On Feb 28, 12:29 am, Ian White GM3SEK wrote:
Jim wrote:
I'll bet the hardware cost of a electronically steered phased array
for HF suitable for ham use is comparable to the hardware cost of a big
tower, rotator, and Yagi.. the phased array just isn't available as an
off the shelf product yet.

The fully steerable phased array that can also handle 1.5kW TX power is
not available as a HAM product off the shelf yet... but we can already
see where the future is headed.

---------
snip
----------
It's all down to the magic phasing box at the centre of the array.
Whatever goes into that box will be hard to design, complicated to
control, and expensive to build... but most four-square owners would be
happy to have even a fraction of those extra capabilities.

I think it might be a bit of time before it's an off the shelf product
(lack of demand is part of the reason).
It also depends a bit on just how good you want the performance to be
(null depth, primarily.. forward gain is not very sensitive to phasing
and amplitude accuracy), and whether you want to make it an antenna
system that just hooks onto an existing rig and PA, essentially
hanging off a single feedline, or whether a higher level of
integration is desirable or feasible. (for instance, rather than power
combining a bunch of solid state amp modules like the current kilowatt
class SSPAs do, put a power module on each antenna)

"the magic box in the center of the array" is more the former model,
and while simple conceptually, in the long run probably isn't the best
way to solve the problem in a system context. For instance, a multi
channel receiver, which can do all the phasing, very precisely, at low
levels, either with analog or digital processing, can give you the
nice deep nulls and adaptation.For TX, though, null depth isn't as
important as maximizing the power squirted in the right direction.

Hardware wise, the design isn't particularly complicated (any of the
current crop of automatic antenna tuners can serve as the building
block). Hard to control is mostly a matter of calibration and the
right algorithms (and, yes, non trivial, but so is building a tetrode
or FET power amp that's stable from 10 to 160, etc.). Expensive is
more a matter of "compared to what".. You can buy a kilowatt autotuner
for $500, so, assuming you needed 8 of them to control 4 elements (a
single L network can only give you 90 degrees of phase shift, and you
need 180).. that's $4K. Probably need some relays and transformers,
as well as controller. I'd figure $6K, today.. But that's "off the
shelf" assemblies. and not purpose designed. Start comparing that to
the $10K to put up a tower and a Yagi (comparing new prices to new
prices, plus building permits, etc.) and the phased array starts to be
competive.

So.. not today, but I'd figure that in 10 years, you'll start to see
real broadband phased arrays (not just 4-8 switched beams in a single
band like a 4 square). They provide a real solution to restricted
space installations, if nothing else, because you can make effective
use of the "volume" of space within a small suburban lot (i.e. the max
theoretical gain from an antenna that fits in a box that's say,
15x30x10 meters is pretty high).

Jim, W6RMK

On 28 Feb, 23:20, wrote:
On Feb 28, 12:29 am, Ian White GM3SEK wrote: Jim wrote:
I'll bet the hardware cost of a electronically steered phased array
for HF suitable for ham use is comparable to the hardware cost of a big
tower, rotator, and Yagi.. the phased array just isn't available as an
off the shelf product yet.

The fully steerable phased array that can also handle 1.5kW TX power is
not available as a HAM product off the shelf yet... but we can already
see where the future is headed.

---------
snip
----------
snipe). They provide a real solution to restricted
space installations, if nothing else, because you can make effective
use of the "volume" of space within a small suburban lot (i.e. the max
theoretical gain from an antenna that fits in a box that's say,
15x30x10 meters is pretty high).

Jim, W6RMK

Jim, this volume approach to gain versus linear length comparison
sounds very interesting but I have not seen any reference to it
anywhere. Basically a gaussian array say for a single feed point
makes full use of volume where other antennas which are of planar form
lose out. With a planar design one can extrapolate gain by number of
elements combined with length e.t.c but I have not seen any such gain
calculation for a volume,can you help me there?
I have found that the number of elements with respect to contained
volume is a good measurement where the array is contained within the
1/2 wave length cubed beyond which it appears to have reached its
maximum. This means having covered a real estate area of half that of
a yagi but with equivalent gain.
Ofcourse one can gang arrays together if one uses multiple feeds
but I haven't personaly pursued that approach as yet.
Glad to see you posting, most informative
Best regards
Art

Jim wrote:
On Feb 28, 12:29 am, Ian White GM3SEK wrote:
Jim wrote:
I'll bet the hardware cost of a electronically steered phased array
for HF suitable for ham use is comparable to the hardware cost of a big
tower, rotator, and Yagi.. the phased array just isn't available as an
off the shelf product yet.

