We would like to thank our participants:
1. Guy Moxey, Wolfspeed
2. Alexey Cherkasov, Leapers Semiconductor
3. Carsten Schreiter, Semikron Danfoss
4. Tobias Keller, Hitachi Energy
5. Ehab Tarmoom (together with Simon Tomkins), Microchip and Richardson RFPD
6. Evangelos Theodossiu, Vincotech
No questions so far so I think all of you uh in the module business area and To me when I first got in touch with modules was just the 35 millimeter half bridge now we have so many different kind of modules that it has quite exciting to see how the modules of today can solve complex systems requirements any of you have specific comments of the
Special modules I see Alex say Alexa you have such great modules in your background yeah yeah that’s right that’s right uh uh well previously uh mostly I worked with the primordials fire Easter diet modules and just this year I switched to this new company where we focus on uh silicon carbide
Mainly also igbt modules but uh main focus steel is silicon carbide so we do it in various packages and well we’re trying to go with the trend step by step and as you all know uh nowadays uh our whole society our whole world is moving forward to this uh zero
Net uh carbon emissions uh by various steps by various ways so uh some choose renewable energy some go with electric vehicles but still all these right now also go with The wideband Gap Solutions like gallium nitride or silicon carbide and in libraries we focus on silicon carbide for various
Applications and like in my presentation uh today uh we also have quite great solutions for the solar industry for solar inverters in particular we see this actually this kind of Technology being you know increasing a lot in in the coming years to be less dependent on fuel energy on fossil energy uh that’s
Why we stick to this uh Exact Solutions to go with the society to a Greener future thank you and like say in the meantime we got a very important physical question I saw it in the meantime uh how do you evaluate the Center Technology and do you plan to introduce it in the
Near future that’s a question to bingo Tech to evangelos yes I hope you can hear me very well we hear you perfectly okay so it’s a very good question um I think Cinder is let’s say well no technology and already used in uh many applications with were high
Requirements in terms of reliability for instance in Automotive so but the disadvantage of this technology are the costs especially the material costs and also the investment that you have to do in new machines and new processes so before to decide to to use Cinder I think you have to answer several questions
Um for instance can I cover the application or requirements in terms of reliability or especially power cycle capability to keep the link to the presentation that I gave before with standard technology or if this is not the case can I prove the standard in the connection technology to fulfill of the
Specification with how to increase the costs and also if not then will be the cost the customer will willing to accept the higher cost using cinder technology and last but not least if we change to Cinder what will be the solution for the wire bonds and the so the decision for cinder
Should be based on the fact that if there is a value Edge if there is a value for the application especially for the customer which should satisfy this higher costs to implement Cinder without to improve also the wire Bonds on the same time will not make sense from my
Point of view because this will be the next weak point you will gain some additional reliability for instance I would estimate Factor 2.4 more power cycle uh capability compared with this new solder although that we have presented but the limiting factor in this case will be the viables firebones failures so the
Question again is is this 2.5 higher power cell capability worth compared with a much higher model cost or is a solution that we have provided presented with this improved solar lawyer yet sufficient enough to satisfy the the requirement so a full say in our cost driven markets we still believe because
They still believe that with our efficient and cost effective solution we can fulfill the reliability requirements but for sure in future there will be application which will need Cinder but as I said before there are several questions to be answered before to go in this direction thank you evangelos
That is for me I guess yeah it’s a very technical question I had fun in the past when the director at uh a different business said the most important physical dimensions is the price of the device and it’s probably for some people true for some others not it depends on who
You’ve talked to I guess but the question here is 90 euro uh in my presentation I put 90 euro as an example at Adder for an end user that buys a 15 kilowatt drive and that’s a difference from Silicon igbts to silicon carbide it’s a good guess it can be 100 can be
110 can also be 80 it’s somewhere at that range I would expect it but what I also showed that this is actually not really important of course silicon cover it is more expensive and it probably remained that way for ever or for a long long time but on the other side the the
Benefits that brings us the freedom and design which I also try to point out in the presentation that can really overcompensate the cost so whether you go for high efficiency and save energy that can pay back in a year or less or depending on your energy cost even fast
ER or if you go for the lower um the volume if you release the heatsink size I think I brought it down to 66 no 66 percent smaller heatsink than with a silicon solution and if you think that thing further means the drive can shrink that means handling is easier so
For a let’s say 70 75 or 90 kilowatt drive you need a forklift or two people to lift it you can just put it at the wall done that gives you essential benefits or you put if they are smaller the drives you can put more on one pellet put more in
The same container so shipping it’s uh cheaper so it triggers a long chain of really benefits and that’s that’s the real use this silicon carbide can bring us in the drives world um and that as I said that overcompensates the the little investment cost that we have on the
