Fun with Math and Driving Yourself Nuts!
5/1/2020 More Evaluations....
Firstly I am struck at how good the signal sounds when listening to it on an outboard receiver tuned to LSB. There is no doubt the signal has presence and does not have the constricted nasal sound of being too narrow as sometimes happens with a crystal filter. I like what I hear.
But we must endeavor to really make some measurements aside from anecdotal stories from either myself or from other hams.
I once had an experience in Vietnam (darn right dangerous) where I had the OOD watch and we had a probe on our line. I collected a squad of about 11 Seabees and we went to investigate. It was about 11PM and as luck would have it there was an explosion and one of our team was seriously injured.
I was in the lead and thus did not see what happened directly. All I heard was a bang and saw a huge light. A board of inquiry was called and there were 12 different stories of what happened.
Finally the one story emerged. We had these huge "pop flares". These were flares about 18 inches long and in the base was a 22 Shell. Hitting the base caused the explosion and the flare is lit off.
As it turned out one the guys carrying the flare tripped over his size 13 feet and when he fell to the ground the flare still in hand was activated. Instead of going straight up in the air it went along the ground and hit another Seabee in the head and severely burned him.
The point is there were 12 different explanations but only one was true.
So we must find the real truth in our evaluation. I am devising a plan to measure the opposite sideband suppression (we don't need 12 different stories) and that may take a day or two to finalize the approach which I will document here.
I am looking to introduce an tone into the microphone circuit and then using the FFT on my Rigol DSO see what I can see while fiddling with R16 and R6.
Stay tuned for more fun and excitement. I see where the emperor (with a small e) does not want Dr. Fauci to testify before Congress -- Yes that indeed would be embarassing to hear about his penchant for blue lights inserted into the body!
73's
Pete N6QW
4/30/2020 ~ Circuit Observations.
This is a good time to stop and reflect on what has been done so far and what may be hiding in the bushes.
Regrettably our emperor (with a small e) see's his two primary enemies as being Data and Science! He hates Data because that contradicts his "gut" and he hates Science because it tells him shining a blue light up his butt will not cure the Novel Coronavirus.
But as hams and homebrewer's we must place an enormous resolve to look at what the data says and what science tells us.
Firstly I have been able to transmit a signal on 40 Meters using the Phasing SSB Transmitter. The anecdotal reports say -- it really sounds good. In several cases I followed up my contacts with emails and have received the formal affirmation that "it sounded good".
There was a bit of a snicker or two as I accompanied the emails with the picture below. For those with a sharp eye you will see a blank circuit board in the lower right hand corner. Ashamedly that only took 5 minutes to make with my $250K CNC mill and is for the IRF510 stage which will be built today. Just chuck up a piece of PCB and push the start button and 5 minutes later you are there.
But this is the point to STOP and collect some Data and to review the Science.
Here are the concerns I have with the rig and some of these issues I am not certain how to proceed. So If you have some thoughts then email me at craponthebench@gmail.com.
The Science tells us that the best you can expect with the 2Q4 is about 41 dB of opposite Sideband Suppression and there was an earlier posted link to an analysis from KK7B which demonstrates that point. Thus from the start we are nowhere near 55 or 60 dB that is achievable with other phasing approaches or a crystal filter. The digital filters in the SDR radios are essentially brick walls so not even close.
So the burning questions are you really achieving opposite sideband suppression and secondly to what degree?
Science and other published works tells us to include a hardware based audio filter in the microphone circuit. The theory of this is in part to aid in the opposite sideband suppression.
KK7B does have an audio filter in the T2 and I have on order a set of parts to build one for installation in this Phasing Rig. I simply lifted Campbell's circuit and through the magic of Digi-Key those 120 mH and 33 mH chokes are headed my way. They are relatively low priced (in the $1 range) so not like floating a loan. Maybe by early next week.
So we do need to know (based on Science) the opposite sideband suppression and how to gather the data to measure if we are even close to 41 dB of suppression.
