开云体育

Hello all,

?

Long time lurker, first time poster here. I recently built a 60m-15m QMX and I am having some trouble getting good reception on 15m and 17m (see attached RF sweeps). I should mention that I am using a Rev 1 board purchased in 2023, but with a mid-band component kit from 2024. Here’s where I’m at:

• I get good power (4-5w at 12V) out on all bands, and my LPF sweeps look like the firmware manual
• 60m-20m RF sweeps look more or less like what’s shown in the manual. I use a QRP Labs 50 ohm dummy load for the sweeps.
• When I connect to an antenna, I can hear FT8 signals on 40m and 20, but not on 17 or 15
• Because the lower bands seem to receive OK, I suspect an issue with L401a, C401 or C403. However, 20m uses L401a and C401 and is OK, but 17 also uses those and is not good.
• I reflowed solder on all 3 taps for L401a and the capacitors, with no change. I’m pretty confident I’ve burned off all the enamel at the junctions at this point. I get continuity at all the points shown in the Rev 1 manual (plus at pin 6 of IC402, which I think makes sense because I’m using all 4 connections for L401 instead of just the 3 with the low band version)
• I did a factory reset, with no change
• I’ve counted the number of turns on 401a more times than I’d care to admit, and keep coming up with the same number I’m supposed to have (24 turns, tapped at 15)
• I confirmed that C401 and C403 are the correct values printed on the component, and don’t appear to have any physical damage

?

Trying to figure out my next steps, but coming up a bit short. All I can think of is to replace capacitors, and possibly re-wind L401a. Or, even less ideally, replace IC402. I don’t have an oscilloscope (or any real expertise to use one), so my ability to diagnose is somewhat limited. Before I start swapping out components, is there anything else I should try looking at? It’s been fun building this kit and I have managed to iron out quite a few kinks along the way by reading posts on this forum. However, this one has me stumped.

?

Thanks in advance

?

Alex, N7ADH

?

Hi Alex,

This is an interesting one.? For more context, it would be helpful if you posted RF sweep for 20m, and LPF sweeps for 20m, 17m, and 15m.

The bandpass filters (RF sweeps) follow the LPFs in the receive path, so they interact.? If the LPFs have a significant attenuation, it will cascade to the BPF.? Also, 20m and 17m use the same BPF turns (all of the turns on L401a), so it is helpful to see where that full filter is centered, so it can be moved appropriately.? 15m has its own BPF, made of just the 15 turns before the tap, so we should be able to center it at the peak of the filter.

?

Note that the Rev 1 board wires IC402 to the BPFs very differently than Rev 2 and above.? Look at both schematics to see.? In Rev1, the capacitors go into the Y inputs of IC402, then the corresponding W output of IC402 feeds the toroids, which are then input to the X inputs of IC402, with then the full L/C filtered signal at the Z output.? In Rev 2 and above, the L/C are combined first, and fed only to the Y inputs of IC402.

?

Since you are using the mid-band kit, which on Rev 1 will use all 4 of both the X and Y inputs, make sure that the 4th input of each is really connected for C403 and the L401a tap at pins 10 and 6 of IC402 (it doesn't show on the schematic, and I don't want to take mine apart to measure :).? But I suspect the board has the connections, otherwise you wouldn't have had holes for C403.? So this is likely not a problem, but the careful inspection may be instructive.

?

Stan KC7XE

Hi Stan, thanks for the speedy reply. Appreciate your insight. I am getting continuity from C403-Pin 10 and from the "15-turn" tap of L401a to Pin 6 of IC402. Seems like that's good, right? I had noted the different layout around IC402 between Rev 1 and Rev 3/4, but wasn't sure how that would affect the build, if at all. Perhaps that is part of the issue though.

?

Here is my RF sweep for 20m:

And the LPF sweeps for 20-15m:

Best,

?

Alex N7ADH

?

?

The combination of all the sweeps you sent show the problem and lead to the solution.

And the rev1 board connectivity is all in place properly, or the sweeps would look way different.

?

First a bit of design background:

Ideally, a bandpass filter will have the frequency of interest centered in the band.? But in the QMX design, some bandpass filters are shared, so this can't fully happen.? For the mid-band QMX, 20m and 17m share a bandpass filter.? So you center the filter between those bands, usually with each band just off the edge of the filter. ? And 15m has its own bandpass filter - BUT it shares a toroid with the 20m/17m filter, so all three bands interact - the 15m turns are used for the 20m/17m filter.

?

Now, from your filter sweeps:

15m (21MHz) filter:? the center of the filter is low, closer to 18MHz, leaving 21MHz with about 18dB of attenuation.? So that filter needs moved up to get 21MHz in its passband.? Moving it up requires less inductance in the toroid (fewer turns, or turns spread out).

?

20m/17m (14MHz/18MHz) filter:? the center of this filter is also low, <13MHz.?? And 18MHz has about 18dB of attenuation.? It needs adjusted upward in frequency so 14MHz is at one side of the passband, or a bit outside the passband if necessary, so that 18MHz is also close to or within the passband, each with just a few dB of attenuation.

