rf everywhere

[RichD]

Wireless is everywhere now, miniaturized to an astounding degree.

Recently, I saw a report on a button size gardening
gadget - stick it in the soil, it reports on moisture.
Bluetooth earphones, etc.

Who's designing these things? In my experience, RF
designers are a rare breed, and with the digital market
vastly larger, they're even rarer.

I'll guess, the IC have been perfected to the no-brainer
level. But still, you need need amps, filters, antenna, plus
issues of noise and layout, yes/no? That stuff isn't obsoleted.

I don't work in this area, but I'm curious, so can anyone
elaborate on what's going on, from a system viewpoint?
What are the chip functions, options, price, trade-offs?
In which situations would you reject them, to roll your own?

Is it simple on/off keying, or more sophisticated? Currently,
in communications theory, sensor networks are a hot topic,
where thousands of sensors are competing for bandwidth,
but for mundane consumer apps, I doubt those issues arise.

I'm looking to pick the brains of any gurus here -

 

[Tim Williams]

Hmmm, not a big deal I suspect.

Build a general purpose RF block for, say, 2.45GHz BT or 802.11(etc), or
whatever. Give it handles to talk with anything (modulations, bit
streams, etc.), design and build it on a particular fab process, and like
magic, anything incorporating that block will also work. Monolithic
inductors can be fabricated with not very good Q at 2.45GHz (I think they
usually peak around Q = 10 or 20 around 5GHz), but enough to do "silicon
oscillators" and stuff. Voltage regulation (bandgap, or old school buried
zener) and temperature compensation are no-brainers, as ICs go. Want a
DDS? Just chuck some more IP at it! Then whatever ancillary function
(moisture, temperature sensor, etc.) simply plugs into this mess of
transistors and functions.

Quite crazy, as all that circuitry is squeezing into a few milimeters of
silicon, when a few decades ago it was, well of course it was migrating to
thick film before monolithic, but before that, it was all machined
cavities, hand-soldered RF transistors, and microstrip everywhere. I
suppose Bluetooth would've taken up a whole rack, back in the 70s, and
that's assuming the computing power to provide whatever spread spectrum,
encoding, error detection, etc. functionality is required.

Tim

 

[Adrian Jansen]

Wireless is everywhere now, miniaturized to an astounding degree.

Recently, I saw a report on a button size gardening
gadget - stick it in the soil, it reports on moisture.
Bluetooth earphones, etc.

Who's designing these things? In my experience, RF
designers are a rare breed, and with the digital market
vastly larger, they're even rarer.

I'll guess, the IC have been perfected to the no-brainer
level. But still, you need need amps, filters, antenna, plus
issues of noise and layout, yes/no? That stuff isn't obsoleted.

I don't work in this area, but I'm curious, so can anyone
elaborate on what's going on, from a system viewpoint?
What are the chip functions, options, price, trade-offs?
In which situations would you reject them, to roll your own?

Is it simple on/off keying, or more sophisticated? Currently,
in communications theory, sensor networks are a hot topic,
where thousands of sensors are competing for bandwidth,
but for mundane consumer apps, I doubt those issues arise.

I'm looking to pick the brains of any gurus here -

You are right, but seems like someone has solved the RF problems once
for each of the useful bands, then its a piece of cake to interface with
sensors and one end and display/alarm at the other.

For an example, the tyre pressure monitor systems at 433 MHz. 10 gram
package, including battery, you screw on a tyre valve. Monitors tyre
pressure and temperature for about 1-2 years of operation. Reports real
time, every minute or so, to in-car readout.

 

[Spehro Pefhany]

You are right, but seems like someone has solved the RF problems once
for each of the useful bands, then its a piece of cake to interface with
sensors and one end and display/alarm at the other.

For an example, the tyre pressure monitor systems at 433 MHz. 10 gram
package, including battery, you screw on a tyre valve. Monitors tyre
pressure and temperature for about 1-2 years of operation. Reports real
time, every minute or so, to in-car readout.

Frequently fails mechanically, causing loss of tire pressure, allows
tire shops to charge for a "rebuild kit" whenever they swap a tire,
requires a trip to the dealer (or specialized equipment/knowledge) to
replace, even with an OEM replacement part.. other than that, they're
just spiffy.


Best regards,
Spehro Pefhany

 

[Trevor]

Frequently fails mechanically, causing loss of tire pressure, allows
tire shops to charge for a "rebuild kit" whenever they swap a tire,
requires a trip to the dealer (or specialized equipment/knowledge) to
replace, even with an OEM replacement part.. other than that, they're
just spiffy.

So pretty much like many of the new gadgets on modern cars, something else
to go wrong that costs you money, even if you never wanted it in the first
place :-(

Trevor.

 

[Cydrome Leader]

Hmmm, not a big deal I suspect.

