Super duper hype fast FET driver?

[John Larkin]

On Sun, 21 Aug 2011 13:26:08 -0700, Joerg <invalid@invalid.invalid>
wrote:

John Larkin wrote:
On Sun, 21 Aug 2011 09:54:50 -0700, Joerg <invalid@invalid.invalid
wrote:

John Larkin wrote:
On Sun, 21 Aug 2011 07:14:47 -0700, Joerg <invalid@invalid.invalid
wrote:

Bill Sloman wrote:
On Aug 21, 12:13 pm, John Larkin
jjlar...@highNOTlandTHIStechnologyPART.com> wrote:
On Sat, 20 Aug 2011 18:27:00 -0700 (PDT),BillSloman



bill.slo...@ieee.org> wrote:
On Aug 21, 9:52 am, Joerg <inva...@invalid.invalid> wrote:
BillSlomanwrote:
On Aug 21, 12:17 am, Joerg <inva...@invalid.invalid> wrote:
BillSlomanwrote:
On 20/08/2011 3:34 AM, Joerg wrote:
BillSlomanwrote:
On 20/08/2011 2:46 AM, Joerg wrote:
snip

143 lines of pompous bilge, none on the subject of fet drivers.
John Larkin loves posting about fet drivers - he can make implausible
claims about how fast and cheap his are, and pose as the expert
electronic engineer that he wants to be accepted as.

John posts actual scope plots. So when he says that a transition happens
in under one nanosecond and proves it with a scope plot I don't know
what you mean by "implausible". Just because some people or datasheets
say it can't be done does not mean it can't be done. Also, he can show
the proof in revenue Dollars.

Now if someone had the opposite sex of the 2N7002 or a PNP with 15V+ and
no "saturation hold" that would be great. The BSS84 and it's siblings
ain't that hot.
Have you tried any pfets?

I did try the BSS84 a few years ago and the results were not enthusing.

My mental model of a mosfet is an infinitely fast piece of silicon
with some capacitance and wire bonds. So the limit on fast switching
is mostly how hard you can drive the gate. Most fets can switch way
faster than the datasheets suggest if you slam them hard enough.
Unfortunately, my 600 ps mystery driver is only good for 6.5 volts
maybe, which that will drive a 2N7002 to an amp or so, fast, but pfets
usually need more drive to turn on hard. But I bet there's one out
there somewhere.

A little DC pre-bias, just below threshold, can buy another volt or
two of enhancement.


I haven't played much with pfets as really fast switches. Right, a
complement to the 2N7002, push-pull against a 7002, with the same
12-cent gate drivers, would be interesting, and might solve your
problem. Just ignore the shoot-through maybe.

I'll put that on my slow-day experiment list. All I need now is a slow
day.

Why doesn't somebody make p-channel gaasfets? The world wonders.

A gold-doped BJT would be nice as well. However, I have not seen any
commercially successful gold or otherwise doped PNP transistors. And I
guess there ain't no market to write home about so I won't hold my breath.

My experience using bjt's as this sort of fast switch has been
disappointing. I even tried some 45 GHz SiGe parts, and they switched
slow. Mosfets are much better, phemts are radically better.


RF transistors are also a problem if they can't get to an amp or more.
Usually you'll need that to swish charges around in a capacitive load.
LDMOS can pulse nicely but the price tag is usually prohibitive. NPX
isn't very useful there either because they won't release SPICE model.
But PolyFet in Camarillo does.



Some day I'll have to see how the BFT92 and the BFG31 behave. But they
are intended as amplifiers so they won't are about saturation effects.
Trick to keep it out of saturation get old quickly when every pF and
every thenth of an inch count.

Yeah. I'd experiment with finding a 2N7002 complement. Now you've got
me interested.

I have the gear (pulsers, drivers, sampling scope) to experiment, if
you come up with some candidate parts.


This one could be a contender but it's Infineon and seems close to
unobtanium, in which case it would not do too much good:

http://www.infineon.com/dgdl/BSA223SP_Rev1.3.pdf?folderId=db3a304412b407950112b408e8c90004&fileId=db3a304412b407950112b42ae5834414

That looks really nice. Low capacitances, dynamite transfer curve,
nice pulsed current. Let's try some.

This looks similar:

http://www.fairchildsemi.com/pf/FD/FDV304P.html

The Fairchild gumdrop parts are mostly excellent.

This looks promising, too:

http://www.vishay.com/docs/71411/tp0610k.pdf

John

 

[Joerg]

John Larkin wrote:

On Sun, 21 Aug 2011 13:26:08 -0700, Joerg <invalid@invalid.invalid
wrote:

John Larkin wrote:
On Sun, 21 Aug 2011 09:54:50 -0700, Joerg <invalid@invalid.invalid
wrote:

John Larkin wrote:
On Sun, 21 Aug 2011 07:14:47 -0700, Joerg <invalid@invalid.invalid
wrote:

Bill Sloman wrote:
On Aug 21, 12:13 pm, John Larkin
jjlar...@highNOTlandTHIStechnologyPART.com> wrote:
On Sat, 20 Aug 2011 18:27:00 -0700 (PDT),BillSloman



bill.slo...@ieee.org> wrote:
On Aug 21, 9:52 am, Joerg <inva...@invalid.invalid> wrote:
BillSlomanwrote:
On Aug 21, 12:17 am, Joerg <inva...@invalid.invalid> wrote:
BillSlomanwrote:
On 20/08/2011 3:34 AM, Joerg wrote:
BillSlomanwrote:
On 20/08/2011 2:46 AM, Joerg wrote:
snip

143 lines of pompous bilge, none on the subject of fet drivers.
John Larkin loves posting about fet drivers - he can make implausible
claims about how fast and cheap his are, and pose as the expert
electronic engineer that he wants to be accepted as.

John posts actual scope plots. So when he says that a transition happens
in under one nanosecond and proves it with a scope plot I don't know
what you mean by "implausible". Just because some people or datasheets
say it can't be done does not mean it can't be done. Also, he can show
the proof in revenue Dollars.

Now if someone had the opposite sex of the 2N7002 or a PNP with 15V+ and
no "saturation hold" that would be great. The BSS84 and it's siblings
ain't that hot.
Have you tried any pfets?

I did try the BSS84 a few years ago and the results were not enthusing.
My mental model of a mosfet is an infinitely fast piece of silicon
with some capacitance and wire bonds. So the limit on fast switching
is mostly how hard you can drive the gate. Most fets can switch way
faster than the datasheets suggest if you slam them hard enough.
Unfortunately, my 600 ps mystery driver is only good for 6.5 volts
maybe, which that will drive a 2N7002 to an amp or so, fast, but pfets
usually need more drive to turn on hard. But I bet there's one out
there somewhere.

A little DC pre-bias, just below threshold, can buy another volt or
two of enhancement.

I haven't played much with pfets as really fast switches. Right, a
complement to the 2N7002, push-pull against a 7002, with the same
12-cent gate drivers, would be interesting, and might solve your
problem. Just ignore the shoot-through maybe.

I'll put that on my slow-day experiment list. All I need now is a slow
day.

Why doesn't somebody make p-channel gaasfets? The world wonders.

A gold-doped BJT would be nice as well. However, I have not seen any
commercially successful gold or otherwise doped PNP transistors. And I
guess there ain't no market to write home about so I won't hold my breath.
My experience using bjt's as this sort of fast switch has been
disappointing. I even tried some 45 GHz SiGe parts, and they switched
slow. Mosfets are much better, phemts are radically better.

