Since the receiver board was partially detached from the chassis I pulled it out and brought it down to the house for a closer look. The first thing I did was find a service manual. It showed the front end filter could be padded down to 6m. However, the filter topology was unlike anything I have seen before.
It appears to be a chain of Pi filters with a series coupling capacitor. I couldn't find any details of the transform used so instead I pulled up the Iowa Hills RF Filter Design software and examined if there was scope to modify this filter to something I could derive values for.
My guess is the inductors on this board were around 60nH with an unloaded Q of 70 based on data I had seen for somewhat similar inductors. While the modification to the layout for a series bandpass filter would be little effort, I quickly established that a series inductor configuration had very little attenuation below 6m so I discarded that approach. No real surprise in hindsight given the small inductors.
Turning to a conventional capacitor coupled bandpass filter I quickly ran up a 3 pole filter with a centre frequency of 52MHz, a bandwidth of 12MHz and standard capacitor values. My initial plan was to run the Local Oscillator on the high side of the received frequency. Attenuation on the image of the local 6m repeater which I would mostly be listening to ( 53.8MHz + 2 x 45MHZ IF) of 143.8MHz was around 35dB. Since there are two of these filters, one before and one after the RF pre-amp, that amounts to 70dB of attenuation.
Which didn't sound like a lot to me if I was transmitting on 2m. My reasonning is 50W at 146.5MHz is 47dBm. Assume this is an ideal transmitter with no phase noise spreading the energy into nearby frequencies. That 47dBm gets attenuated by 35dB before hitting the pre-amp. Again, assume the pre-amp ( an SGA-6489 MMIC) can cope and amplifies this by 20dB before another 35dB of attenuation. Thus the mixer sees a net attenuation of -35dB + 20dB -35dB, or -50dB.
I am guessing but 47dBm from the 2m antenna might result in 0dBm into the 6m feedline. So my 2m transmission is only attenuated to -50dBm at the mixer in the 6m receiver. No big deal if I assume an ideal local oscillator with no phase noise since the mixer is a nice double balanced diode mixer and will probably cope.
But what if the local oscillator has a noise floor that is 60dB below the peak? If the mixer has a loss of 6dB does that mean a 2m signal gets mixed down with 66dB of loss? I don't know but I would be worried that after a lot of work I find every time I transmit on 2m I have to turn off the 6m receiver to avoid the squelch opening.
I've not noticed this issue or de-sensing on my existing 6m radio. I know the existing 6m receiver has much higher Q inductors. It took just minutes to run up a hypothetical filter with a higher Q and a200nH coils and compare it with what I was proposing to build. I concluded the filters I could make using the inductors on the MX800 board would probably suffice.
Comforted by a few minutes work I had a close look at the board. It would appear that by removing all the filter component I could use the inductors and new capacitors with the existing traces.
I am still grappling with the Fractional N synthesize chip. I will continue to look into this but I thought it worthwhile to reinforce my message about filter software with this post. If you're not using the Iowa Hills filter software, a free program, then you should. It took me considerably longer to write this post than it did to model the filters discussed.
Regards
Richard VK6TT
No comments:
Post a Comment