Module side or on the chip side that’s how we see it and I hope that the end user will also see that and and I think it will drive a change in the drives well sooner or later foreign sorry I’ve got a quick question for Carson because it was an excellent topic
I think in Germany today you’re paying I was Electronica a couple of weeks ago and I asked a lot of German colleagues and they’re paying something like 65 Euro cents per kilowatt hour on the electric so that’s it that depends on your contract I’m lucky minus two years
Old not not renewed yet probably next year I look uh not as as lucky as now anyway the it depends also if it’s industrial uh electricity or or residentials or residential it depends on the supplier I think the the mean is somewhat at 30 probably at
The minute 30 cents industry is a little lower but that also goes up so sorry but the point the point being if you’re looking at a kilowatts of drive you mentioned the payback and right now if you do some simple mathematics the payback’s about three months
But depends on how much you pay and I think it also depends on on your application and your application profile so if you have an application I if you look at my presentation I had an epic running profile for a fan application and that is just an assumption of a
Typical fan but if it if your fan is running full time at 50 or 60 then it’s probably even faster so it really depends on the case so if someone is considering and he’s thinking that silicon carburetor is just too expensive for him then my only recommendation is
Com and let’s look at the application and we’ll find it out that’s that’s the thing and I’m sure that in most of the applications will find it’s it’s it brings benefit it pays back easy and it has so many benefits that they probably don’t even see yet in the long run if
You think it further and that’s the big thing here correct okay Nothing is for free the question is addressed to evangelos you improved PC performance by soldering Improvement no change in viral bonding correct that’s the question this is correctly yeah there this was the understanding as I said before when
The question come up with the centering so their focus was to improve this interconnection technology between chip and substrate for sure once you improved once you have Pro improved this uh joint then you have to look also on the because you are as I said before you’re
Moving this big point from the Zola drone to the wire Bond so if you would like to increase further availability then for sure you have to do also something to have a y bonds but for the moment we are that for instance a cinderlink we discussed about Engineering also copper wire bonds but
This is at the moment the finger that we are looking on some new technologies but at the moment not implemented yet but for sure this will be the next step up that we have uh improved the solder join but we see also in the market already some other uh power module manufacturers
They are doing they are implementing also for the wire bonds and new technologies which for sure and the same most of the cinder uh yeah it’s increasing the the power model cost but it’s a question so what which requirement you will you have to fulfill what is the application if the
Application really needs this and then yeah you you have to think also about this thank you evangelos so it’s question to all we see a trend for recommended Gate Drive voltage of SRC power modules to be moving to lower values for latest generational devices will this trend continue
This demands lower power from the gate drive system but what are the specific advantages for the SIC device they want me to take that one as a device guy device manufacturer go ahead um I mean there’s still some inconsistencies in different Gate Drive voltages positive and negative from different suppliers unlike silicon where
Everyone gravitated years ago to exactly the same drives you’re still seeing a few faults difference in operational and apps Max but you’re right they’re all coming into a similar grouping 15 to 18 volts turn on and minus four volts turn off um a lot of discussion in the industry
About zero volt turn off and I think everybody’s silicon carbide can turn off at zero soft switch without shoot through but the inherent nature of silicon carbide is a low threshold so when hot everybody’s devices are hovering around about two volts hot and so that’s pretty close to zero so
That’s why we all recommend hard switching taking negative so there there are some zones of positive and negative Gate Drive that you’re going to see but unfortunately the end of the industry has not yet evolved to standardize on exactly the same voltage so we have to keep everyone on their toes
Um as long as you fully enhance the switch but turn on which typically wants you through Miller and you’re through the other end is 12 13 volts then you can overdrive slightly to lower RDS alt to a certain point but then you may stand a compromise of long-term
Reliability if that was the second part of the question but I think the first part was the more important that yes there still is some difference still are some differences on vgs positive and negative and you will continue to see some small differences thank you guys any additional comments in from the
Thank you of the speakers we have the next question coming I still would like to see smaller packages than the tall packages to use good sic Parts instead of falling back to SI only for smaller footprints especially if you do HP integration I use sic now for higher temperature reliability applications
And faster switching what’s on the horizon most is just larger power usage which is already out of my application range it’s a to me very general question that needs a little maybe more specific things because I remember sometimes it’s good to have enough uh space and get the heat out
If you get the devices smaller and smaller you have more more problems extracting the Heat so it’s just a physical thing from my perspective uh what you do with Packaging if you make it smaller you get a smaller footprint and if you run hot air which is possible with your devices
Then there are two things that drive uh problem zones that need to be solved that’s just from an old man’s perspective that’s that’s the other thing if you have high voltage and it’s so small that you have have no creepage distance uh for the high voltage application yeah
