5 GHz ferrite beads!...

P

Phil Hobbs

Guest
I like to use SiGe microwave transistors for all sorts of off-label
things, as some SED veterans may remember.

The Infineon BFP640 and BFP650 parts aren\'t the absolute fastest any
more--they top out around 40 GHz f_T, whereas the BFP8xx series get up
to over 80 GHz.

However, they have other amazing properties that make up for their
slowness. ;) These include flatband noise voltages of 0.5 nV in 1 Hz;
betas around 250, which is amazing for an RF device; a low 1/f corner;
and, best of all, an Early voltage so large that it\'s hard to measure.

(It\'s 250V or so, AFAICT--you have to measure quickly or the
self-heating of the device makes the curves go the wrong way!) That
lets you run high gain in a single stage, without worrying much about
linearity. They\'re also _amazing_ cascode devices.

However, they do have a tendency to oscillate if you look at them
crosswise, e.g. give them more than a few mA of collector current.

That\'s fine for lots of things, but lately I\'ve been wanting to make
biased-cascode SPAD(*) front ends for time-of-flight applications.

SPADs have a fair amount of capacitance for their speed, so you have to
terminate them in a low impedance, ideally only a couple of ohms.

Packaged parts have inconveniently high inductance for that, but we
should be able to get emitter impedances below 10 ohms up to 3 GHz or
so. Besides sub-nanohenry inductances, that requires fairly high
collector currents, like 10 mA or thereabouts, putting their f_Ts above
20 GHz. (The resistive part of the emitter impedance is the usual
r_E ~ 25 mOhm/I_C.)

Naturally they want to oscillate like crazy, so they need base
stoppers(**). For really fast transistors, I like to use the Murata
BLM15BA005 and BLM15BA010, which are 0402 beads with impedances of 5 and
10 ohms at 100 MHz, respectively.

Those ones have a nice low-Q impedance peak at around 3 GHz, which works
pretty well with ordinary fast things. However, they\'re a bit wimpy for
these SiGe BJTs, which I\'ve seen oscillate at 12 GHz.

Soooooo, imagine my delight when I found that there are ferrite beads
specified by their impedance, not at 100 MHz, but at _5 GHz._ I\'m sure
they\'re old hat to cell phone designers, but I don\'t pal around with any
of them, so they\'re new to me.

For instance, the Murata BLF03VK221SNGD has 220 ohms impedance (mostly
resistive) at 5 GHz, and is still over 100 ohms at 10 GHz. It\'s 0201
size, of course, but LCSC has them, so I can get JLCPCB to solder them
down for me.

Fun.

Cheers

Phil Hobbs

(*) Single-photon avalanche diode

(**) Something resistive-looking that you put in series with the base of
a transistor to keep it from oscillating. Extra points if the
resistance is mostly up at frequencies you don\'t need for your
measurement. (That\'s where ferrite beads come in.)

--
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 Wed, 14 Jun 2023 19:30:46 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

I like to use SiGe microwave transistors for all sorts of off-label
things, as some SED veterans may remember.

The Infineon BFP640 and BFP650 parts aren\'t the absolute fastest any
more--they top out around 40 GHz f_T, whereas the BFP8xx series get up
to over 80 GHz.

However, they have other amazing properties that make up for their
slowness. ;) These include flatband noise voltages of 0.5 nV in 1 Hz;
betas around 250, which is amazing for an RF device; a low 1/f corner;
and, best of all, an Early voltage so large that it\'s hard to measure.

(It\'s 250V or so, AFAICT--you have to measure quickly or the
self-heating of the device makes the curves go the wrong way!) That
lets you run high gain in a single stage, without worrying much about
linearity. They\'re also _amazing_ cascode devices.

However, they do have a tendency to oscillate if you look at them
crosswise, e.g. give them more than a few mA of collector current.

That\'s fine for lots of things, but lately I\'ve been wanting to make
biased-cascode SPAD(*) front ends for time-of-flight applications.

SPADs have a fair amount of capacitance for their speed, so you have to
terminate them in a low impedance, ideally only a couple of ohms.

Packaged parts have inconveniently high inductance for that, but we
should be able to get emitter impedances below 10 ohms up to 3 GHz or
so. Besides sub-nanohenry inductances, that requires fairly high
collector currents, like 10 mA or thereabouts, putting their f_Ts above
20 GHz. (The resistive part of the emitter impedance is the usual
r_E ~ 25 mOhm/I_C.)

Naturally they want to oscillate like crazy, so they need base
stoppers(**). For really fast transistors, I like to use the Murata
BLM15BA005 and BLM15BA010, which are 0402 beads with impedances of 5 and
10 ohms at 100 MHz, respectively.

Those ones have a nice low-Q impedance peak at around 3 GHz, which works
pretty well with ordinary fast things. However, they\'re a bit wimpy for
these SiGe BJTs, which I\'ve seen oscillate at 12 GHz.

Soooooo, imagine my delight when I found that there are ferrite beads
specified by their impedance, not at 100 MHz, but at _5 GHz._ I\'m sure
they\'re old hat to cell phone designers, but I don\'t pal around with any
of them, so they\'re new to me.

For instance, the Murata BLF03VK221SNGD has 220 ohms impedance (mostly
resistive) at 5 GHz, and is still over 100 ohms at 10 GHz. It\'s 0201
size, of course, but LCSC has them, so I can get JLCPCB to solder them
down for me.

Fun.

