|
Hi All,
I have been working on increasing the power out on 40 meters.? Original I had just under 3 Watts using the QRP Labs dummy load.? I first took two turns off of L510 and power went up slightly.? I ended up taking 4 turns off of L513.? I'm now up to 3.75 Watts.? I felt I might get more power if I took more turns off of L513 but I'm starting to get gun shy.? I've included an LPF plot at the final configuration.? Should I leave it alone or try to go for 5 Watts?
?
From the plot it looks like I may have shifted up the cutoff frequency quite a bit.
?
72,
Steve, W2MY
|
Steve, you should look at your output on 40M with a spectrum analyzer and check the levels of the 2nd and 3rd harmonics, which should be down at least 40dB from the main carrier at 7MHz.? By moving out the LPF cutoff, you may have moved the filter such that it doesn't meet those requirements.?? The 2nd harmonic doesn't need 40dB of attenuation, since it is naturally lower - but you can see on your curve that the 2nd harmonic at 14MHz is only getting about 10dB of attenuation - which may or may not be enough.? And you can't see with this simple graph what is happening at 21MHz.
?
A better approach than just removing windings may be to figure out why your filter has that lowered bench of attenuation at the right side of the passband.? The filter design itself is flat all the way across the passband, so that lowered part comes from parasitics or something else.
|
Thanks Stan,
?
OK, maybe I¡¯ll learn something new. ? ?So maybe I should rewind the toroids back to stock, and then investigate the parasitics? ?I have a nanovna but not sure if it will help. ?What components might I check regarding parasitics?
Steve
|
I really don't know.? I have the same problem, to a lesser degree.
Maybe someone else knows what causes this...
Stan KC7XE
|
Steve,
?
When I built my 20m-80m Rev. 1 (tr)uSDX some time ago, I had to remove one turn on a toroid in order to realize 5W (or very close to it) output on one band. I don't recall the band or the exact power before I removed one turn but I know it made a much larger difference in power than you are seeing per turn removed and it did not significantly disturb the 2nd harmonic knee position on the (tr)uSDX LPF.
?
You may or may not find Manuel's (DL2MAN) (tr)uSDX instructional videos to be useful and applicable to your QRP Labs kit but I found them to be very helpful when building my (tr)uSDX and adjusting the LPFs. Here's Manuel's video about using a VNA for LPF analysis and power/efficiency changes, which is probably the video most applicable to QRP Labs projects within Manuel's instructional playlist.
If you need to cross reference Manuel's procedure and the (tr)uSDX design, in general, to your QRP Labs rig, the schematics here might be useful:
I prefer the "Alternative, very comprehensive Schematic, drawn by KD4SGE & WA4ITD" about halfway down the page vs. the original schematics.
?
I have not needed to do anything but tweak a couple of coils (squeeze together or space apart) with either my QMX or QMX+ kits. My QMX does have more even and higher (overall) power output across it's 5 bands than my QMX+ has across 11 bands. I attribute this to more LPF toroid sharing between bands in the QMX+ than the QMX, but I could be wrong.
?
73,
Cal (AD8Q)
|
The through hole parts in the LPF i can replace, but the smd switching transistors would be beyond my skill. ??
|
I guess the problem is to identify the cause of the 6 db dip in my lpf plot. If it is parasitics, then removing them would give me 5 Watts out. ??
|
On 06/12/2024 12:58, Steven, W2MY via groups.io wrote: but the smd switching transistors would be beyond my skill. What I do is use sharp side cutters on the body. Break it then remove the pins. 73 Alan G4ZFQ
|
What I'd like to see is a typical healthy plot of the 40m LPF.? The only plot I've seen is on page 88 of the Operating Manual for the QMX.
I rewound the 40m LPF toroids and it moved the plot to the left slightly.? Now the attenuation appears to be -20 db for 20 meters instead of -10 db when I removed all the turns from L510 and L513.
?
As far as debugging information goes, the only thing I can say is that the plots for 160m and 80m look great and I am getting 6 Watts out on those bands, and on all the other bands the LPF plots are similar to the 40 meter one with the uneven variation of the plateau of -3 to -6 db, and the output on those bands is more like 2 Watts.
?
I've resoldered all the connections.
?
Regards,
Steve
|
On Fri, Dec 6, 2024 at 04:58 AM, Steven, W2MY wrote:
The through hole parts in the LPF i can replace, but the smd switching transistors would be beyond my skill. ??
Yes--I see Steve. Looking at my QMX+, which has its lid off, ATM, it appears to be tough/risky to attempt to connect a VNA to an LPF. Lifting one side of the? 470k SMD 201 resistor might work but I always worry about lifting the pad off the board unless I use hot air to first remove the whole part and then resolder one end. Both ends need to be hot. The toroids and other nearby components make it extra challenging.
?