The fully steerable phased array that can also handle 1.5kW TX power is
not available as a HAM product off the shelf yet... but we can already
see where the future is headed.

---------
snip
----------
It's all down to the magic phasing box at the centre of the array.
Whatever goes into that box will be hard to design, complicated to
control, and expensive to build... but most four-square owners would be
happy to have even a fraction of those extra capabilities.

I think it might be a bit of time before it's an off the shelf product
(lack of demand is part of the reason).
It also depends a bit on just how good you want the performance to be
(null depth, primarily.. forward gain is not very sensitive to phasing
and amplitude accuracy), and whether you want to make it an antenna
system that just hooks onto an existing rig and PA, essentially
hanging off a single feedline, or whether a higher level of
integration is desirable or feasible. (for instance, rather than power
combining a bunch of solid state amp modules like the current kilowatt
class SSPAs do, put a power module on each antenna)

"the magic box in the center of the array" is more the former model,
and while simple conceptually, in the long run probably isn't the best
way to solve the problem in a system context.

That is very true. Unfortunately, the ham market is divided up into
physically separate compartments of transceivers, power amplifiers and
antennas. That is a severe restriction which makes all the technical
challenges much more difficult.

However, we can try to pare the problem down a little.

Another important point is that the concept of 'market demand' is
beginning to break down in ham radio. The big manufacturers are
increasingly challenged by new products that pay no attention to the
market - they spring directly from some individual or small team
deciding they're going to do it.

Then maybe the design is produced as a kit, or manufacturing is taken up
by some lower-tier company that is faster on its feet. Seems good to
me...


For instance, a multi
channel receiver, which can do all the phasing, very precisely, at low
levels, either with analog or digital processing, can give you the
nice deep nulls and adaptation.For TX, though, null depth isn't as
important as maximizing the power squirted in the right direction.

That seems a good place to cut the problem down to size. By all means
continue to use the existing phasing networks for TX, with 4 or 8
switchable directions and fixed phasing; but switch the antennas over to
a totally separate network for RX.

At the lower power levels, the RX network could be much more complex and
versatile, combining the signals from the four (say) antennas with
amplitudes and phasing that could be varied on the fly.

Another way to scale down the problem is not to be too ambitious about
automatic null steering. In ham operating it is often difficult for a
computer to identify which is the wanted signal and which is the QRM, so
maybe let's not try. Semi-automatic null steering definitely would be
within reach, where the user has a control to steer the null direction
manually for the best audible results, and the computer does the math to
select the required network parameters.


So.. not today, but I'd figure that in 10 years, you'll start to see
real broadband phased arrays (not just 4-8 switched beams in a single
band like a 4 square).

Yes, I think maybe so. We already have most of the technology for an
advanced manually steerable RX array, so it's mainly a matter of
integrating the separate parts of it to make a practical design. Someone
just has to decide to do it... and maybe they already have.


--

73 from Ian GM3SEK

On Mar 1, 11:58 am, Ian White GM3SEK wrote:




"the magic box in the center of the array" is more the former model,
and while simple conceptually, in the long run probably isn't the best
way to solve the problem in a system context.

That is very true. Unfortunately, the ham market is divided up into
physically separate compartments of transceivers, power amplifiers and
antennas. That is a severe restriction which makes all the technical
challenges much more difficult.

Dare I say that what is needed is that much overworked phrase
"paradigm shift"? It used to be that a "transceiver" was a radical
new thing, because everyone knew that you needed a separate Tx and
Rx. Oddly, with phased arrays, perhaps the transceiver becomes passe.

However, we can try to pare the problem down a little.

