A diode ring mixer is inherently a non-linear device. Look at the voltage curves for diodes and you will see that there is a floor near the inverse voltage that guarantees that there will be never be instantaneous transition from positive to negative at zero volts (the ideal).

One uses diode ring mixers because they have an immense dynamic range compared to others.

That being said, one must pay attention to the frequencies one feeds into the mixer. There will always be birdies. One must plan on them being outside of the bands being used so that they are not a problem. Figuring out mixer products to the 4th or 5th harmonic on a spreadsheet is very helpful. There will also always be some IMD. But planning proper impedance matching and filtering will always make it tolerable (i.e., down 40 dB or more).

So it doesn't matter whether or not one feeds a square wave or sine wave. (The square wave is more or less preferred as it more easily saturates the diodes). There are inherent problems which must be addressed. The thing that really matters is the level of input signal and where are the gozintas and gozouttas. The RF must be clean and be capable of 140 dB or so gain level. The LO should be around 7 dbM. The IF must be impedance matched and able to take care of reflected waves. SWR is important; most of the noise and distortion products come from improper feedback from the IF; there have been all sorts of solutions to that problem (especially diplexers).

Still, unless they are very noticeable, most ears can adjust fairly quickly to some level of noise and it is not really a problem --
especially with noisy amplifiers surrounding the mixers...


The BITX solution is quite simple: provide a IF from the mixer that guarantees no in-band birdies, and provide broadband amplifiers around it which minimize reflection back into it. Along with a good ladder filter (the same is true there) It works fairly well. There have been some mods done, most prominently the LC filter on transmit of the BITXxxA series. You may find some to be necessary to conform to the transmit laws of your country.

If I were to provide a filter on the DDS into the mixer, I would consider some sort of low Q LC filter, which is really a variant of a diplexer. But it is probably overkill if the input level is right AND it is fed into the proper terminal.

john
AD5YE


---In BITX20@..., <jgaffke@...> wrote :

With a diode ring mixer, those diodes are pretty much either conducting or not conducting. So the action of the mixer is a square wave, even if you feed it a nice clean sine wave.

Given the balanced nature of the mixer, the narrow band crystal filter, and the band pass filter on the other side out toward the antenna, all those birdies should never come up. And these guys are talking broadband noise, not birdies.

Not that I know anything for sure here.

Jerry

---In BITX20@..., <baruchatta@...> wrote :

I thought that a pure sine wave was best. Isn't a square wave a mix of waves? That would result in all sorts birdies and products?

?

When I bought the VFO kit from Qrp-Labs, I also bought a 40 meter low pass filter kit. I installed it in line with the output of the VFO. Pretty simple, The receiver seems quiet, at least as quiet as with the analog tuning. However I don't have a scope and currently I'm experiencing a high level local noise, but without the antenna it's very quiet.?

On Jan 5, 2017, at 9:53 PM,?dvoorhees@...?[BITX20] <BITX20@...> wrote:...I'm experiencing a high level local noise, but without the antenna it's very quiet.?

?

It's quite common for noise sources in the house to go up the coax and come back in from the antenna.

Could be you need a good "common mode choke".

??

??

My totally uninformed opinion: ?I think the Si5351 noise everybody's talking about is likely hash from the processor driving it. ?And that if this less-than-optimal LPF is reducing the noise to where it's as good as the analog VFO, that's good enough. ?The ideal would be to clean up the noise somehow without band specific filters, then we're back to switching only two filters if we want a multiband rig. ?

The diode ring mixer wants to be driven hard. ?The ideal is for two of the diodes to turn on for 1/2 the LO cycle, the other two diodes to turn on for the other half of the cycle. ?Anything less, and changes in the other input signal (from the antenna when receiving) can disrupt the desired action of this commutating mixer. ?A square wave is actually preferred, and a perfectly clean sine wave from the LO will have a totally non-linear response from the diodes anyway since the current through a diode is exponential with voltage, not linear like a resistor.

Of course I may well have totally different opinions once I get a Bitx to play with.

Jerry


---In BITX20@..., wrote :

If your using the QRPlabs VFO for the BitX40, then the LPF you bought for 40 meters is optimized for 7-7.5MHz. You need a LPF for the actual VFO output which is 4.5-5MHz. The 7MHz one is probably helping to clean the output somewhat, but is not optimal.?

On Jan 5, 2017, at 9:53 PM,?dvoorhees@...?[BITX20] <BITX20@...> wrote:...I'm experiencing a high level local noise, but without the antenna it's very quiet.?

?