Better power transistors. (?)

[Sjouke Burry]

Science daily had an article about a big jump in
power transistors.
Article:
> https://www.sciencedaily.com/releases/2019/06/190605100350.htm

 

[Phil Allison]

Sjouke Burry wrote:

Science daily had an article about a big jump in
power transistors.
Article:
https://www.sciencedaily.com/releases/2019/06/190605100350.htm

** But 5V gate IGBTs already exist.

https://www.rohm.com/datasheet/RGPR30BM40HR/rgpr30bm40-e

Threshold nominally 1.7V and full conduction(30A) at 4 to 5V.



...... Phil

 

[Tim Williams]

"Phil Allison" <pallison49@gmail.com> wrote in message
news:24b08102-b146-4783-8bd7-c45a6e1e0362@googlegroups.com...

** But 5V gate IGBTs already exist.

https://www.rohm.com/datasheet/RGPR30BM40HR/rgpr30bm40-e

Threshold nominally 1.7V and full conduction(30A) at 4 to 5V.

There's also teeny MSOPs that handle 400V and 300A (pulsed), with similar
Vge(on).

Both have embarrassing speed. I don't know if anyone makes those at high
voltages either?

Hopefully the article is talking about regular (fast) speed parts, which
would be handy indeed.

Also, hopefully they're not just talking about something chintzy, like a
nice GaN transistor below a depletion mode regular-IGBT. Which, I mean,
that would work, but it won't be as cheap as a proper Si-only part, and
won't switch the load any faster, just take less drive.

Or SiC IGBT, or... can they make IGBTs out of GaN? IGHBT? That'd be fancy
as hell...

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/

 

[Clifford Heath]

On 6/6/19 2:49 pm, Tim Williams wrote:

"Phil Allison" <pallison49@gmail.com> wrote in message
news:24b08102-b146-4783-8bd7-c45a6e1e0362@googlegroups.com...
** But 5V gate IGBTs already exist.

https://www.rohm.com/datasheet/RGPR30BM40HR/rgpr30bm40-e

Threshold nominally 1.7V and full conduction(30A) at 4 to 5V.


There's also teeny MSOPs that handle 400V and 300A (pulsed), with
similar Vge(on).

Both have embarrassing speed.  I don't know if anyone makes those at
high voltages either?

Hopefully the article is talking about regular (fast) speed parts, which
would be handy indeed.

Also, hopefully they're not just talking about something chintzy, like a
nice GaN transistor below a depletion mode regular-IGBT.  Which, I mean,
that would work, but it won't be as cheap as a proper Si-only part, and
won't switch the load any faster, just take less drive.

Assuming you're not limited by source lead inductance. Tom McEwan found
that was so much of an issue he got a patent for matching it with an
inductive gate kicker - I recall Win Hill trying that and switching 1kv
in <2ns a few years back.

Clifford Heath

 

[Tim Williams]

"Clifford Heath" <no.spam@please.net> wrote in message
news:rf2KE.49926$NT5.43907@fx47.iad...

Assuming you're not limited by source lead inductance. Tom McEwan found
that was so much of an issue he got a patent for matching it with an
inductive gate kicker - I recall Win Hill trying that and switching 1kv in
2ns a few years back.

Years? I thought he was working on that a few months ago...

We've been at that point for a while -- three-leaded packages are an
absolute killer and there are a number of Kelvin leaded parts available now
(though sadly none as mainstream as TO-220 and 247*).

*SOT-227/Mini-Bloc is pretty common, but many times more expensive.

This also gives us a lot of great SMTs, with scary ratings (huge switching
V*I area) and very little practical dissipation. You don't get much leeway
during testing!

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/

 

this means AM radio will become even more useless due to the interference

m

 

[Clifford Heath]

On 6/6/19 9:40 pm, Tim Williams wrote:

"Clifford Heath" <no.spam@please.net> wrote in message
news:rf2KE.49926$NT5.43907@fx47.iad...
Assuming you're not limited by source lead inductance. Tom McEwan
found that was so much of an issue he got a patent for matching it
with an inductive gate kicker - I recall Win Hill trying that and
switching 1kv in <2ns a few years back.

Years?  I thought he was working on that a few months ago...

Perhaps also recently, but I'm referring to these two threads:
<https://groups.google.com/forum/#!searchin/sci.electronics.design/US05274271%7Csort:date/sci.electronics.design/pFPMMILyooo/BO0kM7wFoksJ>
and
<https://groups.google.com/forum/#!msg/sci.electronics.design/Oo-EvJIpOsA/81lHmQti6WgJ;context-place=forum/sci.electronics.design>

from 19 years ago

Clifford Heath.

 

[Tim Williams]

"Clifford Heath" <no.spam@please.net> wrote in message
news:jthKE.628$ZD7.384@fx03.iad...