RF transistors are also a problem if they can't get to an amp or more.
Usually you'll need that to swish charges around in a capacitive load.
LDMOS can pulse nicely but the price tag is usually prohibitive. NPX
isn't very useful there either because they won't release SPICE model.
But PolyFet in Camarillo does.


Some day I'll have to see how the BFT92 and the BFG31 behave. But they
are intended as amplifiers so they won't are about saturation effects.
Trick to keep it out of saturation get old quickly when every pF and
every thenth of an inch count.
Yeah. I'd experiment with finding a 2N7002 complement. Now you've got
me interested.

I have the gear (pulsers, drivers, sampling scope) to experiment, if
you come up with some candidate parts.

This one could be a contender but it's Infineon and seems close to
unobtanium, in which case it would not do too much good:

http://www.infineon.com/dgdl/BSA223SP_Rev1.3.pdf?folderId=db3a304412b407950112b408e8c90004&fileId=db3a304412b407950112b42ae5834414


That looks really nice. Low capacitances, dynamite transfer curve,
nice pulsed current. Let's try some.

Tough to get though. Once I tried to cajole Infineon into samples of
another FET, explicitly telling them that I don't want them for free,
would also pay S&H and all that. They didn't afford me as much as a
courtesy call back. Ditched them.

This looks similar:

http://www.fairchildsemi.com/pf/FD/FDV304P.html

The Fairchild gumdrop parts are mostly excellent.

This looks promising, too:

http://www.vishay.com/docs/71411/tp0610k.pdf

I had looked at those a long time ago. The turn-off times are sluggish
yet they are driving them with gusto, for example Rgen 10V/10ohms for
the Vishay. They are smacking a whopping amp into the gate. 10pF and
150ohms load doesn't explain 35nsec turn-off and I bet their marketing
guys would razz them for being too conservative.

But one never knows with FETs.

--
Regards, Joerg

http://www.analogconsultants.com/

 

[John Larkin]

On Sun, 21 Aug 2011 14:21:02 -0700, Joerg <invalid@invalid.invalid>
wrote:

John Larkin wrote:
On Sun, 21 Aug 2011 13:26:08 -0700, Joerg <invalid@invalid.invalid
wrote:

John Larkin wrote:
On Sun, 21 Aug 2011 09:54:50 -0700, Joerg <invalid@invalid.invalid
wrote:

John Larkin wrote:
On Sun, 21 Aug 2011 07:14:47 -0700, Joerg <invalid@invalid.invalid
wrote:

Bill Sloman wrote:
On Aug 21, 12:13 pm, John Larkin
jjlar...@highNOTlandTHIStechnologyPART.com> wrote:
On Sat, 20 Aug 2011 18:27:00 -0700 (PDT),BillSloman



bill.slo...@ieee.org> wrote:
On Aug 21, 9:52 am, Joerg <inva...@invalid.invalid> wrote:
BillSlomanwrote:
On Aug 21, 12:17 am, Joerg <inva...@invalid.invalid> wrote:
BillSlomanwrote:
On 20/08/2011 3:34 AM, Joerg wrote:
BillSlomanwrote:
On 20/08/2011 2:46 AM, Joerg wrote:
snip

143 lines of pompous bilge, none on the subject of fet drivers.
John Larkin loves posting about fet drivers - he can make implausible
claims about how fast and cheap his are, and pose as the expert
electronic engineer that he wants to be accepted as.

John posts actual scope plots. So when he says that a transition happens
in under one nanosecond and proves it with a scope plot I don't know
what you mean by "implausible". Just because some people or datasheets
say it can't be done does not mean it can't be done. Also, he can show
the proof in revenue Dollars.

Now if someone had the opposite sex of the 2N7002 or a PNP with 15V+ and
no "saturation hold" that would be great. The BSS84 and it's siblings
ain't that hot.
Have you tried any pfets?

I did try the BSS84 a few years ago and the results were not enthusing.
My mental model of a mosfet is an infinitely fast piece of silicon
with some capacitance and wire bonds. So the limit on fast switching
is mostly how hard you can drive the gate. Most fets can switch way
faster than the datasheets suggest if you slam them hard enough.
Unfortunately, my 600 ps mystery driver is only good for 6.5 volts
maybe, which that will drive a 2N7002 to an amp or so, fast, but pfets
usually need more drive to turn on hard. But I bet there's one out
there somewhere.

A little DC pre-bias, just below threshold, can buy another volt or
two of enhancement.

I haven't played much with pfets as really fast switches. Right, a
complement to the 2N7002, push-pull against a 7002, with the same
12-cent gate drivers, would be interesting, and might solve your
problem. Just ignore the shoot-through maybe.

I'll put that on my slow-day experiment list. All I need now is a slow
day.

Why doesn't somebody make p-channel gaasfets? The world wonders.

A gold-doped BJT would be nice as well. However, I have not seen any
commercially successful gold or otherwise doped PNP transistors. And I
guess there ain't no market to write home about so I won't hold my breath.
My experience using bjt's as this sort of fast switch has been
disappointing. I even tried some 45 GHz SiGe parts, and they switched
slow. Mosfets are much better, phemts are radically better.

RF transistors are also a problem if they can't get to an amp or more.
Usually you'll need that to swish charges around in a capacitive load.
LDMOS can pulse nicely but the price tag is usually prohibitive. NPX
isn't very useful there either because they won't release SPICE model.
But PolyFet in Camarillo does.


Some day I'll have to see how the BFT92 and the BFG31 behave. But they
are intended as amplifiers so they won't are about saturation effects.
Trick to keep it out of saturation get old quickly when every pF and
every thenth of an inch count.
Yeah. I'd experiment with finding a 2N7002 complement. Now you've got
me interested.

I have the gear (pulsers, drivers, sampling scope) to experiment, if
you come up with some candidate parts.

This one could be a contender but it's Infineon and seems close to
unobtanium, in which case it would not do too much good:

http://www.infineon.com/dgdl/BSA223SP_Rev1.3.pdf?folderId=db3a304412b407950112b408e8c90004&fileId=db3a304412b407950112b42ae5834414


That looks really nice. Low capacitances, dynamite transfer curve,
nice pulsed current. Let's try some.


Tough to get though. Once I tried to cajole Infineon into samples of
another FET, explicitly telling them that I don't want them for free,
would also pay S&H and all that. They didn't afford me as much as a
courtesy call back. Ditched them.


This looks similar:

http://www.fairchildsemi.com/pf/FD/FDV304P.html

The Fairchild gumdrop parts are mostly excellent.

This looks promising, too:

http://www.vishay.com/docs/71411/tp0610k.pdf


I had looked at those a long time ago. The turn-off times are sluggish
yet they are driving them with gusto, for example Rgen 10V/10ohms for
the Vishay. They are smacking a whopping amp into the gate. 10pF and
150ohms load doesn't explain 35nsec turn-off and I bet their marketing
guys would razz them for being too conservative.

But one never knows with FETs.

Looking at the 2N7002 data sheets, one would never expect the sorts of
speeds I'm seeing. All you can do is get some parts and try them.

The ones that switch fast seem to have low Cg, modest Rds-on, and peak
drain current ratings under an amp (which doesn't stop me from going
above an amp.)

I recall Win Hill getting kilovolt edges in a couple of ns from fets
that, from the data sheets, shouldn't do that.