In the simulation of the 2Q4 circuitry using the transformer coupling with the two trim pots in series across the secondary and feeding the 2Q4, the ratio of R6 and R16, impacts the 90 Degree Phase Shift across the audio spectrum. You can run the simulations yourself and look at the phase shift.
The two values that seems to work best in my simulation are 98 Ohms and 321 Ohms. Noteworthy the original applications schematic for the 2Q4 (using vacuum tubes) has a transformer coupling and essentially a 500 Ohm trim pot with center wiper to ground. Thus I am closely following the original circuitry.
Now, how to measure the opposite sideband suppression. While not illegal, DSB transmissions are fully permitted on our ham bands. But it is a waste of spectrum and likely you will incur the wrath of the FLEX Radio based 40 Meter SDR Police about the other sideband being present.
Now for some real observations about R6 and R16. In the Simulation we can see how the Phase Shift Difference is impacted by these resistors. But in practice here is an observation which is real; but has not been measured precisely.
I connected the Phasing Rig to a dummy load and listened for the best quality sound on an outboard receiver tuned to LSB. I repeated keying the microphone while listening to the signal using large over the ear headphones. I then switched the outboard receiver to USB and heard the signal coming through. I then carefully and with a light touch adjusted R16 (trim pot) ever so slightly and noted a reduction of the signal strength -- it was like a deep null.
I then could also see the pattern on the DSO as being reduced in amplitude. That reduction likely would be the removal of the USB component (or at least I hope so). Next I switched back to LSB and the signal strength in LSB did not appear to change. So the adjustment of R16 seemed to reduce the USB component.
Now in the schematic above I simulated a combiner transformer and connected it to the two outputs from the two audio channels. For grins I varied the inductance values up to 1 mH for each winding and that did not seem to impact the shape of the curve. I designated the output as a measuring point and then referenced that point back to the two outputs of the 2Q4.
This might be like drinking the bath water; but the intent was to see what the combined output might look like with reference to the phased outputs of the 2Q4.
The curve below is the result which looks at a frequency range of 1 hertz to 20 kilohertz. The two plot lines represent a reference phase difference angle of the combined output to each of outputs of the 2Q4. It is a decade scale but noteworthy over the audio range -- it is 90 Degrees with respect to the combined singular output to each of the outputs of the 2Q4.
As an exercise for the reader if you change the value of R6 and R16 the 90 Degree IS NOT maintained over the audio range -- so this indeed may be a useful measurement simulation.
As an exercise for the reader if you change the value of R6 and R16 the 90 Degree IS NOT maintained over the audio range -- so this indeed may be a useful measurement simulation.
Does this tell us something important or just something curious like seeing a rainbow?
As to opposite sideband suppression I did a look see by hooking my Rigol Scope using the FFT plots but not sure what I am seeing is what I am seeing --so more time with the Rigol today. A spectrum analyzer would be nice; but I also recently had unexpectedly dropped in my lap one of those new Nano VNA test instruments. That possibly is another tool in the tool kit. Not quite sure (as I haven't played with it as yet) that it would work for this.
So if we can accurately look at opposite sideband suppression then I would feel more comfortable running 1 KW with my newly built phasing transmitter without the fear of the SDR police.
73's
Pete N6QW
Checked into the JNN on April 29th, 2020! Three contacts early this evening.
4/29/2020 ~ On the Air Testing!
I have assembled a station consisting of the Multi-ELMAC , PMR6A Receiver, The Phasing SSB Transmitter running 12 watts, and a manual TR-Switch. I am really stretching the limit; but this just might be a typical mid-1950's SSB setup.
There is a net around mid-day on 7.204 kHz called the Jefferson Noon Net. One of the net stations is usually loud and so I will see if I can check into that net tomorrow.
[BTW I think there is an urban legend to the effect why it is called the Jefferson Noon Net. At one time there was a movement for the Conservative Northern Californians to decouple themselves from the Liberal SoCal Hollywood types. A new state would be formed which would be called the State of Jefferson. There was a referendum several years ago to do this but it did not make it to the ballot. Was the JNN to be like Fox news for the State of Jefferson?]
