Other new kinds of parts--mezzanine packaging for fun and profit...

P

Phil Hobbs

Guest
Apropos of the \'5 GHz beads\' and \'new kinds of parts\' threads, I\'m
considering doing new small parts of our own, based on magic GHz
transistors, but house-trained enough to use in normal boards (with op
amps and so on).

Specifically, it looks like a win to make mezzanine-packaged SiGe BJTs
with beads, on little crenelated boards. They\'d be something like
TMB-package LEDs, but without the plastic overmold.

I was playing around in LTspice, using the Infineon and Murata models
for the BFP640 and the GHz beads that include the package parasitics, it
seems that the main obstacle to getting good stability at high collector
current is the pad capacitance right at the transistor.

With pad capacitances below 0.1 pF or so, the beads make them nice and
stable at all relevant collector currents, but at 0.5 pF, you have to
stay below 3 mA or thereabouts. Series inductance doesn\'t matter much
for stability, or for performance at < 1 GHz, where we mostly live.

So, we\'re getting some boards made with a whole bunch of 5-mm squares,
each containing a fast transistor and two or three of these magic beads.

Dunno yet how we\'ll get them apart--probably routing and snap tabs, or
else V-grooving. And then there\'s the question of how to use them in
production. We\'ll see!

Cheers

Phil Hobbs


--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com
 
On Fri, 23 Jun 2023 17:05:50 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

Apropos of the \'5 GHz beads\' and \'new kinds of parts\' threads, I\'m
considering doing new small parts of our own, based on magic GHz
transistors, but house-trained enough to use in normal boards (with op
amps and so on).

Specifically, it looks like a win to make mezzanine-packaged SiGe BJTs
with beads, on little crenelated boards. They\'d be something like
TMB-package LEDs, but without the plastic overmold.

I was playing around in LTspice, using the Infineon and Murata models
for the BFP640 and the GHz beads that include the package parasitics, it
seems that the main obstacle to getting good stability at high collector
current is the pad capacitance right at the transistor.

With pad capacitances below 0.1 pF or so, the beads make them nice and
stable at all relevant collector currents, but at 0.5 pF, you have to
stay below 3 mA or thereabouts. Series inductance doesn\'t matter much
for stability, or for performance at < 1 GHz, where we mostly live.

So, we\'re getting some boards made with a whole bunch of 5-mm squares,
each containing a fast transistor and two or three of these magic beads.

Dunno yet how we\'ll get them apart--probably routing and snap tabs, or
else V-grooving. And then there\'s the question of how to use them in
production. We\'ll see!

Cheers

Phil Hobbs
Click to expand...

We make a few little boards with the EPC GaN fets on-board, so we can
glob-top them and replace the baby board if anything fails.

https://www.dropbox.com/sh/1f9zm7ys14rx0oe/AAA2n5S9-RjJa8PS3K2aUU8aa?dl=0
 
On 2023-06-23 19:37, John Larkin wrote:
On Fri, 23 Jun 2023 17:05:50 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

Apropos of the \'5 GHz beads\' and \'new kinds of parts\' threads, I\'m
considering doing new small parts of our own, based on magic GHz
transistors, but house-trained enough to use in normal boards (with op
amps and so on).

Specifically, it looks like a win to make mezzanine-packaged SiGe BJTs
with beads, on little crenelated boards. They\'d be something like
TMB-package LEDs, but without the plastic overmold.

I was playing around in LTspice, using the Infineon and Murata models
for the BFP640 and the GHz beads that include the package parasitics, it
seems that the main obstacle to getting good stability at high collector
current is the pad capacitance right at the transistor.

With pad capacitances below 0.1 pF or so, the beads make them nice and
stable at all relevant collector currents, but at 0.5 pF, you have to
stay below 3 mA or thereabouts. Series inductance doesn\'t matter much
for stability, or for performance at < 1 GHz, where we mostly live.

So, we\'re getting some boards made with a whole bunch of 5-mm squares,
each containing a fast transistor and two or three of these magic beads.

Dunno yet how we\'ll get them apart--probably routing and snap tabs, or
else V-grooving. And then there\'s the question of how to use them in
production. We\'ll see!


We make a few little boards with the EPC GaN fets on-board, so we can
glob-top them and replace the baby board if anything fails.

https://www.dropbox.com/sh/1f9zm7ys14rx0oe/AAA2n5S9-RjJa8PS3K2aUU8aa?dl=0
Click to expand...

I remember. Do you pick-and-place those automatically?

(The glob will increase the pad capacitance a bit.)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com
 
On Fri, 23 Jun 2023 20:10:49 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2023-06-23 19:37, John Larkin wrote:
On Fri, 23 Jun 2023 17:05:50 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

Apropos of the \'5 GHz beads\' and \'new kinds of parts\' threads, I\'m
considering doing new small parts of our own, based on magic GHz
transistors, but house-trained enough to use in normal boards (with op
amps and so on).

Specifically, it looks like a win to make mezzanine-packaged SiGe BJTs
with beads, on little crenelated boards. They\'d be something like
TMB-package LEDs, but without the plastic overmold.

I was playing around in LTspice, using the Infineon and Murata models
for the BFP640 and the GHz beads that include the package parasitics, it
seems that the main obstacle to getting good stability at high collector
current is the pad capacitance right at the transistor.

With pad capacitances below 0.1 pF or so, the beads make them nice and
stable at all relevant collector currents, but at 0.5 pF, you have to
stay below 3 mA or thereabouts. Series inductance doesn\'t matter much
for stability, or for performance at < 1 GHz, where we mostly live.

