Dear friends,

After all my "issues" and self-induced catastrophes with the Developer's Edition, which I have working and can now use routinely for QSOs (just got off a voice net using a vacuum tube amp also) -- I was able to pick up Serial No. 141 V2 from a very nice gentleman who had never transmitted over it, and was selling it at a VERY attractive price without a RP4 -- and I had a spare!? ?It arrived in excellent shape and I was very impressed at the cabling and other materials of instruction manual and the case.? ?Everything was very impressive.

Before I commenced to add some of the "further" modifications (aluminum nitride insulators, protective zeners to the final MOSFET gates) -- I wanted to make careful measurements of its performance "out of the box."? ?To avoid losing the whole rig, I added a "polarity protector" out of p-channel enhancement mosfets (4 in parallel,? single digit milliohms of series resistance) that we have a lot of experience with locally.? ? ?I have the same thing now on my DE version.

I created a testing suite in my "ham radio nook" and made measurements.? ?In the course of that, I also made a study of an ICOM 718 as a confirmation of my testing suite.? The ICOM 718 is a workhorse analog transceiver which has had multiple versions of the final amplifier under the same model no.? ?There have been bipolar as well as MOSFET final amps.? ?I believe mine is MOSFET but I haven't taken it apart in so many years I don't know for sure; it has been WINLINK RMS KX4Z for > 4 years, running 24/7/365 on HF, rain, shine, hurricane, whatever.? ??

The testing suite involves:
1. MFJ switching power supply with added powerpole leads
2. Transceiver under test
3. Short coaxial cable to an ancient Radio Shack combined SWR meter/"wattmeter"? (this is essentially an RF voltmeter calibrated for 50 ohms)
4. Short coaxial cable to an ancient Heathkit Cantenna 50 ohm 1kW oil-immersed dummy load, modified to have a resistive (non-compensated) attenuator to a BNC output
5. Short double shielded BNC jumper cable to a 6dB pad to a Siglent SSA3021X spectrum analyser.

The Heathkit homemade attenuator is NOT compensated.? ?Using the Siglent tracking generator I was able to calibrate the response of the attenuator using exactly the same cabling to the RF input of the Siglent.? ? REFERENCED TO THE 3.5 MHZ SIGNAL here are the observed outputs, showing that more signal gets through at higher frequencies:

3.5 MHz? ?(reference)
7 MHz? ?+1.5 dB
14 MHz? + 4 dB
21 MHz? ?+ 6 dB
28 MHz + 8 dB
42 MHz? + 11 dB
56 MHz + 13 dB
63 MHz? + 14 dB
84 MHz? +16 dB

All of the power measurements reported here will be CORRECTED for these offsets where the attenuator passes MORE SIGNAL for higher frequencies (e.g. for harmonics)

The testing scenario was the same for all radios.? ?The only differences were changing the frequency range of the spectrum analyzer, and which bands were tested.? ?Siglent spectrum analzyer set to 10kHz receiver and visual bandwidth (this is fairly narrow and slower, but gives good sensitivity)
CW approximately 10kHz above the bottom of each band was used.? ?The Siglent has a mode where it picks up "peaks" and reports their frequenciy and power level.? ?From those "dBm" measurements, one can then use subtraction from the carrier signal to come up with referenced "dBc"? (dBcarrier - referenced to the desired signal) numbers.? ?For our HF rigs below 30 MHz, spurious responses generally must be -43dBc or weaker.? ?THere are other regs above 30MHz but I haven't reviewed them.

HERE ARE THE REFERENCE RESULTS FOR A VERY OLD ICOM 718
Power level set to "50%"? ?Power measurements below are APPROXIMATE on the highly questionable Radio Shack SWR/power meter intended for CB operators

MHz? ? ? ? Fundamental? ? ? ? ?Any notable Spurs? ? ? ? 2nd harmonic? ? ? ? ? ? ? ? ? ? ? ? ? ?3rd Harmonic
3.510? ? ? 39Watts -18dBm? ? ? ? ?Below -73dBm? ? ? ? ?NONE SEEN above noise? ? ? ?10.5 MHz? -78 dBm? ?-60dBc
7.01? ? ? ? 25Watts? ?-19dBm? ? ? Basically none? ? ? ? ? ?NONE SEEN above noise? ? ? ? 21 MHz? ?-77 dBm? ? -58 dBc
14.010? ? 25 Watts? -18.85dBm? 5.6MHz? -66dBm -47dBC? NONE VISIBLE? ? ? ? ? ? ? ?42 MHz? ? -80 dBm? ? ?=61 dBc
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?16.8 MHz -71dBm -52dBC


COMMENTS:
1.? ?The testing setup appears validated.? ? The results indicate that the ICOM 718 well-worn transceiver has very very little 2nd harmonic, suggesting the push-pull amplifier is very well balanced, and the 3rd harmonic is very well suppressed
2.? ?The Icom does have a couple of notable spurs on its 20 meter band but they are well below the FCC regulation specification.

