Awesome new FET package :) :) :) Not a geek? you won't care

Seeing this big boy almost makes me regret designing my controller around TO-XXX style packages... but maybe it's best this way though since I'm sure to avoid the dangerous availability/pricing trap that might come with this part. :(

liveforphysics said:
bigmoose said:
Luke, GREAT find, I love the package and the specs... except for the 5v max Vgs... it's a blasted logic level FET... I prefer the ol' fashioned ones with Vgs of 20 volts...
Click to expand...
The logic level gate short coming can be decently worked around with the use isolated DC/DC supplies for the gate drivers rather than boot-strapping through right? Still though, at these current levels, you're going to get some serious ground plane stability and resultant gate bounce, and a logic level gate doesn't give you jack-squat of a safe window to play in for bouncing. I also wish they had higher gate voltage needs, but it seems like everything these days is moving towards the low voltage gate. It's great for non-power electronics, and kinda a big hazard for something like a motor controller application.
Click to expand...
As Eric said, this does not look like a logic level device. It's plateau is at the same level as the 4110 (5V), and it's threshold voltages actually look a bit higher than the 4110 so there should be no problem with dV/dt induced turn-on. If fact it should have quite a bit better switching performance, since it's source capacitance to drain capacitance ratio (Cgs/Cgd) is quite a bit higher than the IR's FETs in general (like 0.75 for the 4110 vs 1.31 for the big IXYS part). Now add to this the very likely much lower source and drain inductance of the IXYS part (not on datasheet), and you have a potentially much better switching device! :twisted:

Pat
 
But, logic level and it's negative effects aren't about having a lower ceiling limit, they are about the inherent inductive voltage ripples in the ground plane (between fet driver ground and source leg) giving the potential to bounce into the conductive-transitional reigon at ill times...

It's not about where the voltage limit is on the gate, it's about having the level where the fet switches dangerously low.

At least that's my take on why logic-level gates are great for low current devices, and bad for high current devices (because of the inherent increased inductive effects at high currents.)
 
Camlight,

If you are going to order a bunch of these fets, you can talk to IXYS direct.
I looked at going direct and you only needed to put down ~$2000, which was definitely more parts than I needed.
But if you have a project...

Mindshare Technical Sales is an IXYS rep...they may be able to contact the factory too.
Steve Arnold
480-945-7156
 
liveforphysics said:
But, logic level and it's negative effects aren't about having a lower ceiling limit, they are about the inherent inductive voltage ripples in the ground plane (between fet driver ground and source leg) giving the potential to bounce into the conductive-transitional reigon at ill times...

It's not about where the voltage limit is on the gate, it's about having the level where the fet switches dangerously low.

At least that's my take on why logic-level gates are great for low current devices, and bad for high current devices (because of the inherent increased inductive effects at high currents.)
Click to expand...
I agree about logic level devices being more sensitive to inductance-caused ringing problems, but it's mostly one event during the PWM switching that is a problem: high side FET turn-on event. As this top device turns on, it can pull the bottom FET back into conduction, creating very dangerous shoot-through currents. Having lower gate drive loop inductance, a higher gate theshold voltage, lower gate drive pull-down resistance and lower Cgd/Cgs ratio for your bottom FET will help avoid this problem, as would slowing down the high side FET turn-on too. It took me a long time to add this understanding to my PWM knowledge bank (brain) - wrobinson made me aware of it not even a year ago.

However, I don't understand what point you are trying to explain above Luke? Is there something I am not comparing right between FET datasheets?
 
I think we're on the same page, and trying to say the same thing my friend. I was essentially just trying to say, if it conducts with 5v on the gate, I call it logic level, even if the gate can handle 30v (or whatever) safely.


I disagree that it's only the top-side switch that can cause issues though. If you remember in my battery tester FET stage, I couldn't figure out what on earth was causing my FET stage to suddenly just thermally burst and fail at around 400amps, when at 350amps or so, it was barely luke-warm. Moose advised me to check my gate situation on the scope, and it convinced me to change my whole gate drive and gate drive ground plane layout, and sure enough, never had that problem again. The inductive ripple was causing my gate voltage to take a nice slow trip out of the conductive region and loiter around in the transitional region (I think it stays transitional because current through the fet drops when it enters it, and the flyback voltage causes it to bounce back conductive and out again, maybe a few times for each cycle. If I was running gates that didn't trigger until perhaps 10v, the problem would not have happened until 900amps or so (at least that's how I read it).
 
