News, Updates and Other Minutiae

I have, for quite a while, provided modified versions of the code for my projects free of charge. However, the cost of webhosting for this site has tripled in the last couple of years and the little I make out of the pcbs I sell goes nowhere near covering this cost.

As of October 2017, I will be asking for a donation in exchange for providing custom modified code for my projects to help keep this site available.

February 2018

Updated Australian Band Plan page with the latest info (under 'Misc' heading).

November 2017

A modification plus software to allow the Simple DDS VFO to be re-calibrated as often as you want without having to re-program the PIC has been added to the Simple DDS VFO page. Simple DDS VFO bug problem fixed. Code uploaded to webpage as version 'c'.

October 2017

There is a bug in the Simple DDS VFO code. See Simple DDS VFO page for the problem and updated code. Included an add-on for the Simple VFO project to indicate which step size has been selected. Simple DDS VFO project updated and software file uploaded. The pushbutton on the encoder is used to change step size (step sizes 10Hz, 1kHz and 10kHz).

September 2017

Noise Canceller kits now available.

July 2017

Updated both the DDS Xtal Substitute and DDS Xtal Substitute Mark 2 projects into one project. The original webpages are available to download as PDF's.

May 2017

New project added - my version of the X-Phase noise canceller. PCB available and a kit of parts to follow soon

This is a filter/leveller for use after an AD9850 DDS module, from G3OAG. I will let him explain the circuit in his own words (edited and condensed from our email conversations).

G3OAG DDS Filter/Leveller

I have worked out that levelling the DDS output can be achieved with a simple high pass filter. My DDS, using an e-Bay module and 2N3904 amp stage, gives me 10.3dbm at 10.8Mhz (1.8Mhz + 9Mhz offset) down to 2.3dbm at 38Mhz (29Mhz + 9Mhz )...a difference of around 7.5dbm. Up to now, using a simple T design, three pole high pass filter, I have achieved some degree of levelling with 2 x 470pf caps in series and a coil of around 0.1uh in the middle. This obviously lets the higher frequencies pass relatively unattenuated, and the lower frequencies are attenuated so that at the output of the filter, the level is getting on to the same level as the higher frequencies. After further experimentation, I have ended up with ten components after trying three previously. You can see the output of the final 2N5109 amplifier and splitter for the two DBM's is within 1db across the nine bands. I do not think this circuit is totally repeatable, as there will be various capacitance and inductive differences between different builds. I ended up having to use a damped trap ...the 1.58uH coil and the 150pf with the 220 ohm, to lower the 160m output which as you know is very high compared to the 10m output - in this case a range from 10.8Mhz to 38Mhz owing to the 9Mhz IF offset. However, it seems to function OK, and I'm sure other builders can make the necessary changes.

G3OAG filter/leveller schematic G3OAG filter/leveller output levels

The amp following the e-Bay DDS board which uses a 2N3904, is from the LU5DJV DDS circuit, but they are all similar. This outputs around 10.5dbm at 160m (10.8MHz with the 9Mhz offset) going down to around 2.3dbm at the 10m band...38Mhz. The amplifier following my filter is the standard 2N5109 Class A design, originally in the ARRL Handbook, which has been used everywhere, and gives around +17db to +20db gain. This feeds a power splitter, which gives two equal outputs to feed both the receive and transmit mixers. I was hoping to get the magic +7dbm out from each output, but owing to losses in the filter, I am getting around 5.5dbm...not that it will make much difference. Below are two photos of my DDS and the filter, which is crammed onto the board. The two chips at the top left are 2x ULN2803A which are the drivers for the band relays in the RX input filters and the PA output filters, and also for the mode switching. You can also see at the bottom right the ancient Hatfield 3320 splitter and the two SMA sockets which feed the RX and TX mixers. This is the VU2CNS DDS circuit which works well. This DDS will go into my homebrew 20w rig which I built years ago and which currently uses a Vackar VFO and which I've had hundreds of QSO's with. My new build DDS will go into a new rig.

G3OAG dds G3OAG filter