New 900V SiC MOSFET launched by Cree, 11.5nS switching time
Ash Small, Thu May 14 2015, 12:35AM]c3m0065090j.pdf[/file]
Launched yesterday, from what I can gather. No-one has them in stock yet, so don't know the price.
This could be what I've been waiting for. Pulsed current of 90A, 900V and ~10nS switching times.
Can't wait to see the price.
Re: New 900V SiC MOSFET launched by Cree, 11.5nS switching time
hen918, Thu May 14 2015, 07:09PM
Looks like the ultimate MOSFET. Package looks a bit weird though, although it should be more efficient at high powers, you would have thought the 7L D2PAK would be better suited to lower voltage, higher current applications, where the package is often the limiting factor.
hen918, Thu May 14 2015, 07:09PM
Looks like the ultimate MOSFET. Package looks a bit weird though, although it should be more efficient at high powers, you would have thought the 7L D2PAK would be better suited to lower voltage, higher current applications, where the package is often the limiting factor.
Re: New 900V SiC MOSFET launched by Cree, 11.5nS switching time
Patrick, Thu May 14 2015, 07:45PM
This is a great package and transistor, i finnally have a "DS" that comes off the die, instead of a dam PCB trace with incudtance+Vdrop+current switchng time. now the driving IC can actually see real measurements.
Patrick, Thu May 14 2015, 07:45PM
This is a great package and transistor, i finnally have a "DS" that comes off the die, instead of a dam PCB trace with incudtance+Vdrop+current switchng time. now the driving IC can actually see real measurements.
Re: New 900V SiC MOSFET launched by Cree, 11.5nS switching time
Ash Small, Thu May 14 2015, 09:08PM
Let's see what the price is first. Last time I saw anything similar thay were £40 each from Farnell.
The speed of these is amazing considering the voltage and current capability. Heating must be a problem, hence the wierd package, I presume.
Now we just have to wait and see who has them on the website first, and the all important question.....the price.
Ash Small, Thu May 14 2015, 09:08PM
Let's see what the price is first. Last time I saw anything similar thay were £40 each from Farnell.
The speed of these is amazing considering the voltage and current capability. Heating must be a problem, hence the wierd package, I presume.
Now we just have to wait and see who has them on the website first, and the all important question.....the price.
Re: New 900V SiC MOSFET launched by Cree, 11.5nS switching time
Adrenaline, Fri May 22 2015, 06:19PM
Not too bad:
Digikey
has them for 10.31USD
They do have it in TO-247 too:
http://www.cree.com/~/media/Files/Cree/Power/Data%20Sheets/C3M0065090D.pdf
looks like 9.98USD from Digikey
http://www.digikey.com/product-detail/en/C3M0065090D/C3M0065090D-ND/5253283
Adrenaline, Fri May 22 2015, 06:19PM
Not too bad:
Digikey
has them for 10.31USD
They do have it in TO-247 too:
http://www.cree.com/~/media/Files/Cree/Power/Data%20Sheets/C3M0065090D.pdf
looks like 9.98USD from Digikey
http://www.digikey.com/product-detail/en/C3M0065090D/C3M0065090D-ND/5253283
Re: New 900V SiC MOSFET launched by Cree, 11.5nS switching time
Sulaiman, Fri May 22 2015, 08:15PM
For c.r.t. T.V./Monitor type flyback I'm considering this 1700V SiC mosfet
not quite as fast but very useful......
(I'm not really converted to surface mount yet)
Sulaiman, Fri May 22 2015, 08:15PM
For c.r.t. T.V./Monitor type flyback I'm considering this 1700V SiC mosfet
not quite as fast but very useful......
(I'm not really converted to surface mount yet)
Re: New 900V SiC MOSFET launched by Cree, 11.5nS switching time
Ash Small, Tue May 26 2015, 01:00PM
Mouser has them in stock in UK for £7.21
This is about one sixth the price of the previous generation ones I was looking at, although the others had a slightly better voltage rating and was slightly faster.