Here is a photograph of the testing setup:




In the next post(s) I'll provide testing results with the V2 sBits that I received, as received (with Raspberry pi installed)

Gordon L. Gibby KX4Z

In this post, I'll give information about the initial testing results of the Serial No. 141 V2 sBitx as received.

The testing setup was very similar to that used in the validation tests with the Icom 718? ? The coaxial cable to the radio was a BNC double-shielded cable with appropriate adapter to allow it to connect to the Radio Shack ancient SWR/power meter.? ? The MFJ power supply was set to approximately 12.9 VDC based on its front-panel meter, which has been generally accurate when compared to a digital VOM.? ? Morse code was generated using an external key.? ?

without any further adjustment of the received unit, here were the values recorded:

MHz? ? ? ? Fundamental? ? ? ? ? ? ?Any notable Spurs? ? ? ? ? ? ? ? ? ?2nd harmonic? ? ? ? ? ? ? ? ? ? ? ? ? ?3rd Harmonic

3.510? ? ? 20W 75% -20.9dBm? 5.48MHz -69 dBm? -48dBC? ? ?7MHz -63.5dBm -43dBc? ? ? ? ? 10.5MHz -77 dBm? ?-56 dBc
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?among others?
7.010? ? ? 20W? 72% -21.5dBm? appx -66 dBm or less? ? ? ? ? ? 14MHz =59dBm? -37.5 dBc? ? ? ?-65 dBm? -43.5 dBc

14.01? ? ? 20W? 78%? ?-21.3dBm 8.9MHz? -63 dBm? ? ? ? ? ? ? ? ?28MHz -48 dBm? -27 dBc? ? ? ? ? 42MHz -53dBm? -32 dBc

21.01? ? ? 15W? ?86%? ?-22 dBm? ? ?approx --56dBm or less? ? ?42MHz -43 dBm? ?-21 dBc? ? ? ? ?63 MHz? -52dBm? ?-30 dBc
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? (uncorrected measurement)

28.01? ? ? 12W? ? 40%? ? -23 dBm? ?around -66 dBm? ? ? ? ? ? ? ? 56 MHz -60 dBm? ?-37 dBc? ? ? ? ?84 MHz? -73 dBm? ?-50 dBc
(any higher and non linear)? ? ? ? ? ? (uncorrected measurement)
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ??

These data were obvious of considerable concern.? ?Particularly the existence of a fairly strong SECOND harmonic.? ?Balanced push pull circuits are supposed to null out a second harmonic.? ?This suggess the amplifiers of one or more stages at not well balanced.? ? ?The Third harmonics of some bands are above spec.

Here is a photograph of the testing situation:




And here is one spectrum observed :




Here are the uncorrected results from the Siglent:?



Based on these results, I decided a change was needed to the radio, to be detailed in the next post.

Gordon L. Gibby KX4Z

Based on the previous results, I decided to readjust the IDLING current for the final MOSFET Stage.
Using a digital VOM on its 10A scale, I interposed it in the positive line to the transceiver.??

When receiving with audio at 0%, the current was 0.52-0.53 Amperes
When transmitting SSB with drive at 1 and mike at 1, the current rose to approximately 0.99 Amperes (it changes with temperature as the devices heat up)
When the idle current adjustment was carefully adjusted, down to ZERO idling current, it read approximately 0.81 - 0.83 Amperes.

This indicates the idling current as RECEIVED was approximately 0.16 Amperes.? ?This was close to, but slightly lower than the idling current measurement that was on the very impressive TESTING SHEET that came with the radio, which indicated the idling current had been set to about 0.20 Amperes.

I decided to increase the idling current to approximately 0.36 Amperes (above the minimum and thus I carefully adjusted for a total current of 1.19-1.21 amperes.? ?I then reassembled the system....and in the process I damaged the DSI video cable.? ? Not having a suiitable spare, I was DELIGHTED to learn that I could see the screen of the unit by using an adapter to the HDMI port closest to the USB-C on the Raspberry Pi.? ?Hooray!? ?Able to continue testing.?