John

 

[krw@att.bizzzzzzzzzzzz]

On Sun, 21 Aug 2011 13:36:40 -0700, Joerg <invalid@invalid.invalid> wrote:

krw@att.bizzzzzzzzzzzz wrote:
On Sun, 21 Aug 2011 10:09:16 -0700, Joerg <invalid@invalid.invalid> wrote:

krw@att.bizzzzzzzzzzzz wrote:
On Sun, 21 Aug 2011 09:54:50 -0700, Joerg <invalid@invalid.invalid> wrote:

John Larkin wrote:
On Sun, 21 Aug 2011 07:14:47 -0700, Joerg <invalid@invalid.invalid
wrote:

Bill Sloman wrote:
On Aug 21, 12:13 pm, John Larkin
jjlar...@highNOTlandTHIStechnologyPART.com> wrote:
On Sat, 20 Aug 2011 18:27:00 -0700 (PDT),BillSloman



bill.slo...@ieee.org> wrote:
On Aug 21, 9:52 am, Joerg <inva...@invalid.invalid> wrote:
BillSlomanwrote:
On Aug 21, 12:17 am, Joerg <inva...@invalid.invalid> wrote:
BillSlomanwrote:
On 20/08/2011 3:34 AM, Joerg wrote:
BillSlomanwrote:
On 20/08/2011 2:46 AM, Joerg wrote:
snip

143 lines of pompous bilge, none on the subject of fet drivers.
John Larkin loves posting about fet drivers - he can make implausible
claims about how fast and cheap his are, and pose as the expert
electronic engineer that he wants to be accepted as.

John posts actual scope plots. So when he says that a transition happens
in under one nanosecond and proves it with a scope plot I don't know
what you mean by "implausible". Just because some people or datasheets
say it can't be done does not mean it can't be done. Also, he can show
the proof in revenue Dollars.

Now if someone had the opposite sex of the 2N7002 or a PNP with 15V+ and
no "saturation hold" that would be great. The BSS84 and it's siblings
ain't that hot.
Have you tried any pfets?

I did try the BSS84 a few years ago and the results were not enthusing.


I haven't played much with pfets as really fast switches. Right, a
complement to the 2N7002, push-pull against a 7002, with the same
12-cent gate drivers, would be interesting, and might solve your
problem. Just ignore the shoot-through maybe.

I'll put that on my slow-day experiment list. All I need now is a slow
day.

Why doesn't somebody make p-channel gaasfets? The world wonders.

A gold-doped BJT would be nice as well. However, I have not seen any
commercially successful gold or otherwise doped PNP transistors. And I
guess there ain't no market to write home about so I won't hold my breath.
Hell, just some cheap NPN arrays would be really nice. I can get cheap FET
arrays but not bipolar, anymore. ...

Voila:

http://www.fairchildsemi.com/ds/FF/FFB3904.pdf

At around 50c not very cheap. If it needs to be cheaper you've got to
shop in Asia.

$.50 wouldn't make me too happy (FETs arrays I'm using are $.10) but I could
make it work. HOWEVER, "NPN Multi-Chip General Purpose Amplifier" says it
all.


Well, you didn't say monolithic


http://www.kexin.com.cn/pdf/KC846S.pdf

No useful specs.


That is normal with many Asian suppliers, got to get used to it and test
a lot for yourself. You can sometimes obtain additional data from them
but sometimes you'd have to be married to the CEO's cousin's daughter or
something like that.

"Test a lot?" What does that tell me? It went against my grain to use a
green LED (GaN) at 3.3V. "Because one works..." Well, at least if the next
lot doesn't work we'll know.

http://www.rohm.com/products/discrete/transistor/complex/#03

"Very small package with two transistors."

All those aren't fast though.

Me? I don't need fast. ;-)


Lucky you

Almost all my stuff is RF nowadays.

We buy modules for all that stuff. I wouldn't attempt it with the shoestring
capital budget we're on (we're down a scope and it doesn't appear that they're
going to even replace it).

... Cheap current source/sinks would be nice, too.

That'll be a challenge, I don't think there's a market for those.

Evidently there isn't a market for transistor arrays, either. Perhaps not
even a 2N7002 with a sex change. ;-)


No, there sure ain't :-(

....not that I have anything for the '7002, mind you.

 

[Joerg]

krw@att.bizzzzzzzzzzzz wrote:

On Sun, 21 Aug 2011 13:36:40 -0700, Joerg <invalid@invalid.invalid> wrote:

krw@att.bizzzzzzzzzzzz wrote:
On Sun, 21 Aug 2011 10:09:16 -0700, Joerg <invalid@invalid.invalid> wrote:

[...]

http://www.kexin.com.cn/pdf/KC846S.pdf
No useful specs.

That is normal with many Asian suppliers, got to get used to it and test
a lot for yourself. You can sometimes obtain additional data from them
but sometimes you'd have to be married to the CEO's cousin's daughter or
something like that.

"Test a lot?" What does that tell me? It went against my grain to use a
green LED (GaN) at 3.3V. "Because one works..." Well, at least if the next
lot doesn't work we'll know.

I didn't mean a lot as in production lot, but as in "a lot of testing"

Meaning some of the properties have to be measure. In some designs that
is the only way to succeed because there are no parts that can
"formally" do what you need them to do.

http://www.rohm.com/products/discrete/transistor/complex/#03
"Very small package with two transistors."

All those aren't fast though.
Me? I don't need fast. ;-)

Lucky you

Almost all my stuff is RF nowadays.

We buy modules for all that stuff. I wouldn't attempt it with the shoestring
capital budget we're on (we're down a scope and it doesn't appear that they're
going to even replace it).

A scope? That's scary. I hope that doesn't mean any bad news. You can
nowadays get a lotta scope for $1-2k.

[...]

--
Regards, Joerg

http://www.analogconsultants.com/

 

[Joerg]

John Larkin wrote:

On Sun, 21 Aug 2011 14:21:02 -0700, Joerg <invalid@invalid.invalid
wrote:

John Larkin wrote:
On Sun, 21 Aug 2011 13:26:08 -0700, Joerg <invalid@invalid.invalid
wrote:

John Larkin wrote:
On Sun, 21 Aug 2011 09:54:50 -0700, Joerg <invalid@invalid.invalid
wrote:

John Larkin wrote:
On Sun, 21 Aug 2011 07:14:47 -0700, Joerg <invalid@invalid.invalid
wrote:

Bill Sloman wrote:
On Aug 21, 12:13 pm, John Larkin
jjlar...@highNOTlandTHIStechnologyPART.com> wrote:
On Sat, 20 Aug 2011 18:27:00 -0700 (PDT),BillSloman



bill.slo...@ieee.org> wrote:
On Aug 21, 9:52 am, Joerg <inva...@invalid.invalid> wrote:
BillSlomanwrote:
On Aug 21, 12:17 am, Joerg <inva...@invalid.invalid> wrote:
BillSlomanwrote:
On 20/08/2011 3:34 AM, Joerg wrote:
BillSlomanwrote:
On 20/08/2011 2:46 AM, Joerg wrote:
snip

143 lines of pompous bilge, none on the subject of fet drivers.
John Larkin loves posting about fet drivers - he can make implausible
claims about how fast and cheap his are, and pose as the expert
electronic engineer that he wants to be accepted as.

John posts actual scope plots. So when he says that a transition happens
in under one nanosecond and proves it with a scope plot I don't know
what you mean by "implausible". Just because some people or datasheets
say it can't be done does not mean it can't be done. Also, he can show
the proof in revenue Dollars.

Now if someone had the opposite sex of the 2N7002 or a PNP with 15V+ and
no "saturation hold" that would be great. The BSS84 and it's siblings
ain't that hot.
Have you tried any pfets?

