I may have found the problem! it appears to be in my Measurement equipment — not in the sbitx!

I found another pad that could take high power for a small bit of time. ?It allows completely flat measurements — and I got FAR better responses
on 20 meters. ?

I need to make some more measurements before full information but I am MUCH more encouraged. ?

gordon Kx4z?

Big improvement.
The Heathkit cantenna circuit originally included a simple power measurement circuit consisting of a 100K resistor in series with a diode/capacitor rectifier.? ?This went to its own connector.? ?I had left it in there when I added my own resistive divider network for going to the spectrum analyzer.? ?Apparently that diode through the 100K was able to create IMD products.? ?I should have thought of this before!? ?I snipped out the 100K connection.? ?

I was able to use a modest power 15dB UHF N-connector pad? (that I finally found in my bins)? ?and two 20dB BNC pads to avoid using the heathkit completely.? ?Testing with the Siglent suggests that below 100 MHz, this measurement setup is essentially FLAT, with no compensation needed.? ?Re-testing of the Heathkit cantenna after cutting the connecton to the 100K showed the same compensations were needed as before, checking all the way to 60 MHz.

Measurements with the UHF moded power 15dB UHF pad / 2 x 20 dB 'BND pads and Siglent:
[THESE MEASRUEMENTS ARE RELATIVE POWERS]? ??
14.010? ?Fundamental -12dBm? ? ?28 MHz -51 dBm (-39dBc)? ?42 MHz -51 dBm? ?(-39 dBc)? ?<<----? 10 to 13 dB BETTER than I had measured before
21.010? ?Fundamental approx 15W? 86%? -13dBm? 42 MHz? ?-39dBm? -26 dBc? ? ?63MHz? -43dBm? -30dBc? ? <---- 0 to 5 dB better than I had measured before
28.010? ?Fundamental approx 10W @ 27% (system looks close to max power at that point)? ?-14.4 dBm? ?
? ? ? ? ? ? ? ? ? ?56MHz? ?-51 dBm? ?-36.6dBc? ? ? ? 84 MHz? ?-64dBm? ?-50 dBc? ? ?<--- basically the SAME as what I had measured before

The diode appears to have caused the most trouble on the 20 meter measurements.? ?

For Comleteness, I went back and remeasured all bands using the revised Heathkit Cantenna system (which has no power limitations).? ?It requires compensatory adjustments for its non flat frequency response.? ? To shorten the lines, I'm only showing the dBc, after substracting.
The results generally agreed with the above but are likely less accurate, with deviations up to 6 dB.? ?I think I trust the UHF pad system that I found, more.

[ALL SPECTRUM ANALAYZER MEASUREMENTS ARE RELATIVE? (NOT ABSOLUTE) AS THE ATTENTUATION OF THE SYSTEM IS NOT MEASURED, ONLY COMPENSATED TO APPROX FLATNESS]? ?dBc? means relative to the fundamental carrier; I have already subtracted the corrected dBm
3.510? ?20W? 73%? -20dBm? ? ?7MHz? ?-46 dBc? ? ? ?10.5 MHz? ?-50 dBc
7.010? ?20W? 72%? ?-20dBm? ?14MHz -40 dBc? ? ? ? ?21? MHz? ?-46dBc
14.010 20W? 78%? ?-20dBm? ? ?28MHz? -30dBc? ? ? 42MHz? ?-36 dBm? ? (note the measurement with UHF pad is probably more accurate and got -39dBc for both)
21.010? 15W 86%? -22 dBm? ? ?42MHz? -20dBc? ? 63MHz? ?-26.5 dBc? ? (note the measurement with UHF pad look 6 dB or so better, but still not as good as wished)
28.010? 10W? 27%? -23 dBm? ? 56MHz? ?-37 dBc? ?84MHz? ?-51 dBc? ? (agrees pretty well with UHF pad measurement)


So the diode caused huge problems with the 20m measurement.? ? ?The UHF pad appears to show better measurements than the cantenna and is likely the more correct of the two (since it is truly flat over frequency)? ??

I apologize for not picking this up earlier -- I just didn't think the diode circuit could have an appreciable effect!? My error.

My measurements suggest that on 15m, I need to back off the power even farther as that band looks worse than the others.? ? ?(I have not touched the calibration values that came with this radio, but the previous owner might have, despite stating he had never transmitted.)? Alternatively, I could add a small external filter.? ?The 28 MHz response is much better, and uses the same filter, I believe.? ? Some of these numbers could still stand to have some improvement, and the 2nd harmonic is definitely "there" but these are more reassuring numbers than what I had before.

73
Gordon KX4Z

I had previously made at least one measurement without the RadioShack SWR/power meter in line, ?because I knew it should have one or more diodes in there as well.?

at that time I did not see a difference, but I didn’t at that time realize there was another diode still in the Heathkit Cantenna dummy load ?system

so now I think I should go back and make measurements without the SWR/power meter in the line. ? ?

today is busy, maybe I can get that done tonight or tomorrow….

gordon kx4z?