So, we\'re getting some boards made with a whole bunch of 5-mm squares,
each containing a fast transistor and two or three of these magic beads.

Dunno yet how we\'ll get them apart--probably routing and snap tabs, or
else V-grooving. And then there\'s the question of how to use them in
production. We\'ll see!


We make a few little boards with the EPC GaN fets on-board, so we can
glob-top them and replace the baby board if anything fails.

https://www.dropbox.com/sh/1f9zm7ys14rx0oe/AAA2n5S9-RjJa8PS3K2aUU8aa?dl=0


I remember. Do you pick-and-place those automatically?
Click to expand...

Yes, in a panel. Then we test with the pogo fixture and mark the ones
that pass, then chop them up.

(The glob will increase the pad capacitance a bit.)
Click to expand...

This is pretty low-impedance, high current stuff.



Cheers

Phil Hobbs
Click to expand...
 
John Larkin <jlarkin@highlandSNIPMEtechnology.com> wrote:
On Fri, 23 Jun 2023 20:10:49 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2023-06-23 19:37, John Larkin wrote:
On Fri, 23 Jun 2023 17:05:50 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

Apropos of the \'5 GHz beads\' and \'new kinds of parts\' threads, I\'m
considering doing new small parts of our own, based on magic GHz
transistors, but house-trained enough to use in normal boards (with op
amps and so on).

Specifically, it looks like a win to make mezzanine-packaged SiGe BJTs
with beads, on little crenelated boards. They\'d be something like
TMB-package LEDs, but without the plastic overmold.

I was playing around in LTspice, using the Infineon and Murata models
for the BFP640 and the GHz beads that include the package parasitics, it
seems that the main obstacle to getting good stability at high collector
current is the pad capacitance right at the transistor.

With pad capacitances below 0.1 pF or so, the beads make them nice and
stable at all relevant collector currents, but at 0.5 pF, you have to
stay below 3 mA or thereabouts. Series inductance doesn\'t matter much
for stability, or for performance at < 1 GHz, where we mostly live.

So, we\'re getting some boards made with a whole bunch of 5-mm squares,
each containing a fast transistor and two or three of these magic beads.

Dunno yet how we\'ll get them apart--probably routing and snap tabs, or
else V-grooving. And then there\'s the question of how to use them in
production. We\'ll see!


We make a few little boards with the EPC GaN fets on-board, so we can
glob-top them and replace the baby board if anything fails.

https://www.dropbox.com/sh/1f9zm7ys14rx0oe/AAA2n5S9-RjJa8PS3K2aUU8aa?dl=0


I remember. Do you pick-and-place those automatically?

Yes, in a panel. Then we test with the pogo fixture and mark the ones
that pass, then chop them up.


(The glob will increase the pad capacitance a bit.)

This is pretty low-impedance, high current stuff.


But after they’re tested and diced, do you p&p them onto another board?
Click to expand...

Cheers

Phil Hobbs


--
Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC /
Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics
 
On Sat, 24 Jun 2023 00:35:49 -0000 (UTC), Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

John Larkin <jlarkin@highlandSNIPMEtechnology.com> wrote:
On Fri, 23 Jun 2023 20:10:49 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2023-06-23 19:37, John Larkin wrote:
On Fri, 23 Jun 2023 17:05:50 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

Apropos of the \'5 GHz beads\' and \'new kinds of parts\' threads, I\'m
considering doing new small parts of our own, based on magic GHz
transistors, but house-trained enough to use in normal boards (with op
amps and so on).

Specifically, it looks like a win to make mezzanine-packaged SiGe BJTs
with beads, on little crenelated boards. They\'d be something like
TMB-package LEDs, but without the plastic overmold.

I was playing around in LTspice, using the Infineon and Murata models
for the BFP640 and the GHz beads that include the package parasitics, it
seems that the main obstacle to getting good stability at high collector
current is the pad capacitance right at the transistor.

With pad capacitances below 0.1 pF or so, the beads make them nice and
stable at all relevant collector currents, but at 0.5 pF, you have to
stay below 3 mA or thereabouts. Series inductance doesn\'t matter much
for stability, or for performance at < 1 GHz, where we mostly live.

So, we\'re getting some boards made with a whole bunch of 5-mm squares,
each containing a fast transistor and two or three of these magic beads.

Dunno yet how we\'ll get them apart--probably routing and snap tabs, or
else V-grooving. And then there\'s the question of how to use them in
production. We\'ll see!


We make a few little boards with the EPC GaN fets on-board, so we can
glob-top them and replace the baby board if anything fails.

https://www.dropbox.com/sh/1f9zm7ys14rx0oe/AAA2n5S9-RjJa8PS3K2aUU8aa?dl=0


I remember. Do you pick-and-place those automatically?

Yes, in a panel. Then we test with the pogo fixture and mark the ones
that pass, then chop them up.


(The glob will increase the pad capacitance a bit.)

This is pretty low-impedance, high current stuff.


But after they’re tested and diced, do you p&p them onto another board?

Cheers

Phil Hobbs
Click to expand...

I think they are hand soldered, but I\'m not sure about that.

They are a nuisance to desolder. We have a Metcal tip that is sort of
a pair of giant tweezers, which helps. The people downstairs in
production do magic with things like this.

Once they are tested and globbed and installed on a motherboard, they
are pretty reliable. Now that we\'ve worked the bugs out of the
process, the yield is almost 100%. We\'ve built about 1400 so far.

The T577 is an output pin driver, with totem-pole GaN fets, and we use
it as a component in several products.
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top