Another important point is that the concept of 'market demand' is
beginning to break down in ham radio. The big manufacturers are
increasingly challenged by new products that pay no attention to the
market - they spring directly from some individual or small team
deciding they're going to do it.

I think this has actually been the case for decades. I doubt, for
instance, that the KWM-1 was motivated by some massive pentup demand
for an integrated SSB transceiver.

Then maybe the design is produced as a kit, or manufacturing is taken up
by some lower-tier company that is faster on its feet. Seems good to
me...

This would be true of many things.. the SDR1000 might be an example. A
variety of TAPR kits might be another.



For instance, a multi
channel receiver, which can do all the phasing, very precisely, at low
levels, either with analog or digital processing, can give you the
nice deep nulls and adaptation.For TX, though, null depth isn't as
important as maximizing the power squirted in the right direction.

That seems a good place to cut the problem down to size. By all means
continue to use the existing phasing networks for TX, with 4 or 8
switchable directions and fixed phasing; but switch the antennas over to
a totally separate network for RX.

In fact, for HF, you can probably get away with smaller active
antennas for receive. There's no particular reason why the Tx
antennas and the Rx antennas have to be the same, since you're not
typically receiver noise figure limited. There IS a strong signal IM
problem..so maybe active receive antennas aren't the right solution.
But, there's a lot of convenience possible if your receive antennas
are all something like 6 foot whips.

At the lower power levels, the RX network could be much more complex and
versatile, combining the signals from the four (say) antennas with
amplitudes and phasing that could be varied on the fly.

You can use nifty things like the 4 quadrant vector multipliers from
Maxim, for instance.


Another way to scale down the problem is not to be too ambitious about
automatic null steering. In ham operating it is often difficult for a
computer to identify which is the wanted signal and which is the QRM, so
maybe let's not try. Semi-automatic null steering definitely would be
within reach, where the user has a control to steer the null direction
manually for the best audible results, and the computer does the math to
select the required network parameters.

That would be where I would start. Adaptive nulling is a bit weird to
work with as a user, especially if you expect to control it. And, for
hams, they want a bit more control. What would be cool is to have a
3D panoramic display that somehow indicates not only the frequency
spectrum, but the angle of arrival.
So.. not today, but I'd figure that in 10 years, you'll start to see
real broadband phased arrays (not just 4-8 switched beams in a single
band like a 4 square).

Yes, I think maybe so. We already have most of the technology for an
advanced manually steerable RX array, so it's mainly a matter of
integrating the separate parts of it to make a practical design. Someone
just has to decide to do it... and maybe they already have.

I've got most of both halves (steerable Tx array and steerable Rx
array) in pieces, but there's a lot of practical usability and
integration issues. Sure would be nice to have a wealthy patron and
have lots of free time and a big budget to work on itgrin. I
developed the in-situ calibration algorithms as part of a R&D effort
at JPL, but the resources eventually ran out. Too many projects and
too little time.

Jim, W6RMK

On 1 Mar 2007 21:16:22 -0800, wrote:

Dare I say that what is needed is that much overworked phrase
"paradigm shift"? It used to be that a "transceiver" was a radical
new thing, because everyone knew that you needed a separate Tx and
Rx. Oddly, with phased arrays, perhaps the transceiver becomes passe.

Hi Jim,

There may be life left in the transceiver to perform what you want
anyway. Just think more power.

I can sense that this discussion is leading back to a separation of
driver from amplifier. You could still use a power divider to feed
the remote amps at the various array locations. Say 10W at 10 active
antennas, each with a 100W rating.

This would preserve investment, and create and alternative to the
Henry market. Hams would have two purchase paths instead of
discarding their introductory base station and opting in for N number
of active arrays driven by a specialty item that looks like their old
rig gathering dust in the corner.

Of course, if you are talking about two or three elements, this shifts
the investment plan (but it still presumes you are willing to abandon
a prior investment). The cost-benefit of this is usually a steep
slope populated only by a few early adopters.


In fact, for HF, you can probably get away with smaller active
antennas for receive. There's no particular reason why the Tx
antennas and the Rx antennas have to be the same, since you're not
typically receiver noise figure limited.