[BTW I think there is an urban legend to the effect why it is called the Jefferson Noon Net. At one time there was a movement for the Conservative Northern Californians to decouple themselves from the Liberal SoCal Hollywood types. A new state would be formed which would be called the State of Jefferson. There was a referendum several years ago to do this but it did not make it to the ballot. Was the JNN to be like Fox news for the State of Jefferson?]
The Phasing Transmitter Bread Board now has the Mic Pre-Amp and Amp, the 2Q4 Audio PSN, the RADIG V.2 with the two ADE-1's and the magic combiner transformer. Following that is a Transmit Pre-Driver with a 2N2219 and the Driver Board has a 2N3866 with the LPF installed on that same board. For an Linear Amp I have an RF Brick from an Atlas 210-X, which gets me to about 12 watts. The Driver board puts out about 1/2 watt. I guess I could just whip out a IRF510 Board on the CNC and we could get a few more watts out of the Atlas Brick.
The Phased LO's come from an Arduino Uno R3 and the Si5351 running CLK0 and CLK1 for the phase outputs.
Stay tuned we may actually have a contact in the next day or so.
73's
Pete N6QW
********************************
4/28/2020 ~ More Refinements
So OK the refinements include adding an audio filter ahead of the PSN as KK7B did in his T2 and as was suggested by Greg, my friend down in VK Land. That did not do anything to increase the low frequency response as it seems to be stuck at 400 Hz.
No matter what was changed --no movement on the low end. Any changes made seem to impact the Phase Shift up/down from 90 Degrees not the low end frequency response.
That said I have learned some tricks to make the data more readable such as plotting 10M points and then in the plot settings select show points and you get the curve below. I also learned how to eliminate the amplitude from the plot. This sure makes things more readable!
Thanks again to Greg I was enlightened on how to introduce an array of values into the simulation and plot the same on a single chart. I wanted to show the sensitivity of R6 by only changing its value from 10 Ohms to 321 Ohms. Pretty cool stuff guys -- Thanks Greg!
Next is the actual circuit with the KK7B filter; but I still owe you the microphone pre-amp circuit. The plot above is with the KK7B filter in the simulation. Not seeing any significant changes with the filter probably will not use it with the 2Q4
#TOT Listening to the emperor (with a small e) such as injecting Lysol likely will result in your death. He cannot be trusted!
Evidently I am not alone --his all important (to him) approval rating has dropped by a significant factor.
MAGA ~ Dump Trump in 2020
Evidently I am not alone --his all important (to him) approval rating has dropped by a significant factor.
MAGA ~ Dump Trump in 2020
4/25/2020 ~ Explaining the Circuit Revisions
Yesterday I spent some time exploring what would make the signal better and what hardware changes would be involved.
Firstly thanks to Dave I was supplied a reference authored by Rick Campbell that explores several methods of audio phase shifting with a specific evaluation of the 2Q4. I will make this document available on my website. But I have to load it so you will need to recheck this blog for the link.
http://www.n6qwradiogenius.us/Phasing_SSB.html
http://www.n6qwradiogenius.us/Phasing_SSB.html
Scroll down to the KK7B Link!
Spoiler alert --- the best you can expect from the 2Q4 is 41 dB of opposite sideband suppression. Other methods can provide up to 60 dB of suppression.
So to the purists who much like those illuminati who decried the phase noise characteristics of the Si5351, you can stop now and just find another blog to pick on. But, if you want to explore the very basics of phase shifting to generate SSB signals, then continue to ride along.
With much frustration and lack of results, I moved back to the transformer coupling from the microphone amp to the 2Q4. I had indicated that I sort of found a sweet spot on the 500 Ohm trimmer pot where there seemed to be the best signal output much like the null with a manually adjusted Balanced Modulator.
Noteworthy I found that the apparent sweet spot was located at a point not on the center of the range. LT Spice simulations have confirmed that to be the case and something more in line with 425 ohms total with a split of about 100 to 325 Ohms works best. In my LT Spice simulation I show the pot as two resistors R6 and R16. The phase shift is shown below.
