Posts Tagged ‘linearity’
From Lightning Comes a New Icom IC-7610 (First Transmission)
Wow. What a radio!
One of the most useful (and, to me, amazing) features of this Icom IC-7610, is the IP+ function, which, when turned on, improves the Intermodulation Distortion (IMD) quality by optimizing the direct sampling system performance. This function optimizes the Analog/Digital Converter(ADC) against distortion when you receive a strong input signal. It also improves the Third-order Intercept Point (IP3) while minimizing the reduction of the receiver sensitivity.
In short: I was listening to an s-0 (i.e., no strength-meter movement) weak signal of a DX station, when right adjacent to the frequency came an s-7 signal, wiping out my ability to copy that weak signal. I turned on the IP+ and the distortion of the adjacent signal disappeared, and once again, I heard the weak signal IN THE CLEAR! WOW!
This video is a quick capture of my running the Olivia Digital Mode on HF, on the 30-Meter band. The transmissions are of a two-way Olivia digital-mode radio conversation between station K8CJM and station NW7US on 12 November 2019 (UTC date). K8CJM is located in Dayton, Ohio, and I am located in Lincoln, Nebraska. I’m running the radio at full power. The radio is rated as being able to handle 100% duty cycle at full power. The radio ran cool, no significant heating.
A few months ago, a lightning strike took out my ham radio station. The antenna was NOT connected, but I did not unplug the power supply chain and my computer from the wall. The surge came in through the power mains, and fried my uninterruptable power supply, the interfaces between my PC and radio, and fried the radio. Thankfully, all of that was covered by my homeowner’s insurance policy, less the steep deductible. My insurance covered all of the blown items, and that provided me this chance to obtain a repack version of the Icom IC-7610. I bought an extended four-year warranty.
CAUTION: Check the documentation of your transceiver/transmitter. NEVER run your radio’s power out at a level that exceeds what it can handle in reference to the duty cycle of the mode you are using. Olivia, for instance, is a 100-percent duty cycle mode. Morse code is NOT quite 100% duty cycle. Nor is SSB, a mode that operates with a duty cycle much lower than 100%. Your radio’s manual should tell you the specifications regarding the duty cycle it can handle! If you run more power than your radio can handle with the given duty cycle of the mode in use, you will blow your radio’s finals or in some other way damage the radio! Beware! I’ve warned you!
Compression and ALC!?
Some have noted that it appears that I’ve left on the Compression of the transmitted audio. However, the truth is that compression was not being used (as is proof by carefully taking note of the zero meter movement of the Compression activity). I had the radio set for 20-Meter USB operation on the Sub VFO. Compression was set for standard USB operation. Note also that the radio was transmitting USB-D1, which means the first data/soundcard input to the radio.
Also, some people complain about my use of ALC, because, in their view, ALC (automatic level control) is a no-no for data modes.
The notion that one must NEVER use ALC when transmitting digital modes is not accurate.
Multi-frequency shift keyed (MFSK) modes with low symbol rate–such as the Olivia digital modes–use a single carrier of constant amplitude, which is stepped (between 4, 8, 16 or 32 tone frequencies respectively) in a constant phase manner. As a result, no unwanted sidebands are generated, and no special amplifier (including a transmitter’s final stage) linearity requirements are necessary.
Whether the use of ALC matters or not depends on the transmitted digital mode.
For example, FSK (Frequency-Shift Keying; i.e., RTTY) is a constant-amplitude mode (frequency shift only). In such a case, the use of ALC will NOT distort the signal waveform.
PSK31 does contain amplitude shifts, as an example, therefore you don’t want any ALC action that could result in distortion of the amplitude changes in the waveform.
On the other hand, the WSJT manual says that its output is a constant-amplitude signal, meaning that good linearity is not necessary. In that case, the use of ALC will NOT distort the transmitted signal-amplitude waveform. You can use ALC or not, as you choose when you run WSJT modes, or Olivia (MFSK).