I did try the BSS84 a few years ago and the results were not enthusing.
My mental model of a mosfet is an infinitely fast piece of silicon
with some capacitance and wire bonds. So the limit on fast switching
is mostly how hard you can drive the gate. Most fets can switch way
faster than the datasheets suggest if you slam them hard enough.
Unfortunately, my 600 ps mystery driver is only good for 6.5 volts
maybe, which that will drive a 2N7002 to an amp or so, fast, but pfets
usually need more drive to turn on hard. But I bet there's one out
there somewhere.

A little DC pre-bias, just below threshold, can buy another volt or
two of enhancement.

I haven't played much with pfets as really fast switches. Right, a
complement to the 2N7002, push-pull against a 7002, with the same
12-cent gate drivers, would be interesting, and might solve your
problem. Just ignore the shoot-through maybe.

I'll put that on my slow-day experiment list. All I need now is a slow
day.

Why doesn't somebody make p-channel gaasfets? The world wonders.

A gold-doped BJT would be nice as well. However, I have not seen any
commercially successful gold or otherwise doped PNP transistors. And I
guess there ain't no market to write home about so I won't hold my breath.
My experience using bjt's as this sort of fast switch has been
disappointing. I even tried some 45 GHz SiGe parts, and they switched
slow. Mosfets are much better, phemts are radically better.

RF transistors are also a problem if they can't get to an amp or more.
Usually you'll need that to swish charges around in a capacitive load.
LDMOS can pulse nicely but the price tag is usually prohibitive. NPX
isn't very useful there either because they won't release SPICE model.
But PolyFet in Camarillo does.


Some day I'll have to see how the BFT92 and the BFG31 behave. But they
are intended as amplifiers so they won't are about saturation effects.
Trick to keep it out of saturation get old quickly when every pF and
every thenth of an inch count.
Yeah. I'd experiment with finding a 2N7002 complement. Now you've got
me interested.

I have the gear (pulsers, drivers, sampling scope) to experiment, if
you come up with some candidate parts.

This one could be a contender but it's Infineon and seems close to
unobtanium, in which case it would not do too much good:

http://www.infineon.com/dgdl/BSA223SP_Rev1.3.pdf?folderId=db3a304412b407950112b408e8c90004&fileId=db3a304412b407950112b42ae5834414

That looks really nice. Low capacitances, dynamite transfer curve,
nice pulsed current. Let's try some.

Tough to get though. Once I tried to cajole Infineon into samples of
another FET, explicitly telling them that I don't want them for free,
would also pay S&H and all that. They didn't afford me as much as a
courtesy call back. Ditched them.


This looks similar:

http://www.fairchildsemi.com/pf/FD/FDV304P.html

The Fairchild gumdrop parts are mostly excellent.

This looks promising, too:

http://www.vishay.com/docs/71411/tp0610k.pdf

I had looked at those a long time ago. The turn-off times are sluggish
yet they are driving them with gusto, for example Rgen 10V/10ohms for
the Vishay. They are smacking a whopping amp into the gate. 10pF and
150ohms load doesn't explain 35nsec turn-off and I bet their marketing
guys would razz them for being too conservative.

But one never knows with FETs.

Looking at the 2N7002 data sheets, one would never expect the sorts of
speeds I'm seeing. All you can do is get some parts and try them.

I have seen very zippy transitions from a 2N7002 myself. Not like yours
but then again the fastest scope I have rolls off at 1GHz. Never seen a
P-channel come anywhere close to that. There used to be a BSS83
P-channel that was very good. But it seems to be gone. Believe it or
not, there is also a BSS83 N-channel from another European manufacturer.
Whoops ...

The ones that switch fast seem to have low Cg, modest Rds-on, and peak
drain current ratings under an amp (which doesn't stop me from going
above an amp.)

I recall Win Hill getting kilovolt edges in a couple of ns from fets
that, from the data sheets, shouldn't do that.

Was that shortly before this happened?

http://www.flickr.com/photos/36342938@N02/4808112575/

--
Regards, Joerg

http://www.analogconsultants.com/

 

[krw@att.bizzzzzzzzzzzz]

On Sun, 21 Aug 2011 15:27:05 -0700, Joerg <invalid@invalid.invalid> wrote:

krw@att.bizzzzzzzzzzzz wrote:
On Sun, 21 Aug 2011 13:36:40 -0700, Joerg <invalid@invalid.invalid> wrote:

krw@att.bizzzzzzzzzzzz wrote:
On Sun, 21 Aug 2011 10:09:16 -0700, Joerg <invalid@invalid.invalid> wrote:


[...]

http://www.kexin.com.cn/pdf/KC846S.pdf
No useful specs.

That is normal with many Asian suppliers, got to get used to it and test
a lot for yourself. You can sometimes obtain additional data from them
but sometimes you'd have to be married to the CEO's cousin's daughter or
something like that.

"Test a lot?" What does that tell me? It went against my grain to use a
green LED (GaN) at 3.3V. "Because one works..." Well, at least if the next
lot doesn't work we'll know.


I didn't mean a lot as in production lot, but as in "a lot of testing"

But that means you need a lot of lots! ;-)

Meaning some of the properties have to be measure. In some designs that
is the only way to succeed because there are no parts that can
"formally" do what you need them to do.

In my former life, we would specify what was needed. Sometimes it got
expensive but the groundrules for such designs was "worst case". Life is much
simpler now. ;-)

http://www.rohm.com/products/discrete/transistor/complex/#03
"Very small package with two transistors."

All those aren't fast though.
Me? I don't need fast. ;-)

Lucky you

Almost all my stuff is RF nowadays.

We buy modules for all that stuff. I wouldn't attempt it with the shoestring
capital budget we're on (we're down a scope and it doesn't appear that they're
going to even replace it).


A scope? That's scary. I hope that doesn't mean any bad news. You can
nowadays get a lotta scope for $1-2k.

I don't think it's permanent bad news. I think it's more prudence, given the
Obama recession. I really can't argue much. The scope we lost was a 300MHz
model. The remaining one is only 100MHz (there is another but it's usually
tied up with firmware stuff) and they seem to think that's good enough. Of
course, I'm holding out for a MSO3034. ;-)

I really need other stuff, too. I'm going up to Atlanta for compliance
testing on a design I know will fail ESD. It was never designed to pass (the
JTAG port is exposed on an external connector) but the requirements changed at
the last minute (well, the last week). I really need an ESD gun.

Maybe, after Obama gets the boot. ...or I retire again. ;-)

 

[krw@att.bizzzzzzzzzzzz]

On Sun, 21 Aug 2011 19:17:40 -0400, Jamie
<jamie_ka1lpa_not_valid_after_ka1lpa_@charter.net> wrote:

Joerg wrote:

krw@att.bizzzzzzzzzzzz wrote:

On Sun, 21 Aug 2011 13:36:40 -0700, Joerg <invalid@invalid.invalid> wrote:


krw@att.bizzzzzzzzzzzz wrote:

On Sun, 21 Aug 2011 10:09:16 -0700, Joerg <invalid@invalid.invalid> wrote:



[...]


http://www.kexin.com.cn/pdf/KC846S.pdf

No useful specs.


That is normal with many Asian suppliers, got to get used to it and test
a lot for yourself. You can sometimes obtain additional data from them
but sometimes you'd have to be married to the CEO's cousin's daughter or
something like that.

"Test a lot?" What does that tell me? It went against my grain to use a
green LED (GaN) at 3.3V. "Because one works..." Well, at least if the next
lot doesn't work we'll know.



