An idea for the K2

[WA2JJH]

I use my K2 on CW and SSB. I was wondering how K2 Hams would feel about  
different
grades of SSB adapters.
 
  For $89 the SSB board certainly gives you the most bang for the  buck.
I see a 2.6khz up grade option is offered.
 
  I am working on a true RF proc. I may have to buy a second SSB  board.
After all, if all my mods destroy operation, I would like to have an extra  
SSB board to get back to normal
 
I noticed a larger SSB board would fit. I am going to experiment on the SSB  
board.
I am trying to get an RF speech proc. AM mod/demod . perhaps a socket to  
take an auxillary group of xtals. for a 2nd 7 pole filter, Perhaps even  passband
or IF shift.
 
  I am willing to spend $89 to see if I can build a more deluxe SSB  board .
 
 
_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Geoffrey Mackenzie-Kennedy-2]

Intersting ideas. If you did enlarge the SSB board to get all the bits in,
in which directions would you enlarge the exixting board?
IMHO it would be essential to contain the "new" board in a 'tin' box in turn
mounted on metal, which raises the question of access to test points etc on
the RF board, but a solution might be available.

If you are considering some form of RF Clipping processing, you will need a
second 'IF Filter' anyway. Although some people have got away with it, it is
not a good idea to clip a DSB signal as produced by the balanced modulator.
A lot of weird gurgles can result.

As far as basic SSB 3db bandwidth is concerned, my personal preference when
using small ladder filters, less than 8 crystals, is for a bandwidth between
2.0 kHz and 2.2 kHz. Although wider filters can sound nice, usually the
filter's lower sideband skirt suffers to some degree (upper skirt if
crystals in shunt connection). The result can be anti-social. Having said
that, it is possible to build ladder filters with bandwidths around 5kHz at
5MHz, but various circuit trickery is involved and the skirts are a bit
wide.

Good luck with the project.

73,
Geoff     GM4ESD

----- Original Message -----
From: <[hidden email]>
To: <[hidden email]>
Sent: Tuesday, March 01, 2005 10:07 PM
Subject: [Elecraft] An idea for the K2
>
>   For $89 the SSB board certainly gives you the most bang for the  buck.
> I see a 2.6khz up grade option is offered.
>
> I noticed a larger SSB board would fit. I am going to experiment on the
SSB
> board.
> I am trying to get an RF speech proc. AM mod/demod . perhaps a socket to
> take an auxillary group of xtals. for a 2nd 7 pole filter, Perhaps even
passband
> or IF shift.
>

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Geoffrey Mackenzie-Kennedy-2]

This is a horse of a very different colour!!  We are now talking 'RFCompression/Expansion' rather than 'RFClipping' after the modulator, a little bit like AGC in the receiver or ALC in the Tx, but with low level signals boosted. One approach is to generate a good SSB RF signal (modulator plus the 7 crystal filter) and then play with Compression / Expansion of the RF signal, which can be difficult. One method is to "fast detect" the level of SSB signal and pass back a control voltage to a variable gain amplifier in the audio circuits. Another and prefered method is to use a "Forward AGC" circuit block all at the SSB IF. The VERY BIG danger with RF Compression /Expansion systems is that all the time constants have to be absolutely right otherwise you can have a mess or oscillation. Audio Companders are easier. For RF processing, RF Clipping is very much simpler. If I had to make a choice and did not have time nor the right test equipment, I would go for RF Clipping with two filters. The BIG filter at the output. Perhaps a properly designed 3 or 4 crystal filter after the modulator so that a SSB signal is clipped might be enough, but the results would have to be checked out with a RF Spectrum Analyser or equivalent along with means to check audio quality.

73,
Geoff     GM4ESD
_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Stewart Baker]

Any form of increasing the mic gain internally (eg rf speech clipper) needs to
be carefully thought out. The microphone signal routing and earthing from the
front panel socket thru the RF board and into the KSB2 is far from ideal.
Common mode, series mode - the lot - RF feedback is the likely result.

A groundplane board with the crystal filter laid out to reduce blow-by may be
the way to go. A tin can over the whole assembly may help. The biggest hurdle to
overcome in making changes to the KSB2 is interfacing with the existing firmware
which is far from straightforward.