You certain can have other issues with the gate drive - excess inductance, bad routing, whatever. But if you have the cleanest, textbook-perfect gate drive circuit with zero inductance, dv/dt when the top switch turns on can still cause you grief.

BTW, if you have issues with this FET having a 5V threshold voltage, you should look at the IR datasheets again. 4110s, 4310s, 4115s, and most of the other standard TO-220/-247 IR FETs have threshold voltages around 3V. :) True logic-level FETs are actually much lower, like 1V. Like this guy: http://www.onsemi.com/pub_link/Collateral/NTHD4508N-D.PDF. In any case, since the spec value for this part is min 2.5V, max 5V, it's reasonable to assume most parts will have a nominal value around 3.75V which is a little higher than the IR parts but not much.

Regarding the dv/dt problem, you should read this whitepaper from IR:
http://www.irf.com/technical-info/whitepaper/syncbuckturnon.pdf
It explains the problems and solutions pretty clearly.

The threshold voltage defines where the FET will start conducting in the linear region. The plateau happens at a slightly higher voltage, after the FET has reached full current when the drain-source voltage starts to drop. After the FET is fully saturated, the gate voltage begins to rise again. At this point the FET is turned on and saturated, but driving the gate voltage higher will fully "enhance" the FET to reduce Rdson. Most of the power devices spec their Rdson at Vgs=10V. You can drive it higher but with diminishing returns.
 
liveforphysics said:
I think we're on the same page, and trying to say the same thing my friend. I was essentially just trying to say, if it conducts with 5v on the gate, I call it logic level, even if the gate can handle 30v (or whatever) safely.
Click to expand...
If a gate threshold voltage of 5V means a logic level device to you, then you must be seeing logic level FETs everywhere! Seriously though, check out the IRLB3036 datasheet for what a logic level device looks like from IR at least. It starts conducting at around 2V when hot, compared to around 3V for it's "normal" level brother the IRFB3006 (see figure 3). Now compare this to your IXYS's figure 6, and we see that even the regular level IR part (3006) has lower threshold voltages than the IXYS part. Note that IXYS doesn't make their Y axis logarithmic in figure 6 like IR does in it's figure 3, so the graphs look different.


liveforphysics said:
I disagree that it's only the top-side switch that can cause issues though. [...]
Click to expand...
I didn't mean to say that it's only the turn-on event of the high side switch that causes problems. I'm saying that it's the switching event that's the hardest of all to control in your PWM design, even if you have almost no inductance-induced oscillation in your driver <--> FET gate loop. I agree with your example of how cleaning up and tightening your gate drive loop made the difference in your own design, but even with a super-dooper design without gate drive inductance issues you can still have the dV/dt sporadic turn-on happen if other variables are not taken into account. Can I ask how fast your top-side FETs were turning on in your circuit after you had got rid of the unwanted oscillations?

You see I had everything else in my controller under good control after a number of design revisions, only then to find out that I had hit the switching limitations imposed by the FETs I was using (TO-220 IRFB3006 FETs). Specifically it was the TO-220 package source inductance combined with the crappy 1.3 Cgd/Cgs ratio of the FET that was causing sporadic turn-on of the bottom FET while the top FET was turning on, causing short but bad shoot-though currents when switched too fast. I have now just finished making a new power PCB populated with infinion FETs instead of IR FETs, and I'm hoping that I'll be able to get swtching times a fair bit lower using these FETs since infinion FETs are optimized for higher speed switching (much lower Cgd/Cgs ratio, like 5-7 times better). Inductance issues will be the same, but the bottom infinion FET will be much more resistance to turn-on. I'll post some comparative results when I get around to powering up the new board if the results are significant... or not so significant for that matter.

Pat
 
bigmoose said:
Luke, GREAT find, I love the package and the specs... except for the 5v max Vgs... it's a blasted logic level FET... I prefer the ol' fashioned ones with Vgs of 20 volts...
Click to expand...

I need to retract this criticism. I read Vgs(th) Max=5V as Vgs Max = 5V. When looking at the graphs like you said, the Vgs goes well above logic level. guess I am used to seeing Vgs Max in the Maximum table of the datasheet; and glossed over the "test conditions" that followed Max in the IXYS style. Too old, scanning too fast, and vision going; makes for too many mistakes... Thanks for catching it. I

A FET with very interesting possibilities. Luke, let me know also when Future gets back to you on pricing.
 
12p3phPMDC said:
Camlight,

If you are going to order a bunch of these fets, you can talk to IXYS direct.
I looked at going direct and you only needed to put down ~$2000, which was definitely more parts than I needed.
But if you have a project...

