(tr)uSDX power, efficiency and filter optimisation

Post Reply
Ohwenzelph
 

Posts: 207
Joined: 01 Jan 2022, 03:47

Re: (tr)uSDX power, efficiency and filter optimisation

Post by Ohwenzelph »

Manuel, when you make changes, how big a change do you make at each step? Your cap values for 10 meters were very close to the theoretical ones for that FET.
Faraaz VK4JJ
 

Posts: 215
Joined: 30 Dec 2021, 21:56

Re: (tr)uSDX power, efficiency and filter optimisation

Post by Faraaz VK4JJ »

Ohwenzelph wrote: 10 Apr 2022, 17:42 Looking for the "practical" values for the higher bands: not there, still looking...
They don't exist yet, AFAIK.

Maybe you will be the first to discover and then hopefully publish your results :)

High-band frequencies will always be more difficult to get good optimisation and efficiencies with due to higher core losses.
DL2MAN
 

Posts: 706
Joined: 30 Dec 2021, 19:18
Contact:

Re: (tr)uSDX power, efficiency and filter optimisation

Post by DL2MAN »

Ohwenzelph wrote: 10 Apr 2022, 21:52 Manuel, when you make changes, how big a change do you make at each step? Your cap values for 10 meters were very close to the theoretical ones for that FET.
I ususally don´t mess with the caps. Only with the Toroids.
If I mess with caps, it´s usually only the shunt capacitor, and this I modify by observing the class E Waveforms and modify the capacitor until the class E Waveforms are looking right.
How much you need to change it depends on the waveform.
Usually caps are within the nominal values +/-5%
Ohwenzelph
 

Posts: 207
Joined: 01 Jan 2022, 03:47

Re: (tr)uSDX power, efficiency and filter optimisation

Post by Ohwenzelph »

Thank you, what would you do if you saw the waveform I posted earlier in this theme? If you can tell anything from it, clearly I am not sure what it means. Thanks again.
g7jur
 

Posts: 35
Joined: 18 Feb 2022, 22:49

Re: (tr)uSDX power, efficiency and filter optimisation

Post by g7jur »

SA4DHT wrote: 09 Apr 2022, 09:35
SA4DHT wrote: 02 Apr 2022, 09:54 I have a kit from *Rxxxx* batch 3. I am facing the same issues regarding efficiency as others have reported at 12 V in:
20 m 3.8W 45.7%
30 m 3.4W 50.0%
40 m 2.7W 50.5%
60 m 2.6W 51.5%
80 m 3.8W 51.1%

Started to check 20 m with a high resolution thermal camera and find C12, C19 and C16 getting very hot about 70 deg C. So there energy is wasted. I will try to replace them. I have NP0 here but not sure if they are 50V or 100V. Another question is how the target capacitance is achieved on good versus bad board. Wrong type and voltage is one possibility another if one or two capacitors are used per position in order to get the goal capacitance. MOSFET BS170 are getting hot too at about 40 deg C one beeing at higher temperature so current sharing is not perfect. T2 is getting hot too about 40 deg C. On one of the videos Manuel has posted it seems to me T1 and T2 are winded with bigger diameter wire than 0.4 mm.
In order to solve the above situation on my batch 3 from *Rxxxx* kit:

1) Checked the internal efficiency calculation and found it about 10% too low compared to measuring with external instruments. Measured the trace used as current shunt with a Keithley 197 and found it to be 19 mOhm instead of 17 mOhm as set as default in 8.6 R Shunt. A appropriate instrument is needed in order to measure such low resistances!

2) Ordered Kemet C0G 0805 100V 5% low ESL capacitors and replaced all bandfilter capacitors.

Results after that at 12V in:

20 m 3.09W 68.67%
30 m 3.78W 77.35%
40m 3.23W 80.75%
60m 3.38W 80.48%
80m 6.00W 82.78%

2nd harmonic looks good on all bands. I might try to trim the LX1 toroids to adjust power a little bit, but on 13.8V in I have more than 4W on all bands. U9 is a LM833 and works ok but sideband suppression could be better. Noise is about S3 even with a dummy load on all bands.
I have made some QSOs and I am getting good reports. Audio seems to be ok even using the internal mic.