I didn't mean a lot as in production lot, but as in "a lot of testing"

Meaning some of the properties have to be measure. In some designs that
is the only way to succeed because there are no parts that can
"formally" do what you need them to do.



http://www.rohm.com/products/discrete/transistor/complex/#03

"Very small package with two transistors."


All those aren't fast though.

Me? I don't need fast. ;-)


Lucky you

Almost all my stuff is RF nowadays.

We buy modules for all that stuff. I wouldn't attempt it with the shoestring
capital budget we're on (we're down a scope and it doesn't appear that they're
going to even replace it).



A scope? That's scary. I hope that doesn't mean any bad news. You can
nowadays get a lotta scope for $1-2k.

[...]

Starting at a Ghz bw ?

We don't need anything like that. Who looks at the carrier? ;-)

 

[Phil Hobbs]

On 08/21/2011 07:17 PM, Jamie wrote:

Joerg wrote:

krw@att.bizzzzzzzzzzzz wrote:

On Sun, 21 Aug 2011 13:36:40 -0700, Joerg <invalid@invalid.invalid
wrote:


krw@att.bizzzzzzzzzzzz wrote:

On Sun, 21 Aug 2011 10:09:16 -0700, Joerg <invalid@invalid.invalid
wrote:



[...]


http://www.kexin.com.cn/pdf/KC846S.pdf

No useful specs.


That is normal with many Asian suppliers, got to get used to it and
test
a lot for yourself. You can sometimes obtain additional data from them
but sometimes you'd have to be married to the CEO's cousin's
daughter or
something like that.

"Test a lot?" What does that tell me? It went against my grain to use a
green LED (GaN) at 3.3V. "Because one works..." Well, at least if the
next
lot doesn't work we'll know.


I didn't mean a lot as in production lot, but as in "a lot of testing"


Meaning some of the properties have to be measure. In some designs that
is the only way to succeed because there are no parts that can
"formally" do what you need them to do.



http://www.rohm.com/products/discrete/transistor/complex/#03

"Very small package with two transistors."


All those aren't fast though.

Me? I don't need fast. ;-)


Lucky you

Almost all my stuff is RF nowadays.

We buy modules for all that stuff. I wouldn't attempt it with the
shoestring
capital budget we're on (we're down a scope and it doesn't appear
that they're
going to even replace it).



A scope? That's scary. I hope that doesn't mean any bad news. You can
nowadays get a lotta scope for $1-2k.

[...]

Starting at a Ghz bw ?

Jamie

You can get a Tek 11801C and an SD14 plugin (3 GHz, with built-in FET
probes) for $1k total on eBay. Of course that's a stroboscopic sampling
scope and not a single-shot digitizing one, but it's beautiful for
repetitive signals.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203
Briarcliff Manor NY 10510
845-480-2058

hobbs at electrooptical dot net
http://electrooptical.net

 

[Jamie]

John Larkin wrote:

On Sun, 21 Aug 2011 14:21:02 -0700, Joerg <invalid@invalid.invalid
wrote:


John Larkin wrote:

On Sun, 21 Aug 2011 13:26:08 -0700, Joerg <invalid@invalid.invalid
wrote:


John Larkin wrote:

On Sun, 21 Aug 2011 09:54:50 -0700, Joerg <invalid@invalid.invalid
wrote:


John Larkin wrote:

On Sun, 21 Aug 2011 07:14:47 -0700, Joerg <invalid@invalid.invalid
wrote:


Bill Sloman wrote:

On Aug 21, 12:13 pm, John Larkin
jjlar...@highNOTlandTHIStechnologyPART.com> wrote:

On Sat, 20 Aug 2011 18:27:00 -0700 (PDT),BillSloman



bill.slo...@ieee.org> wrote:

On Aug 21, 9:52 am, Joerg <inva...@invalid.invalid> wrote:

BillSlomanwrote:

On Aug 21, 12:17 am, Joerg <inva...@invalid.invalid> wrote:

BillSlomanwrote:

On 20/08/2011 3:34 AM, Joerg wrote:

BillSlomanwrote:

On 20/08/2011 2:46 AM, Joerg wrote:

snip

143 lines of pompous bilge, none on the subject of fet drivers.

John Larkin loves posting about fet drivers - he can make implausible
claims about how fast and cheap his are, and pose as the expert
electronic engineer that he wants to be accepted as.


John posts actual scope plots. So when he says that a transition happens
in under one nanosecond and proves it with a scope plot I don't know
what you mean by "implausible". Just because some people or datasheets
say it can't be done does not mean it can't be done. Also, he can show
the proof in revenue Dollars.

Now if someone had the opposite sex of the 2N7002 or a PNP with 15V+ and
no "saturation hold" that would be great. The BSS84 and it's siblings
ain't that hot.

Have you tried any pfets?


I did try the BSS84 a few years ago and the results were not enthusing.

My mental model of a mosfet is an infinitely fast piece of silicon
with some capacitance and wire bonds. So the limit on fast switching
is mostly how hard you can drive the gate. Most fets can switch way
faster than the datasheets suggest if you slam them hard enough.
Unfortunately, my 600 ps mystery driver is only good for 6.5 volts
maybe, which that will drive a 2N7002 to an amp or so, fast, but pfets
usually need more drive to turn on hard. But I bet there's one out
there somewhere.

A little DC pre-bias, just below threshold, can buy another volt or
two of enhancement.


I haven't played much with pfets as really fast switches. Right, a
complement to the 2N7002, push-pull against a 7002, with the same
12-cent gate drivers, would be interesting, and might solve your
problem. Just ignore the shoot-through maybe.

I'll put that on my slow-day experiment list. All I need now is a slow
day.

Why doesn't somebody make p-channel gaasfets? The world wonders.


A gold-doped BJT would be nice as well. However, I have not seen any
commercially successful gold or otherwise doped PNP transistors. And I
guess there ain't no market to write home about so I won't hold my breath.

My experience using bjt's as this sort of fast switch has been
disappointing. I even tried some 45 GHz SiGe parts, and they switched
slow. Mosfets are much better, phemts are radically better.


RF transistors are also a problem if they can't get to an amp or more.
Usually you'll need that to swish charges around in a capacitive load.
LDMOS can pulse nicely but the price tag is usually prohibitive. NPX
isn't very useful there either because they won't release SPICE model.
But PolyFet in Camarillo does.



Some day I'll have to see how the BFT92 and the BFG31 behave. But they
are intended as amplifiers so they won't are about saturation effects.
Trick to keep it out of saturation get old quickly when every pF and
every thenth of an inch count.

Yeah. I'd experiment with finding a 2N7002 complement. Now you've got
me interested.

I have the gear (pulsers, drivers, sampling scope) to experiment, if
you come up with some candidate parts.


This one could be a contender but it's Infineon and seems close to
unobtanium, in which case it would not do too much good:

http://www.infineon.com/dgdl/BSA223SP_Rev1.3.pdf?folderId=db3a304412b407950112b408e8c90004&fileId=db3a304412b407950112b42ae5834414


That looks really nice. Low capacitances, dynamite transfer curve,
nice pulsed current. Let's try some.


Tough to get though. Once I tried to cajole Infineon into samples of
another FET, explicitly telling them that I don't want them for free,
would also pay S&H and all that. They didn't afford me as much as a
courtesy call back. Ditched them.



This looks similar:

http://www.fairchildsemi.com/pf/FD/FDV304P.html

The Fairchild gumdrop parts are mostly excellent.

This looks promising, too:

http://www.vishay.com/docs/71411/tp0610k.pdf


I had looked at those a long time ago. The turn-off times are sluggish
yet they are driving them with gusto, for example Rgen 10V/10ohms for
the Vishay. They are smacking a whopping amp into the gate. 10pF and
150ohms load doesn't explain 35nsec turn-off and I bet their marketing
guys would razz them for being too conservative.

