Keyboard Shortcuts
ctrl + shift + ? :
Show all keyboard shortcuts
ctrl + g :
Navigate to a group
ctrl + shift + f :
Find
ctrl + / :
Quick actions
esc to dismiss
Likes
Search
QMX build for [40m/30m] , [20m/17m] & [15m/12m/10m]
#qmx
|
I just joined the group after ordering a QMX kit. If possible I¡¯d like to build it from the outset for the band groups stated above. I accept that this will require total redesign of the filters. For simulation I intend to use QUCS which is now available via Ubuntu Software. And I have a NanoVNA to help with the practical work. ? But before thinking too hard about filter design I¡¯d like to know more about the likely performance of the remaining hardware at higher frequencies. Particularly the Tx output power and Rx noise figure (especially as there is no LNA). To look in to this these things I was thinking that I might first link out the filters with a direct coax connection???and thus measure the raw Tx power and Rx NF available at the BNC before filter losses. And then I wondered if anyone else had already done that? ? Another issue is that my intended radio would be for seven bands. Does anyone know if the firmware can accommodate seven bands, and whether or not the user can determine what the bands will be without having to get modified firmware? |
|
Hi Nick Up to 6 bands are user configurable. There's nothing to stop you defining a band named 1012 with a frequency range 24MHz to 30MHz. It would have to refer to the same BPF and LPF that's all. Remember you have 4 BPF taps and 3 LPF taps. In my opinion 7 amateur bands is probably stretching it and will be challenging!? The TX PA and Receiver are fundamentally the same or similar to QDX. I measured good receive sensitivity on QDX even up to the 8m band (40MHz). TX power drops off slightly with frequency but I usually get at least 4W on 10m. 73 Hans G0UPL On Sun, Aug 13, 2023, 12:32 PM Nick G4IKZ via <nds12321=[email protected]> wrote:
|
|
On 13/08/2023 09:17, Nick G4IKZ via groups.io wrote:
I accept that this will require total redesign of the filters. For simulation I intend to use QUCSNick, Note that others have not been able to successfully simulate this sort of filter. There are significant stray capacitances to make it difficult. One design had an unwanted resonance that ruined 20m. Be prepared for trial and error, fitting BPFs when you know it is working. 73 Alan G4ZFQ |
|
Hello Alan, Nick et Al,
toggle quoted message
Show quoted text
I have a 40/30/20/17/15/12/11/10 QDX on the test bench QRV 40m this minute. I¡¯ve run it most of the day on 12m with some good results after having done a while on 15m and a whole night on 20m (got some spots from a ballon so Rx is working ok). Earlier in the week if done mixes of 17m, 10 and 12. The LPF mix I have is 40/30; 20/17/15; and 12/11/10m. BPF 40/30; 20; 17/15; and 12/11/10m. It¡¯s still NQR exactly but further to be done. LOFs are good with 20m having worst at -50 dB for 2nd Harmonic. All others are very good according to my TinySA though I need Hans to validate the accurate spectrum before I release all and we go for a proper solution for both QDX and QMX (I¡¯ve wound the toroids) and will build a QMX as soon as the addition board arrives here. More to come! 73 Ross Sent from my iPhone 6 On 13 Aug 2023, at 13:22, Alan G4ZFQ <alan4alan@...> wrote: |
|
I should also say all my toroids in QDX are T37-17s and QMX T30-17s. Also BN61-202s (RWTST wound) and, atm, an FT37-61 in the Rx as I found the inductance of the input to the transformer and 2-3pF from the FST output was acting as a parallel resonant at 28MHz (a figure that keeps popping up)
toggle quoted message
Show quoted text
But full specs will be made available soon. 73 Ross Sent from my iPhone 6 On 13 Aug 2023, at 13:58, Afghan Kabulldust via groups.io <kabulldust@...> wrote: |
|
Alan, quite a bit of frustration, believe me! HB was much much easier.
toggle quoted message
Show quoted text
73 Ross Sent from my iPhone 6 On 13 Aug 2023, at 14:54, Alan G4ZFQ <alan4alan@...> wrote: |
|
Alan,
toggle quoted message
Show quoted text
I do a lot of LTSpicing and have manage to simulate Class E Push Pull and have the actual circuit work, however, in this case it is almost impossible to simulate as you so aptly pointed out. The BPF that sort of works, I¡¯ve wound another and have accurately measured the zl on each winding to try to understand what the parasitic capacitances are, a lot does not make sense as it¡¯s very difficult to model all the associated circuitry and it¡¯s L and L values and even the network that it results in. I tried direct C between inductor taps and the connection to the first half of the GST to see if I could reduce the extraneous networks but the resulting BPFs were all very sharp, which should not be, so series caps from circuit traces and FST are really having a big effect. 73 Ross Sent from my iPhone 6 On 13 Aug 2023, at 15:10, Afghan Kabulldust via groups.io <kabulldust@...> wrote: |
|
On 13/08/2023 15:20, Afghan Kabulldust via groups.io wrote:
Alan,Ross, Hans has made some really innovative designs but that BPF is a masterpiece of complicated simplicity. |
|
Having run some simulations of re-banded LPFs I'm now looking at the Rx filtering, and am wondering if would be possible and beneficial to modify that. As it stands the Rx filtering appears to be a single band-switched series resonant circuit - that runs between the Rx port of the antenna switch and the balun at the mixer input. A possible problem with it is that (by definition) if a single band setting is narrow enough to be doing any good then it might not be wide enough to allow band sharing, especially on the higher bands where Rx NF is more important.
