[Lowfer] Homebrewed SSB Filter question

[Jim]

> I want to experiment a little, and try building a SSB filter that
> will pass the USB of my Osc. at 184.322 KHz.

> *** QUESTION *** ; If I use crystals to build the filter, will my
> Center Frequency ( Fc ) need to be 185.822 KHz ?? I have a handfull
> of 184.322 KHz crystals, but that would pass my carrier freq. and
> not the USB audio. Am I correct on this ?? Not sure, so asking.

If you're carrier is (or would be, if you weren't using DSB) at 184.322 kHz, 
you'd need a filter that will "let through" from about 300Hz to 3300Hz above 
that for the USB. So you'd want something with around 3 kHz of bandwidth that 
was centered about 1.8 kHz above the "carrier" frequency. So the center 
frequency would be around 186.122 kHz.

If you had a center frequency of 185.822 kHz and a bandwidth of 3 kHz, you'd 
let through from 0 Hz to 3 kHz on the USB. Depending on your carrier 
suppression (which, in turn is probably going to depend on how balanced your 
DBM is) that may work.

> *** SECONDLY *** ; Since this will be used with a signal of about
> 33mW; what are some thoughts on possibly using Op Amps to design
> a Bandpass filter for the USB audio ?? If this approach will work,
> the Op Amps will also serve as amplifier stages ( I Think ).

An op-amp might work. That tight a filter at that high of a frequency would 
probably be a bit challenging. But I wouldn't say it's impossible without 
running some numbers.

You can certainly build some amplification along with filtering into the 
op-amps. But you'll probably need something else to act as a final stage.

You've mentioned this before and I'll admit that's where I get fuzzy on what's 
going to happen. If you drive something like a simple push-pull pair with the 
output of the op-amp to get to 1 watt (or more if you go the Part 5 route), I 
really don't know what would happen. If you look at something like a two-tone 
test pattern for a SSB signal, it's not the kind of thing that you can push 
through a non-linear stage. So I would assume (and I could well be wrong) 
that you'll probably have to have some type of linear amplification following 
the op-amp. (Unless you can find an op-amp that's good for 1 watt output.)

But, having said that, I wonder if something like a modified high-efficiency 
audio amplifer chip might do the job. That's probably the route I'd look at. 
But again, I'm way beyond where I can speak from experience.

> This is just something I want to experiment with so I can at least
> say I tried. I'll appreciate any replies.

There are two approaches to generating a SSB signal. One is filtering the 
other is phasing. My two cents worth would be that using op-amps, you'd 
probably find it easier to build a phasing system than to build a filter with 
a 185 kHz center frequency and a 3 kHz  bandwidth.

You had mentioned before that you're really doing what amounts to a "direct 
conversion receiver" in reverse and building a direct conversion SSB 
transmitter. And that's not uncommon, it's just you're using a bit different 
approach than I've seen. It's interesting, but you're probably not going to 
find a lot of resources to help you based on any real experience.

So that's why I'd really recommend considering a phasing approach. It has a 
lot of advantages.

1) It's pretty much proven and there are designs (like the T2 transmitter) 
that give you a good starting point.
2) It's flexible--you just feed the oscillator frequency you want into it and 
you get the SSB signal out. (Maybe a bit of tweaking, but not much.) So if 
you want to move to 185 kHz or 183 kHz, all you need to do is change the 
local oscillator frequency that you're feeding into the phasing transmitter.
3. It doesn't take any special filters. Crystal filters are pretty easy to 
build--but they're not always easy to "optimize". If you can deal with that 
by adding an attenuator here or there to help with impedance matching and 
then just stick in another stage of amplification, that's no big deal. An 
active filters with op-amps is an option, but that just seems like a pretty 
high Q filter at a fairly high frequency (relative to most of the active 
filters I've built). You think about a 70 Hz bandwidth center at 700 Hz (a 
decent CW filter), you've got a Q of 10 (700/70). For this you'd have a Q of 
around 60 (180/3). You could use just a high-pass filter instead of a 
bandpass, but I still don't think that's going to simplify things--you need a 
lot of cut off over a tiny chunk of spectrum to block the LSB.

The only real disadvantage of the phasing approach is that it's going to take 
several op-amps to properly build the phasing signals. But once you have 
those, the design is good from LF to SW (and maybe even higher in the case of 
the T2).

Not trying to be a naysayer, but seems like a bit of a challenge. But actually 
you've kind of got me thinking now. I need to finish the Class D and get it 
on the air (I'm bad about partially finished projects). But after that I 
might try to get a SSB rig going also. It's definitely the way to go for the 
more advanced digital modes.

73 de
Jim W4JBM

http://www.geocities.com/w4jbm

"With a soldering iron in one hand, a schematic in the
other, and a puzzled look on his face..."

Working the world from the New Dog Iron Ranch!