Today, I repeated the remaining troublesome measurements, this time without the Radio Shack SWR/power meter (which includes a diode based circuit).
Sbits -->? ?15dB uHF pad? -->? 20dB pad --> 20dB Pad ->? Siglent spectrum analyzer, internal attentuator 10 dB, bandwidth 10kHz

Band? ? ?%power? ? measured dBm? ? ? ? Second Harmonic? ? ? ? ? ? ? ? ? ? ?Third Harmonic
Freq? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?freq dbm? dbc? ? ? ? ? ? ? ? ? ? ? ? ? ? freq dbm? dbc

20m? ? ? ? ?78%? ? ? ? ? ? -11.2dBm? ? ? ? ? ? ?28M? -47.21dBm -36dBc? ? ? ? ? ?42M? -52.7dbm -41.5 dBc
14.010

15m? ? ? ? 86%? ? ? ? ? ? ?-13.14dBm? ? ? ? ? ?42M? -35.1 dBm? ?-22dBc? ? ? ? ? 63M? -41.2dBm? -28 dBc
21.010 MHz

Unfortunately these aren't any better and just a tiny bit worse than previous measurements after I removed the diode from the Heathkit cantenna.??

On a whim, I decided to demonstrate what could be done with a little more filtering.? ?The only thing have I readily available is a 15m dual-section Butterworth filter that I have built as part of our Field Day antenna multiplexer work.? ?It has GREAT attenuation away from 15 meters.? ?So I threw THAT in series between the sBitx and the 15dB pad....and here is what happened:

21.010M? 86%? ? ? ? -13.57dBm? ? ? ? ? ? 42MHz? BELOW -88 dBm? ? ? 63MHz? BELOW -88dBm
Spurs -- were approximately -61 to -63 dBm....which means -48 dBc to -50 dBc --?

The noise floor of the analyzer was -88dBm.? ? The harmonics were therefore BELOW the observable limit and were at least 75 dBc below!? ?

Here is how beautiful the display looked:? ?This is a completely, totally, thoroughly legal spectrum.? ?With filtering, this is possible.? I'll admit, this requires volume for the parts....




I think this makes an interesting additional product possible.? ?A very simple filter system could be marketed as an add-on that would add even one section of additional filtering, or one more low pass section -- and reduce the harmonics quite significantly.? ? Alternatively when I use my vacuum tube amp, with its tuned filters.... it cleans everything up also.??

Now that my errors in measurement have been corrected, we have a much clearer picture of where Serial #141 falls well or not so well.

Gordon KX4Z

OK, my goal has been to document the performance as I received it secondhand from the gentleman who bought it and said he never transmitted with it. ? So I think now I have a good list of where filters need a little tweaking.?

With my limited available time I think next I’m going to pull the main board off, add zeners on the gates, and switch out the insulators for aluminum nitride if I can find my extras! ?

i’ve looked a little bit into the seven pole filters and it looks like I could tweak with some added capacitance at especially beginning and end, and possibly bring 20 m and 15 m closer, ?or into compliance. ?If I lose power at ?18MHZ or 28 MHZ I’ll have to make some trade-offs. ?It is possible I could connect a BNC (after disconnecting the output transformer) to the input of the filter section and map the filters as they are. ?It’s always possible that there has been some damage that I don’t know about, rendering the filters not as effective as others ? ? It was very clear with the filter added to the output that this can be cleaned up to just beautiful!!

I have limited limited time because I’m teaching high school students but I’m going to try to get a little bit done. ?
73
gordon Kx4z?

I'm still very puzzled, even after fixing errors in my experimental setup (principally the DIODE in the Heathkit cantenna!)? that I still appear to have harmonics that are above the limits.? ?Even when using a commercial high power "pad" and no heathkit at all.? ??

I was concerned that the low pass filters might not be robust enough.? ? To study that theory, I realized that I could lift the power output wire from the big transformer of the final stage....and inject spectrum analyzer tracking generator (through a 6dB pad) right there at the beginning of the filter system.? ?I could read the output of the low pass filters by connecting (through a 6dB pad) the RF input of the Siglent spectrum analyzer to the output.? ?I set the DRIVE at 1, and mic gain at 0, and lower sideband on all bands.? ?There was no apparent signal that the spectrum analyzer saw before turning on the tracking generator.??