These should still be plenty fast enough for what I have in mind, though. Anyone else who has been thinking about the 'inductive adder pulse generator' circuits that were being discussed here a while back should take a look at these as well
Ash Small, Tue May 26 2015, 01:00PM
Mouser has them in stock in UK for £7.21
This is about one sixth the price of the previous generation ones I was looking at, although the others had a slightly better voltage rating and was slightly faster.
These should still be plenty fast enough for what I have in mind, though. Anyone else who has been thinking about the 'inductive adder pulse generator' circuits that were being discussed here a while back should take a look at these as well
Re: New 900V SiC MOSFET launched by Cree, 11.5nS switching time
Sulaiman, Tue May 26 2015, 07:19PM
A small but maybe important thing I noticed,
the SiC mosfet reverse gate voltages are much lower than the forward voltage,
some designs may have problems with this.
Sulaiman, Tue May 26 2015, 07:19PM
A small but maybe important thing I noticed,
the SiC mosfet reverse gate voltages are much lower than the forward voltage,
some designs may have problems with this.
Re: New 900V SiC MOSFET launched by Cree, 11.5nS switching time
Mads Barnkob, Wed May 27 2015, 06:12AM
VGSmax Gate - Source Voltage -8/+18 V (Absolute maximum values)
Yes, most driver circuits will have to be changed, preferably to a modern mosfet driver ic that can do those exact voltages, atleast the lower turn off voltage. Like:
Mads Barnkob, Wed May 27 2015, 06:12AM
Sulaiman wrote ...
A small but maybe important thing I noticed,
the SiC mosfet reverse gate voltages are much lower than the forward voltage,
some designs may have problems with this.
A small but maybe important thing I noticed,
the SiC mosfet reverse gate voltages are much lower than the forward voltage,
some designs may have problems with this.
VGSmax Gate - Source Voltage -8/+18 V (Absolute maximum values)
Yes, most driver circuits will have to be changed, preferably to a modern mosfet driver ic that can do those exact voltages, atleast the lower turn off voltage. Like:
Re: New 900V SiC MOSFET launched by Cree, 11.5nS switching time
Ash Small, Wed May 27 2015, 11:38AM
It doesn't mention anything in the datasheet. (link in OP)
I presume you are referring to reverse voltage spikes when used in flyback topology.
This is exactly what I was planning on using them for. Are you not also considering using SIC's for flybacks?
It does mention something about not exceeding -4 volts between gate and source if using the integral freewheeling diode.
I presume you'd need to use the maximum voltages in order to achieve the published switching times. Does this mean blocking the integral freewheeling diode and adding an external Shottky, or something?
Ash Small, Wed May 27 2015, 11:38AM
Sulaiman wrote ...
A small but maybe important thing I noticed,
the SiC mosfet reverse gate voltages are much lower than the forward voltage,
some designs may have problems with this.
A small but maybe important thing I noticed,
the SiC mosfet reverse gate voltages are much lower than the forward voltage,
some designs may have problems with this.
It doesn't mention anything in the datasheet. (link in OP)
I presume you are referring to reverse voltage spikes when used in flyback topology.
This is exactly what I was planning on using them for. Are you not also considering using SIC's for flybacks?
It does mention something about not exceeding -4 volts between gate and source if using the integral freewheeling diode.
Mads Barnkob wrote ...
VGSmax Gate - Source Voltage -8/+18 V (Absolute maximum values)
Yes, most driver circuits will have to be changed, preferably to a modern mosfet driver ic that can do those exact voltages, atleast the lower turn off voltage. Like:
VGSmax Gate - Source Voltage -8/+18 V (Absolute maximum values)
Yes, most driver circuits will have to be changed, preferably to a modern mosfet driver ic that can do those exact voltages, atleast the lower turn off voltage. Like:
I presume you'd need to use the maximum voltages in order to achieve the published switching times. Does this mean blocking the integral freewheeling diode and adding an external Shottky, or something?