Hope this is useful.

73 de Dirk

After reading your post, I set the R Shunt setting to 19. this made thing much better.
I will measure this track, as I can do 4 wire measurements.
Maybe the *Rxxxx* kit has thinner copper foil on the PCB ?
Dirk well done.
Philip G7JUR

Hello all.
Just measured the R shunt by running one amp through it and measuring the voltage drop.
Which is 18.8 mV 18.8 miliohms.
Philip G7JUR
Last edited by g7jur on 18 Apr 2022, 21:38, edited 1 time in total.
DL2MAN
 

Posts: 706
Joined: 30 Dec 2021, 19:18
Contact:

Re: (tr)uSDX power, efficiency and filter optimisation

Post by DL2MAN »

SA4DHT wrote: 02 Apr 2022, 09:54 Measured the trace used as current shunt with a Keithley 197 and found it to be 19 mOhm instead of 17 mOhm as set as default in 8.6 R Shunt. A appropriate instrument is needed in order to measure such low resistances!
While I totally agree about your statement on "appropriate instrument needed", and I don´t have equipment to measure such a low resistance, I would like to ask:

Did you check your REAL Output Power with Oscilloscope over Dummy Load (or any other high precision Power meter) and Input Current, and calculated your real Efficiency ?
Otherwise, adjusting the Setting in Menu 8.6 would just be Cosmetics, but not taking care about the real problem: Low Efficiency.

73 Manuel; DL2MAN
DL2MAN
 

Posts: 706
Joined: 30 Dec 2021, 19:18
Contact:

Re: (tr)uSDX power, efficiency and filter optimisation

Post by DL2MAN »

Ohwenzelph wrote: 11 Apr 2022, 13:59 Thank you, what would you do if you saw the waveform I posted earlier in this theme? If you can tell anything from it, clearly I am not sure what it means. Thanks again.
Very Hard to Tell from your Picture, as there is no units and scale. But let´s look at the timing. If the Scope Photo is right, this would mean: something totally off, as your highest Drain Peak develops, when gate is switched on.

Look at this curve in Comparison:
Class_E.jpeg
Class_E.jpeg (78.19 KiB) Viewed 9776 times
Drain Peak rises and falls in the period while Gate is off. This is 80m (because I had it just sitting on my hard disk), but the principle is the same.

When you look at this document:
https://people.physics.anu.edu.au/~dxt1 ... design.pdf
Page 8 internally, Page 16 according Index,
your Waveform is like upper left example.
This would Indicate to Decrease Shunt Capacitor -> Start with this
and Decrease Serial resonance Cap (-> which I would not do), but Increase Serial resonance Coil Inductance instead.
Do one after another, and document the change of the waveforms.
This will show you the direction.

73 Manuel; DL2MAN
hb9hem
 

Posts: 36
Joined: 07 Jan 2022, 04:39

Re: (tr)uSDX power, efficiency and filter optimisation

Post by hb9hem »

Does anyone have a good Spice model for the amplifier? I built one for the 40m portion, my goal was to play with parasitics in order to find out what would explain the low efficiency. But my model produces barely 2W power, so something is wrong. The efficiency I'm getting is quite ok. Nonetheless, what I have is not quite right...
g7jur
 

Posts: 35
Joined: 18 Feb 2022, 22:49

Re: (tr)uSDX power, efficiency and filter optimisation

Post by g7jur »

DL2MAN wrote: 12 Apr 2022, 07:28
SA4DHT wrote: 02 Apr 2022, 09:54 Measured the trace used as current shunt with a Keithley 197 and found it to be 19 mOhm instead of 17 mOhm as set as default in 8.6 R Shunt. A appropriate instrument is needed in order to measure such low resistances!
While I totally agree about your statement on "appropriate instrument needed", and I don´t have equipment to measure such a low resistance, I would like to ask:

Did you check your REAL Output Power with Oscilloscope over Dummy Load (or any other high precision Power meter) and Input Current, and calculated your real Efficiency ?
Otherwise, adjusting the Setting in Menu 8.6 would just be Cosmetics, but not taking care about the real problem: Low Efficiency.