Cheers

Phil Hobbs

(*) Single-photon avalanche diode

(**) Something resistive-looking that you put in series with the base of
a transistor to keep it from oscillating. Extra points if the
resistance is mostly up at frequencies you don\'t need for your
measurement. (That\'s where ferrite beads come in.)
Click to expand...

I assume that the real part of the impedance vs frequency curve of the
bead keeps low frequency Johnson and Ib noise down but kills the
multi-GHz oscillations. Is that the idea?
 
Am 15.06.23 um 03:26 schrieb John Larkin:
On Wed, 14 Jun 2023 19:30:46 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:


(**) Something resistive-looking that you put in series with the base of
a transistor to keep it from oscillating. Extra points if the
resistance is mostly up at frequencies you don\'t need for your
measurement. (That\'s where ferrite beads come in.)

I assume that the real part of the impedance vs frequency curve of the
bead keeps low frequency Johnson and Ib noise down but kills the
multi-GHz oscillations. Is that the idea?
Click to expand...

Yes, exactly.

At LF, there is no real part of the impedance
worth speaking of and thus no thermal noise.
It can be seen in LTspice simulation; I have
done it for a Würth ferrite bead from the LTspice
library, cannot find it now.

Gerhard
 
On Thu, 15 Jun 2023 04:19:30 +0200, Gerhard Hoffmann <dk4xp@arcor.de>
wrote:

Am 15.06.23 um 03:26 schrieb John Larkin:
On Wed, 14 Jun 2023 19:30:46 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:


(**) Something resistive-looking that you put in series with the base of
a transistor to keep it from oscillating. Extra points if the
resistance is mostly up at frequencies you don\'t need for your
measurement. (That\'s where ferrite beads come in.)

I assume that the real part of the impedance vs frequency curve of the
bead keeps low frequency Johnson and Ib noise down but kills the
multi-GHz oscillations. Is that the idea?

Yes, exactly.

At LF, there is no real part of the impedance
worth speaking of and thus no thermal noise.
It can be seen in LTspice simulation; I have
done it for a Würth ferrite bead from the LTspice
library, cannot find it now.

Gerhard
Click to expand...

Fast transistors tend to oscillate in real life but not in Spice. Wire
bonds maybe?
 
John Larkin <jlarkin@highlandSNIPMEtechnology.com> wrote:
On Wed, 14 Jun 2023 19:30:46 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

I like to use SiGe microwave transistors for all sorts of off-label
things, as some SED veterans may remember.

The Infineon BFP640 and BFP650 parts aren\'t the absolute fastest any
more--they top out around 40 GHz f_T, whereas the BFP8xx series get up
to over 80 GHz.

However, they have other amazing properties that make up for their
slowness. ;) These include flatband noise voltages of 0.5 nV in 1 Hz;
betas around 250, which is amazing for an RF device; a low 1/f corner;
and, best of all, an Early voltage so large that it\'s hard to measure.

(It\'s 250V or so, AFAICT--you have to measure quickly or the
self-heating of the device makes the curves go the wrong way!) That
lets you run high gain in a single stage, without worrying much about
linearity. They\'re also _amazing_ cascode devices.

However, they do have a tendency to oscillate if you look at them
crosswise, e.g. give them more than a few mA of collector current.

That\'s fine for lots of things, but lately I\'ve been wanting to make
biased-cascode SPAD(*) front ends for time-of-flight applications.

SPADs have a fair amount of capacitance for their speed, so you have to
terminate them in a low impedance, ideally only a couple of ohms.

Packaged parts have inconveniently high inductance for that, but we
should be able to get emitter impedances below 10 ohms up to 3 GHz or
so. Besides sub-nanohenry inductances, that requires fairly high
collector currents, like 10 mA or thereabouts, putting their f_Ts above
20 GHz. (The resistive part of the emitter impedance is the usual
r_E ~ 25 mOhm/I_C.)

Naturally they want to oscillate like crazy, so they need base
stoppers(**). For really fast transistors, I like to use the Murata
BLM15BA005 and BLM15BA010, which are 0402 beads with impedances of 5 and
10 ohms at 100 MHz, respectively.

Those ones have a nice low-Q impedance peak at around 3 GHz, which works
pretty well with ordinary fast things. However, they\'re a bit wimpy for
these SiGe BJTs, which I\'ve seen oscillate at 12 GHz.

Soooooo, imagine my delight when I found that there are ferrite beads
specified by their impedance, not at 100 MHz, but at _5 GHz._ I\'m sure
they\'re old hat to cell phone designers, but I don\'t pal around with any
of them, so they\'re new to me.

For instance, the Murata BLF03VK221SNGD has 220 ohms impedance (mostly
resistive) at 5 GHz, and is still over 100 ohms at 10 GHz. It\'s 0201
size, of course, but LCSC has them, so I can get JLCPCB to solder them
down for me.

Fun.

Cheers

Phil Hobbs

(*) Single-photon avalanche diode

(**) Something resistive-looking that you put in series with the base of
a transistor to keep it from oscillating. Extra points if the
resistance is mostly up at frequencies you don\'t need for your
measurement. (That\'s where ferrite beads come in.)

I assume that the real part of the impedance vs frequency curve of the
bead keeps low frequency Johnson and Ib noise down but kills the
multi-GHz oscillations. Is that the idea?
Click to expand...

Yes. The BLF03VK221SNGD looks like a Q=1 LR up to about 1 GHz, then
becomes resistive up to its peak Z at 6GHz, where the capacitance starts to
take over.

--
Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC /
Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top