However the Tx LPFs are already providing good attenuation above the wanted band. So the Rx might be OK with some switched high-pass filters at that place instead. These would comprise a single series C and a single parallel L. The mux for the series caps already does what is needed. So to make it work would just need the switched Ls to be reconfigured as parallel components to ground. Has anyone looked in to this? |
|
Nick, Your idea is spot on,? I think. It allows for establishing a BP width tailored to the multi-band cases. One downside is that a 2 pole HPF (one inductor, one capacitor) will have a rather shallow response slope. Another is the cost of course. ?I would also recommend terminating the HPF with a load resistor. The Tayloe detector has an interesting input impedance property according to Dan Tayloe himself: elevated impedance at or close to the clock frequency, declining to very low values at frequencies removed from the clock.? Literature on this aspect of the Tayloe detector is virtually nil. A resistor might tame that property somewhat. JZ KJ4A? On Thu, Aug 17, 2023, 8:25 AM Nick G4IKZ via <nds12321=[email protected]> wrote: Having run some simulations of re-banded LPFs I'm now looking at the Rx filtering, and am wondering if would be possible and beneficial to modify that. As it stands the Rx filtering appears to be a single band-switched series resonant circuit - that runs between the Rx port of the antenna switch and the balun at the mixer input. A possible problem with it is that (by definition) if a single band setting is narrow enough to be doing any good then it might not be wide enough to allow band sharing, especially on the higher bands where Rx NF is more important. |
|
I agree that the roll-off of a 2-element HPF isn't great. But as frequency gets lower a given dB/octave equates to more dB/MHz. So I think a switched HPF might still be a good way to keep the high-power broadcasters out of the mixer.
Regarding cost I wasn't thinking about extra hardware - just a reconfiguration of the switched C and switched L that are already present. Hopefully just a bit of track-cutting and rewiring to turn the existing switchable bandpass filter in to a switchable high-pass filter. |
|
I'm very eager to see how that works out, Nick!
Good luck!...JZ KJ4A On Thu, Aug 17, 2023 at 10:52?AM Nick G4IKZ via groups.io <nds12321@...> wrote:
|
|
One more thing, Nick:
toggle quoted message
Show quoted text
Because the mux switches you will use to selectively ground the inductors have their parasitic capacitance, you may still see some unplanned strange resonances. Hopefully they won't matter. TANSTAAFL, as the saying goes! JZ On Thu, Aug 17, 2023 at 10:53?AM John Zbrozek <jdzbrozek@...> wrote:
|
|
I also wonder why QDX/QSX are missing some kind of matching. In many amateur designs using that kind of mixer there is usually a matching to 50 Ohm (see example attached, Polish transceiver Husarek). I am sure the matching keeps filters characteristic as designed, maybe even helps with the parasitic resonances.
It is worth to check one day. 73 Pawe? SO8FM. |
|
I agree. And even if this worked out it would only give significant attenuation for signals around the medium wave AM broadcast bands. For anything else it's just going to be few dB. But how else do we avoid NF degradation when sharing simple BPFs for the high bands? Any transmit-only WSPR guys might not mind poor Rx NF. So for them a 7-band Tx might still be good fun.? But those wanting 7-band Rx might not be so happy.
|
|
Pawe?,
toggle quoted message
Show quoted text
This is tricky business! Since Han's designs use the Tayloe circuit as a downconverter to an IF, and not as a direct conversion to baseband detector, the input impedance of that circuit at the receive frequency is somewhere between its extremes, and rather unknown. Maybe I should stop referring to it as a detector and call it a downconverter? Are those zero ohm resistors in your schematics? Dobry dzie¨½ 73 JZ KJ4A On Thu, Aug 17, 2023 at 12:02?PM Pawe? H <pawel@...> wrote:
|
|
Nick,
The Tayloe circuit is incredibly robust, tolerating all manner of egregious sins! It also possesses its own intrinsic BPF property. I would give your idea a try. We might all be pleasantly surprised by the RX performance. JZ On Thu, Aug 17, 2023 at 12:07?PM Nick G4IKZ via groups.io <nds12321@...> wrote:
|
|
John,
I would say that conversion from MHz range to tens of kHz range makes no difference to having 0 IF.? For me the important is that the circuit have correct impedance for the particular band frequency. We are calculating all of these filters for 50 Ohms in and out, right? The matching resistors are these 100 Ohms. There is always connected two to the signal path, so in parallel it gives 50. Zero resistors are another option for the matching, but for some unknown to me reasons the designer choose another option and left them just in case. I assume 1nF caps also take part in some kind of filtering. I also saw the solutions with single series 33 Ohm resistor directly in front of the mixer splitter transformer. 73 Pawe? SO8FM? |
|
Pawe?, What then is the input impedance of the Tayloe downconverter at the receive frequency? You can read Dan Tayloe's very interesting response to Steve K1RF's question here: 73, JZ KJ4A On Thu, Aug 17, 2023, 3:21 PM Pawe? H <pawel@...> wrote: John, |
to navigate to use esc to dismiss