Will to247 fet work instead of to220

Ratking

Ratking

1 kW
Joined
May 20, 2010
Messages
494
Location
Norway
I just read that to247 fets are way better than the smaller to220 package. Distance between the legs are 2.54mm for to220 and 5.2mm for this 400A fet from ixys http://ixdev.ixys.com/DataSheet/DS100221(IXFH-FT400N075T2).pdf

If we could force the legs down in the holes I would guess we would have a pretty tough controller able to withstand alot more abuse. Even if we had to make a new board that copied the layout and just bigger fetholes :lol: it would still be worth it. Just as long as someone comes up with a bulletproof controller that is.

Any input appreciated, as I don't know what to see for regard the different in the fet requirement to drive them.
 
I would say the gate driver circuit might need modification. Also, the current rating of the FET is not the only parameter that needs attention. There are also considerations like Rds-on and switching times. But if you've found a better option than the irfb4110, then great! Let the infineon modding continue :D
 
The spacing on the fet heatsink would change also because the package is bigger... So would it work right out of the box... prob not.
 
Ratking said:
I just read that to247 fets are way better than the smaller to220 package. Distance between the legs are 2.54mm for to220 and 5.2mm for this 400A fet from ixys http://ixdev.ixys.com/DataSheet/DS100221(IXFH-FT400N075T2).pdf

If we could force the legs down in the holes I would guess we would have a pretty tough controller able to withstand alot more abuse. Even if we had to make a new board that copied the layout and just bigger fetholes :lol: it would still be worth it. Just as long as someone comes up with a bulletproof controller that is.

Any input appreciated, as I don't know what to see for regard the different in the fet requirement to drive them.
Click to expand...

As others have already mentioned, spacing of the FET's is the real limitation to populating Infineon controllers with superior TO-247 package FETs. All the infineon controller PCB designs try to minimise parasitic impedance effects by placing only a small gap between adjacent FETs to minimise relative track lengths, together with minimising thermal difference between relative FETs. Another advantage of doing this is of course the minimisation of the length of the board and therefore enclosure dimensions.

Some ES members have managed to populate 6 FET Infineons with TO-247 FETs by using a replacement custom heat-spreader and further 'spread-out' packages. They also extended the FET leads where they wont reach the footprint holes in the PCB on their own. Although this may work in theory and/or in practice, i would be hesitant to try this myself due to the unbalanced inductive impedance of relative FET leads. As far as i can tell, this sort of mod is only easily achieved with 6 FET controllers as the increased relative spacing of packages can still be accommodated within the standard enclosure dimensions. To try this with a 12 FET or greater controller would only compound the spacing/relative-lead-length problem.

Also, with FETs, It doesn't matter what the inflated current rating of the Silicon/Junction is, it's the lead limit that is always the bottleneck. TO-220ABs have a lead limit of 75A, TO-247s have lead limits between ~150A > ~200A

The most important limiting parameters of the particular FET you linked to are;

IXYS IXFH-FT400N075T2 (TO-247IXFH)
VDSS = 75V
Lead Current Limit = 160A RMS (The real Max continuous current limit of the FET. Higher is better)
RDS(on) = 2.3 mΩ typical @ 100A (An indicator of package heat produced for a given current transfer. Lower is better)
TJ = 175°C (Destructive temperature point of the junction. Higher is better)
Total RthJH = 0.36 °C/W (Junction to Heat-sink thermal resistance. The rise in temperature of the junction for a given power dissipation. Lower is better)
[RthJH = RthJC + RthCH]
Qg(on) = 420 nC typical (Total Gate charge. The total amount of energy needed to switch the FET gate on throughout the gate charge cycle. Lower can be better as gate driver requirements are lower and switching time for a limited gate current is faster)
Price = $6.125 (Digikey, Currently a Non stocked item. Lower is better :lol:)
http://ixdev.ixys.com/DataSheet/DS100221(IXFH-FT400N075T2).pdf

A decent TO-247 FET. But i would consider this one a little better.

International Rectifier- IRFP4368 (TO-247AC)
VDSS = 75V
Lead Current Limit = 195A RMS
RDS(on) = 1.46mΩ typical
TJ = 175°C
RthJH = 0.53°C/W
Qg(on) = 380 nC typical
Price = $8.57(Newark), $7.70(Digikey. Currently a Non stocked item)
http://www.irf.com/product-info/datasheets/data/irfp4368pbf.pdf

But compared to an TO-220 IRFB4110, both are a significant improvement if you can settle for a 75V limit.