The dry sizes are small you know all silicon carbide starts at 600 volts 650. the die sizes are very small so you could fit the die into a qfn three by three millimeter package quite easily but to your point thermals and secondly 3.303 by three qfn violates creepage and
Clearance at 650 volts so it’s a bit of a it’s another physical challenge thermals and creepage and clearance if you’re looking at small discrete packages yeah so this question can only be answered by putting up the technical requirements in the application and find out what can be the next possible best solution
You cannot put General things out the way that question was said but in general we half over the years when I was still a younger men the half Bridges the 35 millimeter half Bridge was a standard and that had been changed quite a bit now the next question yeah the gentleman was
Just added a bit more so this is low power lower voltage to be quite honest low power lower voltage 200 volts 100 odd Watts it’s possibly more of a gang type of environment than it is silicon carbide yep okay the next question we are waiting for the next question my backstage is waiting
Guy my deepest sympathy to John’s family and the dwarf speed thank you Lord I will do absolutely will do and what’s always fun to meet John and we had so many great discussions and I was shocked when Jorge told me yeah that he passed away it’s life and we have
To accept life as it is so back to our sic subjects I look I think most of the German people are now out for lunch for dinner for dinner sorry I only students out for lunch now so because you are representing the semicon site now both as you can see some condenser
But you are sitting in in Nuremberg I belong to the to the Nuremberg part yeah I belong to semicron but now we’re semicondent first so okay it’s uh it’s a long time ago when I worked for Harris and we had a number of great meetings in Nuremberg getting the Harris
Especially GE solid state and RCA igbts in front to semicron who wanted to put them and that half Bridges and the big problem was at that time nobody at Harris really has the understanding of uh the wire bonding and bonding the gate was a sick wire at that time there were no
Multiple hats who could sin and sick wire on one station wasn’t very exciting time and I had nice memories to the meeting room in Nuremberg at semicron oh it’s probably a while ago so if we come now here it’s a while ago somehow everything looks different and there’s
More yeah it just does not ended up in business uh it helped me to get a better picture of what possible in the uh power industry I saw all the nice uh half Bridge modules the one where White and the other were black and the black were Siemens
And the white were semicon that was from black side dark side that was before infinian was uh built or taken out of Siemens so we have a question what is your take on gen versus sic devices in the lower voltage range 600 volt do you think advantages and gun devices technology
Will result in gain device taking over sic onset on such applications since gen can switch even faster the question is that even faster switching is an advantage and practical applications Emi is the problem and switching speed still needs to be a little monitored that’s what an element still remembers
I think the space for both Technologies really about 600 650 volt node um gal is showing some very good performance in the sort of lower power switch mode power supplies like cell phone chargers laptop chargers wall cubes where it can switch fast and it’s maximum 600 volts and it’s a few amps I
Think Gan is a very elegant technology is integrated sometimes with the gate driver because it’s lateral so I think that naturally leans itself to that sort of computing consumer type power supply Market but then if you look at higher power 650 600 650 volt which can go up to
You know tens and tens of tens of kilowatts on board charges for example solar for example then obviously the gun limitations on ampacity start to to be noticed and of course um you know you are not switching at megahertz type levels it’s not a quasi-resonant flyback at 10 kilohertz
It’s a hard switched totem pole for example in a PFC so I think there’s a great space for both Technologies doing different applications around that voltage node myself in just one comment to to to to to guys I think from Bingo Tech we always always put the question what is the application
Which will need this High switching device and so we’ll what will be the evaluate for the customer and uh for this and for the moment there are some applications I mean in this low power markets uh gun is already implemented especially for for mobile chargers so you already will find the market gun
Charges or maybe this is also for onboard Chargers this is also something about com so always we have to ask okay what is the application and then for the moment we do not see here the mainstream of the applications needed the size featuring because on the other side you
Have to think also if you can handle the especially let’s example the passive components do you have so the passive components available to handle this High switching um and yeah as I said as I say I think for this 600 65 50 volt yeah both Technologies can be used depending on
The application but the higher for the higher voltage I think here silicon carbide is the the choice for the moment for the next years and let’s see how the gun will develop also for the next years Also let’s add on to that um so as you mentioned that the higher current really
You know silicon carbide is probably the the better solution um you know you look at the RDS on over temperature before now for silicon it’s you know two two and a half times uh for for Gan I believe it’s maybe around two times when you guys you can confirm that
Um the silicon carbide is gonna be a lot lower so when dealing with high voltage High current um really you know the silicon carbide will give you maybe like 700 volt devices that we have in microchip are about 15 increase 15 to 20 increase at 175 C so uh really we’ll keep the
Conduction losses low with a uh you know silicon carbide thank you app gentlemen thank you very much we had a fruitful discussion and I wish you a nice rest of the day and so I want to say goodbye and enjoy the evening or the afternoon wherever you are located at the moment
Thanks a lot thanks a lot thank you thank you bye-bye