Re: New 900V SiC MOSFET launched by Cree, 11.5nS switching time
Mads Barnkob, Wed May 27 2015, 12:28PM
What Sulaiman and I pointed out, is that you can no longer just use a regular UCC37322 MOSFET driver with a GDT to drive these, like many SSTC schematics use(most MOSFETs can take +/- 20Vgate), you need a "special" MOSFET driver IC to handle the correct gate voltages.
Mads Barnkob, Wed May 27 2015, 12:28PM
Ash Small wrote ...
It doesn't mention anything in the datasheet. (link in OP)
I presume you are referring to reverse voltage spikes when used in flyback topology.
This is exactly what I was planning on using them for. Are you not also considering using SIC's for flybacks?
It does mention something about not exceeding -4 volts between gate and source if using the integral freewheeling diode.
I presume you'd need to use the maximum voltages in order to achieve the published switching times. Does this mean blocking the integral freewheeling diode and adding an external Shottky, or something?
Sulaiman wrote ...
A small but maybe important thing I noticed,
the SiC mosfet reverse gate voltages are much lower than the forward voltage,
some designs may have problems with this.
A small but maybe important thing I noticed,
the SiC mosfet reverse gate voltages are much lower than the forward voltage,
some designs may have problems with this.
It doesn't mention anything in the datasheet. (link in OP)
I presume you are referring to reverse voltage spikes when used in flyback topology.
This is exactly what I was planning on using them for. Are you not also considering using SIC's for flybacks?
It does mention something about not exceeding -4 volts between gate and source if using the integral freewheeling diode.
Mads Barnkob wrote ...
VGSmax Gate - Source Voltage -8/+18 V (Absolute maximum values)
Yes, most driver circuits will have to be changed, preferably to a modern mosfet driver ic that can do those exact voltages, atleast the lower turn off voltage. Like:
VGSmax Gate - Source Voltage -8/+18 V (Absolute maximum values)
Yes, most driver circuits will have to be changed, preferably to a modern mosfet driver ic that can do those exact voltages, atleast the lower turn off voltage. Like:
I presume you'd need to use the maximum voltages in order to achieve the published switching times. Does this mean blocking the integral freewheeling diode and adding an external Shottky, or something?
What Sulaiman and I pointed out, is that you can no longer just use a regular UCC37322 MOSFET driver with a GDT to drive these, like many SSTC schematics use(most MOSFETs can take +/- 20Vgate), you need a "special" MOSFET driver IC to handle the correct gate voltages.
Re: New 900V SiC MOSFET launched by Cree, 11.5nS switching time
Wolfram, Wed May 27 2015, 12:31PM
Page 1, Absolute maximum ratings, "VGSmax"
The reverse voltage spikes across the transformer will appear as forward voltage spikes across the MOSFET, and this is what you need to watch out for, these should not exceed VDSmax.
The internal diode on these devices seems to perform worse with negative gate voltage, so if you rely on the diode you should not use negative gate voltages. If you know when the diode is supposed to conduct, you can turn on the MOSFET, this should lower the reverse voltage drop as MOSFETs conduct in both directions, it will effectively shunt the diode.
Switching times shouldn't change much with applied voltage I think. The internal diodes in SiC MOSFETs are very fast but soft, so they will have a high voltage drop (and therefore dissipation) with significant current flowing through them. If you want to improve it, you can add an external parallel diode with lower voltage drop, but this is only neccessary when you have high average reverse current flowing. No series schottky is neccessary.
Wolfram, Wed May 27 2015, 12:31PM
Ash Small wrote ...
It doesn't mention anything in the datasheet. (link in OP)
Sulaiman wrote ...
A small but maybe important thing I noticed,
the SiC mosfet reverse gate voltages are much lower than the forward voltage,
some designs may have problems with this.
A small but maybe important thing I noticed,
the SiC mosfet reverse gate voltages are much lower than the forward voltage,
some designs may have problems with this.
It doesn't mention anything in the datasheet. (link in OP)
Page 1, Absolute maximum ratings, "VGSmax"
I presume you are referring to reverse voltage spikes when used in flyback topology.