I recall builders here reporting they've made VNA connections to LPFs (QMX / QMX+ ?) and made sweeps though, so maybe someone will post a tip or details. I'd like to know a safe way to connect my Nana VNA Ultra to my QMX+ and even my QMX.
?
Cal (AD8Q)
|
On Thu, Dec 5, 2024 at 04:15 PM, Stan Dye wrote:
A better approach than just removing windings may be to figure out why your filter has that lowered bench of attenuation at the right side of the passband.? The filter design itself is flat all the way across the passband, so that lowered part comes from parasitics or something else.
Stan, Steve,
?
I just checked and I have that "bench" on 40m too. I do have better than -20dB at the 2nd harmonic frequency though (attached screenshot), which is the worst attenuation of all my LPFs/bands. Others are about -27dB at the 2nd harmonic or a little better.
?
Cal (AD8Q)
|
Plotting QMX LPF's with a nanoVNA is not something I've done but:
?
With the rig off, there should be zero volts across the 1n4007 diodes, isolating the LPF and making it possible to simply connect it up to the VNA for a two port test of the LPF.? The nanoVNA drives the device under test with a very low level signal, not big enough for the 1n4007's to conduct.? The 47uH inductor blocks any RF, so it will not be seen by the nanoVNA.? I would expect a somewhat similar plot from the nanoVNA as the QMX diagnostics give, but 1n4007 capacitance with the rig off, and the effects of other components upstream and downstream from the filter when the rig is transmitting will make the nanoVNA plot somewhat different.
?
Hopefully somebody who has done this will report their technique and results.
?
Doubt you need to pull that 470k resistor (or the 1n4007's), but two pin surface mount parts are easy enough to pluck off using two soldering irons.? For 3 pin SOT23's, you can first lift the one lead and tip the part up a bit, then go after the other two.? Clipping SOT23 parts (and bigger) with very small fine tipped diagonal cutters or an Xacto knife can work with the right technique, but you risk mechanically ripping the pads off the board if done wrong.? These boards are factory stuffed with no-lead solder, so you need some really hot soldering irons to melt it.
?
Jerry, KE7ER
?
On Fri, Dec 6, 2024 at 09:39 AM, Cal (AD8Q) wrote:
On Fri, Dec 6, 2024 at 04:58 AM, Steven, W2MY wrote:
The through hole parts in the LPF i can replace, but the smd switching transistors would be beyond my skill. ??
Yes--I see Steve. Looking at my QMX+, which has its lid off, ATM, it appears to be tough/risky to attempt to connect a VNA to an LPF. Lifting one side of the? 470k SMD 201 resistor might work but I always worry about lifting the pad off the board unless I use hot air to first remove the whole part and then resolder one end. Both ends need to be hot. The toroids and other nearby components make it extra challenging.
?
I recall builders here reporting they've made VNA connections to LPFs (QMX / QMX+ ?) and made sweeps though, so maybe someone will post a tip or details. I'd like to know a safe way to connect my Nana VNA Ultra to my QMX+ and even my QMX.
?
Cal (AD8Q)
|
On Fri, Dec 6, 2024 at 09:51 AM, Alan G4ZFQ wrote:
What I do is use sharp side cutters on the body.
I use a #15 scalpel blade and sever the leads at the body.? Hold the scalpel near vertical and use a rocking motion to shear the lead.? Pluck off the body then desolder the leads individually.
73, Don N2VGU?
|
On Fri, Dec 6, 2024 at 12:02 PM, Jerry Gaffke wrote:
These boards are factory stuffed with no-lead solder, so you need some really hot soldering irons to melt it.
Thanks for your discussion of the VNA approach, Jerry!
?
When I had to replace my SOIC gates after the finals in my QMX+ blew after experimenting with EEPROM settings in an ATU, I could not get the solder to melt with hot air until I was already far too fearful for the health of the PCB and nearby components, despite first masking surrounding areas with both 3M insulating fiberglass tape covered by 3M stainless steel tape. The technique I developed, which I've since learned is similar to the Chip Quik SMD rework kit, is I first used solder wick and an iron to remove as much of the difficult-to melt Pb-free solder as possible. Then I added some good 'ol regular lead/tin solder to the SOIC legs and more flux too. After "re-alloying" the joints, they melted quickly with hot air and I picked off the SOIC using tweezers like a dream! Nearby components and the board were unaffected.
?
73,
Cal (AD8Q)
?
On Fri, Dec 6, 2024 at 10:43 AM, Cal (AD8Q) wrote:
I just checked and I have that "bench" on 40m too. I do have better than -20dB at the 2nd harmonic frequency though (attached screenshot), which is the worst attenuation of all my LPFs/bands. Others are about -27dB at the 2nd harmonic or a little better.
|
The LPF's on the QMX are not symmetrical.