Now with an increased MOSFET idling current, I re-measured some of the bands as follows:


MHz? ? ? ? Fundamental? ? ? ? ? ? ?Any notable Spurs? ? ? ? ? ? ? ? ? ?2nd harmonic? ? ? ? ? ? ? ? ? ? ? ? ? ?3rd Harmonic

3.510? ? ?20W? ?73%? -20.8 dBm? ?approx -66dBm? ? ? ? ? ? ? ? 7 MHz -62dBm -43dBc? ? ? ? ? ? ? ?10.5? -72 dBm -51 dBc

7.010? ? ?20W? ?72%? ?-21.5 dBm? ? 10.5 MHz -55DBm? ? ? ? ?14MHz -60dBm? -38.5 dBc? ? ? ? ?21? -66 dBm? -44.5 dBc
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? (-44 dBc)


14.01? ? 20W? ? 78%? ? -21 dBm? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?28MHz? -47 dBm? -36dBc? ? ? ? ? 42MHz -52dBm? ?-31 dBc



CONCLUSIONS

1.? The 40 meter 2nd harmonic is 1 dB improved.
2.? ?The 20 meter 2nd harmonic is almost 12 dB improved -- HUGE improvement
3.? ?The 20 meter 3rd harmonic is a real problem.??
4.? ?All of my measurements are subject to question, this is NOT a professional measurement lab.? ?This is just the best that I can do with a very expensive (in my estimation) $1400 spectrum analyzer and reasonable layout of a radio as it might well be utilized in practice using a very popular power supply.? ?

I believe the 20 meter 3rd harmonic has been previously noted to be a problem.

OK.....so there is work still to be done on this radio!? ?I am concerned that the ARRL testing lab would conclude that it is not yet read for their testing and an article.? ?I think manufacturing is a MUCH more difficult task than prototype building (which in itself is a monumental task!).? ? So I think we should work together to make this so that Serila No. 141 (and most other radios) easily meets the requirements right out of the box.? ?Such an excellent design deserves to have this accomplished, yes?

Sincerely,

Gordon L. Gibby KX4Z

Dim-witted me took dozens of tries, but I *finally* found Evan's post on how he dealt with the 20 meter third-harmonic:

/g/BITX20/message/100269

Hi Gordon,

I need to update that thread.? Correcting the shielding helped, but the issue was a bad 50db tap connection.? The home brew tap developed an open connection for the ground on the BNC meter point.?

The sbitx uses a BNC antenna connection.? I assume that the Icom 718 has a UHF SO239.? Could it be the adapter?

73
Evan
AC9TU

Thanks, Evan. It would be extremely unlikely that the adapter would PASS the appropriate energy being developed at the fundamental frequency any differently than the spurs and harmonics that it is also passing.

In weeks and weeks of work on the DE model (that now is? working well) the only "bad connections" that I experienced in the meaurement chain were when solder migration literally disconnected one of the push-pull final MOSFETS!!? ?The drain connection apparently got hot enough to allow plastic-like flow of my low-temp solder!? ? One I switched to the AlN insulators I didn't see this again, which MIIGHT be their better conductivity or just less hours.? ?I've been using that rig now for a bit and no problems.? ?

All I can say SO FAR is that Serial # 141 has problems.? ? Definite problems.? ?Increasing the idling current made big improvements in some of the problems.? ?I suspect I have "bad connections" problems similar to what you observed, that are adversely impacting the workings of the filters.....speculation at this point.? ?Further testing may isolate the culprit.? ?

Manufacturing tolerances (e.g. the variations in threshold of different MOSFETs or the differences in gain of different transistors) and variations in physical alignment, connection make manufacturing quality control difficult (I think; I'm not an expert!).? ? Taking a snapshot random sample of ten produced units and extensively testing them might give an idea of the mean and standard deviation of performance, but this would be a solid week worth of work, I fear.? ? That's the advantage that a big company like ICOM, with an ARMY of engineers and manufacturing engineers, and quality control engineers, has over a small shop like HFSIGNALS.? ?We, in the community, can fill in as volunteers for some of those positions....but then, I'm not well trained there!? ?And we each have different skills and assets we can call upon.

When the FLEX units first came out I believe they had HUGE problems that weren't discovered until people started poking around.? And those were VERY expensive units!!