?

Thankfully both need adjusted in the same direction.? If you have the windings all bunched together, perhaps spreading them out evenly will reduce the inductance enough.? But I suspect you will need to remove a turn from the 15 turn part - this will reduce inductance in both filters sharing the toroid.? So try removing first one turn, then checking the scans, then if they are still too low, remove another turn, which may move the filters too high - but if so you can compensate by squeezing the turns together to increase the inductance again.

?

So why did this happen on your build?? There is up to a 20% tolerance in the toroid manufacture, and you may have gotten one that was out that far.? But I suspect another reason: The rev1 style of routing the signal through the IC402 between inductor and capacitor will add a small amount of capacitance to the filter, and this needs counter-acted by reducing the inductance.? The number of windings on the inductors for the mid-band version (which came out later than rev2) were tuned for the way IC402 is used on the later boards, with less capacitance.?? This would affect mostly the higher frequency bands, due to the smaller capacitances needed in those filters.

?

One additional adjustment to consider:

The LPF scans look good, but the LPF for 15m has 21MHz at or just above the cutoff frequency - which will attenuate 15m a bit, before it even gets to the BPF.? It's not too bad as-is, but you could try and spread out the windings on L510, to move the LPF cutoff frequency up a bit.? If the windings are already fully spread, you could remove one winding, and then squeeze the remaining windings together to keep 21MHz near the cutoff edge. ? Spreading the windings of L513 may also help a bit - but don't get too dramatic here, unless you have means to test the harmonic rejection of the LPFs.

?

I hope this is helpful - making the BPF adjustments should improve your receive on 17m and 15m by almost 18dB, which is 3 full S-units.

Good luck, Stan

Thanks Stan, this is great. The explanation makes perfect sense to me. I ended up removing three (!) turns from the 15m portion of 401a. I think I could possibly benefit from removing one more. Or possibly removing a turn from L510 since the windings are about as equally spaced out as I can make them. In any case, here are the new sweeps:

?

17m still looks pretty bad, and 20m is getting close to the edge. I thought I would try reconfiguring 17m to use BPF 0 instead of 1 (so sharing with 15m instead of 20) and I get the following result:

This seems better than on BPF1, but I'm not sure if setting it up this way would have any unintended consequences. Any thoughts on whether I should try taking another turn off of 401a or call it close enough? Thanks again for your guidance -?

?

Alex

Looks much better now.? If you want to experiment with taking off another turn on L401a or L510, it won't hurt anything.

And leaving it set to share 15/17 rather than 20/17 will also work fine - the only side effect I know of is that you will need to remember to make this setting again whenever you do a factory reset.

Just also remember to evaluate the changes by listening to received signals - is it better, and good enough for your liking?

?

I'm wondering a little what makes the 'top' of the 15m BPF only get up to about -12.? It would be better if that were higher.?? Maybe adjusting L510 would help with that, but I'm not sure what all causes it.? It is a little lower than on your first plot of that band, so something about removing turns may also have affected it.?? It may be interesting to see if bunching the turns together would make that go up by increasing the Q of the filter - if so, removing another turn and bunching them a bit may give better results.? But I am way out of my depth here -? I can only guess at cause/effect.? So hopefully someone else can comment.? Attenuation could be coming from the LPFs or from efficiency of this filter.

?

One note to remember about the RF and LPF scans: the scale on the LPF scans is made by normalizing some set of the passband values to 0dB, and making everything relative from there on each scan; whereas the RF scans measure dB values offset from a common reference (the fixed level of the test signal clk2).? So when the top of the RF scan is down like that, it does indicate a lower signal level getting through.

?

Stan KC7XE

Thanks Stan. I did not realize that the RF dB values are using a common offset. That means that the RX performance on 17 isn't that different between BPF1 and 0 - only about 3dB. I decided to try removing one more turn from 401a, and here are the results (after a few different attempts at bunching/spreading:

Looks basically the same to me. I did try a version with very close bunching of the turns, but that only gave me 1db more of peak at 15m and moved the top of the curve to the left of 21mhz. I am kind of surprised that the BPF shape and amplitude are relatively insensitive to the number of turns on L401a. Makes me think there might be some other factor at play that's driving the filter response. In any case, I think I'm ready to chalk this up to issues with using a Rev 1 board in a different manner than intended. Unless, of course, you and/or others on the forum have any ideas about how I could squeeze a bit more performance out of 15/17m receive. Either way I appreciate your help and the wealth of knowledge on this forum

?

Alex N7ADH

?

?

Those scans look nice.? You have gained at least 6dB on 15 and 17 - maybe more.?

Now to test it in practice and see if you notice the change.

Yes, there is some other 'gain' factor going on; something attenuating the signal on 15/17 apart from the BP filter.? I wish I knew what it was.

Stan KC7XE