I would guess that T501 is driving into an impedance other than 50 ohms resistive,
but close enough to 50 ohms that you would get meaningful plots from a nanoVNA.
?
Jerry, KE7ER
?
?
On Fri, Dec 6, 2024 at 12:02 PM, Jerry Gaffke wrote:
Plotting QMX LPF's with a nanoVNA is not something I've done but:
?
With the rig off, there should be zero volts across the 1n4007 diodes, isolating the LPF and making it possible to simply connect it up to the VNA for a two port test of the LPF.?
|
On Fri, Dec 6, 2024 at 10:43 AM, Cal (AD8Q) wrote:
I just checked and I have that "bench" on 40m too. I do have better than -20dB at the 2nd harmonic frequency though (attached screenshot), which is the worst attenuation of all my LPFs/bands. Others are about -27dB at the 2nd harmonic or a little better.
?
Oops--I forgot to clarify in my last post: Others are about -27dB at the 2nd harmonic or a little more favorable.
|
On Fri, Dec 6, 2024 at 12:18 PM, Jerry Gaffke wrote:
The LPF's on the QMX are not symmetrical.
I would guess that T501 is driving into an impedance other than 50 ohms resistive,
but close enough to 50 ohms that you would get meaningful plots from a nanoVNA.
?
Jerry, KE7ER
Thanks again, Jerry!
?
Manuel (DL2MAN) advises the same in his (tr)uSDX instructional video. He says, due to not being a 50 ohm system,? the LPF passband is not fully accurate from the nanoVNA, but it's good enough for evaluating the attenuation at the 2nd harmonic. At least that's the gist of it, as I recall it.
?
Cal (AD8Q)
|
Steve
?
Here are plots of LPF from 2 QMX+ and 1 QMX
?
Hope they help
?
73
?
Ian G4GIR
?
toggle quoted message
Show quoted text
----- Original Message -----
Sent: 06/12/2024 15:23:27
Subject: Re: [QRPLabs] How many Turns to take off a Toroid? #QMXplus
What I'd like to see is a typical healthy plot of the 40m LPF.? The only plot I've seen is on page 88 of the Operating Manual for the QMX.
I rewound the 40m LPF toroids and it moved the plot to the left slightly.? Now the attenuation appears to be -20 db for 20 meters instead of -10 db when I removed all the turns from L510 and L513.
?
As far as debugging information goes, the only thing I can say is that the plots for 160m and 80m look great and I am getting 6 Watts out on those bands, and on all the other bands the LPF plots are similar to the 40 meter one with the uneven variation of the plateau of -3 to -6 db, and the output on those bands is more like 2 Watts.
?
I've resoldered all the connections.
?
Regards,
Steve
|
On Fri, Dec 6, 2024 at 09:39 AM, Cal (AD8Q) wrote:
I recall builders here reporting they've made VNA connections to LPFs (QMX / QMX+ ?) and made sweeps
I have done this LPF scan with my nanovna.? It is quite simple with the QMX+, if you happened to install your diodes with the cathode side up.? With the QMX you need to tack some wires on the diode pads, so it is a bit harder.?
?
Maybe best shown with a photo; see attached.
- You inject the S11 port signal into the filter at the cathode of the diode.
- You should really do this via a dc blocking capacitor, like .01uF or so.? But in my observations, it makes almost no difference in the filter shape, so I don't do it.? If you don't use the blocking capacitor, however, it will give you false readings on LPFs not tuned to: because somehow the nanovna's direct connection at least partially enables the filter it is attached to, by referencing it to the nanovna's ground.
- You receive the S21 port signal from the antenna connector.
- You need to turn on the QMX, and tune to the band you are testing, so it will 'turn on' the filter via the switching diodes.? BUT BE CAREFUL: disconnect the key and/or otherwise disable transmit, so you don't accidentally bump something and transmit into your nanovna - that would definitely break it.
- You see the filter shape.? Note that there is usually (on mine) a little narrow notch that appears within the band you are tuned to - but this is some artifact, because the actual frequency you are tuned to is at a point bumping up out of that notch back up to the passband level.
One more note: it seems that you should be able to do this scan with the qmx 'off', and connect the S21 port to the other diode's cathode at the output of the filter.? This does give a reasonably good view of the passband design and the cutoff frequency, but the deep cutoff notch and the stop-band are not seen for some reason, at least not on the lower bands.? I don't know exactly why.? But it is better anyway to do the scan with the QMX on and filter selected, because you get a better view of parasitics that happen in the full real filter path.? Oh, and as always, I calibrated the nanovna at then ends of my test cables to give an accurate result.
?
I attached a couple of photos of my setup, and one of my 40m/30m LPF scan.
I hope this is helpful.? Questions are welcome.
Stan KC7XE
?
|
Sorry, images from NanoVNA scan of QMX+ LPF attached.
|
Steven, W2MY
Donald S Brant Jr