But one never knows with FETs.


Looking at the 2N7002 data sheets, one would never expect the sorts of
speeds I'm seeing. All you can do is get some parts and try them.

The ones that switch fast seem to have low Cg, modest Rds-on, and peak
drain current ratings under an amp (which doesn't stop me from going
above an amp.)

I recall Win Hill getting kilovolt edges in a couple of ns from fets
that, from the data sheets, shouldn't do that.

John


Good thing you have some Cg there to help clamp those kilovolt

edges...Assuming they were short to begin with.

Jamie

 

[Jamie]

Joerg wrote:

krw@att.bizzzzzzzzzzzz wrote:

On Sun, 21 Aug 2011 13:36:40 -0700, Joerg <invalid@invalid.invalid> wrote:


krw@att.bizzzzzzzzzzzz wrote:

On Sun, 21 Aug 2011 10:09:16 -0700, Joerg <invalid@invalid.invalid> wrote:



[...]


http://www.kexin.com.cn/pdf/KC846S.pdf

No useful specs.


That is normal with many Asian suppliers, got to get used to it and test
a lot for yourself. You can sometimes obtain additional data from them
but sometimes you'd have to be married to the CEO's cousin's daughter or
something like that.

"Test a lot?" What does that tell me? It went against my grain to use a
green LED (GaN) at 3.3V. "Because one works..." Well, at least if the next
lot doesn't work we'll know.



I didn't mean a lot as in production lot, but as in "a lot of testing"

Meaning some of the properties have to be measure. In some designs that
is the only way to succeed because there are no parts that can
"formally" do what you need them to do.



http://www.rohm.com/products/discrete/transistor/complex/#03

"Very small package with two transistors."


All those aren't fast though.

Me? I don't need fast. ;-)


Lucky you

Almost all my stuff is RF nowadays.

We buy modules for all that stuff. I wouldn't attempt it with the shoestring
capital budget we're on (we're down a scope and it doesn't appear that they're
going to even replace it).



A scope? That's scary. I hope that doesn't mean any bad news. You can
nowadays get a lotta scope for $1-2k.

[...]

Starting at a Ghz bw ?

Jamie

 

[Joerg]

Phil Hobbs wrote:

On 08/21/2011 07:17 PM, Jamie wrote:
Joerg wrote:

krw@att.bizzzzzzzzzzzz wrote:

On Sun, 21 Aug 2011 13:36:40 -0700, Joerg <invalid@invalid.invalid
wrote:


krw@att.bizzzzzzzzzzzz wrote:

On Sun, 21 Aug 2011 10:09:16 -0700, Joerg <invalid@invalid.invalid
wrote:

[...]

http://www.rohm.com/products/discrete/transistor/complex/#03

"Very small package with two transistors."


All those aren't fast though.

Me? I don't need fast. ;-)


Lucky you

Almost all my stuff is RF nowadays.

We buy modules for all that stuff. I wouldn't attempt it with the
shoestring
capital budget we're on (we're down a scope and it doesn't appear
that they're
going to even replace it).



A scope? That's scary. I hope that doesn't mean any bad news. You can
nowadays get a lotta scope for $1-2k.

[...]

Starting at a Ghz bw ?

Jamie




You can get a Tek 11801C and an SD14 plugin (3 GHz, with built-in FET
probes) for $1k total on eBay. Of course that's a stroboscopic sampling
scope and not a single-shot digitizing one, but it's beautiful for
repetitive signals.

Yup. It pays to not always insist on the latest, best, tallest, most
expensive gold-plated gear. I do have a GHz BW scope here but it is
repetitive sampling, HP from the early 90's. Good enough for work and I
do a very wide variety of projects, including pulse stuff.

I never felt the urge to own a 5-10 GSPS super-expensive scope. Sure, at
times it would have been nice but I was always able to make a signal
reptitive enough that the HP displayed it.

One does not need a Lamborghini to haul some stuff from town

--
Regards, Joerg

http://www.analogconsultants.com/

 

[John Larkin]

On Sun, 21 Aug 2011 17:10:08 -0500, "krw@att.bizzzzzzzzzzzz"
<krw@att.bizzzzzzzzzzzz> wrote:

On Sun, 21 Aug 2011 13:36:40 -0700, Joerg <invalid@invalid.invalid> wrote:

krw@att.bizzzzzzzzzzzz wrote:
On Sun, 21 Aug 2011 10:09:16 -0700, Joerg <invalid@invalid.invalid> wrote:

krw@att.bizzzzzzzzzzzz wrote:
On Sun, 21 Aug 2011 09:54:50 -0700, Joerg <invalid@invalid.invalid> wrote:

John Larkin wrote:
On Sun, 21 Aug 2011 07:14:47 -0700, Joerg <invalid@invalid.invalid
wrote:

Bill Sloman wrote:
On Aug 21, 12:13 pm, John Larkin
jjlar...@highNOTlandTHIStechnologyPART.com> wrote:
On Sat, 20 Aug 2011 18:27:00 -0700 (PDT),BillSloman



bill.slo...@ieee.org> wrote:
On Aug 21, 9:52 am, Joerg <inva...@invalid.invalid> wrote:
BillSlomanwrote:
On Aug 21, 12:17 am, Joerg <inva...@invalid.invalid> wrote:
BillSlomanwrote:
On 20/08/2011 3:34 AM, Joerg wrote:
BillSlomanwrote:
On 20/08/2011 2:46 AM, Joerg wrote:
snip

143 lines of pompous bilge, none on the subject of fet drivers.
John Larkin loves posting about fet drivers - he can make implausible
claims about how fast and cheap his are, and pose as the expert
electronic engineer that he wants to be accepted as.

John posts actual scope plots. So when he says that a transition happens
in under one nanosecond and proves it with a scope plot I don't know
what you mean by "implausible". Just because some people or datasheets
say it can't be done does not mean it can't be done. Also, he can show
the proof in revenue Dollars.

Now if someone had the opposite sex of the 2N7002 or a PNP with 15V+ and
no "saturation hold" that would be great. The BSS84 and it's siblings
ain't that hot.
Have you tried any pfets?

I did try the BSS84 a few years ago and the results were not enthusing.


I haven't played much with pfets as really fast switches. Right, a
complement to the 2N7002, push-pull against a 7002, with the same
12-cent gate drivers, would be interesting, and might solve your
problem. Just ignore the shoot-through maybe.

I'll put that on my slow-day experiment list. All I need now is a slow
day.

Why doesn't somebody make p-channel gaasfets? The world wonders.

A gold-doped BJT would be nice as well. However, I have not seen any
commercially successful gold or otherwise doped PNP transistors. And I
guess there ain't no market to write home about so I won't hold my breath.
Hell, just some cheap NPN arrays would be really nice. I can get cheap FET
arrays but not bipolar, anymore. ...

Voila:

http://www.fairchildsemi.com/ds/FF/FFB3904.pdf

At around 50c not very cheap. If it needs to be cheaper you've got to
shop in Asia.

$.50 wouldn't make me too happy (FETs arrays I'm using are $.10) but I could
make it work. HOWEVER, "NPN Multi-Chip General Purpose Amplifier" says it
all.


Well, you didn't say monolithic


http://www.kexin.com.cn/pdf/KC846S.pdf

No useful specs.


That is normal with many Asian suppliers, got to get used to it and test
a lot for yourself. You can sometimes obtain additional data from them
but sometimes you'd have to be married to the CEO's cousin's daughter or
something like that.

