18 Mosfet Infineon Controller

dirtdad

1 kW
Joined
Mar 2, 2008
Messages
309
Just got in a load of Infineons, see a sample of them below. 72V 65A 18 4310 Mostfets top. The middle houses either the 72V 45A or 48V 30A, and the cute little bug at the bottom is a 36V 20A model:
inf1.jpg
I had them all made with CA direct connect plugs and with a pile of mating shielded pins. I could only get the complete connector for the hall sensor harness. But it works this way.

I am interested in both extremes. I am claiming one of the 36V 20A for myself for a later project, I see it as being good for highly pedal assist/long range/stealth projects, and its cheap. But the beast at the top is what I am excited about. Infineon decided on 18 4310 Mosfets instead of 12 4110s for a few reasons. They claim 18 4310s will have as much current capacity as 12 4110s. I don't know enough about the ratings of Mosfets to evaluate that claim. They also claim that more smaller fets will result in better heat management by distributing the same current over more Fets and more PCB real estate. And 18 4310s are cheaper than 12 4110s.

I promptly installed it on my eChopper / test mule that is constantly being rearranged with new stuff. Just a quick update on the bike. I have the Crystalyte 5304 back on it. There is no way I am going to try 65A with my BMC 600W. The Boa 4.25" tire fits in back with no problem, and looks...well, big:
inf3.jpg
View attachment 1
But the controller is another matter. I am going to have to figure out a better place to install it. For now it is just strapped down. I still have my stand alone CA I will have to switch over to a direct connect.
inf4.jpg
As for performance. Well, its 72V 65A, waddaya expect, it screams! The new Infineon boards feature soft start, which allows you to modulate the throttle better at low speeds. I used to have to hit the throttle and coast if I wanted to do under 10mph but no more, I can apply a steady low throttle and operate much smoother. It is also possible to program the LVC on the chip through an RS232 interface, I plan on getting that as soon as I can. The 48V 30A can handle 36V, so dialing it back to a 36V LVC setting will result in a much beefier 36V controller than the little 10A runt. The 48V can handle 60V too, etc.

The performance bottleneck always moves around as you upgrade, and this is no exception. Turns out my Yesa BMS are set up to sustain 40A max. They can deliver 8 second bursts of 60A. The Cycle Analyst confirms this. Max current is 40A most of the time, but when I look at my MaxAmps at the end of a ride it is always 60A.

So while I cannot report on what it is like to pull 65A constantly, I can say that 40A with 60A bursts is extremely powerful. It has not increased my top speed like current increases have in the past. Looks like 40mph top speed is it. Acceleration and hill climbing are excellent however. I out-accelerate cars at traffic lights all the time. I have a benchmark hill by my house I call "popcorn hill" for how climbing it has blown fets like popcorn in the past. My previous best was about 13mph at the top with a 45A Infineon (which only ever managed about 30A for some reason, then it finally blew). The 18FET beast blows over the hill at over 20mph, which is significant on a grade that peaks at around 14%.

On one hand part of me wants to upgrade to higher output batteries. Yesa is working on them. But on the other hand, this makes a pretty well balanced team - batteries, controller and motor - and the power is all I need, really. Infineons have proven reliable, and the new changes to the board should make them even more so. The 18FET will be my full time controller mated to the 5304 so I will be doing durability testing. The BMC durability testing will be moved to another bike that is not done yet. I want to put the BMC on a pedal electric bike.
 
DD, Neat report.

what is the dimensions of the big controller?

what's the retail on those puppies?

Like your current crankset config, btw...

peace,

Len
 
It may just be my eyes, but the high current wiring on the 72v65 looks a bit anaemic. What gauge is it ?
 
The big controller is 9" x 3.25" x 1.75". Battery and Phase wires are all 10 or 12g. My batts came with 8g. The wire is plenty for the phase wires. I need to shorten the battery wires. If I could pull full time 65A I would make battery wires all 8g into the controller.
 
Lenk42602 said:
what's the retail on those puppies?

Len, follow your nose herefor pricing.

DD, please don't take my AWG comment as being excessivley critical, I don't pay for the wiring on my setup so I have always leaned toward oversized wiring :) Just a thought though, if you would change out the feeders for 8 AWG @ 65a, maybe it would be worth getting the next batch from the factory like that ? Sooner or later one of your customers is going to push your controllers hard up against their limmits and better to slightly over engineer if it does not affect your unit pricing too much.

Kudos for getting the CA wiring integrated from the factory, it is great to see distributors finally getting serious about listening to the customer and evolving their product.
 
Thanks for the report.
A bank of 18 4310's ought to handle 100 amps. I bet you could tweak it higher.
There should be no problem with heating.
Sounds like you need a pack of A123's :twisted:
And some metal gears for that BMC motor...

The weak point will become the torque arms.