Gordon KX4Z

Gordon,

The push-pull design should provide a significant reduction of the 2nd harmonic.? The second harmonic seems to be an issue for 40 and 20 meters.? I would look at the driver and finals to see if there is an imbalance.??

For the 20-meter 3rd harmonic, I would check the LPF switching circuits to see if the diodes select the correct one and isolate the other filters.? For another issue, Ashhar Farhan suggested increasing the bias current for the pin diodes.? I'm not sure it is the issue here, as higher bias is needed for lower frequencies (due to the junction charge draining too quickly at low frequencies).

Happy Hunting!
73
Evan
AC9TU

Ashhar Farhan,

You beat me to the bias comment.? Were the issues that you saw on 20 meters as well??

73
Evan
AC9TU

Dear Evan and Ashhar —

no, I had not made the suggested change in the drive to the “pin” diodes — I merely evaluated the unit as received. ?I wasn’t expecting to find the issues at all!!

A) since perhaps increasing the diode bias is a more important update than I had realized, I’ll do that and then make Re- measurements.
B) ?I should also sand/grind some of the connections between case parts and make separate measurements there also if I can. ?

Thanks! ?This radio is worth it

gordon Kx4z?

OK.? ?Updated results.
FIRST -- I had forgotten or missed (although I made comments about it!) Ashhar's prescription to add 470 ohms in parallel with the drive resistors for the "pin" diodes.
I did that today -- on both sides.? ? On R201/202 and on R202/203, one 470 ohm for each pair.? ? This should raise the bias current on each PIN diode to approx 75 mA

ERROR CORRECTION:? previously I stated that increasing the idle bias current made a substantial improvement in the 2nd harmonic of the 20m CW signal.? ?My notes show a SUBTRACTION error to get an "improved" -36dBc.? ?The correct subtraction is -26dBc? (-47dBm versus -21 dBm carrier) which is basically UNCHANGED from the original measurements.

TONIGHT -- I made new measurements using the SAME SETUP, after adding in the new resistors.? ?I also made certain to tighten the case screws "snug".


3.510? ?20W? 73%? -20.6dBm? ? ? ? ? 7MHz -66dBm? -45.4 dBc? ? ? ? ? ? ? ?10.5MHz -77 dBm? -56.4 dBc
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?slightly better? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?slightly better -- and both PASS

7.010? ?20w? ?72%? ?-21.5dBm? ? ? ? ?14MHz -60 dBM? ?-38.5dBc? ? ? ? ? ? 21MHz? ?-66dBm? ? -44.5dBc? ? ? ?little change

10.11? ?~20W 81%? ?-21 dBm? ? ? ? ? ? 20.2MHz -57.3dBm? -36.3dBc? ? ? ? 30.3MHz -62 dBm? -41 dBc

14.010 20W? ?78%? ? ?-21 dBm? ? ? ? ? 28MHz? -48 dBm? ?-27dBc? ? ? ? ? ? ? 42MHz? ?-52.4dBm? ?-30.4 dBc? ?little change

18.1 MHz 15W 88%? ?-22dBm? ? ? ? ? 36.2MHz? -65 dBm? -43 dBc? ? ? ? ? ? 54.3 MHz -82.4 dBm? ?-60dBc? ? ?PASS


1.? What is monotonous is the LACK of change in most of these numbers despite all the things I have tried.? ?It is eerie how much they repeat.
2.? I presume that the 20meter and 18.1 MHz use the same filter.? ?The 18.1 MHz PASSES while the 20 meter fails miserably.? ?Is the filter out of adjustment?? That the 18.1 MHz signal PASSES certainly shows that there is adequate signal islolation and that a workable filter can get as much as -60 dBc!? ?This is a much better result

3.? Do the 40 meter and 30 meter use the same filter?? ?I was expecting a similar improvement of the 30 meter signal -- but it might be even worse than the 40 meter!? ?(I had to interpolate my attenuator corrections so this might be off a dB)? ? ??

It doesn't seem that the PIN diodes can be blamed for all this.? ? The ICOM 718 has (with my measrueemnt setup) undetectable 2nd harmonics but these 2nd harmonics are as strong as -27 dBc (20 meters) and generally running in the -36 to -45 range on most bands.? ?While that is certainly weaker than the fundamental, it doesn't compare to the ICOM 718!? ?It makes you think that we may have 2nd harmonics being generator in the earlier stages, possibly the 4-transistor driver. ?