A schematic of my setup:


There could be a mistake that I'm not recognizing, so here is a PHOTO of my setup:


When I measured the frequency response of the filters, they looked very very effective.? ?Here is a table of the -1dB point, the 2nd harmonic point, the -24 db point, and the 3rd harmonic point.? ? Frequently at the tails my measurements were "in the noise" with my settings (approx 10kHz VBW and probably 20dB of internal attentuator -- it didn't dawn on me to adjust that)

BAND? ? ? -1db point? ? ? ?2nd H. Freq? ? -24 dB point? ? ?3rd H. freq
80 M? ? ? ? ? 5.4 MHz? ? ? ? ? -22dB? ? ? ? ? ? ~ 7.06MHz? ? ?~-60 dB
40M? ? ? ? ? 11.4 MHz? ? ? ? ? -21 dB? ? ? ? ? ~ 14.6 MHz? ? ?-51 dB
20M? ? ? ? ? ?16.4 MHz? ? ? ? -44dB? ? ? ? ? ? ?21.6 MHz? ? ? ? -57 dB
15M? ? ? ? ? ~3dB@...? ? ?-21 dB? ? ? ? ? ? ?43 MHz? ? ? ? ? -57 dB? ?(I had to change the Siglent to go to much higher frequencies to get this one measured)

Those look (to me) like rather impressive filtering!!!? ?When your 3rd harmonic frequency (as proxied by a tracking generator, tiny signal) is being attenuated by 50-60dB....how in the world can you end up with a signal stronger than -43dBc???

It appears to me that there are a few possibilities:
A)? The Siglent is measuring improperly when measuring the power from the sBitx.? ?(However, it sees a used ICOM 718 as completely clean)
B)? With full transmittter power,? something in my MEASUREMENT system is generating harmonics -- clearly I had some issues there, but now I'm able to make measurements using ONLY commercial pads.... and the Siglent....-- and besides, this didn't seem to happen with the Icom 718??
C)? With full transmitter power, something in the SBITX is generating harmonics.

Does anyone see why it wouldn't be (C)??? ?Am I missing something?? ? ?Maybe I'm doing something stupid again???

Here are photos captured as I mapped out the various filters - they look very impressive to me.
Gordon Gibby KX4Z

What about the DIODES in the RF power supply that generates the HV to back bias the switching diodes?? ?They are connected through a 47pf to the main output of the amp.? ?could THEY be generating stronger harmonics???

Other DIODES to consider are the "pin" diodes 1N4007, and also the diodes in the SWR/power measuring circuit.
I don't know if the rf power supply 1n4148s are the problem, but the MFJ QSK unit uses DC 600 volts developed from the power supply of the amplifier to run it.... See:? ?

Just not sure where the harmonic energy is coming from....with the filters looking NICE.? ??
Of course, ANYTHING generated by the "antenna connector"-sided diode would have zero filtering....
Gordon KX4Z

pr maybe Gordon,

I have not seen others post this type of problem.? Even my DE has compliance with all bands.? I would suggest a component issue with one of the mixers that does not provide a symmetrical response that increases the 2nd harmonic.? Or maybe it is an issue with the driver.? Have you looked at the spectrum of the driver feeding the PA?? The driver is also a push-pull class AB amp.? An imbalance there would cause a signal to be amplified as if it is part of the base signal.??

73
Evan
AC9TU

开云体育

I think I had no further problem with the DE when I was finished and I suspect that if I went back to it it would be even cleaner I haven’t gotten rid of the diode in the Heathkit Cantana?

开云体育

To answer your question: I have not looked at the earlier stages. However this radio passes on some bands but fails on others. ?

开云体育

And of course the DE had relays….

After pondering this over night, I'm starting to more and more suspect that it has to be some diode that is AFTER the filters, since the filters look very impressive.
Even if you had a 3rd Harmonic EQUAL to the fundamental, those filters would push it almost 15 dB below the specification.? ? It is unbelievable that my 15meter 3rd harmonic of -28dBc is the result of anything BEFORE the filter.? ?Just unbelievable that it could be the drive, or final or anything!? ?Because a 60dB filter would wipe it out!

Also, in this post?/g/BITX20/message/105519? ?I showed proof that with some filtering AFTER the last diodes (an external filter) the harmonics can be? easily reduced to nearly undetectable (with a given spectrum analyzer setting).? ?This again validates my measurement setup.? ?If I can reduce the harmonics to undetectable....they don't seem to be the result of some failure of the measurement system shielding or linearity or anything else.

One of the most likely candidates for a diode past the filter is the 1N4007 used as a "PIN" diode at the output side of each filter.? ?These were not marketed as PIN diodes meeting any given harmonic suppression specification.? ?It could be possible that the intrinisic layer is of differing composition in different manufacturing outputs, and this might explain why some bands do well and others not in my particular Serial # 141.? ? And perhaps there are resonances or something in the circuitry?? ?