I'd say the input of people on this forum had a lot of influence on the design of this.
Never underestimate the power of the forum, Luke...
 
fechter said:
A bank of 18 4310's ought to handle 100 amps. I bet you could tweak it higher.
And some metal gears for that BMC motor...
The weak point will become the torque arms.
I'd say the input of people on this forum had a lot of influence on the design of this.
Never underestimate the power of the forum, Luke...
Wow, over 100A. Let see, that could be in the range of 7500W / 10HP. Nice. Gary Goodrum reports melting X5 phase wires with 100A. Doctorbass reports phase wires on his X5 were 12g with thin Teflon style shielding so the did not look bigger. I don't know what size Gary's phase wires were, maybe smaller? Maybe the 12g phase wires were a running change made to X5s?

I would suggest that torque arms might become a wear item on such a set up. Even with EV techs custom steel torque arms, I have found them deformed around the axle over time. And I use 2.

Yes, this forum had a definite influence, and a good one. By propagating this CA interface it establishes a de facto standard for any other company that might want to make a Cycle Analyst competitor. Hopefully the same thing can be established for all connectors for the controller/motor/battery.
 
Ha Ha ... YOU SO CRAZY!

http://www.comcycle-usa.com/ RULES!

NICE! :wink:

btw ... These controllers are programmable! (He He) :shock:

Welcome to the WONDERFUL WORLD of INFINEON Retail!

AWESOME FIRST STEP!

PS ... My Ol' Lady want's the 9" Controller! :roll:

BIG_9_INCH.jpg
 
Knuckles said:
Ha Ha ... YOU SO CRAZY!

http://www.comcycle-usa.com/ RULES!

NICE! :wink:

btw ... These controllers are programmable! (He He) :shock:

Welcome to the WONDERFUL WORLD of INFINEON Retail!

AWESOME FIRST STEP!

So is http://www.comcycle-usa.com/ the official place to buy these from now?

I thought you also had a lot to do with these too.
 
fechter said:
I thought you also had a lot to do with these too.

I actually do not (currently) sell the 18-fet Infineon.
I have not had the opportunity to examine it in detail.
I do know that the power resistors are arranged completely different than the 12-fet version.
Also, the 18-fet pcb uses an LM393 voltage comparator as well as the LM317 VR.
It is a very different pcb layout from the 12-fet Infineon that I already understand fairly well at this point.

As indicated by Pete ... dirtdad has it for sale here ... http://www.comcycle-usa.com/ProductInfo.aspx?id=5081525
 
Nice!!!

18FET should get the job done. Like the idea of a beefy controller.

Whats up with the 3 wire battery connection?

Just make it 2

Yes I understand the need to have a hot to the brains but why not have it wired internaly off the board?

If I bot one I would end up just making a 2 wire batt connector and soldering the two hots together before the connector end.

Anyways can't have everthing.

Nice work looking forward to hearing more about how these work out.
 
Microbatman said:
Yes I understand the need to have a hot to the brains but why not have it wired internaly off the board?

If I bot one I would end up just making a 2 wire batt connector and soldering the two hots together before the connector end.

Anyways can't have everthing.

Nice work looking forward to hearing more about how these work out.

I have come to like the idea of bringing the brain hot wire out of the box. I do just crimp the two hots in the battery connector, but I splice a handlebar mounted on/off switch into the brain hot. The switch and its wires can be small, you are not cutting battery power. A switch on the brain wire is a kill switch for most runaway conditions, if not all. When you turn off the controller, you eliminate its tiny 3W constant drain and let the controller cool completely without having to disconnect the battery. The switch also turns off a CA direct connect attached to the controller if you have one. A stand alone CA is always on and lit up unless you disconnect the battery. And anything else you may be running by tapping into power on the CA would also turn off like lights, etc. so you end up with a nice way to shut things down.
 
dirtdad,

From my post here ... http://endless-sphere.com/forums/viewtopic.php?f=2&t=5713&start=150#p114045

Knuckles said:
solarbbq2003 said:
knuckles you gotta chat with fechter on his critical frequency chip speed discovery!

Just to clarify ... The "critical frequency" issue was associated with the old (original) pic-micro based "crazy" controller.
I suspect the archaic software used on that old pcb was very sloppy.

The Infineon is all-new and is (most likely) running entirely different and efficient code.
The Infinon code on the 27-mhz XC846 chip has proven acceptable with PUMA and BMC at very high rotor rpm.

So it's not just the MCU speed (in mhz) it is also a matter of the MCU software code.

Bottom line ... Infineon works PERFECTLY WELL with the BMC at 72V+ ...
Maybe dirtdad can report actual no-load and road testing of the Infineon/BMC at higher voltages ...

That would be the best way to put any doubt to rest.

How are the BMC and Infineon getting along at 72V?

Is popcorn hill now just a "speed bump" for you?
 
