Switching tiny small capacitors to a VCO

[Joel Kolstad]

I have a VCO that using varactor tuning internally along with a "hang inductor
on this pin" approach to set the center frequency. In my case, that center
frequency is around 400MHz and the inductor turns out to be ~22nH (working
backwards from the data sheet, internally the varactors are between 5.17pF and
7.44pF as you change the tuning voltage).

I'd like to extend the tuning range of the VCO, and figured that, as the
varactors have a range of ~2.27pF, if I could switch in a ~2pF or so capacitor
in parallel with the inductor, the tuning range would be extended. I soldered
down a 2pF cap and verified that the VCO range does drop as predicted, so the
only problem is: How do you switch in something so small as a 2pF cap? Most
devices I'm aware of -- MOSFETs, a discrete varactor diode, etc. -- tend to
have 10x as much parasitic capacitance and therefore appear as a short to the
2pF cap without providing any significant switching action.

Another problem is that the oscillator's output is ~1.5Vpp (centered at ground
due to the inductor), so I expect that things like a PIN diode would just
self-bias and also turn into a pretty good short.

Any ideas?

Thanks,
---Joel Kolstad

 

[Didi]

... Most devices I'm aware of -- MOSFETs, a discrete varactor diode, etc. -- tend to

have 10x as much parasitic capacitance and therefore appear as a short to the
2pF cap without providing any significant switching action.
...
Any ideas?

Teledyne relais is the only thing I can think of in that situation...
They have them TO-99 (or sort of) sized, I think polarised exist as
well,
and IIRC they specified 0.3 pF capacitance - perhaps that would be
OK for you. I also found a much cheaper version - and somewhat
larger - but I don't have the capacitance figure at hand, it may have
been 10x the Teledyne one, and then may be not. I can find out if
you want me to, I'll have do some digging.

Dimiter

 

[[email protected]]

I have a VCO that using varactor tuning internally along with a "hang inductor
on this pin" approach to set the center frequency. In my case, that center
frequency is around 400MHz and the inductor turns out to be ~22nH (working
backwards from the data sheet, internally the varactors are between 5.17pF and
7.44pF as you change the tuning voltage).

I'd like to extend the tuning range of the VCO, and figured that, as the
varactors have a range of ~2.27pF, if I could switch in a ~2pF or so capacitor
in parallel with the inductor, the tuning range would be extended. I soldered
down a 2pF cap and verified that the VCO range does drop as predicted, so the
only problem is: How do you switch in something so small as a 2pF cap? Most
devices I'm aware of -- MOSFETs, a discrete varactor diode, etc. -- tend to
have 10x as much parasitic capacitance and therefore appear as a short to the
2pF cap without providing any significant switching action.

Another problem is that the oscillator's output is ~1.5Vpp (centered at ground
due to the inductor), so I expect that things like a PIN diode would just
self-bias and also turn into a pretty good short.

Any ideas?

Thanks,
---Joel Kolstad

Why don't you just use a hyperabrupt varactor with a wider range?

 

[Joel Kolstad]

Why don't you just use a hyperabrupt varactor with a wider range?

Because the varactors in question are inside the VCO IC that I have no way of
changing.

 

[Jamie]

Because the varactors in question are inside the VCO IC that I have no way of
changing.

Ah, a #120 drill bit and a Buck Roger laser cutter
with a big magnifier glass should do it!

 

[John Larkin]

I have a VCO that using varactor tuning internally along with a "hang inductor
on this pin" approach to set the center frequency. In my case, that center
frequency is around 400MHz and the inductor turns out to be ~22nH (working
backwards from the data sheet, internally the varactors are between 5.17pF and
7.44pF as you change the tuning voltage).

I'd like to extend the tuning range of the VCO, and figured that, as the
varactors have a range of ~2.27pF, if I could switch in a ~2pF or so capacitor
in parallel with the inductor, the tuning range would be extended. I soldered
down a 2pF cap and verified that the VCO range does drop as predicted, so the
only problem is: How do you switch in something so small as a 2pF cap? Most
devices I'm aware of -- MOSFETs, a discrete varactor diode, etc. -- tend to
have 10x as much parasitic capacitance and therefore appear as a short to the
2pF cap without providing any significant switching action.