Nowhere in this am I advocating running your audio really high, thinking that the ALC will take care of it. I am not saying that. I am saying that some ALC is not going to be an issue. You MUST not overdrive any part of the audio chain going into the transmitter!
Transmit audio out of the sound card remains at a constant amplitude, so there will be no significant change in power output if you adjust your input into the radio so that the ALC just stops moving the meter, or, you can have some ALC meter movement. You can adjust your audio to the transmitter either way.
If the transmitter filters have a significant degree of ripple in the passband then you may find that RF power output changes with the selected frequency in the waterfall when there is no ALC action. Allowing some ALC action can permit the ALC to act as an automatic gain adjustment to keep the output power level as you change frequencies.
Linear and Non-Linear
Regarding linear and non-linear operation (amplifiers, final stages): While a Class-C amplifier circuit has far higher efficiency than a linear circuit, a Class-C amplifier is not linear and is only suitable for the amplification of constant-envelope signals. Such signals include FM, FSK, MFSK, and CW (Morse code).
If Joe Taylor’s various modes (in WSJT software) are constant-envelope signals, than class-C works, right? At least, in theory.
Some Additional Cool History
The digital mode, Thor, came out of DominoEX when FEC was added. Here is an interesting history of FSQ that seems to confirm that FSQ is like MFSK, so no problem with a bit of ALC.
The following is from https://www.qsl.net/zl1bpu/MFSK/FSQweb.htm
History – Let’s review the general history of Amateur MFSK modes. The first Amateur MFSK mode developed anywhere was MFSK16, specified by Murray Greenman ZL1BPU, then first developed and coded by Nino Porcino IZ8BLY in 1999. Before MFSK16 arrived, long-distance (DX) QSOs using digital modes were very unreliable: reliant, as they were, on RTTY and later PSK31. MFSK16 changed all that, using 16 tones and strong error correction. Great for long path DX, but nobody could ever say it was easy to use, never mind slick (quick and agile)!
Over the next few years, many MFSK modes appeared, in fact too many! Most of these were aimed at improving performance on bands with QRM. Most used very strong error correction, some types a poor match for MFSK, and these were very clumsy in QSO, because of long delays.
The next major development, aimed at easy QSOs with a slick turnaround, was DominoEX, designed by Murray Greenman ZL1BPU and coded by Con Wassilieff ZL2AFP, which was released in 2009. Rather than using error correction as a brute-force approach, DominoEX was based on sound research and achieved its performance through carefully crafted modulation techniques that required no error correction. The result was a simpler, easier to tune, easily identified mode with a fast turn-around.
DominoEX is widely used and available in many software packages. A later development by Patrick F6CTE and then Dave W1HKJ added FEC to this mode (THOR) but did not add greatly to performance, and at the same time eroded the fast turn-around. The final DominoEX- related development was EXChat, a version of DominoEX designed specifically for text-message style chatting. While completely compatible with DominoEx, it operates in ‘Sentence Mode’, sending each short over when the operator presses ENTER. EXChat was developed by Con ZL2AFP and released in 2014.
Back in 2013, Con ZL2AFP developed an MFSK mode for LF and MF which used an unusual decoding method pioneered by Alberto I2PHD: a ‘syncless’ decoder, which used a voting system to decide when one tone finished and another began. The first use of this idea was in JASON (2002), which proved to be very sensitive, but very slow, partly because it was based on the ASCII alphabet. The new mode, WSQ2 (Weak Signal QSO, 2 baud) combined the syncless decoder with more tones, 33 in total, and an alphabet specially developed by Murray ZL1BPU, which could send each lower case letter (and common punctuation) in just one symbol, resulting in a very sensitive (-30 dB SNR) mode with a 5 WPM typing speed.
In the subsequent discussion in late 2014, between the developers ZL2AFP and ZL1BPU, it was realized that if the computer had enough processing power to handle it, WSQ2 could be ‘sped up’ to become a useful HF chat mode. This required a large amount of development and retuning of the software to achieve adequate speed was involved, along with much ionospheric simulator and on-air testing used to select the most appropriate parameters.
