EMRFD Message Archive 4533
Message Date From Subject 4533 2010-04-20 07:07:47 Tim Multiplier chains and filters and transmitted fundamentals I am building a series of VFO's to match my R2 and homebrew transmitter.
For the low bands (80M-20M) I have a basic J310 VFO at half the frequency, followed by a J310 buffer and a low-pass filter, a two-diode multiplier, and another J310 buffer with low-pass filters before and after the last buffer. Very much like the Kanga UVFO or the schematic at the end of EMRFD.
Side note... I've been getting some Russian variable caps that have built-in reduction drives off of E-bay and am very impressed with them. They're tiny and they're clearly superior to the Japanese variables and drive mechanics. Just IMHO. I set them up to cover the bottom 80 or 100 kHz of each band, put a big rubber/plastic knob (from McMaster-Carr) and the result is a very nice tuning rate on CW with excellent feel.
What I note about the Kanga UVFO and the schematic at the end of EMRFD, is that there's no bandpass to remove the fundamental. For example for my 80M VFO, I start out with a VFO at 1.75MHz. It gets low-passed and then multiplied up to 3.5 MHz. The two-diode multiplier is very good but there's still a little of the fundamental at 1.75MHz left... clearly visible on the scope as the even/odd humps having slightly different heights. My gut feeling is that the asymmetry is not in the diode multiplier but in the waveform out of the low-pass still not being quite symmetrical.
Is this "noticeable on the scope" artifact going to do anything to my receiver or my transmitted signal? My gut feeling is that just like we aren't supposed to be radiating harmonics, we shouldn't be radiating subharmonics either. If I'm having a CW QSO on 3.5 MHz I don't want to be sending anything at all on 1.75MHz. But traditional amp chains seem to have only low-pass filtering.
The FCC regs clearly have numerical limits on harmonic content. In my 30+ years of being a ham and building/fixing old stuff I've even managed to get a couple of OO postcards telling me about my harmonics. But what if any rules/regulations define how much bandpass filtering I need to put on to get rid of the 1.75MHz oscillator fundamental? Is the traditional class C transmitter chain with low-passes good enough at getting rid of that fundamental?
4534 2010-04-20 08:54:00 Tayloe Dan-P26412 Re: Multiplier chains and filters and transmitted fundamentals My rough experience is that even if the sine wave on the scope is
slightly non-perfect, the harmonic will be at least 30 dB down. If the
harmonics are better than than, the signal will appear clean.
The Ft. Tuthill design used a 1/3 frequency VFO that was tripled to 3.5
MHz so that the TX would not pull the VFO. A wide, fixed tuned bandpass
filter was used to clean up the unwanted sub harmonic content. You
might want to take a look at it.
- Dan, N7VE
4536 2010-04-20 14:55:41 ajparent1 Re: Multiplier chains and filters and transmitted fundamentals My experience with the UVFO is that the filters seems to be peaked low pass and the fundamental was way down due to the balanced diode doubler. I got better than 45db down and that was before the I/Q hybrid. A simple high pass was enough to clean it up to the point where I know it was there but hard to measure without a Spectrum analyzer.
If your driving a DBM with it and looking there the signal will
look horrid due to the diodes switching.
4538 2010-04-20 20:00:15 KK7B Re: Multiplier chains and filters and transmitted fundamentals Hi Tim,
I measured the subharmonic output at J3 in the UVFO circuit in EMRFD figure 12.78. The 7 MHz subharmonic output is 30 dB below the 14 MHz output from the VFO circuit board. That vfo circuit has a single pi section low-pass filter before the diode doubler, and only broad-tuned low-pass filters on the output, so you are mostly seeing the performance of the diode doubler. There is some peaking in the low-pass filters, but they are not the old familiar "peaked low-pass" response with a single peak. They have a flat-topped response around the frequency of interest, and are about 6 dB down an octave lower. See figure 9.65 for simulations I did of similar filters used in receive LNAs.