The hardware for R6 and R16 is two 500 ohm pots connected as variable resistors in series with the common connection to ground. One pot is set at 100 Ohms and the other at about 323 Ohms. I find that if you power on the rig (without the microphone connected) and listen on an adjacent receiver and slightly tweak the 100 ohm pot in either direction there is a spot where the background noise drops and it sort of nulls. Connect the microphone and it is a clean signal.
There is a microphone pre-amp circuit and I will provide a schematic on the next posting.
Significant is that the flat part of the curve is from 400 Hz to about 2300 Hz. So that will get the 40M FLEX guys upset that the signal has no lows. I have not cracked that nut on the low end.
But the bottom line is that I have been able to simulate what has been built. Later I will post the full schematic including some capacitor changes to provide a signal boost.
The hardware for R6 and R16 is two 500 ohm pots connected as variable resistors in series with the common connection to ground. One pot is set at 100 Ohms and the other at about 323 Ohms. I find that if you power on the rig (without the microphone connected) and listen on an adjacent receiver and slightly tweak the 100 ohm pot in either direction there is a spot where the background noise drops and it sort of nulls. Connect the microphone and it is a clean signal.
There is a microphone pre-amp circuit and I will provide a schematic on the next posting.
Significant is that the flat part of the curve is from 400 Hz to about 2300 Hz. So that will get the 40M FLEX guys upset that the signal has no lows. I have not cracked that nut on the low end.
But the bottom line is that I have been able to simulate what has been built. Later I will post the full schematic including some capacitor changes to provide a signal boost.
#TOT Listening to the emperor (with a small e) may result in your death. He cannot be trusted!
*******************
4/25/2020 ~ Circuit Revisions
After fiddling with the phase inverter circuit and becoming frustrated with the results (taps off of the collector and emitter) I went back to how the Model 350 2Q4 was installed with tube circuits.
The original 2Q4 circuit had a transformer with the primary in the plate circuit and each leg on the secondary fed the PSN. A 500 Ohm pot is connected across the secondary with the center wiper connected to ground. The photo below shows the 600:600 Ohm Modem transformer fitted to the DC Rail and the collector of the 2N3904. I bridged the secondary with a small 500 Ohm Trimmer pot and the wiper to ground.
Boom more output and a bonus. The question comes up of how to adjust the 500 Ohm Trimmer Pot. I connected my scope to the small RF amp stage and then speaking into the microphone I adjusted the pot to the extreme ends and there was a max signal on the scope. But approximately mid range was a smaller signal point. It was quite obvious. This was just like nulling out a carrier in a balanced modulator. That is where it is set.
I am hitting about 5 Volts PTP which translates to about 60 milliwatts. Now there is no LPF on the output so we need to make that measurement after a filter is installed.
Today was neaten up things day.
On the left we have the audio board and mic jack followed by the RADIG V.2 and on the right is the one transistor RF amp. The Arduino/Si5351 is in the background.
73's
Pete N6QW
4/24/2020
PSA: Do Not Drink Lysol! Your government has misled you. But instead build something!
More tests today with the Phasing Transmitter. Much progress yesterday and the plan is to attempt to improve the SSB signal. The proof of concept verifies it works; but now we need to make it work so it is suitable for use on air on the 40 Meter Band.
I went back to the LT Spice Simulation and am looking to better balance the outputs off of the collector and emitter of the 2N3904. The other issue is than the signal will lack lows as the simulation shows the 90 degree shift drifts way off into space at about 350 Hz. The max is about 2.3 kHz. So the trick is to address how to boost the lows and yet have the 90 Degree shift. It may be the inherent values in the 2Q4. But several post ago I posted the values with part numbers. So a homebrew version may be needed.
I made some incremental parts value changes
and now see about 13 Milliwatts on peaks. It sounds better but the scope pattern still looks a bit rough
Thus more work is needed but I am encouraged by the works so far. Still am looking for lower end response and that has eluded me so far.
One possible approach is to change the values in the LT Spice simulation for just the 2Q4 and then we can evaluate if indeed the long pole in the tent is the 2Q4. Stay Tuned!
73's
Pete, N6QW.