International Rectifier IRFB4110 (TO-220AB)
VDSS = 100V
Lead Current Limit = 75A (http://irf.custhelp.com/cgi-bin/irf...?p_faqid=373&p_created=1060813068&p_topview=1)
RDS(on) = 3.7mΩ typical
TJ = 175°C
RthJH = 0.902°C/W
Qg(on) = 150 nC typical
Price = $4.83 (Digikey), $5.44 (Newark)
http://www.irf.com/product-info/datasheets/data/irfb4110pbf.pdf


I'll leave the comparative summary of what all this data means for later (or someone else). Im sick of writing right now :mrgreen:

EDIT: For those still wanting a 100V ceiling like an IRFB4110, heres one of the best upgrades in TO-247 packaging - 2.0mΩ RDS(on) http://www.irf.com/product-info/datasheets/data/irfp4468pbf.pdf
 
Does the 75V limit mean 80V will destroy it? 84V? My 20s lipo is 74V nominal...

These FETs look significantly better than the 4110. Is the package sizing/spacing REALLY such a hard issue to get around? Why aren't we adapting the infineon boards to work with these FETs instead of moving to higher and higher numbers of 4110s? It seems to me a 12, 18, 24... 36FET with these slightly larger FETs would perform very very well. Good enough to warrant trying to solve the spacing problem. Am I wrong?
 
dozentrio said:
Does the 75V limit mean 80V will destroy it? 84V? My 20s lipo is 74V nominal...
Click to expand...

Yes and dont forget any ripple ontop of your nominal voltage. playing on the safe side max voltage for a 75v fet should be around 65v

These FETs look significantly better than the 4110. Is the package sizing/spacing REALLY such a hard issue to get around? Why aren't we adapting the infineon boards to work with these FETs instead of moving to higher and higher numbers of 4110s? It seems to me a 12, 18, 24... 36FET with these slightly larger FETs would perform very very well. Good enough to warrant trying to solve the spacing problem. Am I wrong?
Click to expand...

I think yes, but only as a 6 fet controller as already been said or Maybe 12 fet with some major layout/heat sink mods.
 
Why not use a 12 fet board, but only use 6 of the 247's? Then the space issue is solved and you just have to figure out how to make the legs fit? Or will the 12 fet controller not run with only single fets instead of 6 pairs?
 
For those still wanting a 100V ceiling like a TO-220 IRFB4110, here's one of the best upgrades in TO-247 packaging - 2.0mΩ RDS(on) http://www.irf.com/product-info/datasheets/data/irfp4468pbf.pdf

The max i would personally push 75V FETs is 18S LiFePO4(~65.7V hot off the charger), or 16S LiCoO2/LiMn2O4 (~67.2V hot off the charger). As has been said, it gives some margin for ripple and manufacturing tolerance.
 
John in CR said:
Why not use a 12 fet board, but only use 6 of the 247's? Then the space issue is solved and you just have to figure out how to make the legs fit? Or will the 12 fet controller not run with only single fets instead of 6 pairs?
Click to expand...

I dont think there will be anything to gain i.e 12 smaller fets or 6 larger fets, the 12 smaller fets will handle roughly what 6 larger fets can and no mods to the board.
 
What if you took an 18fet and made it into a 12 fet. There would be something to gain.
 
icecube57 said:
What if you took an 18fet and made it into a 12 fet. There would be something to gain.
Click to expand...

This maybe a possible option as the 18fet board is longer but there will still have to be some major modding to the PCB layout ( I have never seen a 18fet board so not 100% if the driver circuit/layout will suit ) I think if I were to go to a 12 fet layout using these larger packaged fets I would use a 6 fet board and design a slave fet board stacking each pair of fets one in front of the other and use 2 heatsink spreader bars and rehouse the whole controller into a more suited box ( Its something I was considering a while ago until I started looking at the price of the fets ) I could not justify spending about $100 on something to try, I still may try this ( when/if I blown up one of my controllers ) but using the smaller package fets as they are a tad cheaper and it will make the controller not that much bigger than a standard 6 fet.
 
If I were to do such a conversion of an existing PCB, I'd peel off the existing FET traces, and lay down copper PCB repair tape for the drain and source traces for the larger FETs. Then drill holes as needed for the leads of the larger FETs, solder in as needed, then run identical-length (shielded?) gate wires from the driver outputs of the existing PCB circuitry.

It'd be better to do it with copper sheet, perhaps, for better high-current conductivity, but the repair tape already has adhesive on it to make attaching it to the PCB easy. You could then solder copper sheet (or bars) to the repair tape, if desired.
 
On thinking about it some more, i think the 18 FET TO-220 pcb would be a great donor board for a 12 FET TO-247 upgrade/modification......

One in three FET footprints would simply be omitted to make space for the wider FET packaging. There would be minimal added lead length.

Whereas Phase current was previously limited to 3 x 75A packages = 225A (TO-220 package limit), it is now increased to 2 x 195A = 390A (TO-247 package limit). A 73.3% increase in current capability. EDIT: a few TO-220 FET's are spec'd with 120A lead limits - There is some contentiousness about the value of this rating though as the same manufacturers application documents spec this package type with a 75A wire bond limit between the die and the lead.