Build a general purpose RF block for, say, 2.45GHz BT or 802.11(etc), or
whatever. Give it handles to talk with anything (modulations, bit
streams, etc.), design and build it on a particular fab process, and like
magic, anything incorporating that block will also work. Monolithic
inductors can be fabricated with not very good Q at 2.45GHz (I think they
usually peak around Q = 10 or 20 around 5GHz), but enough to do "silicon
oscillators" and stuff. Voltage regulation (bandgap, or old school buried
zener) and temperature compensation are no-brainers, as ICs go. Want a
DDS? Just chuck some more IP at it! Then whatever ancillary function
(moisture, temperature sensor, etc.) simply plugs into this mess of
transistors and functions.

Quite crazy, as all that circuitry is squeezing into a few milimeters of
silicon, when a few decades ago it was, well of course it was migrating to
thick film before monolithic, but before that, it was all machined
cavities, hand-soldered RF transistors, and microstrip everywhere. I

years ago I was given a box of microwave "plumbing" from what may have
been a broadcast engineer. The stuff would have worked with microwaves or
hydraulic fluid. The guy who made the stuff seemed to be really good with
a jewelers saw, copper pipe, brass discs rods and solder.

 

[[email protected]]

Hmmm, not a big deal I suspect.

Build a general purpose RF block for, say, 2.45GHz BT or 802.11(etc), or
whatever.  Give it handles to talk with anything (modulations, bit
streams, etc.), design and build it on a particular fab process, and like
magic, anything incorporating that block will also work.  Monolithic
inductors can be fabricated with not very good Q at 2.45GHz (I think they
usually peak around Q = 10 or 20 around 5GHz), but enough to do "silicon
oscillators" and stuff.  Voltage regulation (bandgap, or old school buried
zener) and temperature compensation are no-brainers, as ICs go.  Want a
DDS?  Just chuck some more IP at it!  Then whatever ancillary function
(moisture, temperature sensor, etc.) simply plugs into this mess of
transistors and functions.

I remember working on making bluetooth in a "single chip"
we had a working radio and build an evolution of an existing SOC to
stack
on top of it in a single package

Everything worked great when we tested the first samples, but then
the
software guys started running their code in ROM then the sensitivity
dropped

turned out that the ROM being in a different corner of the SOC coupled
noise
into the radio VCO inductors, but the RAM where the test code was run
didn't

Quite crazy, as all that circuitry is squeezing into a few milimeters of
silicon, when a few decades ago it was, well of course it was migrating to
thick film before monolithic, but before that, it was all machined
cavities, hand-soldered RF transistors, and microstrip everywhere.  I
suppose Bluetooth would've taken up a whole rack, back in the 70s, and
that's assuming the computing power to provide whatever spread spectrum,
encoding, error detection, etc. functionality is required.

Tim

I worked on one of the very first bluetooth implementations, it was;
a DSP, a flash, an FPGA, an RF chip, a saw filter, a whole bunch of
passives
it was probably 5*5cm PCB fully packed on both sides

-Lasse

 

[Cydrome Leader]

I remember working on making bluetooth in a "single chip"
we had a working radio and build an evolution of an existing SOC to
stack
on top of it in a single package

Everything worked great when we tested the first samples, but then
the
software guys started running their code in ROM then the sensitivity
dropped

turned out that the ROM being in a different corner of the SOC coupled
noise
into the radio VCO inductors, but the RAM where the test code was run
didn't



I worked on one of the very first bluetooth implementations, it was;
a DSP, a flash, an FPGA, an RF chip, a saw filter, a whole bunch of
passives
it was probably 5*5cm PCB fully packed on both sides

In the 1988 to 1990s ish time, there was a story in popular mechanics or
popular science about a digital ghost canceler for television signals that
bounced off buildings. It was huge PCB made using an array of DSPs and
have to have pounds of gold plated ceramic chips on it. It was a pretty
looking board, that must have screamed at like 16MHz or something like
that.

What would that take these days, to basically subtract patterns from a
NTSC signal? A couple chips?

 

[Adrian Jansen]

Frequently fails mechanically, causing loss of tire pressure, allows
tire shops to charge for a "rebuild kit" whenever they swap a tire,
requires a trip to the dealer (or specialized equipment/knowledge) to
replace, even with an OEM replacement part.. other than that, they're
just spiffy.


Best regards,
Spehro Pefhany

Really I was commenting on the RF stuff. Certainly that seems to work
as well as needed. Whether the rest of the design is as good as the RF
section is a different kettle of fish.

The aftermarket units using sensors like Tyredog seem to have a learning
mode to accomodate sensor changes without special tools.

Personally I would be happy if the system just warned that a tyre is
going down, before it wrecks the tyre. Identifying which tyre is at
fault is a secondary, and usually very easy, job.

 

[MrTallyman]

years ago I was given a box of microwave "plumbing" from what may have
been a broadcast engineer. The stuff would have worked with microwaves or
hydraulic fluid. The guy who made the stuff seemed to be really good with
a jewelers saw, copper pipe, brass discs rods and solder.