73 Manuel; DL2MAN
Totally correct, when I have time I will measure the shunt track. and investigate in greater detail.
SA4DHT
 

Posts: 19
Joined: 10 Jan 2022, 14:13

Re: (tr)uSDX power, efficiency and filter optimisation

Post by SA4DHT »

DL2MAN wrote: 12 Apr 2022, 07:28
SA4DHT wrote: 02 Apr 2022, 09:54 Measured the trace used as current shunt with a Keithley 197 and found it to be 19 mOhm instead of 17 mOhm as set as default in 8.6 R Shunt. A appropriate instrument is needed in order to measure such low resistances!
While I totally agree about your statement on "appropriate instrument needed", and I don´t have equipment to measure such a low resistance, I would like to ask:

Did you check your REAL Output Power with Oscilloscope over Dummy Load (or any other high precision Power meter) and Input Current, and calculated your real Efficiency ?
Otherwise, adjusting the Setting in Menu 8.6 would just be Cosmetics, but not taking care about the real problem: Low Efficiency.

73 Manuel; DL2MAN
Yes, I measured the output power with an oscilloscope across a 50 Ohm dummy load and the input power with a multimeter. The output power was matching the internal measurement, but the input power transmitting minus idle power in relation to the output power gave a somewhat better efficiency. After that I made a 4-wire measurement of the shunt trace. As I wrote this contributed only to about 10%. The main improvement has been the replacement of all bandfilter capacitors to good quality ones. I will even try to change C7 + C23 in case these are the wrong type too.

73, de Dirk SA4DHT
6k2kus
 

Posts: 8
Joined: 19 Mar 2022, 02:14

Re: (tr)uSDX power, efficiency and filter optimisation

Post by 6k2kus »

Today I received a kit from Aliexpress and constructed it. After the test I found the following result out of the box. What I have done new is to have 19 turns for L15 instead of 22 as Manuel updated:

Efficacy: around 80%
Power: around 5W from 12V input (more on the 80m band)
2nd, 3rd harmonics below -53dB

On all bands.
It's crazy and of course the triumph of (tr)uSDX.
Beofre I had *Rxxxx* 3rd batch. I could not obtain 50% efficiency and had to compromise the power with harmonic suppression -43dB.

I have done nothing different this time. So that was not our fault. Just I wanted to report. Maybe I can do more finetuning but I am just satisfied.
Ohwenzelph
 

Posts: 207
Joined: 01 Jan 2022, 03:47

Re: (tr)uSDX power, efficiency and filter optimisation

Post by Ohwenzelph »

"19 turns for L15 instead of 22 as Manuel updated"
Please, Where was this update?
Ohwenzelph
 

Posts: 207
Joined: 01 Jan 2022, 03:47

Re: (tr)uSDX power, efficiency and filter optimisation

Post by Ohwenzelph »

W6LKB "Can you explain how you built the transformer and where you connected it exactly?"
viewtopic.php?p=1686#p1686
https://drive.google.com/file/d/1t6T_75 ... sp=sharing
Find 20220321 truSDX_E_ Filterabgleich(1).docx
Open short and load standards to calibrate nanovna at ends of cables
Open short and load standards to calibrate nanovna at ends of cables
00D8A80D-3BE1-46B2-B2DE-CDEA50A26E73.jpeg (204.88 KiB) Viewed 9488 times
Might be important at uhf/vhf and not at hf, I do not know.
4:1 Adapter for Manuel's adapter
4:1 Adapter for Manuel's adapter
76239C69-C787-4E7C-A2B6-CFD9380F9B59.jpeg (97.2 KiB) Viewed 9488 times
4 turns one wire from the top, 4 turns second wire from the bottom, connected as shown. (No I do not understand it)
Maybe a 5:1 would be better? (10 ohm to 50 vs "12.5" ohm to 50 )
(Didn't have BN73-202 so glued some 73 toroids instead.)
6k2kus
 

