This is an interesting presentation. When I use multiband radios like an Elecraft KX1 I always use an antenna tuner and a long wire antenna. It is true a resonant high dipole will give the best results. In the case you use it portable and you like to use it on more as one band it is more convinient with a tuner and a End Feed Half Wave wire Antenna. You will get surprising results even with this very simple antenna system. My experience guide me to use batteries with QRP transceivers. When you operate low power you like to stand out with a very clean and easy to read signal. I work mostly in cw mode. It is interesting to learn the technology build into this radio. The PA stage and the math behind it is amazing.
A comment about using FET power amps: MOSFETs like the BS170 have two characteristics that are important with regard to efficiency. First there is the on-resistance, and its relationship the gate-source voltage. And second is the gate charge, because the gate isn't fully on or off until you've charged or discharged most of that gate charge. For the on-resistance, it is important to drive the gate HARD. The minimum on resistance is not reached until you get the gate voltage up to 10V. But the gate drive in this radio uses 74ACT00 NAND gates, which only go to a maximum of 5V, where the on resistance is about four times what it is at 10V. This is something that could use some improvement. For the gate charge, this is why there are pads for up to four BS170s and also for a DPAK package MOSFET. Guido used two 74ACT00 gates in parallel, which gives it a very high instantaneous current, to charge and discharge the gates as quickly as possible. But also, BS170s have a low gate charge compared with bigger "power" MOSFETs, making them harder to drive, which is why they provided multiple options for how many FETs and which type. I believe the use of three BS170s in parallel was found to have the best efficiency overall. Based on these factors, I think the best approach would be, instead of driving them with 5V logic, using some higher-voltage logic or discrete transistor push-pull amp that can drive ~100 mA both directions should be good. Note that this current would only be on for a few nanoseconds at each edge of the square wave, so this wouldn't have a big effect on the overall power required by the radio, and a complementary pair like 2N3904/3906 should be up to the task.
That's a very important point you made , I bet this would also improve the ssb distortion. Maybe have to try this , could also switch the gates with another fet with the drain limited to 10v.
I've been studying the design of this radio, because at first it did not make any sense to me, but then I found some notes by Guido, the guy who wrote all of the firmware for it, and it's very clever at making the components do what they aren't intended to do. And while Dr. Fong describes one way to do this, it is not exactly how it is done in this radio. My first question was, how does he suppress the unwanted sideband? How is this possible without a crystal filter? The answer is, digitally. What he has done is to implement an image rejecting mixer (look it up on Wikipedia) numerically. This involves running the microphone input, with some amplification, into an ADC (which uses the sigma-delta method Dr. Fong describes), which he then runs through a Hilbert transform using the ATmega328P as a digital signal processor. The Hilbert transform phase shifts the whole audio bandwidth by 90 degrees. So now he has, numerically, the initial (I) phase and quadrature (Q) phase of the modulating signal, already technically SSB, although it hasn't been mixed with the RF yet. I'll get to that. Now here's the clever part: he digitally converts this from I and Q, which are in Cartesian coordinates, into phase and amplitude, or polar coordinates, using arctangent(I/Q) to get the phase, and sqrt(I^2 + Q^2). So that's the fundamental innovation - separating the phase and amplitude components. Next, the synthesizer chip, the Si5351A, does not have the capability of directly modulating phase, but it CAN have its frequency adjusted very quickly. So when you want to advance the phase, you just raise the frequency for a short period, then return to the center frequency, and likewise when you want to retard the phase, you lower the frequency for a short period, then return to center frequency. So this chip is always sending out a square wave that is phase modulated around the specified carrier frequency. This square wave is essential to getting high-efficiency, because the FETs are feeding a tank circuit, turning on very close to the negative-going zero-crossing, so there is very little loss when they turn on, and then turning them off very close to the zero-current point, which again results in very low loss in the transistors. But all we seem to have here is a phase-modulated signal, whose power is proportional to the voltage input to the FET drains. And this is the second innovation: if you take a PWM output from the ATmega328P, and use that to regulate the voltage feeding the FETs, you have amplitude modulation. Now, if you only used the amplitude modulation, this would be an AM transmitter. But since the RF is already phase modulated to do just one sideband, the result is SSB, at very high efficiency. It's a genius design, and not as tricky to calculate as Dr. Fong thinks - the ATmega328P is NOT a powerhouse DSP chip; it's a general purpose microcontroller.