#TOT -- The emperor's (with a miniscule e) latest miracle cure will kill you!
**************************
4/23/2020 ~ Initial Transmit Testing!
Test with Transmit Audio! Eureka -- working!
I need more microphone gain so that is the next build. But these tests are very promising!
73's
Pete N6QW
Vote Him Away … #TOT. 47000 Dead is not a perfect response.
4/22/2020 ~ More Parts to the Audio PSN Board
Below is the schematic I am using and if you want the LT Spice simulation send me an email to
While the LT Spice simulation looks pretty good my test plan is to use the NE5534 op-amps and employ my Rigol DSO to look at the two outputs ala Lissajous figures to see how the phase shift holds up over the range 270 to 2700 Hertz. I have a test oscillator that I need to fix first before doing that test so just a bit of a program delay.
Noteworthy the phase shift is very sensitive to the collector and emitter resistors in the mic amp stage. R24 is 770 ohms and R4 is 232 Ohms. I built those resistors using a fixed resistor and a trim pot connected as a variable resistor. I have manually set those values as determined by the simulation; but further tweaking may be required once the circuit is running.
Proof of the 90 Degree Phase Shift at 300 Hz.
#TOT
Vote Him Away!
******************
4/21/2020 ~ A Visit to KK7B's 1993 QST Article.
In April of 1993 (23 years ago) , QST published a seminal article authored by KK7B on the construction of a multimode Phasing type SSB Transmitter. The very first graphic shows a block diagram of the transmitter.
The author states this block diagram has been in the ARRL Handbook for over 40 years and even mentions the Audio Phase Shift Network is the B&W Model 350 2Q4 --only the transmitter used tubes!
The blocks following the PSN will utilize my RADIG V2.0 board --that part is already built. [ADE-1's and FT-37-43 combiner.)
In Campbell's rig he uses a MMIC (MAR-2) to develop a whopping 3 milliwatts. I will use a relay steered amplifier block. That said when I get around to using Op-amps my choice would be to follow this article only use NE5532's and perhaps an AG303-86G MMIC as the amplifier. (He guys I am using what is in the bins).
****************************
4/20/2020 ~ Firstly, Some Management Tips!
Hey emperor (with a diminishing e) here are a couple of management tips for you.
The 7 P's: Proper Prior Planning Prevents Piss Poor Performance.
We certainly know you did nothing in February to plan for this as now the US represents 33% of the world's infected cases and 25% of the deaths.
This is followed by...
If you fail to plan, you plan to fail!
The blame finger is pointing right at you, emperor, and not China, not the WHO nor the US Governor's! BTW the VP should not skate through this.
Just finished cutting the board for the audio section and this includes the 2N3904, space for the 2Q4 and the two op-amp stages. The photo below shows the plan for the original stage that used the 2Q4 with a couple of transistors.
********************
News Flash ~ Friend Dean, KK4DAS has just completed the build of my Simple SSB Transceiver project and has made his 1st QSO today on that rig! He was excited and so was I. See Dean's Blog https://kk4das.blogspot.com.
Info on the Simple SSB can be found at www.n6qw.com.
The Phasing Type SSB Transmitter, Receiver or Transceiver has a strong underpinning of Mathematics to make it all play. Basically the ability to null out a carrier as well as favor one sideband over another comes down to relationships between and among those old trigonometric functions: Mr. Sine and Mr. Cosine. Essentially the way signals are combined following those trigonometric functions results in a "selected sideband".
But there are circuits that enable the "sum" or "difference" signals that are shifted by 90 Degrees which in effect makes the math work. These circuits are complex.
The reason behind the complexity is that the signal amplitude and the phase shift must hold not for just one frequency but for several thousand frequencies as you would have in the typical voice range from 200 to 2700 Hz -- my friend that is 2,500 frequencies.
There are many ways to create the signals shifted by 90 Degrees and these include simple networks such as the 2Q4 that is a mere four resistors and four capacitors. Not just any eight components; but very specific components with tight tolerances.
