Dear all,
See the attached picture of my HF QDX for the 20 meter band.  Every band from 10 to 17 meters are perfectly in specification.  It is only the 20 meter band that is out significantly.  The QDX works 100% on 15 meters and I have already made some confirmed contacts with it on the band.  (tx is 4,5w @ 9v).
So, I know you all will be saying its the bandpass filter, and specifically the last solder joint on the BPF (for 20 meters), I have re-floated the solder 3 times on the bandpass filter and for good measure bridged ic3 pin 6 to the 20 meter pin on the BPF with a piece of wire.  No difference on the graph as attached.  I have done all the tests on IC3 for continuity as well as checking for 2.5v from pin 3 to 6.  All are positive.
Has anyone else seen the same issue.
73
Timo
ZR6TM

Timo,
I haven't built a upper-HF QDX yet.
Have you considered the state of that band's LPF? It's in the circuit for both Rx and Tx.
Is the LPF used by 20m "shared" with another band?

--Al
WD4AH

Hello, again Timo,

The first thing is to verify that you are running the sweeps when the QDX is connected to a 50-ohm dummy load.

The sweep is not that bad, so that I would verify the operation.  Do you get decodes on 20 meters?  If so, it may not be worth potentially damaging the QDX to get optimum performance.  Due to the QRP output, the QDX will hear many more stations than stations listening to the QDX.

If you feel it is necessary to adjust the 20 meters reception, start with verifying that C30 is 39pF.  Then see if you can squeeze the last 3 turns of L12 together to increase the inductance.  If that does not help, you can try adding a turn onto the end of the 20 turns, making it a total of 21.

I have not yet built my High-Frequency version of the QDX, so the above is based on the original QDX.
73
Evan
AC9TU

Hi Evan,
I should stop building QDX's, hahaha.  
73 buddy - yes my thinking exactly - I don't want to start messing up the board - AGAIN.  I actually have not tried it on the 20 meter band yet.  Will see tonight when the band opens a bit more.  Will check C30 but quite positive I have 39pf cap in correctly.   Yes - I have squeezed the last few turns together - no change.  I was think perhaps I need to add 3 more turns or so ??
Will feedback.
73
Timo.  

Not 100% sure but I think the LPF for 17 and 20 are together.  Perhaps Evan will be able to answer this question more accurately.  
73
Timo

Timo and Al,

I expect an LPF problem to show up as a low overall value in the RF Sweep.  Note that Timo's sweep for 20 meters does get to 0 DB, and the peak is higher than desired.  To me, that indicates the low pass filter is OK.  There could be interaction, but it does not make sense from a theoretical standpoint.  Also, my interpretation of Timo's post is that the output is 4.5 watts which again tests the LPF as being OK.

I must remind everyone that I am an Electrical and Industrial Controls Engineer.  I have not done much RF design work.  Others with more experience and knowledge might want to chime in.
73
Evan
AC9TU

Timo,

You asked, "I was think perhaps I need to add 3 more turns or so ??"

My response is that it should be too many, but taking off turns is much easier than adding more.  You can try the total of 23 turns and if it overshoots in the other direction or reduces the signal values on the left of the sweep, then remove a turn at a time to see how that impacts the sweep.  Note that you can unwind the turns and leave the extra wire connected until you get what you want.

73
Evan
AC9TU

All,
To educate myself I did an Elsie plot of the 20m LPF used in the QDX-hHF (my designation) model.

I haven't tried doing a BPF filter yet. The assembly manual indicates the 20m LPF is not shared with other bands.
Timo's plot at 14MHz is so low that I would suspect it to be the problem. But I see your point Evan.

Maybe someone who has a "good" 20m QDX-hHF could share their RF plot.

--Al

Well, it is surely bringing in the FT8 20-meter contacts.  This makes it even more difficult to understand why the RF graphs looks so bad.
73
Timo
 

Timo,

I would not mess with it unless the 20-meter output is less than 4 watts into a 50 Ohm dummy load.  If that output is low, the LPF could be part of the issue, as Al suggested.

It would be an interesting experiment to see what adding 3 turns to the final 3 turns would do to the sweep results.

There could be an issue with IC3.  Of the 4 pins of IC3 used to select the correct capacitor, only pin 10 should be at 2.5 volts.  The others should be close to 0.  You can compare the voltage of the other band pins when the corresponding band is selected to verify what the off voltage should be.

73
Evan
AC9TU

Thanks Even, will do.  I presume that is during the TX phase where only one pin will be 2.5V, correct.  What is concerning is the fact that I don't know how fantastic the RX could be if the 20 meters where testing correctly.  Currently it is reading between 15 to 20 contacts in each 15 second RX over, which is really cool, but can probably be better.  
73
Timo

Timo,

You DO NOT need to transmit to measure the voltage on IC3.  IC3 is not a part of the transmitter circuits.

My experience with my QDX is that if I can hear them, I can only connect with them if the SNR is over -20db.  Most of the time, it needs to be -15db or better.

73
Evan
AC9TU

Thank you Hans, Evan and Al for all the help and input.  As always your input is highly appreciated.  This issue is then resolved. 
73
Timo
ZR6TM 

No no, I was wrong!
Sorry...
--
73, Oleh, UY2ZA

Hi,
I am using the first BPF for 20m only, whereas I added 22p to C30 and added 2 turns to L12. Still not good but better. 
For 17m I use a 15m BPF which is nicelly optimized for 15m and at 17m only - 5dB! 

73 Bojan S53DZ

Hi,

Just finished building my high-band QDX and got caught out by the dodgy looking RF sweep on 20m - all other sweeps and built-in diagnostics look good on the other bands. But, as Hans says, the performance on 20m is extremely good regardless!

It might be worth adding this as a note to the documentation or troubleshooting guide?

73's
Jim G7FRI

It looks to me that what Hans wrote on Jan 7 is not true. It looks to me that 17m shares the bandpass filter with 15m. Both show around 20 - 22 MHz passband.
The 20m is on its own, passing 17.5 MHz upwards. Perhaps we just need to change C30.
How easy is it to get the through hole capacitors out of the board?
73 Brian VK4BAP

Brian,