The reverse voltage spikes across the transformer will appear as forward voltage spikes across the MOSFET, and this is what you need to watch out for, these should not exceed VDSmax.
The internal diode on these devices seems to perform worse with negative gate voltage, so if you rely on the diode you should not use negative gate voltages. If you know when the diode is supposed to conduct, you can turn on the MOSFET, this should lower the reverse voltage drop as MOSFETs conduct in both directions, it will effectively shunt the diode.
I presume you'd need to use the maximum voltages in order to achieve the published switching times. Does this mean blocking the integral freewheeling diode and adding an external Shottky, or something?
Switching times shouldn't change much with applied voltage I think. The internal diodes in SiC MOSFETs are very fast but soft, so they will have a high voltage drop (and therefore dissipation) with significant current flowing through them. If you want to improve it, you can add an external parallel diode with lower voltage drop, but this is only neccessary when you have high average reverse current flowing. No series schottky is neccessary.
Re: New 900V SiC MOSFET launched by Cree, 11.5nS switching time
Sulaiman, Wed May 27 2015, 05:44PM
Today I did a quick check on the 1700V device that I mentioned above,
I only used the component test function of my 'scope at work (two wire, c50Hz 8Vrms via a resistor, X=voltage, Y=current)
when Vgs = low; Vd > Vs ... very high resistance, Vd < Vs ... looks like a diode but with an unusually high forward voltage
when Vgs = high; looks like a short circuit
8 Vrms gate-source does nothing ... no gate breakdown at c -11 V peak
from the datasheet, Vgs abs. max. -10 /+25, Recommended -5 /+20
so I was probably stressing the gate but not too badly
(I think that I read somewhere that mosfet gates accumulate damage ... unreliable memory/no references)
Sulaiman, Wed May 27 2015, 05:44PM
Today I did a quick check on the 1700V device that I mentioned above,
I only used the component test function of my 'scope at work (two wire, c50Hz 8Vrms via a resistor, X=voltage, Y=current)
when Vgs = low; Vd > Vs ... very high resistance, Vd < Vs ... looks like a diode but with an unusually high forward voltage
when Vgs = high; looks like a short circuit
8 Vrms gate-source does nothing ... no gate breakdown at c -11 V peak
from the datasheet, Vgs abs. max. -10 /+25, Recommended -5 /+20
so I was probably stressing the gate but not too badly
(I think that I read somewhere that mosfet gates accumulate damage ... unreliable memory/no references)
Re: New 900V SiC MOSFET launched by Cree, 11.5nS switching time
mister_rf, Wed May 27 2015, 08:45PM
Done my home work.
As Wolfram had suggested, I have tested some SiC MOSFET transistors for the body diode’s conduction.
I have measured the MOSFET's body diode forward drop for different values of the gate voltage. Ids = 10 Amps
The results.
mister_rf, Wed May 27 2015, 08:45PM
Done my home work.
As Wolfram had suggested, I have tested some SiC MOSFET transistors for the body diode’s conduction.
I have measured the MOSFET's body diode forward drop for different values of the gate voltage. Ids = 10 Amps
The results.
Re: New 900V SiC MOSFET launched by Cree, 11.5nS switching time
Wolfram, Thu May 28 2015, 10:48AM
Excellent, that's exactly what I had hoped. Even if the diode is very soft, the FET conducts in reverse as well and shunts the diode when it's on. This means that these are excellent for pulse density modulation, where we rely on FETs to conduct significant current in both directions when on.
Wolfram, Thu May 28 2015, 10:48AM
mister_rf wrote ...
Done my home work.
As Wolfram had suggested, I have tested some SiC MOSFET transistors for the body diode’s conduction.
I have measured the MOSFET's body diode forward drop for different values of the gate voltage. Ids = 10 Amps
The results.
Done my home work.
As Wolfram had suggested, I have tested some SiC MOSFET transistors for the body diode’s conduction.
I have measured the MOSFET's body diode forward drop for different values of the gate voltage. Ids = 10 Amps
The results.