There's no reason to expect you need a large transmit antenna for
solid state amps either. The native source resistance of a transistor
is quite low, and has to be transformed UP to match 50 Ohms. If you
had a radiation resistance of only several ohms (a very short
radiator) all you have to pay attention to is cutting Ohmic loss and
providing flexible inductance. Unfortunately, this may be a
performance killer - but if you are demanding multiband performance,
you will have to answer for this for any size array element.

There IS a strong signal IM
problem..so maybe active receive antennas aren't the right solution.

The same transceiver that survives IM would still handle it from
several phase active array elements. As you can see, redesigning a
new driver eventually leads you back to the gear you have. If you now
demand a separate receiver, separate driver, and separate active array
antennas, costs rise faster by the number of connections.

Adaptive nulling is a bit weird to
work with as a user, especially if you expect to control it. And, for
hams, they want a bit more control.

Programmable oscillators that shift immediately and start at any point
in the cycle (absolute phase AND frequency control) would be miles
further down the road.

These devices are NOT synthesizers NOR linear multipliers. They are
variable frequency look-up tables driving a DAC. You send the chip a
command of the frequency you want, and the phase angle; send the "go"
command, and the oscillator shifts from its existing frequency and
phase immediately to the new one (immediately being within one table
cell lookup).

When I looked into these 10 years ago, they (driver/DAC combination)
operated from the KHz up to about 16MHz and cost less than one XTAL
ordered to a single frequency. You still have the elemental clock osc
XTAL for processor and driver, but those litter the world for pennies.

Using this kind of technology is about the only reason I would trash
my current rigs and go for a custom driver (but it would have to
inhabit each array antenna's amp and thus render it a complete
transmitter. This would in turn cast all the features (like SSB
generation) into each of those elements. The unit cost of this would
climb because of feature creep, not component pricing.

What would be cool is to have a
3D panoramic display that somehow indicates not only the frequency
spectrum, but the angle of arrival.

This again, argues for small adaptive antennas. Now you need them in
layers.

73's
Richard Clark, KB7QHC

In fact, for HF, you can probably get away with smaller active
antennas for receive. There's no particular reason why the Tx
antennas and the Rx antennas have to be the same, since you're not
typically receiver noise figure limited.

Adaptive nulling is a bit weird to
work with as a user, especially if you expect to control it. And, for
hams, they want a bit more control.

Using this kind of technology is about the only reason I would trash
my current rigs and go for a custom driver (but it would have to
inhabit each array antenna's amp and thus render it a complete
transmitter. This would in turn cast all the features (like SSB
generation) into each of those elements. The unit cost of this would
climb because of feature creep, not component pricing.

What would be cool is to have a
3D panoramic display that somehow indicates not only the frequency
spectrum, but the angle of arrival.


When living in a residential area, as I do, canceling or reducing local qrm
and man made noise is your first interest!
Using an array is may be the only way to go. Either by using analog means
(phasers) or using digital signal processing. In DSP a lot more is possible.

I use an array of 2 small active loop antennas.
The two phase coherent receivers are made using the front-ends of two
Elecraft K2´s and a Delta 44 soundcard in the PC. The software makes
the rest of the dual (SDR) receiver. No expensive hardware needed.
See: http://www.pa0sim.nl/Software.htm
(one of the windows shows the phase difference (TDOA), which helps
eg. selecting the correct phase for the phaser)

Even two antennas can be very effective against local qrm and even non
local qrm. I sure would like to try four antennas HI. It is easy to scale it
to four antennas/receivers, but I still have my qrl .....

73 de Jan PA0SIM

On 2 Mar, 10:41, "Jan Simons PA0SIM"
wrote:
In fact, for HF, you can probably get away with smaller active
antennas for receive. There's no particular reason why the Tx
antennas and the Rx antennas have to be the same, since you're not
typically receiver noise figure limited.

Adaptive nulling is a bit weird to
work with as a user, especially if you expect to control it. And, for
hams, they want a bit more control.

Using this kind of technology is about the only reason I would trash
my current rigs and go for a custom driver (but it would have to
inhabit each array antenna's amp and thus render it a complete
transmitter. This would in turn cast all the features (like SSB
generation) into each of those elements. The unit cost of this would
climb because of feature creep, not component pricing.