Another problem is that the oscillator's output is ~1.5Vpp (centered at ground
due to the inductor), so I expect that things like a PIN diode would just
self-bias and also turn into a pretty good short.

Any ideas?

Thanks,
---Joel Kolstad

Use...

a. the drain of a small-geometry gaasfet

b. one or two schottky diodes

c. one low-capacitance PIN diode, with adequate off drive. Skyworks?

d. A MAX1474 with a series padder cap. Maybe.

e. A real switch, one of the Bourns things that looks like a trimpot.


Actually, I think that a typical PIN diode doesn't need back-bias.
It's so slow that, at 400 MHz, it won't even rectify. Backbias would
lower the off capacitance, of course.

John

 

[Joel Kolstad]

Thanks John, those are all great ideas!

 

[[email protected]]

Use...

a. the drain of a small-geometry gaasfet

b. one or two schottky diodes

c. one low-capacitance PIN diode, with adequate off drive. Skyworks?

d. A MAX1474 with a series padder cap. Maybe.

e. A real switch, one of the Bourns things that looks like a trimpot.


Actually, I think that a typical PIN diode doesn't need back-bias.
It's so slow that, at 400 MHz, it won't even rectify. Backbias would
lower the off capacitance, of course.

John

Or

(f.) a GaAs MMIC T/R switch like a Hittite HMC226 in a SOT-26 package

http://www.hittite.com/product_info/product_specs/switches/hmc226.pdf

 

[John Larkin]

Or

(f.) a GaAs MMIC T/R switch like a Hittite HMC226 in a SOT-26 package

http://www.hittite.com/product_info/product_specs/switches/hmc226.pdf

They don't specify "off" capacitances, or any capacitances actually,
and it doesn't look like they can be calculated from the datasheet.
Most RF type parts are shockingly underspecified for their actual
electrical behavior.

I used one Hittite part, an 8:1 GaAs mux, that was specified "DC-4
GHz" in big bold text. Turns out that they consider "DC" to be 200 MHz
and up, and there are in fact series capacitors inside. So during our
conference call, trying to find a workaround, I asked the designer the
value of the series caps. He said, smugly, that he couldn't tell me
because it was proprietary.



Hey, this is interesting:

|
|
|
+---------------+
| |
| |
| |
C1 L
| |
| |
+-----+ |
| | |
C2 / SW |
| / |
| | |
| | |
+-----+---------+----GND


Where C2 is the parasitic switch capacitance. L makes the whole mess
look infinite when the switch is open. Some algebra required.

John

 

[Joel Kolstad]

I used one Hittite part, an 8:1 GaAs mux, that was specified "DC-4
GHz" in big bold text. Turns out that they consider "DC" to be 200 MHz
and up, and there are in fact series capacitors inside.

Join the club... we were bitten by the same problem with Hittite HMC349
switches, also specified as "DC-4GHz" but also containing series capacitors
on the internal terminations (something that *is* shown in the tiny
schematic drawing of the part, if you squint just right). We redesigned the
board to use Macom crossover switches and external terminations instead!
The quantity was ~1,000, so probably not much of a sting to Hittite, but we
certainly let their reps know our displeasure.

That being said, in general Hittite makes some good stuff. Maybe it's just
their marketing weasels who play these specsmanship games...

---Joel

 

[Mark]

Join the club... we were bitten by the same problem with Hittite HMC349
switches, also specified as "DC-4GHz" but also containing series capacitors
on the internal terminations (something that *is* shown in the tiny
schematic drawing of the part, if you squint just right). We redesigned the
board to use Macom crossover switches and external terminations instead!
The quantity was ~1,000, so probably not much of a sting to Hittite, but we
certainly let their reps know our displeasure.

re your original question...

consider switching some additional inductance in series with the low
side of the inductor...

Mark

 

[Joel Kolstad]

re your original question...

consider switching some additional inductance in series with the low
side of the inductor...

Thanks, but there's a similar problem with that approach: A 22nH inductor at
400MHz is 55 ohms reactive. If you hang a switch off the end of that
inductor, a capacitance of 7.2pF is -55 ohms and the two resonate. So while I
think that's a workable approach, you'd presumably still want a low
capacitance switch, somewhere in the ballpark of <1pF.

