Someone posted this at Class D forum :
http://www.diyaudio.com/forums/class-d/165091-power-gan-fets-now-digikey.html#post2156002
http://epc-co.com/epc/documents/datasheets/EPC1011_datasheet_final.pdf
Digi-Key - 917-1005-6-ND (Manufacturer - EPC1011)
Very low capacitance, very high transconductance (almost 10S at even below 2A) !!!
Small problem with the bare die, but can be solved.
Larger problem potentially with the high gate current.
Have you already tried them, Nelson ?
Anyone interested in a GB ?
Patrick
http://www.diyaudio.com/forums/class-d/165091-power-gan-fets-now-digikey.html#post2156002
http://epc-co.com/epc/documents/datasheets/EPC1011_datasheet_final.pdf
Digi-Key - 917-1005-6-ND (Manufacturer - EPC1011)
Very low capacitance, very high transconductance (almost 10S at even below 2A) !!!
Small problem with the bare die, but can be solved.
Larger problem potentially with the high gate current.
Have you already tried them, Nelson ?
Anyone interested in a GB ?
Patrick
I have not tried them. They look like nice switches, but what
do you suppose the dissipation capability will be?
😎
do you suppose the dissipation capability will be?
😎
That depends on how you package them.
You get an LGA bare die. The rest is in your own hands.
Properly packaged, I expect them to do 100W. The die itself is not small at 3.5x1.4mm.
Actually I should not say FETs. They should be called GaN power transistors.
Looks very linear above 5A, and negative tempco as a bonus.
Patrick
You get an LGA bare die. The rest is in your own hands.
Properly packaged, I expect them to do 100W. The die itself is not small at 3.5x1.4mm.
Actually I should not say FETs. They should be called GaN power transistors.
Looks very linear above 5A, and negative tempco as a bonus.
Patrick
Well they are glass everywhere except for the solder pads, so unless you have some nifty processing equipment on hand you'd be faced with aligning a direct metal interface to the part. Probably copper, polished, and cold.
You missed the point.
The die is grown on Si wafer, and the bump pads have no solder balls.
No one forces you to use flip chip.
You can even mount it to a TO247 or whatever else.
Patrick
The die is grown on Si wafer, and the bump pads have no solder balls.
No one forces you to use flip chip.
You can even mount it to a TO247 or whatever else.
Patrick
Well, no. I can't mount it to a TO247. Maybe someone with some more equipment could. Maybe a group buy could be arranged for 100,000 pieces. I'm not super familiar with these processes, but I think usually lead wires are bonded to the top of the die and the bottom is etched and plated or something so that it may be soldered to a thermal pad. Anything lower performance than that would probably be a waste of time.
<<Actually I should not say FETs. They should be called GaN power transistors.>>
They are FETs, Enchancement mode MOSFETs to be precise.
Max. power dissipation can be calculated very easy.
If max. continious drain current specified as 12A, and the max. Ron
is 0.025 Ω (we need the worst case here), then the max. power dissipated
for this case will be Pmax=Id²Ron =3.6W.
The chip actually can withstand currents up to 40A (the pulse
requirement for this is only because of the small case employed).
Coupled with the capacitances figures and the transconductance rough estimate of ~16S@2.5A, practically zero (or negative) current tempco at our
currents of interest, that makes these little creatures really remarkable devices.
Clearly, this is just the first steps of GaN technology.
They are FETs, Enchancement mode MOSFETs to be precise.
Max. power dissipation can be calculated very easy.
If max. continious drain current specified as 12A, and the max. Ron
is 0.025 Ω (we need the worst case here), then the max. power dissipated
for this case will be Pmax=Id²Ron =3.6W.
The chip actually can withstand currents up to 40A (the pulse
requirement for this is only because of the small case employed).
Coupled with the capacitances figures and the transconductance rough estimate of ~16S@2.5A, practically zero (or negative) current tempco at our
currents of interest, that makes these little creatures really remarkable devices.
Clearly, this is just the first steps of GaN technology.
It would be nice if they could package them up.
I understand that you can get something packaged in a TO247
with as few as a thousand. It's probably the set up charge
that bites.
😎
I understand that you can get something packaged in a TO247
with as few as a thousand. It's probably the set up charge
that bites.
😎
Member
Joined 2009
Paid Member
The data sheet says high electron mobility, but GaN on it's own isn't too good if I remember. I suspect this isn't a regular MOSFET where electrons are created under a gate oxide but this a HEMT - a device that forms a 2-dimensional electron 'gas' at the interface between two different layers (probably GaN and AlGaN) where it can exhibit high mobility. HEMTs have been made in the past for microwave use where they exhibit extraordinary speed.
GaN, to me, has always been a little weird...I much prefer Fibonaccis and a few other things.
...oops! Wrong forum...😀
...oops! Wrong forum...😀
If we are going to ask for a bigger case🙂, we better ask for TO-264 - this one will have more realistic 180-200W rating, than the smaller case TO-247.
Can they put a solder pad on the bottom of a die after dicing or is that something that is usually arranged for at the wafer level?
Naa, a dab of paste on each pad, hit it with the air pencil for a second or 2 and shazaam. Bumps 😀
The die ain't small but it ain't big either. We're really going to have to die bond to the back side to dissapte any heat. Go to IRF.com and check out DirectFETs. A copper "can" with the die bonded to it (The Drain) and solder bumps for the Source and Gate 😀 A little paste on the board and,,, You get it
I'm not sure that would be enough for us though. I think a big copper "can" will cuase a disimilar expansion ratio and crack the die??? Die bonding isn't hard either though. Just higher temp paste and a hot oven. Now a wire bonder ain't somethin any of my friends have in their garage... Hmmm
Last edited:
> The die ain't small but it ain't big either. We're really going to have to die bond to the back side to dissipate any heat.
Now this is a real pro. This is exactly what I have in mind.
> Now a wire bonder ain't somethin any of my friends have in their garage.
If you had been designing wire bonders and die bonders in the 80's and the 90's, then you would have had a few such friends.
We'll see what we can do.
😉
Patrick
Now this is a real pro. This is exactly what I have in mind.
> Now a wire bonder ain't somethin any of my friends have in their garage.
If you had been designing wire bonders and die bonders in the 80's and the 90's, then you would have had a few such friends.
We'll see what we can do.
😉
Patrick
Last edited:
You'll get that done and they'll release them in TO220s for linear audio applications in the next week, 5 bucks a pop.
Not unless Nelson orders 100k from them.
If they would offer them in TO220 for 5 bucks, then everyone here is happy.
Patrick
If they would offer them in TO220 for 5 bucks, then everyone here is happy.
Patrick
Might not hurt to get them on the phone and ask what the plans are regarding optimizing and releasing parts for AF linear. Maybe this particular series of part is so tuned for switching compared to what might be possible with linear in mind that there is no particular point in packaging them for linear, market or not. Then again maybe you'd have a hard time convincing them that such a transistor is even necessary for audio.
By the way, if you meant that you have worked on such equipment, how do they prepare the dice for bonding to a thermal pad?
By the way, if you meant that you have worked on such equipment, how do they prepare the dice for bonding to a thermal pad?
Last edited:
Only Nelson has the reputation and the possible quantities to ask a manufacturer for such things.
DIYer will never order 10k of the stuff. So there is no motivation for them.
Patrick
DIYer will never order 10k of the stuff. So there is no motivation for them.
Patrick
Just one, in the '90s. (Die Bonder) 😀
Did mostly Test and Front End in the '80s
I shared a cube with the Wire Bonder guy... Definately more difficult than Die Bond and Bumps 😀
- Status
- Not open for further replies.