What would be cool is to have a
3D panoramic display that somehow indicates not only the frequency
spectrum, but the angle of arrival.

When living in a residential area, as I do, canceling or reducing local qrm
and man made noise is your first interest!
Using an array is may be the only way to go. Either by using analog means
(phasers) or using digital signal processing. In DSP a lot more is possible.

I use an array of 2 small active loop antennas.
The two phase coherent receivers are made using the front-ends of two
Elecraft K2´s and a Delta 44 soundcard in the PC. The software makes
the rest of the dual (SDR) receiver. No expensive hardware needed.
See:http://www.pa0sim.nl/Software.htm
(one of the windows shows the phase difference (TDOA), which helps
eg. selecting the correct phase for the phaser)

Even two antennas can be very effective against local qrm and even non
local qrm. I sure would like to try four antennas HI. It is easy to scale it
to four antennas/receivers, but I still have my qrl .....

73 de Jan PA0SIM- Hide quoted text -

- Show quoted text -

Jan,
the above is beyond me. I see it as an ideal addition to a Gaussian
array where you have two receive points or more on the antenna such
that the receiver could make the best choice which is polarity of
choice. At the same time allows you to disconnect two feed points to
concentrate on the desired feed. Of course all feeds can be connected
for recieve anyway to counteract fade. I believe that there is a niche
in ham radio for what you have there.
Thanks for sharing
Cheers and beers
Art

On Mar 1, 10:32 pm, Richard Clark wrote:

I can sense that this discussion is leading back to a separation of
driver from amplifier. You could still use a power divider to feed
the remote amps at the various array locations. Say 10W at 10 active
antennas, each with a 100W rating.

This probably the most practical thing.. and in fact what I'm working
on prototyping. The challenge is in the amplifiers (although the
recent FCC ruling doing away with the anti-CB amplifier rules will
help)

Most inexpensive amplifiers do not have "well behaved" properties at
RF (i.e. they change all their parameters as they heat up).. this
makes dealing with phasing and mutual coupling a bit challenging If
the output Z changes, then the network matching it to the element
needs to change..if the gain and phase through the amp changes, then
the drive needs to be adjusted.

Fortunately, the problems are solvable, at least in a theoretical
sense.


This would preserve investment, and create and alternative to the
Henry market. Hams would have two purchase paths instead of
discarding their introductory base station and opting in for N number
of active arrays driven by a specialty item that looks like their old
rig gathering dust in the corner.

Yes.. this is exactly the growth path I would envision.


There's no reason to expect you need a large transmit antenna for
solid state amps either. The native source resistance of a transistor
is quite low, and has to be transformed UP to match 50 Ohms. If you
had a radiation resistance of only several ohms (a very short
radiator) all you have to pay attention to is cutting Ohmic loss and
providing flexible inductance. Unfortunately, this may be a
performance killer - but if you are demanding multiband performance,
you will have to answer for this for any size array element.

Indeed, yes.. fortunately, you already need to have an adjustable
impedance transformer (because the feed point Zs change with frequency/
steering), and even more fortunately, you don't need a broadband
match.. All you need is a few kHz, so a single L and C might do it,
"good enough".

There IS a strong signal IM
problem..so maybe active receive antennas aren't the right solution.

The same transceiver that survives IM would still handle it from
several phase active array elements. As you can see, redesigning a
new driver eventually leads you back to the gear you have. If you now
demand a separate receiver, separate driver, and separate active array
antennas, costs rise faster by the number of connections.

But it turns out that you want a different kind of phasing for receive
than for transmit (not only a different pattern, but it turns out
you'd like to do it a different way... null formation being one
reason).. that pushes you away from a simple adjustable LC phasing
network for the receive array. For receive, you'd also like it fast
(and, potentially, multiple beams at once).


Adaptive nulling is a bit weird to
work with as a user, especially if you expect to control it. And, for
hams, they want a bit more control.

Programmable oscillators that shift immediately and start at any point
in the cycle (absolute phase AND frequency control) would be miles
further down the road.