Perhaps also recently, but I'm referring to these two threads:
https://groups.google.com/forum/#!searchin/sci.electronics.design/US05274271%7Csort:date/sci.electronics.design/pFPMMILyooo/BO0kM7wFoksJ
and
https://groups.google.com/forum/#!msg/sci.electronics.design/Oo-EvJIpOsA/81lHmQti6WgJ;context-place=forum/sci.electronics.design

from 19 years ago

The recent work was Win's SiC pulser. But I see this predates me; good
memory. ;-)

Current link to patent:
https://patents.google.com/patent/US5274271A/en

The table is excellent, rare that articles on these subjects actually
disclose the values used. A bit curious what the layout looked like. The
inductors are small, I wonder if a microstrip construction would do.
1N5408s don't exactly have low inductance themselves (maybe 5nH when mounted
flush over ground plane?). Much better should be possible today, say with
SMC or TO-277 or other low-profile diodes.

SiC schottky also interesting (for the shock line, not so much the final
avalanche bit), but basically impossible to find outside of leaded packages,
grumble.

Hm, so dynamic avalanche is the same thing that BJTs do, but in a PN
junction, and since it's not three-layer punch-through*, it takes extreme
conditions to trigger (rapid dV/dt, peak voltage >> Vrrm)?

https://aip.scitation.org/doi/10.1063/1.1947915 seems to suggest that's
vaguely true: the avalanche mechanism does indeed have a speed limit
(despite what the TVS manufacturers will tell you), and exceeding that speed
limit makes for interesting physics.

*Or whatever the exact mechanism is, I'm not sure I've seen an analysis of
it.

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/

 

[Clifford Heath]

On 7/6/19 10:21 am, Tim Williams wrote:

"Clifford Heath" <no.spam@please.net> wrote in message
from 19 years ago
The recent work was Win's SiC pulser.  But I see this predates me; good
memory. ;-)

I think I triggered the discussion, because I was very interested in
McEwan's low-tech UWB implementations at the time. My OP doesn't seem to
exist in Google's archives.

Current link to patent:
https://patents.google.com/patent/US5274271A/en

That's where he uses a common rectifier as an SRD. PIN structure, so why
not! But it's not the patent where he generates the 1000V step from a
MOSFET. I was referring to this:
<https://patents.google.com/patent/US5332938A/en>

The rectifier-as-SRD patent also says this, which I don't find discussed
elsewhere:
"The present inventor has also built a current-mirror driver circuit
(not disclosed in this application) that uses MOSFETs to generate -1 kV,
2 ns voltage steps to drive transmission line 12 with a 1 Watt power
supply."

The table is excellent, rare that articles on these subjects actually
disclose the values used.

That's a pattern with McEwan. His stud-finder doesn't just give enough
for a full schematic, it even mentions preferred chip manufacturers for
the common CMOS he used: <https://patents.google.com/patent/US5457394A/en>

Hm, so dynamic avalanche is the same thing that BJTs do, but in a PN
junction, and since it's not three-layer punch-through*, it takes
extreme conditions to trigger (rapid dV/dt, peak voltage >> Vrrm)?

They're just bloody big diodes, so they need a big whack.

McEwan and LLNL got mired in an international patent dispute with Time
Domain Systems, which prevented us from getting all the nice impulse
radar goodies that might have flowed from McEwan's work. I'm sure TDS
were legally justified, but I detest them on ethical grounds, because
they failed to deliver any earth-shaking technology from it. Now that
enough time has passed and the patents are expiring, people here should
get busy building all the nice things.

Clifford Heath.

 

[Winfield Hill]

Clifford Heath wrote...

On 7/6/19 10:21 am, Tim Williams wrote:
"Clifford Heath" <no.spam@please.net> wrote in message
from 19 years ago
The recent work was Win's SiC pulser.  But I see this
predates me; good memory. ;-)

I think I triggered the discussion, because I was very
interested in McEwan's low-tech UWB implementations at the
time. My OP doesn't seem to exist in Google's archives.

Current link to patent:
https://patents.google.com/patent/US5274271A/en

That's where he uses a common rectifier as an SRD. PIN
structure, so why not! But it's not the patent where he
generates the 1000V step from a MOSFET. I was referring
to this:
https://patents.google.com/patent/US5332938A/en

The rectifier-as-SRD patent also says this, which I don't
find discussed elsewhere:
"The present inventor has also built a current-mirror driver
circuit (not disclosed in this application) that uses MOSFETs
to generate -1 kV, 2 ns voltage steps to drive transmission
line 12 with a 1 Watt power supply."

Dunno about that one, but explored my computer, found a
SPICE exploration I did Sept 2000, using McEwan's recipe
in patent 5,332,298 where he achieved 900V 36A switching
in 2ns with a Motorola type 4N100. No datasheet found.
I used Motorola's SPICE model for their MTP3N100, and
SPICE was indeed able to switch 900V in 2ns, documented in
"900V 30kW 2ns with a 3N100". McEwan did amazing stuff.