Tests proved that the idea not only worked well, but it also had marked advantages over existing HF MFSK modes, even DominoEX. As expected, the new mode was found to have superior tolerance of signal timing variation, typically caused by multi-path reception, and would also receive with no change of settings over a wide range of signaling speeds.
So this is how FSQ came about. It uses the highly efficient WSQ character alphabet, IFK+ coding, the same number of tones as WSQ (33), but runs a whole lot faster, up to 60 WPM, and uses different tone spacing. The symbol rate (signaling speed) is modest (six tones per second or less), but each individual tone transmitted carries a surprising amount of information, resulting in a high text transmission speed. And it operates in ‘Chat’ (sentence) mode, which allows the user to type as fast as possible since they type only while receiving.
The ability to send messages and commands selectively has opened a huge array of communications possibilities.
What Makes FSQ Different
Incremental Keying – FSQ uses Offset Incremental Frequency Keying (IFK+), a type of differential Multi-Frequency Shift Keying (MFSK) with properties that make it moderately drift-proof and easy to tune. IFK+ also has excellent tolerance of multi-path reception.
IFK was developed by Steve Olney VK2XV. IFK+ (with code rotation) was proposed by Murray Greenman ZL1BPU and first used in DominoEX. IFK+ prevents repeated same tones without complex coding and provides improved rejection of propagation-related inter-symbol interference. In the context of sync-less decoding, the IFK+ code rotation also prevents repeated identical tones, which could not have been detected by this method.
Efficient Alphabet – In FSQ, a relatively high typing speed at a modest baud rate comes about because the alphabet coding is very efficient. All lower case letters and the most common punctuation can be sent in just one symbol and all other characters (the total alphabet contains 104 characters) in just two symbols. (The alphabet is listed below). This is a simple example of a Varicode, where it takes less time to send the more common characters. The character rate is close to six per second (60 WPM), the same as RTTY, but at only 1/8th of the baud rate. (RTTY has only one bit of information per symbol, 7.5 symbols per character, and wastes a third of its information on synchronization, and despite this, works poorly on HF).
No Sync – Another important factor in the design of FSQ is that no synchronizing process is required to locate and decode the received characters. Lack of sync means that reception is much less influenced by propagation timing changes that affect almost all other modes since timing is quite unimportant to FSQ; it almost completely eliminates impulse noise disruption, and it also contributes to very fast acquisition of the signal (decoding reliably within one symbol of the start of reception). Fast acquisition removes the need for the addition of extra idle characters at the start of transmission, and this leads to a very slick system. Add high resistance to QRM and QRN, thanks to the low baud rate, and you have a system so robust that it does not need error correction.
See you on the bands!
7-bit Barker Code and Matched Filter
Teaching Evan the basics of radar signal processing with this baby-block 7-bit Barker code and its matched filter.
Improving 50-MHz Transmit Capability
Over the past week in the evenings, I’ve managed to cobble a little PA together for the 50-MHz transverter. It’s a near-clone of the PA in the Elecraft XV50 using the Mitsubishi RA30H0608M. Last night, I got it all hooked up and installed in the cabinet. It broke into oscillation when keyed on CW.
The oscillation was about 50.040 MHz. I reasoned that it might be the PA output coupling into the nearby TX RF bandpass filters which are followed by 12 dB of gain before returning to the PA. So, tonight, I added a litle shield between the BPF and the PA board. That seemed to clear things up and I got about 10 watts out. There is a 6-dB pad between the last driver stage and the PA, so I should be able to get it up to about 18-20 watts. But, the first goal will be to check the linearity on SSB.
In other transmit capability news, some boxes and heatsinks arrived for the W6PQL PAs for 903 and 1296 this afternoon.
