Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me.
nickskethisniks Guru Joined: 17/10/2017 Location: BelgiumPosts: 458
Posted: 09:40am 25 Oct 2024
Copy link to clipboard
Print this post
Well, I'm close to start building the powerboards soon, but...
I will probably use a different mosfet (hy5110) and gateresistor/bead combination. And after reading wiseguys research about the ds beads, I still have some concerns about using the beads. So I'm doing a little bit of research and testing myself and later after populating a powerboard with the same experiments I will share my findings.
The matter concerns me after reading some application notes where they point not to use "randomly" beads because they can also worsen things. I know you have several decades of expierence designing power electronics Wiseguy and I trust your gut and research. But It's the part were application notes state to design an application so the mosfet stay out of the avalance region, allthough the energy is very low it can still degrade the mosfet overtime.
Personaly I would feel more sure when we can prevent it, you allready hinted (a solution) the use of a 10/15R parallel with the bead, something I didn't understand first when I read about it in an application note. But it is litteraly a resistor soldered on the mosfet leg over the bead.
So why not spending the extra effort after spending hours of work taking time to solder those resistors in?
Revlac even mentioned it to use some beads in his mppt controller, I hope you probe the mosfet because voltages are higher and the bead might be counter productive. I think the use of a bead will be a good thing, but maybe another bead or some extra measurements to prove it's contribution.
So, I allready said I'm doing some testing, well, I tested 4 different beads, with 3 I had about the same results as you. But then the last one, really a tiny bead, this tested wayout better then all the rest. I think the peak halved compared to the other ones, and then with a snubber 1n/10R it was really low compared to the starting point.
Tests were done under load ( no transformer or inductor attached) and with load, most of the time a load with high impedance made things better, a low impedance worsened the shoot true during turn off transient. A snubber could greatly reduce the peak of the pulse without load but not so under load. Adding some 100nF close to the mosfets between positive/negative rail helped but not more then a few volts.
So, I will devote a new thread on the subject after populating an inverter powerboard. But I think I want to say with my post is that there probably are better suited beads to use, or maybe not. I found out it was in some cases until load was applied... Sorry if I opened a pandora box again.
Here you can see the beads, 3 low are ali express and the one most top is a mouser part, the lowest one performed best in my test setup.
These were the beads: The best one was the only 3.2*1.4*1.2 mm
https://nl.aliexpress.com/item/1005004112003854.html?spm=a2g0o.order_detail.order_detail_item.3.35616d762UwNZk&gatewayAdapt=glo2nld Edited 2024-10-25 21:35 by nickskethisniks
wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156
Posted: 04:02am 27 Oct 2024
Copy link to clipboard
Print this post
My take on Ferrite bead use and behaviour. The bead placed on the Drain lead of a MOSFET is worthy of a little more discussion and analysis as suggested By Nicks.
Ferrite beads were initially created and introduced for a few reasons, most of them associated with higher frequencies. Designers could then exploit their specific behaviours.
At HF's they have inherent properties of impedance, inductance, wide band and low loss. My decision to use ferrite beads are mainly due to their high impedance behaviour which usually increases with frequency often up to the GHz region.
The inductance and impedance property of a Ferrite bead will reduce with increased DC bias current. At the DC bias current levels in an inverter the bead's inductance will be extremely low during the FET on time and the bead will be fully saturated.
For the low inverter switching frequencies from 20 - 50kHz the beads have relatively low internal loss as their saturation is instantaneously reached as the FET turns on, so relative to the total switching period their is minor bead heat produced.
It is near impossible for a circuit to have zero inductance at the Drain of the MOSFET, even the Drain leg of the FET introduces inductance, so high speed switching on and especially off is usually accompanied by a decayed "ringing" of the waveform at the Drain of the FET in the MHz range. To reduce the ringing, a bead placed on the Drain lead increases the HF impedance, thereby minimising the circuits Q significantly and the ringing can be reduced to near zero. The main resonance is from the inherent internal capacitance of the FETs Drain and the Drain circuits total inductance.
An added benefit of the reduced HF ringing is the reduction of the harmonics and EMI generation from the otherwise very fast switching edges. The other benefit is that when paralleling MOSFETs there can be tendencies for oscillations to occur where paralleled MOSFETs can appear to behave more like a power multivibrator (at very high frequencies in the 10's to 100's of MHz) than the true current sharing that was desired and aimed for by paralleling the MOSFETs.
A minor negative side effect of the bead is the energy storage during the FETs on period, this energy discharges during the FETs off time. As there is no current path with the FET now off, the inductor voltage increases until one of two things happen, the FET's Drain avalanches internally and tames the voltage "spike" or the voltage peak is not high enough to avalanche the FET.
Due to the high permeability of the "soft" ferrite bead material there is virtually no energy storage of concern and whilst avalanching a MOSFET in a design can be a problematic issue, the very minor energy storage of a small bead can be safely ignored. In my experiments I found that a voltage spike of ~ 20 - 40V could be encountered and simply placing a 4.7 - 12 ohm resistor across either end of the wire passing through the bead could reduce the voltage peak by up to ~ 80% with some minor heating of the resistor.