Posts: 8
Joined: 19 Mar 2022, 02:14

Re: (tr)uSDX power, efficiency and filter optimisation

Post by 6k2kus »

Ohwenzelph wrote: 15 Apr 2022, 09:26 "19 turns for L15 instead of 22 as Manuel updated"
Please, Where was this update?
You may find it in his clip:
https://youtu.be/eC-IvL586K8?t=314

Maybe my expression is misleading: a new Aliexpress kit has thicker wire so he reduced the turns. This should not be crucial since previous kits worked fine.
Ohwenzelph
 

Posts: 207
Joined: 01 Jan 2022, 03:47

Re: (tr)uSDX power, efficiency and filter optimisation

Post by Ohwenzelph »

Thanks. He measured the supplied wire at 0.54 mm instead of 0.4mm, roughly 24 to 23 awg, (though my supposed 24awg measures ~0.56mm which is thicker than it's supposed to be, maybe it has thick enamel). Thicker wire is supposed to create more inductance in a coil "because the wires are closer together"? Maybe it's just that the kits are overall good and perhaps nothing to do with L12. I was hoping taking 2 turns off L12 would fix all my efficiencies but maybe no..
va3rr
 

Posts: 54
Joined: 30 Dec 2021, 21:39
Location: FN25eg
Contact:

Re: (tr)uSDX power, efficiency and filter optimisation

Post by va3rr »

It appears Owen Duffy has been following the L15 discussion

https://owenduffy.net/blog/?p=24648#more-24648
Slava Ukraini!
Ohwenzelph
 

Posts: 207
Joined: 01 Jan 2022, 03:47

Re: (tr)uSDX power, efficiency and filter optimisation

Post by Ohwenzelph »

So, should I replace L12 with a BN43-202 with 4 turns? If a FT37-43 works, what is the advantage?
Faraaz VK4JJ
 

Posts: 215
Joined: 30 Dec 2021, 21:56

Re: (tr)uSDX power, efficiency and filter optimisation

Post by Faraaz VK4JJ »

I am sure Manuel will be along to comment, but please hold off doing changes at the moment. Owen's article is interesting but he has also caveated that some further changes may need to be made downstream. And this theoretical article needs testing:

"The next, and necessary step, is to try a prototype in the application circuit… though I do not have the transmitter and will not be doing that."

He is kindly posting me a BN43-202 core to experiment with as I just did a big Mouser order myself!
DL2MAN
 

Posts: 706
Joined: 30 Dec 2021, 19:18
Contact:

Re: (tr)uSDX power, efficiency and filter optimisation

Post by DL2MAN »

L15 is not prone to future changes from my side.
With 22Turns there is neither Core Saturation nor Current Limitations in a qrp application.
How do I know ?
When looking at it with thermal camera: it stays completely cool while doing its job.

Don't over-engineer it, especially because it works the way it is.

But hey: Feel free to experiment ;)

73 Manuel DL2MAN
g7jur
 

Posts: 35
Joined: 18 Feb 2022, 22:49

Re: (tr)uSDX power, efficiency and filter optimisation

Post by g7jur »

Hello all.
Just measured the R shunt by running one amp through it and measuring the voltage drop.
Which is 18.8 mV 18.8 miliohms.
g7jur
 

Posts: 35
Joined: 18 Feb 2022, 22:49

Re: (tr)uSDX power, efficiency and filter optimisation

Post by g7jur »

Rewound L15 using 0.5 mm wire and 19 turns, efficiency up by 5 to 7 %.

Philip G7JUR
Ohwenzelph
 

Posts: 207
Joined: 01 Jan 2022, 03:47

Re: (tr)uSDX power, efficiency and filter optimisation

Post by Ohwenzelph »

The nanovna is very useful for suppressing the initial harmonic. The 4:1 transformer is not a complete or final answer. A scope is easier to use for lower frequencies (probably a 300 MHz one will be needed for the hi bands setup) and has helped on 80 meters but I have not had luck with 20 meters, though I have tried.