Not many folks making hard coax runs anymore.

Semi-rigid and a few others abound.

 

[Cydrome Leader]

Not many folks making hard coax runs anymore.

this stuff was pretty darn old.

are there power levels where they stil use wavegides and the like?

 

[[email protected]]

this stuff was pretty darn old.

are there power levels where they stil use wavegides and the like?

In countries still using analog TV, the UHF final amplifier is often
implemented with klystron in the 100 kW range. The waveguide is quite
large, due to the low frequency.

DVB-T digital TV transmitters typically operate with only 1-10 kW,
consisting of multiple redundant solid state modules, so there is not
much need for waveguides any more.

 

[Don Pearce]

are there power levels where they stil use wavegides and the like?

As much to do with frequency as power level. I've recently completed
the design of a Ka band transceiver for satellite (30GHz uplink, 20GHz
down). It is a domestic product for delivery of broadband to rural
areas. All of the internal RF filtering is done in waveguide, as is
the external combining of the transmit and receive signals into a
single horn. The transmitter power is just 3W.

d

 

[Don Pearce]

How do you get that to the antenna without waveguide? Coax losses are
much higher than waveguide, and is less likely to have problems since
there is no dielectric to break down.

Andrew's Heliax is pretty low loss, and good for VHF and UHF runs.
Dielectric amounts to nothing more than a thin spiral spacer - the
rest is all air.

d

 

[Don Pearce]

Heliax for 10 KW? Ever had the filter fail on the compressor and get
water in Heliax? I had a stupid SOB for a boss 30 years ago who was too
sheap to replace filtes on schedule and ruined a piece of 3" Heliax used
at 4 GHz. Waveguide is better at high power, and better than Heliax. I
had over 1700 feet of it at one TV transmitter. It carried about 195 KW
of RF to the top of the tower. We had to maintain a set pressure of dry
nitrogen on the waveguide to keep from compressing the sync pulses.

No, not at 10kW. At 100W or so it is fine. On a long run I would
always opt for waveguide - tower dimensions permitting.

d

 

[T]

How do you get that to the antenna without waveguide? Coax losses are
much higher than waveguide, and is less likely to have problems since
there is no dielectric to break down.

Hardline! In essence it's a solid shield with a center conductor. That's
enogh for RF into the 900MHz and 1.2GHz range.

 

[Cydrome Leader]

Heliax for 10 KW? Ever had the filter fail on the compressor and get
water in Heliax? I had a stupid SOB for a boss 30 years ago who was too
sheap to replace filtes on schedule and ruined a piece of 3" Heliax used
at 4 GHz. Waveguide is better at high power, and better than Heliax. I
had over 1700 feet of it at one TV transmitter. It carried about 195 KW
of RF to the top of the tower. We had to maintain a set pressure of dry
nitrogen on the waveguide to keep from compressing the sync pulses.

what exciting things happens when you get moisture ingress?

how hot can coax or waveguides run at high powers? I've never been by
large transmitters, so the concept of anything but a power cord running
warm is just strange to me.

 

[Cydrome Leader]

There are many types of hardline, and more in use by CATV systems
than anything else. I engineered in that field for four years and
designed extensions and upgrades. I also designed community loop
systems.


Specify a brand, type length, frequency and power level. There is a
reason they have so damn many trunk amplifiers in a system.


I found a link to some 5.5" heilical air dielectric that doesn't look
too bad, but the specs stop before a GHz. look at the power handling
spec at .5 MHz, and at 894 MHz. It drops from 1890 KW to 43 KW. The
insetion loss goes from .0045 dB to .215 dB over that range.

http://www.rfsworld.com/dataxpress/Datasheets/pdf/?q=HCA550-50J

how does one terminate such cables? What sort of wiring goes on inside a
giant transmitter or the antenna end of such a monster?

 

[[email protected]]

Why is there less loss with bigger air dielectric line at UHF?

In a coaxial cable the EM field propagates in the dielectric, not
in/on the inner and/or outer conductor. The outer surface of the inner
conductor and the inner surface of the outer conductor will just
confine the EM field into the dielectric as does the inner surface of
a waveguide.

For this reason, the quality of the dielectric is critical, dry air is
nearly as good as vacuum at lower frequencies (below a few hundred
GHz).

In addition, the metallic surfaces should have infinite conductivity,
but of course, skin effect and oxidation will degrade the performance.

The 15 cm coax is not very usable at frequencies well above 1 GHz,
since the EM field starts to propagate in some strange waveguide mode
above these frequencies.

For lower frequencies, coaxal cables work OK even with much larger
dimensions, such as 50 cm coaxial "cables" at a 500 kW short wave
station.

 

[[email protected]]

What is the advantage (if any) to using co-ax v. open-wire feed for
above HFBC example?

Coax has the fields contained within the cable, open-wire does not.
Structures, people, and critters in proximity to coax don't matter.
It's a different matter with open-wire. Open wire usually requires a
BalUn, as well.