Good. Frankly, there's not much else to look for practical purposes. We already have enough clarity in this area.

I'm ok with keeping one hand on the volume but I did think about making a simple passive AGC circuit. If I use a LND150 (depletion + enhancement mode MOSFET, normally halfway on when VGS=0V) and a Schottky diode to make a passive AF shunt AGC, I just need a simple audio amp to complete the whole thing. It's similar to QCX AGC module but the gate bias circuit becomes unnecessary.

Are you guys now trying to feed white noise to the mixer and do the spectral analysis of the I/Q raw signal output for the mixer's frequency response??? (I don't recommend fake impulse for this.)

开云体育

I should say not just the DC but probably 1/f noise, too. The 1/f noise is probably one of the leading enemies of analog/RF engineers.

There probably is a DC component there because of the differential amplifier and the LPF. Any slight imbalance of the input bias, offset, gain, etc. could create a small DC. It's tiny but we are also looking at receivers capable of weaker than -130dBm.

Mine is NanoVNA V2 Plus4 but high impedance measurement is reasonable. When I measured the velocity factors of my coax cables, I measured with the other end open, shorted, and mismatch terminated and calculated the velocity factor from the frequencies that created impedance extremum. They agreed fairly precisely. Since a lot of hams are using NanoVNA mostly as an antenna analyzer, so even a cheap counterfeit might do, but there are situations where the latest model helps.

Based on everything we've seen here, QMX's BPF architecture is very reasonable. Even a very low Q series resonance BPF is hugely beneficial because the mixer's input impedance has a large synergistic effect to sharpen the overall preselector performance. If anyone still argues that the BPF is unnecessary, taking advantage of the mixer's intrinsic band-pass filter characteristics requires that the source impedance be high enough all across the stopband frequencies. A series-resonance BPF, even if the Q is very low, will do exactly that.

That does not mean that the stray capacitance issue and other existing issues are gone. But solutions to those things should best be done without compromising the preselector performance.

from what you described I think you are seeing the annoying DC offset at the center frequency that is common to SDRs and quadrature sampling detectors.

Agreed, attaching TinySA Ultra or something to the mixer input port should tell if there is any significant LO leak to the input port. But I doubt it is large enough to affect the VNA, as the clock is only driving the counter and decoder. It's not a typical SBM or DBM situation. Also, if there is a strong feed through, the VNA measurements would be affected by more frequencies not just one point. But it is still a possibility. More likely, all it takes to see a very high impedance at the input port is precise lining up of the LO and VNA's frequencies.

However, we are getting sidetracked to an unimportant matter here. That peak is irrelevant in actual receiver where the mixer is driven by a source of relatively low impedance. Also, in QMX, we care about the IF frequency offset and its image frequency not the LO frequency like DC receivers.

Also, keep in mind we are only looking at the input impedance and not the overall mixer response. The traces we saw are not the overall frequency response of the mixer. If the signal source is ideal current source with infinite impedance, those two responses may be proportionate within some limits. We are driving the mixer with a source with finite (in fact, low) and varying impedance. The actual frequency response is expected broader near the receiving frequency but narrower farther away than what appears on the impedance traces, thanks to the BPF.

The double balanced Tayloe circuit does a great job of suppressing LO feedthrough.

A few thoughts

Solder suckers do have a place - especially for initially removing solder piles that would consume ‘a mile’ of wick. ?Avoid using them fully vertically?

Wick is great but doesn’t always leave clear any small plated through (via) holes. ?Fluxing the hole and trying again helps but still not always?
A sliver of wood down the hole while heat is applied WILL clear those holes so you can then wick it fully clean

There is an investment in equipment, learning and practicing hot air rework techniques but it’s worth it.

A suitable digital microscope is also a great aid (but yes it’s another investment)

Any large component removed by any of these means should be considered a write -off due to the thermal stress so chop off the leads ?for simpler one-by-one removal if the board is at risk from overheating?

Due to the above it is often useful to cut PCB tracks to check a large component is faulty before ripping it out and finding it wasn’t. ? ?(Cut where repair is easiest not right by a tiny pad ! )?