Excellent, that's exactly what I had hoped. Even if the diode is very soft, the FET conducts in reverse as well and shunts the diode when it's on. This means that these are excellent for pulse density modulation, where we rely on FETs to conduct significant current in both directions when on.
Re: New 900V SiC MOSFET launched by Cree, 11.5nS switching time
Ash Small, Thu May 28 2015, 12:35PM
On the subject of drivers, the driver linked to by Mads above has max rise time of 40nS and typical fall time of 8nS, max 20nS.
The IXDD614SI has typical rise time 25nS, max 35nS and typical fall time of 18nS, max 25nS.
Does anyone know of any suitable drivers with faster rise and fall times? How does the rise and fall times of the driver affect a switch with rise and fall times around 5nS?
Ash Small, Thu May 28 2015, 12:35PM
On the subject of drivers, the driver linked to by Mads above has max rise time of 40nS and typical fall time of 8nS, max 20nS.
The IXDD614SI has typical rise time 25nS, max 35nS and typical fall time of 18nS, max 25nS.
Does anyone know of any suitable drivers with faster rise and fall times? How does the rise and fall times of the driver affect a switch with rise and fall times around 5nS?
Re: New 900V SiC MOSFET launched by Cree, 11.5nS switching time
Wolfram, Thu May 28 2015, 02:07PM
Rise time of the driver is usually quoted into some capacitance, usually a few nanofarads. These SiC devices have very low input capacitance, so the rise time will be somewhat better than the quoted figure. Due to the low gate charge, you don't need to deliver a lot of current to turn it on in a given time. This means that the most powerful driver won't neccessarily be the fastest one, a smaller driver is more likely to have lower internal delays, so a 2-5 A driver will probably be the fastest. The UCC27524 looks like a promising candidate.
If you want to switch actual currents at these time scales, then layout will be extremely critical. What sort of currents are you switching, and what sort of load are you driving?
Wolfram, Thu May 28 2015, 02:07PM
Rise time of the driver is usually quoted into some capacitance, usually a few nanofarads. These SiC devices have very low input capacitance, so the rise time will be somewhat better than the quoted figure. Due to the low gate charge, you don't need to deliver a lot of current to turn it on in a given time. This means that the most powerful driver won't neccessarily be the fastest one, a smaller driver is more likely to have lower internal delays, so a 2-5 A driver will probably be the fastest. The UCC27524 looks like a promising candidate.
If you want to switch actual currents at these time scales, then layout will be extremely critical. What sort of currents are you switching, and what sort of load are you driving?
Re: New 900V SiC MOSFET launched by Cree, 11.5nS switching time
Ash Small, Thu May 28 2015, 04:15PM
Without going into too much detail in this thread, I'm planning on putting a lot of current through them for periods of ~100nS, through a one turn primary, and into an inductive adder type pulse generator type circuit.I'm al;so planning on paralleleng the switches. I've not got into the maths too deeply yet, I was waiting for some suitable devices to be available at a reasonable price. The current plan is to use toroidal iron powder cores with distributed gap and flyback topology.
The second post on this page has a link to an imnductive adder pulse generator and a sketch of my proposed circuit:-
The circuit linked to is push-pull topology and uses a MetGlas core.
Ash Small, Thu May 28 2015, 04:15PM
Wolfram wrote ...
If you want to switch actual currents at these time scales, then layout will be extremely critical. What sort of currents are you switching, and what sort of load are you driving?
If you want to switch actual currents at these time scales, then layout will be extremely critical. What sort of currents are you switching, and what sort of load are you driving?
Without going into too much detail in this thread, I'm planning on putting a lot of current through them for periods of ~100nS, through a one turn primary, and into an inductive adder type pulse generator type circuit.I'm al;so planning on paralleleng the switches. I've not got into the maths too deeply yet, I was waiting for some suitable devices to be available at a reasonable price. The current plan is to use toroidal iron powder cores with distributed gap and flyback topology.
The second post on this page has a link to an imnductive adder pulse generator and a sketch of my proposed circuit:-
The circuit linked to is push-pull topology and uses a MetGlas core.
Print this page