As I found out in my early investigations with the 2Q4 simulation what you hang ahead of the Phase Shift Network and what you hang on following the network impacts the amplitude and actual phase shift performance. At this point I am only investigating the 2Q4. Expectations (and a goal) are at least 40 dB of opposite sideband suppression.
But there are more complex circuits using many more components and having inputs and outputs at 0, 90, 180, and 270 Degrees. These are referred to as Polyphase networks. I have two in the garage which I bought from QRP Labs. They are complex networks and require complex circuitry ahead and following them. Those are for later experimentation.
These more complex circuits offer much greater than 40 dB of opposite sideband suppression and would be better suited to receiver application where expectations of 60 dB are well within what is possible with such "better" circuits.
Other methods also include the use of Op Amps for the phase shifting as championed by KK7B in the R2 and T2 Boards. Those too will likely be investigated. But getting ones feet wet with the simple 2Q4 seems like a good starting place for me personally.
Now with the help of some savvy hams around the world I now have a basic circuit which has been simulated in LT Spice. Drop me an email at the address above if you would like the LT Spice Simulation directly
The simulation involves essentially a microphone input stage which provides two outputs to feed the second piece which is the 2Q4 PSN . The output from the Phase Shift Network feeds two Op Amp Amplifier stages so that you have appropriate signal levels to drive the two ADE-1's that are on my RADIG V2 board. Sideband shifting will be done in software by shifting the Phase of CLK0 and CKL1 in the Si5351. You got to love those Arduino's!
The ADE-1's make all of the magic happen as the audio level 90 Degree out of phase signals are mixed with the two phased outputs at the operating frequency from the Si5351 and then combined in the magic combiner toroid and further amplified.
Note the PSN is sensitive to the in/out impedances of what is hung on its ports. A slight resistor change can impact the signal amplitude as well as the phase shift. With a bit of tweaking I have arrived at compromise setting which maintains the 90 degree shift over the audio spectrum but at a small expense to the amplitude at the lower frequencies. Mind you the amplitudes of the two signals track at all frequencies. But the amplitude for both signals is less at the lower frequencies.
This amplitude issue may have to be treated with some filtering that say would pass the lows but somewhat attenuate the higher frequencies. Likely it is not a singular filter but several in cascade The desired end result is a flat curve with no phase shift. So more may be involved. than just my starting circuit. (BTW Four resistors have changed since this was embedded.)
This is my starting circuit. It begins with a 2N3904 configured as a phase inverter (outputs off of the collector and emitter) which is direct (DC) coupled to the Phase Shift Network. As mentioned earlier the collector resistor and the emitter resistor values directly impact the amplitude and phase shift. In the hardware build there may have to be trim pots included in the collector and emitter circuits to fine tune the values.
The output side of the PSN feeds two Op Amp amplifiers. In this case I used the OP07 as it was in the LT Spice Catalog; but I intend to use NE5534 in the hardware build. My LT Spice is two years old and I am fearful of upgrading it you know like the Arduino IDE --you lose everything.
So how does this look when simulated? I took three shots, 500 Hz, 1500Hz and 2500 Hz which are shown below.
At 1500 Hz the amplitudes of both signals are just about neck and neck whereas one of the signals at 500 Hz and 2500 Hz appears a bit larger. These are not large differences but more of a slight difference.
Now how do you know that the phase difference is fairly constant and how can you see the amplitude levels over the audio range? Thanks to W3JDR, I was shown a way to do this in LT Spice. (Man when you know stuff you can do stuff!) The lower plot is the Phase Shift over the Range 200 to 2700 Hertz. The tangent line is 90 Degrees so these 8 components are really doing a super job. But again I want to stress that resistors external to the PSN have a dramatic effect on the results. The variation over the flat part of the curve from 90 Degrees is 0.1dB
Now the upper curve is the amplitude which shows less amplitude at lower frequencies. But each horizontal grid line is 0.3dB --so we are looking at maybe 2 dB up until about 1500 Hz. So this is where flattening the curve with audio filtering will have to be addressed. Thanks Joe!
Again send me an email if you want this specific simulation file.
73's
Pete N6QW