73
Stewart G3RXQ

On Wed, 2 Mar 2005 08:11:54 -0000, Geoffrey Mackenzie-Kennedy wrote:

> Intersting ideas. If you did enlarge the SSB board to get all the bits in,
>
> in which directions would you enlarge the exixting board?
> IMHO it would be essential to contain the "new" board in a 'tin' box in turn
> mounted on metal, which raises the question of access to test points etc on
> the RF board, but a solution might be available.
>
> If you are considering some form of RF Clipping processing, you will need a
> second 'IF Filter' anyway. Although some people have got away with it, it is
> not a good idea to clip a DSB signal as produced by the balanced modulator.
> A lot of weird gurgles can result.
>
> As far as basic SSB 3db bandwidth is concerned, my personal preference when
> using small ladder filters, less than 8 crystals, is for a bandwidth between
> 2.0 kHz and 2.2 kHz. Although wider filters can sound nice, usually the
> filter's lower sideband skirt suffers to some degree (upper skirt if
> crystals in shunt connection). The result can be anti-social. Having said
> that, it is possible to build ladder filters with bandwidths around 5kHz at
> 5MHz, but various circuit trickery is involved and the skirts are a bit
> wide.
>
> Good luck with the project.
>
> 73,
> Geoff     GM4ESD
>
> ----- Original Message -----
> From: <[hidden email]>
> To: <[hidden email]>
> Sent: Tuesday, March 01, 2005 10:07 PM
> Subject: [Elecraft] An idea for the K2
>
>> For $89 the SSB board certainly gives you the most bang for the  buck.
>> I see a 2.6khz up grade option is offered.
>>
>> I noticed a larger SSB board would fit. I am going to experiment on the
> SSB
>> board.
>> I am trying to get an RF speech proc. AM mod/demod . perhaps a socket to
>> take an auxillary group of xtals. for a 2nd 7 pole filter, Perhaps even
> passband
>> or IF shift.
>>
>
> _______________________________________________
> Elecraft mailing list
> Post to: [hidden email]
> You must be a subscriber to post to the list.
> Subscriber Info (Addr. Change, sub, unsub etc.):
> http://mailman.qth.net/mailman/listinfo/elecraft
>
> Help: http://mailman.qth.net/subscribers.htm
> Elecraft web page: http://www.elecraft.com

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[John Magliacane]

--- [hidden email] wrote:

> I use my K2 on CW and SSB. I was wondering how K2 Hams would feel about  
> different
> grades of SSB adapters.

I would sharpen up the skirts on the audio chain (300 - 2700 Hz) with
active filters prior to the compressor.  (There is no reason for any energy
outside this spectrum to be passed along (and influence the operation of)
the compressor and the balanced modulator, since the crystal filter is too
narrow to pass it, anyway.)

Add some gain (accomplished through the active filtering stated above), and
provide a mic gain control (on the board).

It might be nice to replace the NE602/NE612 with a switching-style
balanced modulator for the following reasons:

* The BFO (VXO) injection voltage into the NE612 varies with the BFO
setting and USB/LSB switching.  The NE612, being a linear mixer, is
sensitive to this.  It's DSB output voltage varies according to the BFO
injection voltage, as does it's carrier suppression.

* A well-designed switching mixer is inherently self-balanced (assuming
matched resistors, minimal stray capacitance, and low offset voltage
OP-AMPs are used), eliminating the need for the carrier balance control
and one less adjustment for builders.

* More dynamic range (limited only by the output swing of the audio feeding
the modulator and the supply voltage to the analog switches), leading to
even less SSB TX distortion (IMD) and carrier suppression than is
possible now.

Good luck with the project!


73, de John, KD2BD


=====
Visit John on the Web at:

        http://www.qsl.net/kd2bd/


       
               
__________________________________
Celebrate Yahoo!'s 10th Birthday!
Yahoo! Netrospective: 100 Moments of the Web
http://birthday.yahoo.com/netrospective/
_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Geoffrey Mackenzie-Kennedy-2]

I believe that you would be proven correct about the RF feedback.

My thought about a tin box totally enclosing the PCB stems from an
observation that the skirts of the varicap "controlled" filter set for say
700Hz bandwidth are a little wider with the KSB2 installed than when it it
is not, and noticeable from approx 40db below passband on down. I made no
measurements because I was Birdy hunting/killing at the time. If this
blow-by is due to leakage between P2 and P3 or through the switching diodes
(off) on the KSB2, then the tin box would have little if any effect. But the
KSB2 does come close to forming a bypass route just due to its proximity to
the post mixer amp, RF board filter etc, and my gut feeling is that a box
around the KSB2 or its replacement would be a good start. The box would also
help to keep out the RF from nearby oscillators.