That's available now.. it's a DDS. And you can buy a radio off the
shelf that has these capabilities (the FlexRadio SDR1000), although,
there are a couple difficulties with the flex (for one, they didn't
bring the sync input to the DDS out to the connector). You can buy a
eval board from Analog Devices for $200.

ordered to a single frequency. You still have the elemental clock osc
XTAL for processor and driver, but those litter the world for pennies.

Interestingly, in a very much higher budget arena (deep space comms
and ranging), they're also doing this. Until recently, you had to
special order the crystal for your spacecraft radio (with 18+ month
lead time!!), so if you had a channel reassignment, it was a real
problem. (One of the Mars Rovers and the Mars Reconnaissance Orbiter
are on the same channel.. wasn't supposed to be a problem because the
rover was not planned to survive long enough)




Using this kind of technology is about the only reason I would trash
my current rigs and go for a custom driver (but it would have to
inhabit each array antenna's amp and thus render it a complete
transmitter. This would in turn cast all the features (like SSB
generation) into each of those elements. The unit cost of this would
climb because of feature creep, not component pricing.

Kind of depends where you divide up the building blocks. The raw
exciter cost (DDS plus mod plus D/A converter for modulation) is quite
low.


Jim

On 2 Mar 2007 11:33:57 -0800, wrote:

Hi Jim

Most inexpensive amplifiers do not have "well behaved" properties at
RF (i.e. they change all their parameters as they heat up).. this
makes dealing with phasing and mutual coupling a bit challenging If
the output Z changes, then the network matching it to the element
needs to change..if the gain and phase through the amp changes, then
the drive needs to be adjusted.

If you are interested in a design for yourself, and maybe a production
run of a few hundred, then look to the HF/10M/6M/2M surplus repeater
market from Motorola and RCA designs of the 70s and 80s. Dirt cheap
decks with enough elbow room to make mods.

Fortunately, the problems are solvable, at least in a theoretical
sense.

One doesn't usually see fortunately paired with theoretical.
ironically this is how this thread started - was with a pig in the
poke explanation passing for science

All you need is a few kHz, so a single L and C might do it,
"good enough".

So, this daydream is on one band only?

But it turns out that you want a different kind of phasing for receive
than for transmit (not only a different pattern, but it turns out
you'd like to do it a different way... null formation being one
reason).. that pushes you away from a simple adjustable LC phasing
network for the receive array. For receive, you'd also like it fast
(and, potentially, multiple beams at once).

Ah, there's always a big but (as PeeWee Herman would say).

Programmable oscillators that shift immediately and start at any point
in the cycle (absolute phase AND frequency control) would be miles
further down the road.

That's available now.. it's a DDS. And you can buy a radio off the
shelf that has these capabilities (the FlexRadio SDR1000), although,
there are a couple difficulties with the flex (for one, they didn't
bring the sync input to the DDS out to the connector). You can buy a
eval board from Analog Devices for $200.

OK, Analog Devices is a star performer. I built a tube version of
this DDS back in '68 when it was called a coherent detector (could
have been called many names depending on where you developed the audio
output).

However, this is NOT what I was referring to, as that is distinctly
different. This is a software controlled oscillator whose frequency
and phase is immediately settable (within on clock, this is in the
nanosecondS range). If you need to phase each array element
independently to phase steer the combined system (also to take care of
phase matching through mutual coupling), the software solution spring
immediately to the front for a solution.

However, as I pointed out, it means N individual drivers following
those oscillators so that each can be phase controlled for the system
combination. The downside finds our current investment sitting in the
corner again, and our investing more design effort into SSB
generation, and hence, feature creep.

ordered to a single frequency. You still have the elemental clock osc
XTAL for processor and driver, but those litter the world for pennies.

Interestingly, in a very much higher budget arena (deep space comms
and ranging), they're also doing this. Until recently, you had to
special order the crystal for your spacecraft radio (with 18+ month
lead time!!), so if you had a channel reassignment, it was a real
problem. (One of the Mars Rovers and the Mars Reconnaissance Orbiter
are on the same channel.. wasn't supposed to be a problem because the
rover was not planned to survive long enough)

The software oscillator I described in the previous post would solve
that for the same cost as the custom XTAL.