I noticed that in the document references that Guido's and Manuel's contributions are reversed. Guido is the software guru who made most of this possible, including the SSB. Manuel did the amp. BTW, I laugh at the idea that someone won't buy or build a transmitter with only 30 dB of harmonic attenuation. It takes just a few small caps and toroids to get another 30 dB of harmonic attenuation on top of that. Low-pass filters are dirt cheap, and quick and easy to build. Something easy that should be fun for a ham to build. Thee seems to be some confusion of the very nature of ham radio. Ultimately, the harmonic content of any rig, commercial, kit, experimental, or completely homebrew is ALWAYS the ham's responsibility. The FCC gives hams the right to do whatever they want to their rigs, including design and build them from scratch, because it's ham radio, not super-CB. The FCC created the Amateur Radio Service so that radio hobbyists had someplace to learn about RF. Their intent is stated in the first few paragraphs of the Part 95 rules and regulations. Anyone who thinks their rig's harmonic content is the responsibility of some equipment manufacturer is sadly mistaken. The FCC WANTS hams to learn to properly maintain their own gear. Hams would be well advised to learn, as they are held responsible whether they do or not. Finally, feel free to review the (tr)uSDX PA tuning video on UA-cam from Manuel, DL2MAN if you want more power from your rig. The rig's PAs are very similar, (if not identical component numbers,) since Manuel designed them both. In a nutshell, power output can be decreased or increased by playing with a particular toroid coil's winding's positions, or for major changes, increased by removing a turn on the correct toriod. Six watts is a good place to stop as far as power output is concerned, and that's only if you've already tuned the output for good efficiency, say 80 to 90% and your antenna has a reasonable SWR. If you tune to 6W you will want to closely regulate the power supply voltage. If you had 11.5 volts on the output FETs when you tuned it up, then applied 14 volts from your car battery your FETs won't last long with the 9 or 10 W output you'll get due to the higher voltage on the FETs.
love how this makes a good cb! at 14;02 it even list that! cb radios are out of sight price wise for one that can do fm ssb and am. you then have then get a power supply . this has you covered just add antenna and go! glad this ham is not ham is not anti cb!
I was very interested to hear of your experience with swapping out the 3 x BS170s for an IRF510 and the fact that this change did not result in an increase in output power. The IRF510 would not be my first choice, it has 180pf of gate capacitance where the 3xBS170 have a combined gate capacitance of about 75pf, that extra capacitance is probably shunting a lot of the drive signal to ground, Also the IRF510 does not perform very well on 12v, it works, but it's not a stellar performance. Raise the supply voltage to 24v and it's a different story, the Bitx boys worked this out a few years ago. So my choice would be a small RF power fet designed for HF operation at 12v, I will try a Mitsubishi RD16HHF1. I understand that Mitsubishi has discontinued this device but I also understand that they are still available, I found some at A$8.50 each. I will let you know how ,it works out.
3 or 4 BS170s work just fine. and they're cheap as dirt, a plus for a rig with no SWR protection. They can produce 6 or 7 watts simply by playing with the turns on one of the toroids for each band. Watch the (tr)uSDX PA tuning video on UA-cam from Manuel, DL2MAN if you want more power from your rig. If you want 10W or more, try the IRF510 on a heatsink with a small step-up switcher. Still cheap to replace the output FET if you pop it, and you might use a switcher with current limit to further protect the PA FET. Note that the increased capacitance will start to limit the output on the higher bands since the TTL driver can't charge and discharge the FET's gate capacitance fast enough.
Great presentation Ed. I’m having so much fun like you with this little radio. I used some suggested settings and it been stable and like you said if you don’t ask nobody seems to notice audio issues. Thanks for the deep dive on the operation theory and how the FET switching works. Sounds like right now it’s a compromise but the near future will be exciting for this method.
Aquí es dónde se confirma el dicho que si antiguamente te gastabas un montón de dinero en comprar la emisora la antena tendría que ser muchísimo más cara, ahora te compras un aparato qrp por escaso dinero y la antena te sale más cara y si te oyen bien y tú los escuchas y te preguntan con cuántos watios sales al aire y tú le dices que con 5 o menos de 10watios te regalan un suplemento de mentiroso muy nutritivo, menos mal que la técnica y la comunicación en Red demuestran que se puede hacer, gracias por el video y saludos
I have owned a uSDX Plus for several years. Amazing little radio with excellent receiver performance for the price.
I am glad Mr Fong shed light on Harmonics at Pwr.., Lot's of folks out there ripping apart Transceivers without knowing this. Thank You
This is an interesting presentation.
When I use multiband radios like an Elecraft KX1 I always use an antenna tuner and a long wire antenna.
It is true a resonant high dipole will give the best results.