Thanks for the report.

edit: posted too late at night :)
 
can anyone confirm if xie cheng are using the good quality irf4310 fets, anyone got any close up pics of the fets they are using?
I suggested to one person to try a xie cheng and he had a fet failure after a fairly short time ( could be many reasons for that of course) but just wondering which fets they are puting in, I'm assuming that similar to the 4110 there are chinese and mexican made ones but both having irf stamp, wondering if similar issue will arise as with the irf4110 fets which can be different quality depending on source.
 
also has anyone got access/contact at tiujuana irf plant, anyone can get fets directly from them?
since its seems very difficult for chinese to access the irf4110 from there, it would be very useful to be able to send to xie cheng or whoever to instal in the controllers, I think probably the only way its going to be possible to get controllers with good really good fets is to get from the mexican plant and send to the chinese controller factories
 
solarbbq2003 said:
also has anyone got access/contact at tiujuana irf plant, anyone can get fets directly from them?
since its seems very difficult for chinese to access the irf4110 from there, it would be very useful to be able to send to xie cheng or whoever to instal in the controllers, I think probably the only way its going to be possible to get controllers with good really good fets is to get from the mexican plant and send to the chinese controller factories

i don't know why anyone would use the 4310 when the 4110 has a lower rds-on and lower miller capacitance, making it a much better choice. i don't know how cheap you are getting the fets, but i have been getting the 4110s for about 2.20 in qty 100 for prime parts with recent date codes from http://www.irtronix.com and higher qty the price goes down. i can give you a contact name if you are interested. i know justin actually bought the parts from digikey for $5 apiece and sent them to china to be installed in crystalyte controllers, before i told him about my source for the fets. my contact there has been great, though i have no idea what you expect to pay for the parts in qty 1000. it couldn't hurt to have one more source.

i'd like to get my hands on one of these new controllers but not in my budget right now.
 
So is this guy a 4110 shister or what?
Mr. goodbuy711 ... http://stores.ebay.com/ALLPARTSPIPE

50 - irfb4110 mosfets for $75 USD free shipping ...
http://cgi.ebay.com/50PCS-IRFB4110-FB4110-POWER-MOSFET-TO-220_W0QQitemZ320288552242QQcmdZViewItemQQptZBI_Electronic_Components

:roll:
 
If their shipping Thru Hong Kong, Knuckles, I'll bet they ain't Mexican!
otherDoc
Hi, Bob Mcree! Glad you are back!
 
Knuckles said:
So is this guy a 4110 shister or what?
Mr. goodbuy711 ... http://stores.ebay.com/ALLPARTSPIPE

50 - irfb4110 mosfets for $75 USD free shipping ...
http://cgi.ebay.com/50PCS-IRFB4110-FB4110-POWER-MOSFET-TO-220_W0QQitemZ320288552242QQcmdZViewItemQQptZBI_Electronic_Components

:roll:

I looked at those 4110 on Ebay...but I was affraid that they won't be reliable....so I'm I right? Have anyone try them?
 
if you look closely at the parts shown in the ebay ad the part number starts with FB4110 while all the mexican ir fets i have on the shelf are labeled IRFB4110. The date code and the IR logo are also in the wrong place, and the embossed batch code at the bottom on the mexican fets is missing completely. Why would anybody pay $1.50 for questionable parts coming from some garage in Hong Kong and put them in a controller they will depend on, when you can get prime parts from an IR authorized distributor with ISO traceability in the US for about 75 cents more? Well, i know why some people would, but why would one of us?

If anybody wants to buy a batch of these and send a few to me i will test them and let you know if they act like the real thing, just to satisfy my own curiosity. It seems like it is just too easy to put any label you want on a junk fet; i would never use fets that did not have a reputable pedigree.
 
I bought a pack of them. He's honest, just gets confused. Sent me an extra pack (didn't need 40 FETs) without me paying for it. I sent the one pack back to him.

Haven't had a chance to modify my controller yet.

DK
 
Deepkimchi said:
I bought a pack of them. He's honest, just gets confused. Sent me an extra pack (didn't need 40 FETs) without me paying for it. I sent the one pack back to him.

Haven't had a chance to modify my controller yet.

DK

i hope they work out for you, and that he is as honest about the product as he is about shipping something. i have heard that there are counterfeits and legitimate devices fabricated in asia that do not meet the same test parameters as the ones made in mexico. the parts i buy come in an IR Mexico tube with the date code on it and all 50 ic's in the tube match the code on the tube. they come from irtronix which is an authorized agent of IR. IMHO changing the fets in a controller is way too much work to do with parts that might not be up to the task.
 
Global Manufacturing Capabilities
International Rectifier is a global manufacturer with major ISO-9001-certified wafer fabrication and assembly facilities located in Newport, Wales; Santa Clara, CA; El Segundo, CA; Leominster, MA; Temecula, CA; Mesa, AZ; and Tijuana, MX

Hmmm ... No facilities in China.
 
Knuckles said:
Global Manufacturing Capabilities
International Rectifier is a global manufacturer with major ISO-9001-certified wafer fabrication and assembly facilities located in Newport, Wales; Santa Clara, CA; El Segundo, CA; Leominster, MA; Temecula, CA; Mesa, AZ; and Tijuana, MX

Hmmm ... No facilities in China.

thanks for the info. i guess this means any that originate in asia are probably not real IR products...
 
Back
Top