"Test a lot?" What does that tell me? It went against my grain to use a
green LED (GaN) at 3.3V. "Because one works..." Well, at least if the next
lot doesn't work we'll know.

http://www.rohm.com/products/discrete/transistor/complex/#03

"Very small package with two transistors."

All those aren't fast though.

Me? I don't need fast. ;-)


Lucky you

Almost all my stuff is RF nowadays.

We buy modules for all that stuff. I wouldn't attempt it with the shoestring
capital budget we're on (we're down a scope and it doesn't appear that they're
going to even replace it).

Makes no sense. Equipment is cheap, and people are expensive.

John

 

[John Larkin]

On Sun, 21 Aug 2011 15:27:05 -0700, Joerg <invalid@invalid.invalid>
wrote:

krw@att.bizzzzzzzzzzzz wrote:
On Sun, 21 Aug 2011 13:36:40 -0700, Joerg <invalid@invalid.invalid> wrote:

krw@att.bizzzzzzzzzzzz wrote:
On Sun, 21 Aug 2011 10:09:16 -0700, Joerg <invalid@invalid.invalid> wrote:


[...]

http://www.kexin.com.cn/pdf/KC846S.pdf
No useful specs.

That is normal with many Asian suppliers, got to get used to it and test
a lot for yourself. You can sometimes obtain additional data from them
but sometimes you'd have to be married to the CEO's cousin's daughter or
something like that.

"Test a lot?" What does that tell me? It went against my grain to use a
green LED (GaN) at 3.3V. "Because one works..." Well, at least if the next
lot doesn't work we'll know.


I didn't mean a lot as in production lot, but as in "a lot of testing"

Meaning some of the properties have to be measure. In some designs that
is the only way to succeed because there are no parts that can
"formally" do what you need them to do.


http://www.rohm.com/products/discrete/transistor/complex/#03
"Very small package with two transistors."

All those aren't fast though.
Me? I don't need fast. ;-)

Lucky you

Almost all my stuff is RF nowadays.

We buy modules for all that stuff. I wouldn't attempt it with the shoestring
capital budget we're on (we're down a scope and it doesn't appear that they're
going to even replace it).


A scope? That's scary. I hope that doesn't mean any bad news. You can
nowadays get a lotta scope for $1-2k.

Or a Rigol, for $350.

John

 

[John Larkin]

On Sun, 21 Aug 2011 19:14:07 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

On 08/21/2011 07:17 PM, Jamie wrote:
Joerg wrote:

krw@att.bizzzzzzzzzzzz wrote:

On Sun, 21 Aug 2011 13:36:40 -0700, Joerg <invalid@invalid.invalid
wrote:


krw@att.bizzzzzzzzzzzz wrote:

On Sun, 21 Aug 2011 10:09:16 -0700, Joerg <invalid@invalid.invalid
wrote:



[...]


http://www.kexin.com.cn/pdf/KC846S.pdf

No useful specs.


That is normal with many Asian suppliers, got to get used to it and
test
a lot for yourself. You can sometimes obtain additional data from them
but sometimes you'd have to be married to the CEO's cousin's
daughter or
something like that.

"Test a lot?" What does that tell me? It went against my grain to use a
green LED (GaN) at 3.3V. "Because one works..." Well, at least if the
next
lot doesn't work we'll know.


I didn't mean a lot as in production lot, but as in "a lot of testing"


Meaning some of the properties have to be measure. In some designs that
is the only way to succeed because there are no parts that can
"formally" do what you need them to do.



http://www.rohm.com/products/discrete/transistor/complex/#03

"Very small package with two transistors."


All those aren't fast though.

Me? I don't need fast. ;-)


Lucky you

Almost all my stuff is RF nowadays.

We buy modules for all that stuff. I wouldn't attempt it with the
shoestring
capital budget we're on (we're down a scope and it doesn't appear
that they're
going to even replace it).



A scope? That's scary. I hope that doesn't mean any bad news. You can
nowadays get a lotta scope for $1-2k.

[...]

Starting at a Ghz bw ?

Jamie




You can get a Tek 11801C and an SD14 plugin (3 GHz, with built-in FET
probes) for $1k total on eBay. Of course that's a stroboscopic sampling
scope and not a single-shot digitizing one, but it's beautiful for
repetitive signals.

Cheers

Phil Hobbs

Invest another $500 or so and get 20 GHz TDR.

John

 

[krw@att.bizzzzzzzzzzzz]

On Sun, 21 Aug 2011 17:21:25 -0700, John Larkin
<jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Sun, 21 Aug 2011 17:10:08 -0500, "krw@att.bizzzzzzzzzzzz"
krw@att.bizzzzzzzzzzzz> wrote:

On Sun, 21 Aug 2011 13:36:40 -0700, Joerg <invalid@invalid.invalid> wrote:

krw@att.bizzzzzzzzzzzz wrote:
On Sun, 21 Aug 2011 10:09:16 -0700, Joerg <invalid@invalid.invalid> wrote:

krw@att.bizzzzzzzzzzzz wrote:
On Sun, 21 Aug 2011 09:54:50 -0700, Joerg <invalid@invalid.invalid> wrote:

John Larkin wrote:
On Sun, 21 Aug 2011 07:14:47 -0700, Joerg <invalid@invalid.invalid
wrote:

Bill Sloman wrote:
On Aug 21, 12:13 pm, John Larkin
jjlar...@highNOTlandTHIStechnologyPART.com> wrote:
On Sat, 20 Aug 2011 18:27:00 -0700 (PDT),BillSloman



bill.slo...@ieee.org> wrote:
On Aug 21, 9:52 am, Joerg <inva...@invalid.invalid> wrote:
BillSlomanwrote:
On Aug 21, 12:17 am, Joerg <inva...@invalid.invalid> wrote:
BillSlomanwrote:
On 20/08/2011 3:34 AM, Joerg wrote:
BillSlomanwrote:
On 20/08/2011 2:46 AM, Joerg wrote:
snip

143 lines of pompous bilge, none on the subject of fet drivers.
John Larkin loves posting about fet drivers - he can make implausible
claims about how fast and cheap his are, and pose as the expert
electronic engineer that he wants to be accepted as.

John posts actual scope plots. So when he says that a transition happens
in under one nanosecond and proves it with a scope plot I don't know
what you mean by "implausible". Just because some people or datasheets
say it can't be done does not mean it can't be done. Also, he can show
the proof in revenue Dollars.

Now if someone had the opposite sex of the 2N7002 or a PNP with 15V+ and
no "saturation hold" that would be great. The BSS84 and it's siblings
ain't that hot.
Have you tried any pfets?

I did try the BSS84 a few years ago and the results were not enthusing.


I haven't played much with pfets as really fast switches. Right, a
complement to the 2N7002, push-pull against a 7002, with the same
12-cent gate drivers, would be interesting, and might solve your
problem. Just ignore the shoot-through maybe.

I'll put that on my slow-day experiment list. All I need now is a slow
day.

Why doesn't somebody make p-channel gaasfets? The world wonders.

A gold-doped BJT would be nice as well. However, I have not seen any
commercially successful gold or otherwise doped PNP transistors. And I
guess there ain't no market to write home about so I won't hold my breath.
Hell, just some cheap NPN arrays would be really nice. I can get cheap FET
arrays but not bipolar, anymore. ...

Voila:

http://www.fairchildsemi.com/ds/FF/FFB3904.pdf

At around 50c not very cheap. If it needs to be cheaper you've got to
shop in Asia.