So I am asking, IF you have 80% efficiency or better on 20 meters, what are the values of your caps and inductors? With details, if at all possible.

If enough people with 20 meter efficiencies at or above 80% will respond, then maybe we can help everyone with this.
Thanks.
DL2MAN
 

Posts: 706
Joined: 30 Dec 2021, 19:18
Contact:

Re: (tr)uSDX power, efficiency and filter optimisation

Post by DL2MAN »

Ohwenzelph wrote: 19 Apr 2022, 02:41
So I am asking, IF you have 80% efficiency or better on 20 meters, what are the values of your caps and inductors? With details, if at all possible.
I have 80% Plus Efficiency on 20m. With the published Values from my Schematics. With my own 2 Reference Builds, with 2 chinese Kit Builds. Just not with the 1 *Rxxxx* Kit. But some users here have reported, they wiped off all Caps and re-build it with known good caps, and they could get good efficiency.

This indicates, that there might be a misplacement on the *Rxxxx* Boards. Unfortunately I do not have all the Information to investigate or proof it and my last communication with them ended up in a big fight, as I was and still am very upset about now at least 3 Batches with issues. And maybe more, that we don´t know of yet.
I have suggested / requested *Rxxxx* to stop all sales, until the Problems are fixed.
They said, they would be only fulfilling open and paid orders, and people would be willing to buy it anyway.
However, this still means, that there will probably be more and more Kits out there with unsolved issues.... Not happy about that.


73 Manuel; DL2MAN
SP9TKW
 

Posts: 57
Joined: 03 Jan 2022, 21:11
Location: Krakow

Re: (tr)uSDX power, efficiency and filter optimisation

Post by SP9TKW »

As I always was interested in a classic band option I will start rebuilding my *Rxxxx* RF board to that plan. Here are the steps I am planning to take. I will start process withing next day or so.

1. Measure and record current power/efficiency/turns on the core for 80/40/20
2. change L15 to 19turns and 0.6mm wire (maximum turns which will fit on that core)
3. Measure and record new power/efficiency/turns on the core for 80/40/20
4. remove all caps and inductors for all bands
5. resolder new caps in new locations in line with "classic band" layout
6. measure and record capacitance in situe for 80/40/20 band
7. install newly rewinded inductors. Use 0.6mm wire (improve Q) for 40 and 20 LPF
8. use Manuel's instructions from schematic regarding number of turns
9. use T37-2 (red) for 80 and 40, use T37-6 (yellow) for 20m band
10. measure and record power and efficiency before any ajustments
11. adjust inductance LX2 for harmonic suppresion for 80/40/20 using nanoVNA method
12. measure and record power and efficiency
13. adjust inductance LX1 for power/efficiency using 4:1 transformer and nanoVNA
14. measure and record power and efficiency

All power and efficiency measurements to be taken at 12V and 13.8V

As I will have time to do it only at late evenings, it will take couple of days before I publish full set of measurements.

Please feel free to suggest additional steps/measurements. I do not have oscilloscope I'm affraid just multimeter and nanoVNA.

73 Marek
PD0MVP
 

Posts: 9
Joined: 01 Jan 2022, 12:50

Re: (tr)uSDX power, efficiency and filter optimisation

Post by PD0MVP »

I also don't have a NanoVNA but I tried to look with my Icom 7300 where the harmonics of 7.1 and 14.2 were.
I found these on the 14.2 and 28.4 mHz respectively.
These were on the bandscope of the 7300 so exactly on the right frequency (used CW).
I may not be measuring this correctly but I have no other option at the moment.
What strikes me is that at 6.5 mHz the efficiency is more than 90% with a power of around 5 watts.
At 6 mHz even above 5 watts and almost 100% effective.
You would almost think to remove a winding from the toroid.
I can't test lower on the 20 meter band, because at 13 mHz the relay switches to another band.
But also here, the lower in frequency the better the efficiency and the output power also becomes higher, at 13 mHz even more than 7 watts.
Post Reply

Users browsing this forum: No registered users and 2 guests