I found that PIN diodes are well known for producing harmonics.? ?There are IEEE papers on it that i can't get to because of a paywall.
I found specifications for high quality PIN diodes that guarantee 80 dB reduction in harmonics.??
I found a discussion on PIN diodes and their possibilities of making harmonics:??https://www.quanticpmi.com/application-notes-how-to-specify-pin-diode-switches\

开云体育

Gordon,

Did you try increasing the forward bias, as Ashhar Farhan has suggested?

73
Evan
AC9TU

Gordon,

The diode you listed in your post has a max reverse voltage of 50 volts.? That is 13 volts short of the 63 volts I calculate for 40 watts in a pure resistive 50-ohm load.? I have found that all pin diodes are hundreds or thousands of dollars for a minimum purchase required to replace the eight diodes.

Maybe go back to relays for LPF selection and change the program to keep the relays on for the selected band to allow for QSK CW operation?? I believe that you did this for your DE.

Another chain of diagnosis might be the voltage doubling circuit (as you pointed out in a prior post).? There may be an issue with the doubler voltage not being filtered enough, or the capacitor is on the high side of the ESR spec.

As I have stated before, I am NOT an RF engineer, just an industrial controls engineer tinkering with radio stuff.
73
Evan
AC9TU

Yes, thanks, Bill and Evan.
The 1N4007 is certainly not specified or intended for PIN service ; it just happens to have an intrinsic layer. Not sure how thick, and it’s one very big reason I’m concerned that it might be the cause… but a lot of my other hypothesis of turned out to be incorrecr

i’m not so worried on the input side of the filter now that I’ve looked at the filter and I am amazed at how good they are, I’m much more worried about the outside of the filter!! ?I think this might be why I never ever worried really about the harmonics on the DE. As I remember all of my writing stay were about spurious submissions, and who knows some of those might have been due to the Cantenna!! Eventually I’ll go back and check it again but I had managed to get it much better.

I think once you select the band, the diodes stay in the correct settings? I’m not certain about that. But it wouldn’t be too terribly difficult to replace the output side diode’s with relays….. just would take more space!!

yes, I did the upgrade to give more current to the diodes—-BRFORE These measurements

if folks are going to use the one in 4007 and it turns out to be the issue, then careful selection is going to become necessary.?

gordon

Gordon,

The voltage doubler is applied to both input and output diodes.? Maybe the 10-meter filter is turning on enough to allow the harmonics through?? Just a guess.

All,
Has anyone else seen the harmonics issue with the V2?? Ashhar Farhan has stated that all V2 sbitx are tested with a spectrum analyzer for signal purity before shipment.

73
Evan
AC9TU

Again, this unit was purchased from another person. I don’t think that person made any alterations, I can’t see any alterations and nothing looks burnt

I made up a small clip lead to try to short out the diets. This is certainly not optimal, because it will act like a small antenna. But it makes an easy way to allow me to sort out some diodes ?and see what happens

So using my UHF pad, and a bunch of BNC attenuators as well, on 21.02 MHz with Drive at 84% here were the values

fundamental: -6. ?Dbm
Second harmonic: -28 DBM
third harmonic -34 DBM

these are similar to some of my previous measurements for this band which is probably the worst

applying the shorting clips to the input Diode was no improvement and maybe one Db worse

applying the shorting clips to the output diode made a 6 dB improvement at the third harmonic

fundamental: -6 dBm
Second harmonic -29 DBm
third harmonic -40 DBM

I am out of time, so I can’t do much more, but I suspect these numbers would be better with a small short rather than this large clip. However it did not completely eliminate the problem. But it’s another clue. ?

Evan -- so in this case, I am *using* the 10meter filter.? ? ?Its knee is at 32-ish MHz and appears to be used for 15 meters as well.? ? At least I hope I am!

These data seem to prove that we have at least a? -28dBc spur (harmonic) being formed due to the conduction of the output diode for the 10m filter!
There are likely OTHER sources, that are responsible for the remaining -34dBc 3rd harmonic.? ? ?Need to get all of this down below -43 dBc.? And for a radio of this advanced design, would be nice to have it significantly more impressive than just meeting requirements.? ?

The material I quoted cited "non linear passive components" -- is it possible that the capacitances of the reverse biased other diodes are also changing in a non-linear manner with instantaneous voltage (because the RF voltage is a varying percentage of the back biasing DC voltage) -- and that THIS is also changing the response in a non-linear manner and introducing harmonics?? ? Seems possible!? ? ?Ashhar wisely included a voltage doubler circuit, but that still means there is a significant percent change in the NET reverse bias on each part of a cycle.? ?What about either replacing the output diodes with RELAYS or using a much much higher reverse voltage?? ? ?

I just don't know exactly what next is the best step.? ? These diodes I don't think are ever switched "under load"? so a relatively modest relay could replace them.? ? at 40 watts, they should see currents only in the 1-2A range (even with 2:1 SWR).? ??

Anyone else made measurements on their V2?? ? Or have ideas?
Gordon KX4Z