If you are looking for a design and a market, I cannot think for the
life of me why that hasn't happened yet.

73's
Richard Clark, KB7QHC

On Mar 2, 12:15 pm, Richard Clark wrote:

If you are interested in a design for yourself, and maybe a production
run of a few hundred, then look to the HF/10M/6M/2M surplus repeater
market from Motorola and RCA designs of the 70s and 80s. Dirt cheap
decks with enough elbow room to make mods.

Or, perhaps the HF Superpacker Pro.. 100+Watts..20dB or so gain.


Fortunately, the problems are solvable, at least in a theoretical
sense.

One doesn't usually see fortunately paired with theoretical.
ironically this is how this thread started - was with a pig in the
poke explanation passing for science

A heck of a lot better than "theoretically there is NO solution..."
girn

All you need is a few kHz, so a single L and C might do it,
"good enough".

So, this daydream is on one band only?

1 adjustable L, 1 adjustable C.. covers all bands. For ham
applications, you only need to have the match and phase adjust at one
frequency at a time, and over a fairly small bandwidth, so the Q of
the matching circuit can be high. Not like a generic wideband phased
array where you need to have a wideband match, and do something like
true-time-delay processing.



OK, Analog Devices is a star performer. I built a tube version of
this DDS back in '68 when it was called a coherent detector (could
have been called many names depending on where you developed the audio
output).

However, this is NOT what I was referring to, as that is distinctly
different. This is a software controlled oscillator whose frequency
and phase is immediately settable (within on clock, this is in the
nanosecondS range).

That is precisly what a DDS does.. It has a phase accumulator where
you can adjust the phase increment per clock. The output of the phase
accumulator goes to a sine lookup table and then to a DAC. A typical
part might have a 48 bit phase accumulator and a 12 bit DAC.

Check out, for instance, the AD9858..

You might not be able to update the phase increment or absolute phase
in a nanosecond, but it's pretty fast (there's some pipeline delay
too.).. I'd say you could clock in a new configuration and have the
new RF appear no more than 100ns later.

If you need to phase each array element
independently to phase steer the combined system (also to take care of
phase matching through mutual coupling), the software solution spring
immediately to the front for a solution.

Well, software for the calculations, but perhaps not for all the RF
processing. You still need to adjust Ls and Cs for the match, unless
you're willing to design a fairly unusual amplifier: ideally, it would
act like a current source with a lot of voltage compliance that can
tolerate a very reactive load.. so you're essentially synthesizing the
L and/or C with an active device.. doable, but not too hot on power
efficiency these days.

Interestingly, in a very much higher budget arena (deep space comms
and ranging), they're also doing this. Until recently, you had to
special order the crystal for your spacecraft radio (with 18+ month
lead time!!), so if you had a channel reassignment, it was a real
problem. (One of the Mars Rovers and the Mars Reconnaissance Orbiter
are on the same channel.. wasn't supposed to be a problem because the
rover was not planned to survive long enough)

The software oscillator I described in the previous post would solve
that for the same cost as the custom XTAL.

Well.. not the same cost as a custom rock. The rock may have a long
delivery time, but it's not particularly more expensive than a
standard frequency. The DDS doesn't come for zero power, either, so
you have a tradeoff of more DC power for flexibility (power
consumption is a very big deal in deep space exploration). And, then,
there's also the whole radiation tolerance issue.


If you are looking for a design and a market, I cannot think for the
life of me why that hasn't happened yet.

Systems like this exist, but not in the ham market. Oddly, it seems
that hams balk at forking out more than $100K for a system that does
all this. The challenge is in getting it in a ham friendly format.
The hardware's not expensive, it's all that software and system
integration. But, 10 years from now, when the early adopters have
cobbled together their systems from bits and pieces, and some of the
concepts become more familiar, I think you'll see someone make the
investment to do the development to create an off the shelf product.
Clearly, there is SOME market for $10K ham widgets or things like
IC7800s and big towers with arrays of SteppIRs wouldn't exist.





73's
Richard Clark, KB7QHC