Lastly MOSFETs, since the 80's have focussed on improving the avalanche ratings of available devices. Whilst Trench FETs (most common) are not as good as Planar FETs, their avalanche energy rating is still usually very large. The HY5608 I recommended has a manufacture tested safe maximum avalanche energy rating (not repetitive) of nearly 3 Joules from an inductance of 0.5 millihenries. This is the type of avalanching that really should be avoided. Compare this to the typical ferrite bead inductance of 2uH with very low volumetric storage levels, safe energy dissipation perhaps in the few millijoule range could be expected.
To date I have not encountered or been aware of any MOSFET failures through using a ferrite bead placed on the drain, but I am aware of significant levels of EMI interference reduction and visibly very clean switching waveforms. I am hoping that Nicks investigations will also mirror what I found. Edited 2024-10-27 14:09 by wiseguyIf at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike
KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679
Posted: 04:28am 27 Oct 2024
Copy link to clipboard
Print this post
I also used smaller ferrite beads compared to wiseguys original design, they appear to work as intended but need further testing - just no time the moment.
That would be interesting if you get the time, and the only thing I would suggest is changing the probe setup.
IMHO you need to remove the probes ground lead and the long nose clamp. Both need to be very short and ground isolated probes (differential probe) with two inputs (even if one probe is the trigger input if its ground lead is connected)
Without that, the waveform and amplitudes at these frequencies cannot be fully trusted.
I do find the very fast switching of the FETS in this design can cause more RF interference, which I expected and wiseguy also mentioned, a small price to pay for having such low heating of the FETS.
EDIT
Dam! just posted and then I see wiseguys posted, I haven't read his yet - so disregard if anything contradicts wiseguys post. _ Edited 2024-10-27 14:48 by KeepISIt's all too hard. Mike.
analog8484 Senior Member Joined: 11/11/2021 Location: United StatesPosts: 108
Posted: 05:39pm 27 Oct 2024
Copy link to clipboard
Print this post
Your finding of the smallest bead being more effective (assuming the same material) is consistent with my experience. The bead effectiveness decreases quickly with the gap distance from the pin going through the bead.
Also, the Ali beads you used all appear to be Ni-Zn material type which I have found to be less effective than Mn-Zn ones. The Ni-Zn beads are more effective in higher frequency ranges. The Mn-Zn ones are more effective at lower frequency ranges closer to the lower switching frequency range of inverters compared to typical SMPS.
Lastly, I have learned to avoid getting beads from Ali as they have inadequate specs and they can differ significantly from one purchase to the next.
nickskethisniks Guru Joined: 17/10/2017 Location: BelgiumPosts: 458
Posted: 12:24pm 29 Oct 2024
Copy link to clipboard
Print this post
Ok, thanks for the comments, I will get some more from mouser to do some more testing with, it's the end of the month so I can top up my cash for projects. I need a few more parts to start, should by a reel of those HY5608 mosfets.
Wich is the prefered software to load in the nanoverter? Ok, found latest version 7.7. Edited 2024-10-30 02:05 by nickskethisniks
wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156
Posted: 08:11am 30 Oct 2024
Copy link to clipboard
Print this post
Message to those waiting patiently on inverter boards They are finally on their way to Australia now, UPS dragged their heels for 4 days (no movement of item) and JLCPCB took over a week to manufacture so despite my best efforts to have them delivered close to my return from holidays I have waited impatiently for a week now ! Not much longer I hope.
The LCD display PCBs have provision to mount an onboard 5V reset monitor to avoid the Nano losing its mind. The Nano controller also has provision on board for a 5V reset controller for the same reason. The Front panels and 2 stage controller cards both have a 4 way connector now for manual second stage switching and the 2nd stage LED display from the front panel. The power PCBs have all the resistors flat to the PCB now.If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike
Cpoc Regular Member Joined: 28/05/2024 Location: PortugalPosts: 78
Posted: 10:17am 30 Oct 2024
Copy link to clipboard
Print this post
Thank you for the update.
wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156
Posted: 06:38am 07 Nov 2024
Copy link to clipboard
Print this post
Will miracles ever cease, after 2 prompts from me the PCbs were languishing in Chek Lap Kok for 4 days, Melbourne for 6/7, the long awaited PCBs finally arrived here this afternoon. Lots of work in front, but hope to get them packed & shipped out by Monday. We will be away for nearly 4 weeks starting next Friday, so must get it all done ASAP.
FYI Tracking - not updated since 5th not very impressive service, I don't like UPS ! Over a week to manufacture + 2 weekends, shipper notified on 25th for parcel pick up delivered 7th....
Edited 2024-11-07 16:49 by wiseguyIf at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike
Cpoc Regular Member Joined: 28/05/2024 Location: PortugalPosts: 78
Posted: 11:04am 07 Nov 2024
Copy link to clipboard
Print this post
Ever since covid I find now shipping take twice as long to receive and now costs more than 2 X the price.
This is everywhere in the world now. Transport costs is getting out of hand where sometimes it cost more to ship and item than the item itself.
Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me.
Print this post