In the case you use it portable and you like to use it on more as one band it is more convinient with a tuner and a End Feed Half Wave wire Antenna.
You will get surprising results even with this very simple antenna system.
My experience guide me to use batteries with QRP transceivers. When you operate low power you like to stand out with a very clean and easy to read signal. I work mostly in cw mode.
It is interesting to learn the technology build into this radio. The PA stage and the math behind it is amazing.
You got that right.
Making a contact half way around the world is quite a shock to be sure.
A comment about using FET power amps: MOSFETs like the BS170 have two characteristics that are important with regard to efficiency. First there is the on-resistance, and its relationship the gate-source voltage. And second is the gate charge, because the gate isn't fully on or off until you've charged or discharged most of that gate charge.
For the on-resistance, it is important to drive the gate HARD. The minimum on resistance is not reached until you get the gate voltage up to 10V. But the gate drive in this radio uses 74ACT00 NAND gates, which only go to a maximum of 5V, where the on resistance is about four times what it is at 10V. This is something that could use some improvement.
For the gate charge, this is why there are pads for up to four BS170s and also for a DPAK package MOSFET. Guido used two 74ACT00 gates in parallel, which gives it a very high instantaneous current, to charge and discharge the gates as quickly as possible. But also, BS170s have a low gate charge compared with bigger "power" MOSFETs, making them harder to drive, which is why they provided multiple options for how many FETs and which type. I believe the use of three BS170s in parallel was found to have the best efficiency overall.
Based on these factors, I think the best approach would be, instead of driving them with 5V logic, using some higher-voltage logic or discrete transistor push-pull amp that can drive ~100 mA both directions should be good. Note that this current would only be on for a few nanoseconds at each edge of the square wave, so this wouldn't have a big effect on the overall power required by the radio, and a complementary pair like 2N3904/3906 should be up to the task.
That's a very important point you made , I bet this would also improve the ssb distortion. Maybe have to try this , could also switch the gates with another fet with the drain limited to 10v.
I've been studying the design of this radio, because at first it did not make any sense to me, but then I found some notes by Guido, the guy who wrote all of the firmware for it, and it's very clever at making the components do what they aren't intended to do. And while Dr. Fong describes one way to do this, it is not exactly how it is done in this radio.
My first question was, how does he suppress the unwanted sideband? How is this possible without a crystal filter? The answer is, digitally. What he has done is to implement an image rejecting mixer (look it up on Wikipedia) numerically. This involves running the microphone input, with some amplification, into an ADC (which uses the sigma-delta method Dr. Fong describes), which he then runs through a Hilbert transform using the ATmega328P as a digital signal processor. The Hilbert transform phase shifts the whole audio bandwidth by 90 degrees. So now he has, numerically, the initial (I) phase and quadrature (Q) phase of the modulating signal, already technically SSB, although it hasn't been mixed with the RF yet. I'll get to that. Now here's the clever part: he digitally converts this from I and Q, which are in Cartesian coordinates, into phase and amplitude, or polar coordinates, using arctangent(I/Q) to get the phase, and sqrt(I^2 + Q^2). So that's the fundamental innovation - separating the phase and amplitude components.
Next, the synthesizer chip, the Si5351A, does not have the capability of directly modulating phase, but it CAN have its frequency adjusted very quickly. So when you want to advance the phase, you just raise the frequency for a short period, then return to the center frequency, and likewise when you want to retard the phase, you lower the frequency for a short period, then return to center frequency. So this chip is always sending out a square wave that is phase modulated around the specified carrier frequency.
This square wave is essential to getting high-efficiency, because the FETs are feeding a tank circuit, turning on very close to the negative-going zero-crossing, so there is very little loss when they turn on, and then turning them off very close to the zero-current point, which again results in very low loss in the transistors.
But all we seem to have here is a phase-modulated signal, whose power is proportional to the voltage input to the FET drains. And this is the second innovation: if you take a PWM output from the ATmega328P, and use that to regulate the voltage feeding the FETs, you have amplitude modulation. Now, if you only used the amplitude modulation, this would be an AM transmitter. But since the RF is already phase modulated to do just one sideband, the result is SSB, at very high efficiency.
It's a genius design, and not as tricky to calculate as Dr. Fong thinks - the ATmega328P is NOT a powerhouse DSP chip; it's a general purpose microcontroller.
Where can we get the bits to install the battery - that'd be great !!
Great little radio, worked 61 countries in 2 months with FT8. Only issue I found is lack of PA protection, the BS170s don't like high SWR!.
exactly so G1M from Xiegu will stand all sorts of abuse so it won't fail - RD16HH used as output mosfet in that case.