Mindshare Technical Sales is an IXYS rep...they may be able to contact the factory too.
Steve Arnold
480-945-7156
Click to expand...
Interesting...thanks for the contact info! :)
 
My rep called me back and said price/availability info will be released for them this week. :)
 
Awesome, looking forward to that info...!
 
Luke, let us know what you find on this part with respect to pricing/availability from the rep. Ask them if they have a date for a 100 volt part if you can. This thing is pretty much like Qty (4)ish IRFB4110's in a single isolated package. Like you said, locating sufficient capacitance is going to be the design problem in packaging.
 
I got this skinny on this part.

There will be no 100v part. :( :( :(

It was a custom package done for a customer in Florida for some 48v UPS. They have zero bulk stock, and an infinity waiting date for the next batch. They are willing to do a batch of them in a quanity starting at 1,000 pieces. Will get price data on this 1,000lot quanity soon.

However, they do have sample-stock, and I managed to secure 6 pieces for samples. I ordered the 6 samples price-unknown... so hopefully they don't turn out to be like $200 each or something stupid, but I'm prepared to pay whatever they charge me to get my hands the samples. It's kinda funny to think about, but the little 6-fet controller I make from the samples will be both the most costly e-bike controller, and most powerful e-bike controller. :)
 
While bummed about it not being a standard part (why create a data sheet? :evil: :evil: :evil: ), I am really looking forward to hearing what the price is!
 
liveforphysics said:
I got this skinny on this part.

There will be no 100v part. :( :( :(

It was a custom package done for a customer in Florida for some 48v UPS. They have zero bulk stock, and an infinity waiting date for the next batch. They are willing to do a batch of them in a quanity starting at 1,000 pieces. Will get price data on this 1,000lot quanity soon.

However, they do have sample-stock, and I managed to secure 6 pieces for samples. I ordered the 6 samples price-unknown... so hopefully they don't turn out to be like $200 each or something stupid, but I'm prepared to pay whatever they charge me to get my hands the samples. It's kinda funny to think about, but the little 6-fet controller I make from the samples will be both the most costly e-bike controller, and most powerful e-bike controller. :)
Click to expand...

How many of these fets are in each UPS? 6 would be nice, and even nicer if the UPS has some advanced cells and is low priced. :mrgreen:
 
OMG! $11.68 per FET in a quanity of 1,000 on a 3 month lead time!!! Hell yes!!!! I love the way companies jump to try to win Microsoft's supplier contracts. :)

I've got 6 samples on their way to me right now. :)


$11.68 each!!!!
 
$11.68? :shock: :shock: :shock:
Wow, time to revisit that datasheet and figure out the mechanical stuff (mounting heat sinks, etc.)!!!
Maybe Mindshare can score some samples for me as I would love to do some testing. :mrgreen:
 
So how plug and play will these be on an infineon board? ? what sorts of issue might one encounter when fitting say 6 of these in a pack of smokes controller, say a 6fet box. Would i be right in saying that a 6fet of these would be like a 24fet of 4110? would just need a big sink or would water cooling be better?
 
Lets see how 6 samples come out, then... forget group buy on 1000... get them to quote 10K and 100K runs :)

-Mike
 
hydro-one said:
So how plug and play will these be on an infineon board? ? what sorts of issue might one encounter when fitting say 6 of these in a pack of smokes controller, say a 6fet box. Would i be right in saying that a 6fet of these would be like a 24fet of 4110? would just need a big sink or would water cooling be better?
Click to expand...


Yep, that summarizes it pretty well. Each one of these is right about the same as 4x4110s in parallel, but this has package advantages over them in leg current limits, and cooling, and the isolated sink surface makes for simple and durable sinking/mounting.

So, yeah, you get a 24 fet equalivlant controller in a package the size off a 6 fet.
 
oh that is SWEET!!!!! I could imagine having these right in/on the frame. makes a nice big sink. do you think 6 of these would run a collosus or halmotor/mattmotor ? I would love to try!!!!

Keep us updated Luke!! This is definetely what we need / direction to go with this.......allround better stuff!!

mike
 
hydro-one said:
oh that is SWEET!!!!! I could imagine having these right in/on the frame. makes a nice big sink. do you think 6 of these would run a collosus or halmotor/mattmotor ? I would love to try!!!!

Keep us updated Luke!! This is definetely what we need / direction to go with this.......allround better stuff!!

mike
Click to expand...

I wouldn't want to run less than a 12 fet of these for a collosus motor. That would be roughly equal to a 48 fet infinion, which should be able to push that motor decently. :)
 
You must log in or register to reply here.
Back
Top