--
Thanks,
- Win

 

On Friday, 7 June 2019 02:22:25 UTC+2, Tim Williams wrote:

"Clifford Heath" <no.spam@please.net> wrote in message
news:jthKE.628$ZD7.384@fx03.iad...
Perhaps also recently, but I'm referring to these two threads:
https://groups.google.com/forum/#!searchin/sci.electronics.design/US05274271%7Csort:date/sci.electronics.design/pFPMMILyooo/BO0kM7wFoksJ
and
https://groups.google.com/forum/#!msg/sci.electronics.design/Oo-EvJIpOsA/81lHmQti6WgJ;context-place=forum/sci.electronics.design

from 19 years ago

The second reference is Mikes analysis of McEwans pulse circuit. He references this web page:

http://www3.sympatico.ca/add.automation/mcewan/mcewan.htm

But of course the link is broken

Anyone got a copy of it?

Regards

Klaus

 

[Winfield Hill]

klaus.kragelund@gmail.com wrote...

On Friday, 7 June 2019, Tim Williams wrote:
"Clifford Heath" wrote

Perhaps also recently, but I'm referring to these two threads:
https://groups.google.com/forum/#!searchin/sci.electronics.design/US05274271%7Csort:date/sci.electronics.design/pFPMMILyooo/BO0kM7wFoksJ
and
https://groups.google.com/forum/#!msg/sci.electronics.design/Oo-EvJIpOsA/81lHmQti6WgJ;context-place=forum/sci.electronics.design

from 19 years ago

The second reference is Mikes analysis of McEwans pulse circuit.
He references this web page:

http://www3.sympatico.ca/add.automation/mcewan/mcewan.htm

But of course the link is broken

Anyone got a copy of it?

I might. I saved three of Mike Monett's html files, dated
19 Sept, 2000. But they don't deal with the 5274271 patent,
but instead the 5332298 MOSFET switching patent. Going over
Mike's document and my notes, Mike showed that McEwan's patent
couldn't work, but my SPICE modeling showed that it could do
900V 2ns high-current falling edges. But just now I noticed
that I had used McEwan's 4nH value for source inductance, but
datasheet SPICE models say 13nH for a standard TO-247 package.

What's current FCC status on allowing UltraWideband pulsing?


--
Thanks,
- Win

 

[Winfield Hill]

Winfield Hill wrote...

klaus.kragelund@gmail.com wrote...

Anyone got a copy of it?

I might. I saved three of Mike Monett's html files, dated
19 Sept, 2000. But they don't deal with the 5274271 patent,
but instead the 5332298 MOSFET switching patent. Going over
Mike's document and my notes, Mike showed that McEwan's patent
couldn't work, but my SPICE modeling showed that it could do
900V 2ns high-current falling edges. But just now I noticed
that I had used McEwan's 4nH value for source inductance, but
datasheet SPICE models say 13nH for a standard TO-247 package.

I did my modeling on 23 Sept, but now don't remember
doing it, let alone Mike's response after posting it.


--
Thanks,
- Win

 

On Wednesday, 12 June 2019 22:14:39 UTC+2, Winfield Hill wrote:

klaus.kragelund@gmail.com wrote...

On Friday, 7 June 2019, Tim Williams wrote:
"Clifford Heath" wrote

Perhaps also recently, but I'm referring to these two threads:
https://groups.google.com/forum/#!searchin/sci.electronics.design/US05274271%7Csort:date/sci.electronics.design/pFPMMILyooo/BO0kM7wFoksJ
and
https://groups.google.com/forum/#!msg/sci.electronics.design/Oo-EvJIpOsA/81lHmQti6WgJ;context-place=forum/sci.electronics.design

from 19 years ago

The second reference is Mikes analysis of McEwans pulse circuit.
He references this web page:

http://www3.sympatico.ca/add.automation/mcewan/mcewan.htm

But of course the link is broken

Anyone got a copy of it?

I might. I saved three of Mike Monett's html files, dated
19 Sept, 2000. But they don't deal with the 5274271 patent,
but instead the 5332298 MOSFET switching patent. Going over
Mike's document and my notes, Mike showed that McEwan's patent
couldn't work, but my SPICE modeling showed that it could do
900V 2ns high-current falling edges. But just now I noticed
that I had used McEwan's 4nH value for source inductance, but
datasheet SPICE models say 13nH for a standard TO-247 package.

Would be interesting if you had a copy

What's current FCC status on allowing UltraWideband pulsing?

For CISP14, the emission limits are quasi peak and average. So if you keep the average energy low enough, I think you can do whatever pulse you like

I believe the FCC15 has the same measuring definition

Cheers

Klaus