$.50 wouldn't make me too happy (FETs arrays I'm using are $.10) but I could
make it work. HOWEVER, "NPN Multi-Chip General Purpose Amplifier" says it
all.


Well, you didn't say monolithic


http://www.kexin.com.cn/pdf/KC846S.pdf

No useful specs.


That is normal with many Asian suppliers, got to get used to it and test
a lot for yourself. You can sometimes obtain additional data from them
but sometimes you'd have to be married to the CEO's cousin's daughter or
something like that.

"Test a lot?" What does that tell me? It went against my grain to use a
green LED (GaN) at 3.3V. "Because one works..." Well, at least if the next
lot doesn't work we'll know.

http://www.rohm.com/products/discrete/transistor/complex/#03

"Very small package with two transistors."

All those aren't fast though.

Me? I don't need fast. ;-)


Lucky you

Almost all my stuff is RF nowadays.

We buy modules for all that stuff. I wouldn't attempt it with the shoestring
capital budget we're on (we're down a scope and it doesn't appear that they're
going to even replace it).

Makes no sense. Equipment is cheap, and people are expensive.

You can't lay off a scope. ;-)

 

[Joerg]

John Larkin wrote:

On Sun, 21 Aug 2011 15:27:05 -0700, Joerg <invalid@invalid.invalid
wrote:

krw@att.bizzzzzzzzzzzz wrote:
On Sun, 21 Aug 2011 13:36:40 -0700, Joerg <invalid@invalid.invalid> wrote:

krw@att.bizzzzzzzzzzzz wrote:
On Sun, 21 Aug 2011 10:09:16 -0700, Joerg <invalid@invalid.invalid> wrote:

[...]

http://www.kexin.com.cn/pdf/KC846S.pdf
No useful specs.

That is normal with many Asian suppliers, got to get used to it and test
a lot for yourself. You can sometimes obtain additional data from them
but sometimes you'd have to be married to the CEO's cousin's daughter or
something like that.
"Test a lot?" What does that tell me? It went against my grain to use a
green LED (GaN) at 3.3V. "Because one works..." Well, at least if the next
lot doesn't work we'll know.

I didn't mean a lot as in production lot, but as in "a lot of testing"

Meaning some of the properties have to be measure. In some designs that
is the only way to succeed because there are no parts that can
"formally" do what you need them to do.


http://www.rohm.com/products/discrete/transistor/complex/#03
"Very small package with two transistors."

All those aren't fast though.
Me? I don't need fast. ;-)

Lucky you

Almost all my stuff is RF nowadays.
We buy modules for all that stuff. I wouldn't attempt it with the shoestring
capital budget we're on (we're down a scope and it doesn't appear that they're
going to even replace it).

A scope? That's scary. I hope that doesn't mean any bad news. You can
nowadays get a lotta scope for $1-2k.

Or a Rigol, for $350.

That may be bit skimpy. I often find myself debugging and finding things
on the digital side, where everyone was 110% sure it couldn't possibly
be a software issue. That's mostly SPI and other serial buses where 2ch
won't really work. Also, I found that 100MHz BW ain't enough. Glitches
in systems where a 8051 screams along at 80MHz or so are specterally
above that.

But IIRC there was that Dave Jones tuning

--
Regards, Joerg

http://www.analogconsultants.com/

 

[Joerg]

krw@att.bizzzzzzzzzzzz wrote:

On Sun, 21 Aug 2011 17:21:25 -0700, John Larkin
jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Sun, 21 Aug 2011 17:10:08 -0500, "krw@att.bizzzzzzzzzzzz"
krw@att.bizzzzzzzzzzzz> wrote:

On Sun, 21 Aug 2011 13:36:40 -0700, Joerg <invalid@invalid.invalid> wrote:

krw@att.bizzzzzzzzzzzz wrote:
On Sun, 21 Aug 2011 10:09:16 -0700, Joerg <invalid@invalid.invalid> wrote:

[...]

http://www.rohm.com/products/discrete/transistor/complex/#03
"Very small package with two transistors."

All those aren't fast though.
Me? I don't need fast. ;-)

Lucky you

Almost all my stuff is RF nowadays.
We buy modules for all that stuff. I wouldn't attempt it with the shoestring
capital budget we're on (we're down a scope and it doesn't appear that they're
going to even replace it).
Makes no sense. Equipment is cheap, and people are expensive.

You can't lay off a scope. ;-)

Sure you can. On Ebay. And the upside is the scope, once sold, won't
turn around and serve you with a lawsuit about half a gazillion hours of
overtime that the scope feels it should have been paid for.

--
Regards, Joerg

http://www.analogconsultants.com/

 

[krw@att.bizzzzzzzzzzzz]

On Sun, 21 Aug 2011 17:37:08 -0700, Joerg <invalid@invalid.invalid> wrote:

John Larkin wrote:
On Sun, 21 Aug 2011 15:27:05 -0700, Joerg <invalid@invalid.invalid
wrote:

krw@att.bizzzzzzzzzzzz wrote:
On Sun, 21 Aug 2011 13:36:40 -0700, Joerg <invalid@invalid.invalid> wrote:

krw@att.bizzzzzzzzzzzz wrote:
On Sun, 21 Aug 2011 10:09:16 -0700, Joerg <invalid@invalid.invalid> wrote:

[...]

http://www.kexin.com.cn/pdf/KC846S.pdf
No useful specs.

That is normal with many Asian suppliers, got to get used to it and test
a lot for yourself. You can sometimes obtain additional data from them
but sometimes you'd have to be married to the CEO's cousin's daughter or
something like that.
"Test a lot?" What does that tell me? It went against my grain to use a
green LED (GaN) at 3.3V. "Because one works..." Well, at least if the next
lot doesn't work we'll know.

I didn't mean a lot as in production lot, but as in "a lot of testing"

Meaning some of the properties have to be measure. In some designs that
is the only way to succeed because there are no parts that can
"formally" do what you need them to do.


http://www.rohm.com/products/discrete/transistor/complex/#03
"Very small package with two transistors."

All those aren't fast though.
Me? I don't need fast. ;-)

Lucky you

Almost all my stuff is RF nowadays.
We buy modules for all that stuff. I wouldn't attempt it with the shoestring
capital budget we're on (we're down a scope and it doesn't appear that they're
going to even replace it).

A scope? That's scary. I hope that doesn't mean any bad news. You can
nowadays get a lotta scope for $1-2k.

Or a Rigol, for $350.


That may be bit skimpy. I often find myself debugging and finding things
on the digital side, where everyone was 110% sure it couldn't possibly
be a software issue. That's mostly SPI and other serial buses where 2ch
won't really work. Also, I found that 100MHz BW ain't enough. Glitches
in systems where a 8051 screams along at 80MHz or so are specterally
above that.

That's been my point when management (and the other engineer) suggest that
100MHz is enough. Gotta have at least 3X, 5x is better.

It's just so weird to see the words "8051 screams along" run together like
that. ;-)

But IIRC there was that Dave Jones tuning

For SPI and particularly I2C, I'd really like to see both the analog and data
domains, particularly for triggering.

 

[Vladimir Vassilevsky]

Joerg wrote:

krw@att.bizzzzzzzzzzzz wrote:


You can't lay off a scope. ;-)



Sure you can. On Ebay. And the upside is the scope, once sold, won't
turn around and serve you with a lawsuit about half a gazillion hours of
overtime that the scope feels it should have been paid for.

Well there are the socialist kind of trade unionized scopes that don't
work; so they won't sell from Ebay. However, they ask for expensive
genuine parts and an authorized shop for repair

VLV