I also measured the subharmonic output of the Sleeping Bag Radio transmitter shown on pages 12.42 and 12.43. The 3.5 MHz subharmonic output is 55 dB below the desired 7 MHz output at the antenna connection. That circuit has a modestly good sine wave driving the doubler, and the doubler drives a single-tuned circuit. After that everything is broadband.
You are correct that a good sine wave into the diode frequency doubler is needed to acheive good cancellation the fundamental. In a transmitter, I usually add some selectivity after the doubler--see the single-tuned circuit in Figure 12.72.
I expect that a pi-section low pass filter before the diode doubler and a double-tuned circuit after would reduce the subharmonic output below the level where I could see it on my spectrum analyzer.
In receiver applications I've had very good results with a 5 section low-pass filter and +16 dBm of fundamental drive into a diode doubler with bias resistor, as shown in figure 5.43. That directly drives a quadrature hybrid and IQ diode ring mixers, with no amplifiers at the signal frequency. That resulted in very clean receiver.
In any case, the spurious signals in these direct conversion circuits are an octave away from the desired signal, and easy to suppress with basic filters.
A few years ago several of my students bought Russian variable capacitors
4540 2010-04-21 06:27:50 Tim Re: Multiplier chains and filters and transmitted fundamentals 4542 2010-04-21 06:59:20 victor Re: Multiplier chains and filters and transmitted fundamentals I suggest to connect a 100 to 500 Ohm trimmer to the doubler input, the ends connected to the two doubler inputs (phase and anti-phase) and the trimmer center pin to ground and adjust it for better balance by measuring the fundamental frequency signal at the doubler output with another receiver or spectrum analyzer. You should be able to improve the fundamental rejecti 4544 2010-04-22 11:15:01 Tim Re: Multiplier chains and filters and transmitted fundamentals 4545 2010-04-22 11:44:07 Tim Re: Multiplier chains and filters and transmitted fundamentals Hi Victor!
The more I think about your suggestion the more I realize how clever it is.
We have a diode doubler that's perfect. We feed a signal at "f" into it with some distortion (largely 2nd harmonic, "2f".) Because of the presence of the 2nd harmonic at the input, we have lopsided odd-even peaks from the doubler - a component at "f".
Look at how insidious the above all sounds. Because the input to the doubler has a undesired 2f component, the output of the doubler has an undesired component at f.
Your suggestion interpreted one way, is to take some of the input, add it in to the output with right sign and magnitude, and it'll cancel out the f component in the output.
Or (alternative viewpoint), take the right fraction and sign of the 2f output, add it back into the input, and it'll cancel the undesired 2f coming into the doubler.
Really clever. Like the snake eating itself!
4548 2010-04-22 18:47:12 ajparent1 Re: Multiplier chains and filters and transmitted fundamentals A way to measure it is any receiver. I used to use a DX302
which was ok or my older AX190(with PLL LO to cover 3.5 to 30)
for that. The S meter and an external step attenuator does the calibrated work.
Since then I do have an agile (tunes .5mhz to 160mhz) receiver that uses a ad8307 detector and no agc. With attenuators and the dynamic range of that radio I can see a range of 70db. The receiver is basically a variant of EMRFD figure 7.60 using the ad8307 in place of the NE604. The LO used to be a dual loop PLL and is now a SI570 LO source. That reduced 20cu inches of PLL to a chip, micro and LCD.
The only difference with this and the SA is rather than scan I can manually sweep a band and see what and where the levels are. Simpler and works. Also makes an ok deaf
4555 2010-04-23 09:28:15 drmail377 Re: Multiplier chains and filters and transmitted fundamentals Be careful with LTSpice (or other simulation packages) when using FFT. There can be very significant changes in your FFT results depending on your sample rate, windowing function etc. In LTSpice often issuing the Spice directive .option plotwinzize=0 will turn off data compression, improving your results at the cost of a larger .raw file. I am no expert in this area with LTSpice, as such I suggest you upload your simulation to the LTSpice Yahoo Group in the Files > Temp area and post a request for suggestions