Content-Transfer-Encoding: 8bit
Message-ID: <5.1.0.14.0.20020814113217.00aba170@gemini.herts.ac.uk>
Date: Wed, 14 Aug 2002 12:35:11 +0100
To: rsgb_lf_group@blacksheep.org
From: "James Moritz" <j.r.moritz@herts.ac.uk>
Subject: Re: LF: Oscillator Stabilty for QRSS?
In-reply-to: <000a01c242fb$2b68ace0$7211f4cc@jsm>
References: <6a.23b88f69.2a798441@aol.com> <00a301c238e9$2266e7e0$389437d2@xtr743187>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

At 11:56 13/08/2002 -0700, you wrote:
>Perhaps someone who has already 'been there' can offer their advice.
>I would like to start building the exciter stages for what will be my 
>136KHz tx (hopefully we will have access to the band within the year!). 
>The DJ1ZB exciter (modified by DF3LP) is the circuit that I am 
>questioning. It mixes two xtal controlled 4MHz oscillators, 136 KHz apart, 
>to extract the 136KHz signal. Can I build two VFO's, somehwere around 1MHz 
>and set them 136KHz apart, that will  be stable enough to handle QRSS work?


Dear Steve, LF Group,

Mixing 2 VFOs or VXOs together to obtain a lower difference frequency does 
nothing to improve the stability - in fact it makes it worse unless you can 
guarantee that both oscillators have exactly the same drift, which they 
won't. The crystals mixed together works quite well because the crystals 
have quite good stability to start with, but the 136kHz difference 
frequency will be much less stable than a single crystal oscillator 
operating on, or divided down to, 136k. The same applies to VFOs - it would 
be much better to make a single 136k VFO, and apply temperature 
compensation to get near zero tempco than it would be to make 2 HF VFOs and 
try and get identical tempcos.

I used a 13.6MHz VFO divided by 100 - this has the advantage that it is 
easy to measure the VFO frequency to 1Hz resolution, and therefore the 
output frequency to 0.01Hz resolution. It was fairly easy to get the output 
frequency to stay within a few Hz, which is quite adequate for QRSS3 
contacts, but you would not want to leave it running all night unattended, 
or use it with more than 10s dot lengths. The crystal mixing scheme has 
comparable stability, and is somewhat easier to make than a stable VFO. If 
you could accept a narrow tuning range, perhaps 50Hz at 136k, replacing the 
VFO with a VXO would be a viable solution. A simple crystal oscillator 
seems to give adequate stability for QRSS30 provided the ambient 
temperature is fairly constant. To improve on this, you need a TCXO or, 
better, an OCXO - this is also true for PLL or DDS synthesisers, which are 
only as stable as their reference oscillators, of course.

My own solution to a stable drive frequency was simply to buy an old 
synthesised signal generator. These are fairly cheap now, especially if you 
get one that does not cover VHF/UHF. I got a Racal 9084, which covers 10k - 
104MHz - I divide the 13.6MHz output by 100 to give 0.1Hz resolution at 
136k. The internal OCXO gives stability better than 1 part in 10^7, it 
stays well within 0.01Hz at 136k. It has many other uses in the shack apart 
from being a VFO of course. Bear in mind if you buy one of these old units, 
you will probably have to repair it at some stage! Quite a few people have 
used an HF rig with the output divided down in a similar way, but unless 
the rig has a high stability reference, the stability will be that of the 
"ordinary crystal" reference oscillator

The requirements for Jason are a bit different - you need digital selection 
from 1 of 16 frequencies. The stability is somewhat relaxed by the 
differential nature of Jason, but the frequency shifts need to be quite 
accurate to preserve the coding correctly. This makes an analogue FSK 
scheme quite tricky. You could use a DDS scheme like G4JNT and SM6LKM. In 
my case, I take a 4.5kHz signal generated by the sound card output from 
Jason, and mix it up to 136kHz with the Racal sig gen. You could generate 
QRSS in a similar way if you wished - the drawback is the need to calibrate 
the sound card frequency errors.

Cheers, Jim Moritz
73 de M0BMU