73,
Geoff    GM4ESD

On Wed.2nd March 2005, Stewart G3RXQ wrote:

Any form of increasing the mic gain internally (eg rf speech clipper) needs
to
be carefully thought out. The microphone signal routing and earthing from
the
front panel socket thru the RF board and into the KSB2 is far from ideal.
Common mode, series mode - the lot - RF feedback is the likely result.

A groundplane board with the crystal filter laid out to reduce blow-by may
be
the way to go. A tin can over the whole assembly may help. The biggest
hurdle to
overcome in making changes to the KSB2 is interfacing with the existing
firmware
which is far from straightforward.

73
Stewart G3RXQ

On Wed, 2 Mar 2005 08:11:54 -0000, Geoffrey Mackenzie-Kennedy wrote:
> Intersting ideas. If you did enlarge the SSB board to get all the bits in,
>
> in which directions would you enlarge the exixting board?
> IMHO it would be essential to contain the "new" board in a 'tin' box in
turn
> mounted on metal, which raises the question of access to test points etc
on
> the RF board, but a solution might be available.
>





_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Bill Coleman-2]

On Mar 1, 2005, at 5:07 PM, [hidden email] wrote:

> I use my K2 on CW and SSB. I was wondering how K2 Hams would feel about
> different grades of SSB adapters.

I've voiced this several times, but I'll reiterate.

I think the KSB2 is a great value at $89. It works, and gives
reasonable performance. My K2 is probably unique in that I am using the
"1.9" filter values as designed by KI6WX. (the bandwidth is closer to
2.0 - slightly narrower than the original filter values, but (most
importantly) with a lower passband ripple).

However, the KSB2 leaves a little bit to be desired.

1) it doesn't have a great VOX. The VOX works better with higher
compression levels. If your mike has a low output, it may not trip.
Although the K2 has enough overall gain (and there's another KI6WX mod
to increase the tx gain), a bit of mike amplification would be an
improvement.

2) the filter passband is kinda sloppy. It's actually amazingly good
for its use of seven inexpensive crystals. Most radios use precision
crystal filters that cost as much as the KSB2 or more. But, I'd be
willing to pop for a more expensive unit that had a quality SSB filter
with steep skirts and very little passband ripple.

3) single filter bandwidth. Even though the K2 has OP1-5, indicating
that perhaps there are selections that can be made, there's only one
filter. Two filter bandwidths would be nice - a wide and narrow, like
most radios.

Of course, there is very limited space in the K2 for the KSB2 or
filters. The KSB2 is already the hardest module to build, with the
highest component density.


Bill Coleman, AA4LR, PP-ASEL        Mail: [hidden email]
Quote: "Not within a thousand years will man ever fly!"
             -- Wilbur Wright, 1901

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Stewart Baker]

Probably as you say that a tin box around the whole assembly would be an
overkill. I was thinking that the crystals/caps in the filter could be spaced
out more, and the whole area ground-planed top and bottom. I agree that coupling
to the main board maybe a route for blow-by, and if space could be found a
screen fitted to the underside of the board may bring some improvement. It
should be possible to estimate the degree of leakage caused by P2/P3, and the
signal diodes. Maybe PIN diodes would give higher isolation.
With the excellent phase noise performance of the K2 it is a shame that some of
this benefit may be being thrown away in the SSB filter. I am given to
understand by people who know a lot more than me about ladder filters that there
is more to be had.

73
Stewart G3RXQ

 On Wed, 2 Mar 2005 18:54:54 -0000, Geoffrey Mackenzie-Kennedy wrote:

