EMRFD Message Archive 6481

Message Date From Subject
6481 2011-07-21 10:59:58 Bill Carver Re: Cartesian loop exciter - QSK
I am switching to PIN, but run 28V vacuum relays from a MOSFET current regulator and 200 VDC supply and total switching time, including bounce, is 1.6 mS. I'm allowing 2.0 mS and it's no impediment to QSK.

DSP processing is the biggest hit to QSK today, delay from input to output is many mS.

W7AAZ
6484 2011-07-21 15:51:25 Dan Mills Re: Cartesian loop exciter - QSK
On Thu, 2011-07-21 at 11:59 -0600, Bill Carver wrote:
> I am switching to PIN, but run 28V vacuum relays from a MOSFET current
> regulator and 200 VDC supply and total switching time, including
> bounce, is 1.6 mS. I'm allowing 2.0 mS and it's no impediment to QSK.

Have you seen the circuit by K5CX using a handful of bog standard
silicon rectifier diodes to do QSK at 1.5KW PEP? It uses a 6A6 and a
handful of 4007s with 350V or so reverse bias to hold them off and a few
tens of mA forward for on.

No idea what it does for RX IMD.....

But yea, QSK with relays is not actually that hard a problem at least at
any speed a mere mortal is going to manage.

> DSP processing is the biggest hit to QSK today, delay from input to
> output is many mS.

Depends on how you do it, but don't forget that the group delay through
a decent quartz filter can be several miliseconds without any DSP
involved.

With DSP you get about 2ms for the decimation filters in the converters
at the input and output, plus whatever the internal algorithms need.

Still QSK at reasonable speeds (20 - 30 WPM) should be entirely feasible
with some modest amount of DSP in the way.

Regards, Dan.
6485 2011-07-21 17:15:39 Bill Carver Re: Cartesian loop exciter - QSK
"
-----Original message-----
From: Dan Mills dmills@exponent.myzen.co.uk

Have you seen the circuit by K5CX using a handful of bog standard
silicon rectifier diodes to do QSK at 1.5KW PEP? It uses a 6A6 and a
handful of 4007s with 350V or so reverse bias to hold them off and a few
tens of mA forward for on.
No idea what it does for RX IMD....."

Yes, worked him several times, he was a great op. The "off" capacitance of the big rectifiers is pretty high. 40 pF or so for the 6A. 1N4007s are something like 10 pF, don't remember exact number now. "Real" PIN diodes are a few pF even with KW power capability, but essentially impossible to purchase them. But LTSPICE models tell me that the suckout, even with 40 pF, isn't too bad and far enough from the operating frequency that its not a major problem in the switch between amplifier and antenna.

Receiver isolation is degraded by diode capacitance, but guess multiple shunt-series 1N4007s will work. Some fairly astute folks say RX IMD is not a problem with a few hundred mils of forward current.

Your right about group delay through narrow filters.

W6JL has a rig with DDS LO, DC receiver, homebrew VFO driving a 6CL6/4CX250 500W amplifier....full QSK and bitched at my lid-like operating when I tweeked the tuner running 0.1W prior to sending "BK BK".
6486 2011-07-21 17:50:30 Dan Mills Re: Cartesian loop exciter - QSK
On Thu, 2011-07-21 at 18:15 -0600, Bill Carver wrote:

>
> 40 pF or so for the 6A. 1N4007s are something like 10 pF, don't
> remember exact number now. "Real" PIN diodes are a few pF even with KW
> power capability, but essentially impossible to purchase them. But
> LTSPICE models tell me that the suckout, even with 40 pF, isn't too
> bad and far enough from the operating frequency that its not a major
> problem in the switch between amplifier and antenna.

IIRC Ameritron had them available as a spare for their QSK board, not
cheap however.

None of the normal suppliers carry the bigger ones, but kits & parts
have some small examples listed.

40pf wont be an issue on topband, but could get a bit boring on 6M.

Actually the most complex thing with that circuit is probably making
sure none of the chokes resonate anywhere that matters (A bit like a
valve PA in that respect).

> Receiver isolation is degraded by diode capacitance, but guess
> multiple shunt-series 1N4007s will work.

Once you get a few diodes down the chain, you could probably go to
'4148s or something to get lower capacitance. Use the 4007s for the
first stage where there is real power to hold off.

> Your right about group delay through narrow filters.

One of the better DSP tricks is actually to implement a IIR allpass
network to equalise the filter group delay (at the cost of yet more
delay), also makes SSB easier to listen to, particularly if you are
running a 1.8K filter!

> W6JL has a rig with DDS LO, DC receiver, homebrew VFO driving a
> 6CL6/4CX250 500W amplifier....full QSK and bitched at my lid-like
> operating when I tweeked the tuner running 0.1W prior to sending "BK
> BK".

ROFL, where did this "Lid" thing come from anyway, I think it is
American but I am afraid I don't get the reference..

73's Dan.
6488 2011-07-21 19:49:12 Bill Carver Re: Cartesian loop exciter - QSK
"LID" = bad operator. Goodness, am I dating myself? I was a novice in 1958 and everyone knew what a LID was back then.

I've stroked all the distributors, nobody has the PIN I wanted. I'll try Ameritron. I'll use 6A10 if I have to. 40 pF from antenna to TX tank isn't too big a deal up to 10m. THe suckout models to less than 2 dB on 10m, and that's about 1 MHz below the frequency the amplifier tank is tuned to.

I am bypassing the whole RF choke business, both for the tube and for the PIN diodes. I'm feeding DC in at the 50 ohm end of the pi-L through a 500 KHz highpass filter. Inductors about 18 uH, no resonances at HF, capacitor is 3300 pF. Capacitor RF current is high, had to hunt to find a Russian doorknob on eBay that had low ESR for the 5A of RF current. I'm shunt feeding DC to the PIN diodes with the same kind of highpass network......no need for "chokes".

Of course I did have to use a 10 KV vacuum capacitor to handle RF + DC on the plate tuning capacitor and 6 KV vacuum cap for the loading capacitor.

W7AAZ
6493 2011-07-22 05:07:37 Dan Mills Re: Cartesian loop exciter - QSK
On Thu, 2011-07-21 at 20:49 -0600, Bill Carver wrote:
> "LID" = bad operator. Goodness, am I dating myself? I was a novice in
> 1958 and everyone knew what a LID was back then.

I had figured that out from context, but was trying to work out the
derivation.

> I've stroked all the distributors, nobody has the PIN I wanted. I'll
> try Ameritron.

They list the following two diodes:

MA4PH301 - Ameritron P/N 303-0301 $10.95
4P4001 - Ameritron P/N 303-4001B $13.35

Regards, Dan.