET's source pin (in purple), and one power MOSFET's source pin (in yellow) when the tactile switch opens (i.e., your vapcap has clicked and you withdraw it). These are their vulnerable nodes and the ones where failure likely occurs:
The significant values: In this case, the max voltage on the heater MOSFET source pin was 34.8V in the above run, but I had other instances on the bench this evening where it went over 50V; the max spec for this pin is 40V. It can tolerate some over-voltage, but not a huge amount, and you'd like to stay further below that upper limit. Sorry I missed this until now.
This is the same setup, same measurements, but with the addition a "flyback" diode to the heater circuit, installed in reverse on one of the heater MOSFET source pins. This diode acts like a release valve and allows that spike of voltage caused by the remaining inductance energy to bleed back into the battery (which can easily absorb it) instead of frying the MOSFET:
Max voltage is down to 26.8V, and the pulse is dampened. The post-switch closing ripple in the power supply MOSFET, in yellow, is now completely gone, too.
I have some more testing to do, but so far, so good.
Just wanted to share some good news on this. Cheers
looks like you hit the bulls eye.
Hi all,
I knew it was too easy. I'm back with a few more turns to this MOSFET failure saga.
Development #1: I discovered my first solution (a single Schottky diode to drain the circuit via one MOSFET's source pin) did not actually solve the problem. It was an improvement over nothing, but I found that inductive spikes of >40V were still possible, and sure enough, I generated a few of those and killed yet more MOSFETs in the space of a few hours.
Back to Google.
[Time passes.]
Development #2: Eureka!! I found the correct solution:
https://www.eng-tips.com/viewthread.cfm?qid=426309
Which can be summed up as: "use two 24V Zener diodes. Reverse bias them as in attempt #1. Put them in a different part of the circuit. Success!"
Here are some new before and after traces (both traces heater MOSFETS, not power MOSFETs):
Before (max volts circled in red):
After (max volts circled in red):
This one really feels like the right solution - that 2nd trace looks perfect, and a 20V load against a 40V upper limit provides for a HUGE margin of safety.
It's a bit tougher to retrofit than the first solution, but I think I have a handle on that, too.
More tomorrow - bed time now.
Cheers,
Great to hear! I have a 15mm ordered to compliment my 14mm, hell im sure I’ll have some crazy custom enclosure built someday and have a 14, 15 and 16mm coil all in one .Loving my 14 mm coil, no problems with it at all...until I tried to install a temporary heat shield as directed earlier in the thread. There's a reason I pay other people to do stuff like this for me. Apparently I messed something up because I couldn't get it to heat up. mr_c was very responsive and gave me a couple suggestions but no go. I sent it back for repairs and all the current safety upgrades and decided to swap to a 15 mm coil.
We got it back last Friday, the turnaround time was incredible, fantastic customer service! We're still getting used to it but so far it's working exactly how we hoped. It's noticeably slower than the 14 mm coil, I'm still adjusting to the extra heating time, but the hits seem to be more thoroughly heated. It's along the lines of cooking something on the stove on medium vs medium-high heat. It takes a bit longer but the added time lets the heat penetrate everything a little bit better. I'm a bit of a flavor chaser, too, generally hitting before the click, so slowing it down is a bit more forgiving.
So yeah, we love it. My wife said having it back feels like a luxury. Huge thanks, @mr_cfromcali! Great IH, great service. We really appreciate it.