I noticed that in the document references that Guido's and Manuel's contributions are reversed. Guido is the software guru who made most of this possible, including the SSB. Manuel did the amp. BTW, I laugh at the idea that someone won't buy or build a transmitter with only 30 dB of harmonic attenuation. It takes just a few small caps and toroids to get another 30 dB of harmonic attenuation on top of that. Low-pass filters are dirt cheap, and quick and easy to build. Something easy that should be fun for a ham to build. Thee seems to be some confusion of the very nature of ham radio. Ultimately, the harmonic content of any rig, commercial, kit, experimental, or completely homebrew is ALWAYS the ham's responsibility. The FCC gives hams the right to do whatever they want to their rigs, including design and build them from scratch, because it's ham radio, not super-CB. The FCC created the Amateur Radio Service so that radio hobbyists had someplace to learn about RF. Their intent is stated in the first few paragraphs of the Part 95 rules and regulations. Anyone who thinks their rig's harmonic content is the responsibility of some equipment manufacturer is sadly mistaken. The FCC WANTS hams to learn to properly maintain their own gear. Hams would be well advised to learn, as they are held responsible whether they do or not.
Finally, feel free to review the (tr)uSDX PA tuning video on UA-cam from Manuel, DL2MAN if you want more power from your rig. The rig's PAs are very similar, (if not identical component numbers,) since Manuel designed them both. In a nutshell, power output can be decreased or increased by playing with a particular toroid coil's winding's positions, or for major changes, increased by removing a turn on the correct toriod. Six watts is a good place to stop as far as power output is concerned, and that's only if you've already tuned the output for good efficiency, say 80 to 90% and your antenna has a reasonable SWR. If you tune to 6W you will want to closely regulate the power supply voltage. If you had 11.5 volts on the output FETs when you tuned it up, then applied 14 volts from your car battery your FETs won't last long with the 9 or 10 W output you'll get due to the higher voltage on the FETs.
Hi, do you have a part number for the battery?
He made a quick comment about adding more power to the uSDX. Could you tell us more?
love how this makes a good cb! at 14;02 it even list that! cb radios are out of sight price wise for one that can do fm ssb and am. you then have then get a power supply . this has you covered just add antenna and go! glad this ham is not ham is not anti cb!
If I'm not mistaken, that processor is what the Arduino Uno uses.
I was very interested to hear of your experience with swapping out the 3 x BS170s for an IRF510 and the fact that this change did not result in an increase in output power. The IRF510 would not be my first choice, it has 180pf of gate capacitance where the 3xBS170 have a combined gate capacitance of about 75pf, that extra capacitance is probably shunting a lot of the drive signal to ground, Also the IRF510 does not perform very well on 12v, it works, but it's not a stellar performance. Raise the supply voltage to 24v and it's a different story, the Bitx boys worked this out a few years ago.
So my choice would be a small RF power fet designed for HF operation at 12v, I will try a Mitsubishi RD16HHF1. I understand that Mitsubishi has discontinued this device but I also understand that they are still available, I found some at A$8.50 each. I will let you know how ,it works out.
3 or 4 BS170s work just fine. and they're cheap as dirt, a plus for a rig with no SWR protection. They can produce 6 or 7 watts simply by playing with the turns on one of the toroids for each band. Watch the (tr)uSDX PA tuning video on UA-cam from Manuel, DL2MAN if you want more power from your rig. If you want 10W or more, try the IRF510 on a heatsink with a small step-up switcher. Still cheap to replace the output FET if you pop it, and you might use a switcher with current limit to further protect the PA FET. Note that the increased capacitance will start to limit the output on the higher bands since the TTL driver can't charge and discharge the FET's gate capacitance fast enough.
Great presentation Ed. I’m having so much fun like you with this little radio. I used some suggested settings and it been stable and like you said if you don’t ask nobody seems to notice audio issues. Thanks for the deep dive on the operation theory and how the FET switching works. Sounds like right now it’s a compromise but the near future will be exciting for this method.
Aquí es dónde se confirma el dicho que si antiguamente te gastabas un montón de dinero en comprar la emisora la antena tendría que ser muchísimo más cara, ahora te compras un aparato qrp por escaso dinero y la antena te sale más cara y si te oyen bien y tú los escuchas y te preguntan con cuántos watios sales al aire y tú le dices que con 5 o menos de 10watios te regalan un suplemento de mentiroso muy nutritivo, menos mal que la técnica y la comunicación en Red demuestran que se puede hacer, gracias por el video y saludos
IS this radio FCC Certificated
First!
*_perk_*