The Skyworks diodes that John suggested look pretty good, e.g., their SMP1345
series (http://www.skyworksinc.com/products_detailpop2.asp?pid=6942). The
fact that some Skyworks rep shoved a sample parts kit in my hand at some
random trade show is also working in their favor right now...

---Joel

 

[James Arthur]

Hey, this is interesting:

|
|
|
+---------------+
| |
| |
| |
C1 L
| |
| |
+-----+ |
| | |
C2 / SW |
| / |
| | |
| | |
+-----+---------+----GND


Where C2 is the parasitic switch capacitance. L makes the whole mess
look infinite when the switch is open. Some algebra required.

John

I was going to suggest:
' +------+
' +-------------> >----o----| |
' | | | VCO |
' --- C1 ( | |
' --- 2pF ( +------+
' | (
' o----------+ | L2
' SW | | |
' ...|....... | ===
' . o---+ . | GND (?)
' . | | . |
' . --- | . (
' . --- | . ( L1
' . | / . (
' . | / . |
' . | | . |
' . +---+ . |
' ...|....... |
' === ===
' GND GND

where L1 is chosen to resonate Cstray of SW, no algebra required.

This technique (either version) allows use of a heftier, lower
resistance PIN switch, improving circuit Q.

James Arthur

 

[PeteS]

Thanks, but there's a similar problem with that approach: A 22nH inductor at
400MHz is 55 ohms reactive. If you hang a switch off the end of that
inductor, a capacitance of 7.2pF is -55 ohms and the two resonate. So while I
think that's a workable approach, you'd presumably still want a low
capacitance switch, somewhere in the ballpark of <1pF.

The Skyworks diodes that John suggested look pretty good, e.g., their SMP1345
series (http://www.skyworksinc.com/products_detailpop2.asp?pid=6942). The
fact that some Skyworks rep shoved a sample parts kit in my hand at some
random trade show is also working in their favor right now...

---Joel

Macom (now part of Tyco) also has some very interesting PIN switches.

Cheers

PeteS

 

[Joel Kolstad]

I was going to suggest:

....

Good suggestion James... I had thought of that but had dismissed it since I
was thinking L1 would be too small, plus I didn't want to have to start
dealing with the problem of inductors only being available in fixed sizes and
having to worry about the variations in switch capacitance from one unit to
the next. Upon reflection, however, it seems I must have calculated L1
incorrectly... checking it now, typically it would be some tens of nH, which
is reasonable.

---Joel

 

[Michael A. Terrell]

Thanks, but there's a similar problem with that approach: A 22nH inductor at
400MHz is 55 ohms reactive. If you hang a switch off the end of that
inductor, a capacitance of 7.2pF is -55 ohms and the two resonate. So while I
think that's a workable approach, you'd presumably still want a low
capacitance switch, somewhere in the ballpark of <1pF.

The Skyworks diodes that John suggested look pretty good, e.g., their SMP1345
series (http://www.skyworksinc.com/products_detailpop2.asp?pid=6942). The
fact that some Skyworks rep shoved a sample parts kit in my hand at some
random trade show is also working in their favor right now...

---Joel

Joel, it works well if you use a diode to short out part of the
inductor. Microdyne used this method for 20 years in their 1100AR ,1200
1400 and 700 series receivers at up to 510 MHz. Reverse bias the diode
to turn it off, and forward bias it with about 10 mA to shunt the end of
the inductor to ground. I have a pile of scrap modules, but none of the
VCO boards. We switched it to three diffrent lengths instead of two, to
reduce phase noise problems at the ends of the band segments. I did a
lot of work with these, but I don't have any schematics or other
documentataion.


--
Service to my country? Been there, Done that, and I've got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

 

[LVMarc]

I have a VCO that using varactor tuning internally along with a "hang inductor
on this pin" approach to set the center frequency. In my case, that center
frequency is around 400MHz and the inductor turns out to be ~22nH (working
backwards from the data sheet, internally the varactors are between 5.17pF and
7.44pF as you change the tuning voltage).