> I believe that you would be proven correct about the RF feedback.
>
> My thought about a tin box totally enclosing the PCB stems from an
> observation that the skirts of the varicap "controlled" filter set for say
> 700Hz bandwidth are a little wider with the KSB2 installed than when it it
> is not, and noticeable from approx 40db below passband on down. I made no
> measurements because I was Birdy hunting/killing at the time. If this
> blow-by is due to leakage between P2 and P3 or through the switching diodes
> (off) on the KSB2, then the tin box would have little if any effect. But the
> KSB2 does come close to forming a bypass route just due to its proximity to
> the post mixer amp, RF board filter etc, and my gut feeling is that a box
> around the KSB2 or its replacement would be a good start. The box would also
> help to keep out the RF from nearby oscillators.
>
> 73,
> Geoff    GM4ESD
>
> On Wed.2nd March 2005, Stewart G3RXQ wrote:
>
> Any form of increasing the mic gain internally (eg rf speech clipper) needs
> to
> be carefully thought out. The microphone signal routing and earthing from
> the
> front panel socket thru the RF board and into the KSB2 is far from ideal.
> Common mode, series mode - the lot - RF feedback is the likely result.
>
> A groundplane board with the crystal filter laid out to reduce blow-by may
> be
> the way to go. A tin can over the whole assembly may help. The biggest
> hurdle to
> overcome in making changes to the KSB2 is interfacing with the existing
> firmware
> which is far from straightforward.
>
> 73
> Stewart G3RXQ
>
> On Wed, 2 Mar 2005 08:11:54 -0000, Geoffrey Mackenzie-Kennedy wrote:
>> Intersting ideas. If you did enlarge the SSB board to get all the bits in,
>>
>> in which directions would you enlarge the exixting board?
>> IMHO it would be essential to contain the "new" board in a 'tin' box in
> turn
>> mounted on metal, which raises the question of access to test points etc
> on
>> the RF board, but a solution might be available.

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Stewart Baker]

 On Wed, 2 Mar 2005 08:00:59 -0800 (PST), John Magliacane wrote:

>
>
> --- [hidden email] wrote:
>
>> I use my K2 on CW and SSB. I was wondering how K2 Hams would feel about
>> different
>> grades of SSB adapters.
>>
> I would sharpen up the skirts on the audio chain (300 - 2700 Hz) with
> active filters prior to the compressor.  (There is no reason for any energy
> outside this spectrum to be passed along (and influence the operation of)
> the compressor and the balanced modulator, since the crystal filter is too
> narrow to pass it, anyway.)
>>
>> Reducing the values of the coupling capacitors in the audio chain makes a big
>> difference (what purpose an audio response that goes down to 30Hz ?!)  
>
> Add some gain (accomplished through the active filtering stated above), and
> provide a mic gain control (on the board).
>
>> Get rid of the mic attenuator circuitry (does anyone use it ?) and add an
>> adjustable mic gain of 10dB (well rf filtered) to the board.

>> Replace the existing compressor IC with a later version that has an
>> adjustable threshold level.

>> Totally revamp the VOX circuitry and add Anti-VOX. Make both hardware
>> adjustable.

> It might be nice to replace the NE602/NE612 with a switching-style
> balanced modulator for the following reasons:
>
> * The BFO (VXO) injection voltage into the NE612 varies with the BFO
> setting and USB/LSB switching.  The NE612, being a linear mixer, is
> sensitive to this.  It's DSB output voltage varies according to the BFO
> injection voltage, as does it's carrier suppression.
>
> * A well-designed switching mixer is inherently self-balanced (assuming
> matched resistors, minimal stray capacitance, and low offset voltage
> OP-AMPs are used), eliminating the need for the carrier balance control
> and one less adjustment for builders.
>
> * More dynamic range (limited only by the output swing of the audio feeding
> the modulator and the supply voltage to the analog switches), leading to
> even less SSB TX distortion (IMD) and carrier suppression than is
> possible now.
>
>> I would like to know more about using a switching mixer for the BM.
>> If the circuitry does not get too complex or introduce problems it would be
>> worth trying. Please let me have more information.

>> 73
>> Stewart G3RXQ

> Good luck with the project!
>
>
> 73, de John, KD2BD
>
>
> =====
> Visit John on the Web at:
>
> http://www.qsl.net/kd2bd/
>
>
> __________________________________
> Celebrate Yahoo!'s 10th Birthday!
> Yahoo! Netrospective: 100 Moments of the Web
> http://birthday.yahoo.com/netrospective/
> _______________________________________________
> Elecraft mailing list
> Post to: [hidden email]
> You must be a subscriber to post to the list.
> Subscriber Info (Addr. Change, sub, unsub etc.):
> http://mailman.qth.net/mailman/listinfo/elecraft
>
> Help: http://mailman.qth.net/subscribers.htm
> Elecraft web page: http://www.elecraft.com

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Stewart Baker]

 On Wed, 2 Mar 2005 20:45:11 -0500, Bill Coleman wrote:

>
>
> On Mar 1, 2005, at 5:07 PM, [hidden email] wrote:
>
>> I use my K2 on CW and SSB. I was wondering how K2 Hams would feel about
>> different grades of SSB adapters.
>>
> I've voiced this several times, but I'll reiterate.
>
> I think the KSB2 is a great value at $89. It works, and gives
> reasonable performance. My K2 is probably unique in that I am using the
> "1.9" filter values as designed by KI6WX. (the bandwidth is closer to
> 2.0 - slightly narrower than the original filter values, but (most
> importantly) with a lower passband ripple).
>
> However, the KSB2 leaves a little bit to be desired.
>
> 1) it doesn't have a great VOX. The VOX works better with higher
> compression levels. If your mike has a low output, it may not trip.
> Although the K2 has enough overall gain (and there's another KI6WX mod
> to increase the tx gain), a bit of mike amplification would be an
> improvement.

>> I think that "it doesn't have a great VOX", is a bit of an understatement.
>> It is not adjustable in threshold, it is insensitive, and has no Anti-VOX.
>> It lets an otherwise excellent transceiver down.
>> 73
>> Stewart G3RXQ
>
> 2) the filter passband is kinda sloppy. It's actually amazingly good
> for its use of seven inexpensive crystals. Most radios use precision
> crystal filters that cost as much as the KSB2 or more. But, I'd be
> willing to pop for a more expensive unit that had a quality SSB filter
> with steep skirts and very little passband ripple.
>
> 3) single filter bandwidth. Even though the K2 has OP1-5, indicating
> that perhaps there are selections that can be made, there's only one
> filter. Two filter bandwidths would be nice - a wide and narrow, like
> most radios.
>
> Of course, there is very limited space in the K2 for the KSB2 or
> filters. The KSB2 is already the hardest module to build, with the
> highest component density.
>
>
> Bill Coleman, AA4LR, PP-ASEL        Mail: [hidden email]
> Quote: "Not within a thousand years will man ever fly!"
> -- Wilbur Wright, 1901
>
> _______________________________________________
> Elecraft mailing list
> Post to: [hidden email]
> You must be a subscriber to post to the list.
> Subscriber Info (Addr. Change, sub, unsub etc.):
> http://mailman.qth.net/mailman/listinfo/elecraft
>
> Help: http://mailman.qth.net/subscribers.htm
> Elecraft web page: http://www.elecraft.com

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[John Magliacane]

> I would like to know more about using a switching mixer for the BM.
> If the circuitry does not get too complex or introduce problems it would
> be worth trying. Please let me have more information.
>
> 73
> Stewart G3RXQ

Imagine that you apply audio to a transformer that has a center tapped
secondary winding.  (With respect to the center tap, the output of the
transformer will consist of two independent audio signals, equal in
amplitude, but 180 degrees out of phase with one another.)

Now imagine that you ground the center tap, and connect a SPDT switch to
the free ends of the secondary.  The center pole of the switch doesn't
connect to the transformer, but serves as the "output" of this circuit.

Finally, imagine that the switch is toggled back and forth at an RF
rate while audio is applied to the transformer.

The end result will be a double sideband suppressed carrier RF signal
on the output of the switch.

In real life, we use OP-AMPs to provide the two out-of-phase audio signals
instead of a transformer, and CMOS analog switches, such as a CD4066 or
CD4053 (or better) in place of the mechanical SPDT switch and apply RF
to the control pins.

The end result is the same.

I use this approach *A LOT* for doubly-balanced mixing, synchronous
detection, phase detection, and modulation through the MF spectrum with
excellent results.

A quick Google search didn't turn up many simple examples of this approach,
but "Bat Detectors" (ultrasonic downconverters used to chase bats), and
"Cave Radios" (LF SSB transceivers with magnetic loop antennas used for
underground explorers) that do appear on the net use them quite often.
I used a switching demodulator and modulator in my 1200 baud BPSK Pacsat
Modem (http://www.amsat.org/amsat/articles/kd2bd/Pacsat_Modem/).

More involved examples can be found in Software Defined Radio designs
utilizing Commutating Detectors (so-called Tayloe Detectors).

A high-quality DSB modulator is quite simple compared to these designs.


73, de John, KD2BD


=====
Visit John on the Web at:

        http://www.qsl.net/kd2bd/


       
               
__________________________________
Celebrate Yahoo!'s 10th Birthday!
Yahoo! Netrospective: 100 Moments of the Web
http://birthday.yahoo.com/netrospective/
_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com