PAUL HEWITT wrote:
>Greetings Ian
>I'm afraid the speed up circuit you
>mention does not work very well for the
>vacuum relays. It indeed does speed
>things up but the circuit doesn't
>release the relay fast enough for true
>QSK. The test setup I used is as
>follows; your circuit with a TIP-142 +
>50uf charge cap and an IRF-620 to pull
>it low. The IRF-620 has the gate loaded
>with 10k and is keyed with a TTL signal
>@10-20hz. The relay is a Kilovac HC-!
>with coil voltage of 26.5.
>
>Here's the bench mark for the relay
>keyed with the IRF-620 only and no
>clamping diode(.1uf 300v ceramic cap
>instead), 26.5vdc.
[...]
>
>Your circuit with 10uf;
>Make time, 1.44ms, same 600ns of bounce
>Release time, 1.48ms
>Remake time of NC contacts, (9vdc
>applied to contacts),10ms
>
That's the value I'm using for QSK. I'm not using either a clamping
diode or a .1uF, because a nice feature of this circuit is that the
speedup cap absorbs the spike on 'break'.
I measured pretty much the same timings as Paul did. Some relays bounced
on make, others didn't, but all were settled by 2.0ms... let's say
2.5-3.0 for safety against production spreads.
But before we talk about the meaning of these results, let's set some
timescales. The 'make' period of a single dot at 40wpm is 30ms. A
typical transceiver will delay the RF until 10ms after key-down, but it
will end the RF only a few ms after key-up. Therefore it's the
transceiver that is responsible for most of the dot-clipping on fast
QSK, not the amp.
In my amp, I allow 3.0ms for the relay, and then switch on the EBS so
the amp is completely ready in 4ms after receiving the PTT signal. This
switching all happens in the transceiver's dead time, and it allows for
some possible delays in getting the PTT signal out.
Paul is correct that key-up is the problem area - but not due to the
relay. After key-up, the PTT line typically goes high straight away, but
the RF runs on for a few milliseconds more. This RF overrun varies
between transceivers. To be quite certain the RF really has finished,
the amp has to hold the relay in for at least 10ms, so the relay's own
break time is not an issue. However, most transceivers give an AGC thump
just after key-up, so they lose this receive time anyway.
The remake time of 10ms is not a practical problem at any keying speed
the transceiver (or the operator) can support - see below.
>This slow release/remake time was
>discussed in the description by K6XX.
>The relay supply has to charge the cap
>before it will release, bigger
>cap=faster make=slower release.
>
That's not quite correct - a bigger cap does *not* give a faster make,
only a slower release. The purpose of the cap is to apply 2x the rail
voltage to the relay for the first few milliseconds. Beyond a certain
minimum, more capacitance will not give a higher voltage, so it won't
increase the make speed.
K6XX used a bigger capacitor than is really necessary to speed-up the
make for a typical relay coil... and then suffered the longer break
time. As well as the measurements, I did some modeling of the build-up
and decay of current, which gave a lot of insight into the capacitor
values required.
The trick is to use the *minimum* capacitor value that will give the
available speed-up on make, and this will give the fastest possible
break. On non-QSK amps, you can deliberately use larger values to extend
the break time.
>I don't believe you're going to hear
>very much between Morse characters with
>these remake times.
Sure you can - it's working right here with a pair of HC-1s. Bear in
mind that the purpose of QSK is not to regain the whole receive time
between dots (we're not doing radar here). It's only to make the
operator aware that *something* is happening on the frequency, so some
loss of receive time is tolerable.
With the transceivers I have here (FT-1000MP and IC-746) the limitations
on QSK are all in the clicks and thumps from the transceivers
themselves. The amp does its switching in the transceiver's dead times,
so it's completely transparent.
>This circuit could eliminate hot
>switching for the clunky open frame
>relays in older amps
Or VHF amps with coaxial relays.
Speeding-up regular antenna relays is probably *the* most important
application of the K1KP/K6XX circuit. If you can bring the make time
inside 10ms, you can eliminate hot switching with most modern
transceivers.
>The same relay keyed with a 33vdc
>un-regulated supply closes in 2.1ms and
>remakes in 2.05ms, no extra parts!!
>More than four dozen relays have been
>measured here with the same results,
>(ITT Jennings RJ1A was faster but died
>after 750K operations). The published
>make times of these relays are
>GUARANTEED maximums and include
>aerospace applications involving
>multiple G's. Typical make times here on
>earth are much less.
>
>Bottom line, measure it's speed first,
>it could be OK for your application
>right out of the box!
It could well be. The main reason I use the speed-up circuit with vacuum
relays is because it's already there, for the non-QSK amps. The other
reason is to provide extra time for transceivers with slow PTT outputs.
That brings me to one final warning about this whole speedup/QSK thing
(and I'm sure both Rich and Paul will agree)...
To make fast RF switching work, you *must* use the fastest possible PTT
output from the transceiver! Don't ever use the relay option - it wastes
valuable milliseconds that you can't afford. Some transceiver PTT relays
are so slow that the RF output has already started *before* the PTT
signal reaches the amp (duh...)
If the fast transistor-switched PTT output of your transceiver won't
handle the voltage/current needs of your amp, then modify the amp with a
fast transistor switch.
--
73 from Ian G3SEK 'In Practice' columnist for RadCom (RSGB)
Editor, 'The VHF/UHF DX Book'
http://www.ifwtech.co.uk/g3sek
|