I don't know, folks.? ? Anyone else have actual DATA or suggestions on how to fix this?? ? From my measurements, 80 meters and 17 meters are good.? Is Serial # 141 just an outlier, and how am I so unlucky??

73
Gordon KX4Z

Gordon,

I did not expect that increasing the PIN diode bias would help.? I agree that there is a good chance that the imbalance causing the even harmonics could be in the driver stage.? I suggest measuring the voltage drop on the emitter resistors (R19, R20, R45, R48)?to see if one of the transistors is not pulling its fair share of the load.? You could also check the signals on the resistors with an oscilloscope to compare wave shapes.

73
Evan
AC9TU

Friends,

AFter our recent club experience taking VA6AM's excellent Triplexer product, and tuning a kit, and then expanding it into a 5-band multiplexer, I learned a lot more about adjusting toroids and HOW MUCH they can vary merely by pinching or expanding the turns.? ?Could it be that our low pass filters aren't exactly what we think they are?

I'm wondering if there would be value in mapping the characteristics of each of the Low Pass Filters.? ??

I think by lifting pin 1 of the output transformer, one could gain access to the input side of the fitlers.
Because they are diode switched, and the voltage created from the minute signals of the spectrum analyzer tracking generator are tiny, one might need to connect a 9V battery to the "HV" terminal to provide a source of back bias for the diodes.

However, I think (with drive = 1 of course) that a coaxial connector could then be solder into the input side of the filters, and using the spectrum analyzer one could map the filter response (at least at a source impedance of 50 ohms from the tracking generator).

What do you think?? ??

Gordon KX4Z

Another thing I could try, is just "squeezing" together the winddings of the coils that make up the 20meter (and likely 17meter) low pass filter to increase their inductance and drive the knee of the LPF as low as it can go with current componnents.? ? Then re-measure and see how that turned out.? ?

Ideas?

Gordon KX4Z

Gordon,

Here is the "ideal" filter response based on simulation:



It looks like -26DBc is the best to be expected from the filter.? I would look at the drive to see what it is doing.? If the push-pull is balanced, the 2nd harmonic should be significantly less than the 3rd.? If you have a high Z input for your SA, you could use it to see the spectrum at the secondary of T5.? The other option would be to look at the two waveforms with your oscilloscope at each driver stage emitters.? Compare R20 to R48 and then to R19 and R47.? With a dual trace scope, look at r20 vs R19 and R45 vs R48.

73
Evan
AC9TU

开云体育

Evan, do you have second and third harmonic measurements for your sBitx for various bands?

Gordon,

I have a DE version.? All of the 3rd harmonics are more than -60DBc down.

The second harmonics are all in the noise floor except for 20 meters.? That one is -57DBc.

I measured with a TinySA Ultra using the harmonic measurement function.? This is at 15 watts output into a 25-watt DC to 3GHz inline attenuator with a step attenuator to adjust for best readings on the TinySA.? I believe that my prior problems were measurement errors on my part.

Note that some close-in spurs are just under -50DBc, most?likely related to the switching regulator.

I still suspect there is an issue with the v2 driver.

73
Evan
AC9TU

Thanks, Evan.

I hope I haven't made a mistake in any of this.? ?But having measured the ICOM 718 and seeing its 2nd harmonic in the noise and the 3rd harmonic way way below FCC regs is reassuring.? ??

So now I picked out the third LPF from the left, which had the next to least turns, and from the software that would appear to be be used for frequencies between 10.5 and 18.5....and scrunched the turns as much as possible, and separated the toroid by as much as their soldered leads would allow.... and when I remeasured, it was less than a dB of any difference.

That was the easiest, simplest thing I could do.??
I agree that I don't particularly love the driver, but I think it looks like what we upgraded the DE to...and I think I did better there.

a)? is the raw signal much much worse now?
b)? is there feed through via the 10 meter filter?

Gordon KX4Z

Gordon,

I must apologize.? I do not understand your questions.? Did you want me to measure the driver spectrum?? I will need to work on setting up my high-impedance probe to test that part of the transceiver, as I have not done it yet.? I have not needed it.? Instead, I use my KiPrim DS1202 scope for rough FFT estimates.? It is not as accurate as the TinySA Ultra, but it is good enough to trace through a driver.

Please ask the question again with more detail on where and what you would like me to measure.

73
Evan
AC9TU