The standard alloy heat-spreader bar could simply be re-drilled for the different FET packaging and the gate resistors replaced with appropriate values (calculating for, and omitting the un-needed ones)

18_FET_PCB_PACKAGE_UPGRADE.JPG
 
With careful tweaking i think it would work. YOU HEAR THAT LYEN I WANT A 12 FET with the 247 packager. Boost you took the mental picture Right out of my head. If you bend the fet legs wouldn it compromise the package and current capability. I know the legs are long and can be manipulated but I wouldnt want it to reduce fet life buy trying to make it fit into a 220 configuration. Its the medium profile controller with balls I always wanted. With 6G phase and battery wires. ::SPLOOGE::
 
As long as the leads are bent with a gradual arc then its fine. They are malleable enough for a couple of bends before they fatigue and crack. Also, the TO-220 footprints would likely need the holes drilled out slightly larger to fit the bigger TO-247 leads. Although you'd loose the through hole plating by doing this, its no problem, as you would ensure the junction of each pin and associated power-bus trace was soldered on its relevant side of the PCB.
 
I think I'll just try to find a scooter hubmotor/controller combo that has 247's in the controller, and pay the say price before shipping as a controller alone. :mrgreen:
 
boostjuice said:
Good luck....Ive only ever seen cheap mainland china controllers populated with TO-220 FETs.
Click to expand...

How many 3k-7k scooter hubmotor/controller combos have you seen? Someone in the motorcycle section had a scooter controller with 247´s recently, and my guess is that may be part of the reason they run those scooters only at 60v.
 
I've read the post you talk about, and I also found the site where they sell the complete scooter. Haven't had the time to give them a mail, but I think thats the way to go IF they are cheap enough. I wonder what kind of fets Kelly Controllers use. They have some large controllers for bldc motors.

Regarding to247 on a to220 board: If we need to make too much modifications to the original board to make them fit, couldn't we just make a very thick and solid board containing only the power\fets and just route the fet drive signal to each fet?
But then again, is there any larger fet package that is superior to to247 if I choose to make myself a power board?
I'm looking for a cheap fet that Infinion can drive without any problem. If anyone could point out what the maximum gatecharge Infinion can produce, its just a matter of looking in a bunch of datasheets :)
 
Ratking said:
But then again, is there any larger fet package that is superior to to247 if I choose to make myself a power board?
I'm looking for a cheap fet that Infinion can drive without any problem. If anyone could point out what the maximum gatecharge Infinion can produce, its just a matter of looking in a bunch of datasheets :)
Click to expand...
There are some amazing TO-264 case FETs out there that have incredible current, power, and thermal resistance specs but they're typically quite expensive. IMHO, TO-247 is a great compromise between gate driving requirements and the advantages of greater power handling and lower thermal resistance/RDS-on.

Regarding your Infinion question...
The greater the gate charge, the hotter the FET will become. :) There's no point you can call a maximum. You'll have to do the classic juggling of cost, availability, case size, gate charge, RDS-on, thermal resistance specs to see which FET will work best for your application.
 
CamLight said:
There are some amazing TO-264 case FETs out there that have incredible current, power, and thermal resistance specs but they're typically quite expensive.
Click to expand...
https://www.onsemi.com/PowerSolutions/product.do?id=NTY100N10
I've used some of those; they're what I used for my final repair of the brushed ScootNGo controller, and they worked fine for all the time I used it after that, until the controller chip itself died. :)

If I have enough left, I might convert a 6-FET over to them just to see if it does anything to help. Or maybe whichever controller I blow up next. :)
 
boostjuice said:
On thinking about it some more, i think the 18 FET TO-220 pcb would be a great donor board for a 12 FET TO-247 upgrade/modification......

One in three FET footprints would simply be omitted to make space for the wider FET packaging. There would be minimal added lead length.

Whereas Phase current was previously limited to 3 x 75A packages = 225A (TO-220 package limit), it is now increased to 2 x 195A = 390A (TO-247 package limit). A 73.3% increase in current capability.

The standard alloy heat-spreader bar could simply be re-drilled for the different FET packaging and the gate resistors replaced with appropriate values (calculating for, and omitting the un-needed ones)

18_FET_PCB_PACKAGE_UPGRADE.JPG
Click to expand...


I really like this idea. I still think there are better solutions with bigger fet packages, but this one is easy and fairly cheap. It may just be enough to get my 6500w turnigy alive.
If Lyen want to sell a 18fet board without the fets, its just for us to solder on 12 247 fets :D . Anyone care to tell me if this is doable or not? I would rather use this "slim" controller instead of the 36fet brick.
 
That's the present Plan with our thread Building the Best Controller.
It's not set in stone but I don't see an issue with supporting both devices on the pcb. :)

~KF
 
You must log in or register to reply here.

Similar threads

Jeremy Harris
Simple BLDC controller
13 14 15 16 17
Replies
407
Views
86,001
GrowleyMonster
G
Back
Top