I'd like to extend the tuning range of the VCO, and figured that, as the
varactors have a range of ~2.27pF, if I could switch in a ~2pF or so capacitor
in parallel with the inductor, the tuning range would be extended. I soldered
down a 2pF cap and verified that the VCO range does drop as predicted, so the
only problem is: How do you switch in something so small as a 2pF cap? Most
devices I'm aware of -- MOSFETs, a discrete varactor diode, etc. -- tend to
have 10x as much parasitic capacitance and therefore appear as a short to the
2pF cap without providing any significant switching action.

Another problem is that the oscillator's output is ~1.5Vpp (centered at ground
due to the inductor), so I expect that things like a PIN diode would just
self-bias and also turn into a pretty good short.

Any ideas?

Thanks,
---Joel Kolstad

Joel,

Why not build your own uhf vco and make have the voc tuning sensitivity
you need? If this is a caommmercial endeavor, I woul dbe delighted to
assisit, having many years of experience and several patent dealing with
wide band and LINEAR tuning VCO's! [email protected] for further discussion.

Best regards

Marc Popek

 

[[email protected]]

They don't specify "off" capacitances, or any capacitances actually,
and it doesn't look like they can be calculated from the datasheet.
Most RF type parts are shockingly underspecified for their actual
electrical behavior.

I used one Hittite part, an 8:1 GaAs mux, that was specified "DC-4
GHz" in big bold text. Turns out that they consider "DC" to be 200 MHz
and up, and there are in fact series capacitors inside. So during our
conference call, trying to find a workaround, I asked the designer the
value of the series caps. He said, smugly, that he couldn't tell me
because it was proprietary.



Hey, this is interesting:

|
|
|
+---------------+
| |
| |
| |
C1 L
| |
| |
+-----+ |
| | |
C2 / SW |
| / |
| | |
| | |
+-----+---------+----GND


Where C2 is the parasitic switch capacitance. L makes the whole mess
look infinite when the switch is open. Some algebra required.

John

This all is nice theoretically
But in real world swept frequency synthesizer designs there's usually a
tight switching speed specification. That's the amount of time it takes
for the frequency synthesizer to switch from the lowest frequency to
the highest frequency. BTW, I think Joel is using the AD4360-7 VCO/PLL
combo chip.

http://www.analog.com/en/prod/0,2877,ADF4360%2D7,00.html

 

[Robert Baer]

Join the club... we were bitten by the same problem with Hittite HMC349
switches, also specified as "DC-4GHz" but also containing series capacitors
on the internal terminations (something that *is* shown in the tiny
schematic drawing of the part, if you squint just right). We redesigned the
board to use Macom crossover switches and external terminations instead!
The quantity was ~1,000, so probably not much of a sting to Hittite, but we
certainly let their reps know our displeasure.

That being said, in general Hittite makes some good stuff. Maybe it's just
their marketing weasels who play these specsmanship games...

---Joel

Well, you never heard of "DC" frequencies?
Some even consider FM as being in the DC frequency range.

 

[Chris Jones]

I have a VCO that using varactor tuning internally along with a "hang
inductor
on this pin" approach to set the center frequency. In my case, that
center frequency is around 400MHz and the inductor turns out to be ~22nH
(working backwards from the data sheet, internally the varactors are
between 5.17pF and 7.44pF as you change the tuning voltage).

I'd like to extend the tuning range of the VCO, and figured that, as the
varactors have a range of ~2.27pF, if I could switch in a ~2pF or so
capacitor
in parallel with the inductor, the tuning range would be extended. I
soldered down a 2pF cap and verified that the VCO range does drop as
predicted, so the
only problem is: How do you switch in something so small as a 2pF cap?
Most devices I'm aware of -- MOSFETs, a discrete varactor diode, etc. --
tend to have 10x as much parasitic capacitance and therefore appear as a
short to the 2pF cap without providing any significant switching action.

Another problem is that the oscillator's output is ~1.5Vpp (centered at
ground due to the inductor), so I expect that things like a PIN diode
would just self-bias and also turn into a pretty good short.

Any ideas?

Thanks,
---Joel Kolstad

Which chip are you using? Is it a differential (symmetrical tank)
oscillator or a single-ended one? Is there any deliberate off-chip
capacitance across the inductor that you could absorb into the
off-capacitance of a switched capacitor bank?

Chris