Revisiting the 50-MHz Transverter
I recently resurrected the 50-MHz transverter project and have made good headway getting it working. On Friday night, I began the process of tapping holes in the PA module heat spreader. But, despite using plenty of “cutting fluid” (3-in-1 oil), I managed to break a (well-used) tap on the first hole. Since the maximum (linear) power out of the driver stage is 200 mW, I embraced my inner QRPer and put the PA project aside to give the transverter a try in the CQ WW VHF contest.
The 6-meter Yagi had come down in favor of Yagis for 222 and 432 when the loaner FT-736R showed up. So, I scampered up onto the roof and moved Yagis around. I had hoped that this moment would be accompanied by changing out RG-8 coax for LMR-600 and LDF4-50A that are taking up space in the shop and shack. But, I was not ready to commit to cutting that and I still don’t have LMR-600UF for the rotator loops. Plus, I should replace the rotator at the same time. That amounted to too much work for the available time. I really just need to bite the bullet and install a rotating mast for the VHF antennas that’s not so precarious.
Got everything hooked up late on Saturday afternoon, but had to tend to some domestic concerns and was QRT until later in the evening when I heard my neighbor W4EE calling CQ on six SSB. Did not know that he had six! Apparently, this is a new thing for him, too. He was surprised that I didn’t vibrate his radio off the desk like I usually do! Told him I was running 200 mW and everything made sense.
Ended up working a few other locals on Sunday including N3UM, who moved me from 6-meter CW to 2-meter SSB for a quick chat. He just completed the N1DPM active bias mod to his Mirage B2518G, so was eager for an audio report. Sounds good! He said my B3016G sounded good on-frequency, but I haven’t gotten the mod actually inserted into mine yet. Probably that Kenwood (TS-700S) audio making up for the amp’s inferiority…
Eventually, I will be posting more details on the circuit here. This is one of those projects that I would not encourage anyone to duplicate as I have constructed it. However, there may be useful features.
Apologies for the bad pun in the title. I was going to call it “Microwave progress” but could help myself when I thought of this one. Here are some notes from tinkering over the past few days.
After learning that the 1152-MHz LO power was -11 dBm, I inserted a MAR-3 MMIC on the 1296RSU transverter board. W1GHZ shows an MAR-6 on his board, but the ever-astute N3UM noticed that the P1dB for the MAR-6 is 1 dBm, which is well below the nominal 7 dBm level of the ADE-5 mixer. Unfortunately, the MAR-3 has about 12 dB of gain vs the MAR-6′s 20 dB. A little bit of skullin’ lead me to my stash of SGA-4586′s (suggested as an economical front end by W9SZ), which can do >20 dB gain with a P1dB point of 16 dBm or so. Perfect.
Unfortunately, with an 144-MHz IF drive of about 2 dBm, the 1296-MHz output was totally trashed with various mixing products. I spent about 20 minutes searching frantically for the 3-dB SMA attenuator I’d purchased at Dayton. But, I never found it—a sign that my organizational scheme has lost control or that I never actually bought the attenuator. Either are possible. So, I did the next logical thing—I added a Pi-network attenuator between the SGA-4586 “LO boost” MMIC and the ADE-5 mixer. If the Dremel tool didn’t make it’s inventor independently wealthy, it should have. The only SMD resistors that I had on-hand that were realistically appropriate (39.6 and 130 ohms) yielded about 6-7 dB of attenuation, so the effective gain of the SGA-4586/attenuator cascade was probably around 13-15 dB, yielding somewhere around 2-3 dBm of LO. Blech. That’s essentially the same as the MAR-3. The 1296-MHz output with 2 dBm of drive at 144 Mhz was about 2.5 dBm.
The data sheet for the ADE-5 suggests that the conversion loss increases as the LO drive level falls bellow 7 dBm (shocking!), but there are not enough data to show how precipitously it deteriorates. However, essentially 0.5 dB of overall power gain does not jibe with the amount of gain in the system, which should be closer to 20 dB. From this standpoint, perhaps the mixing products were due to IF overdrive instead of LO overdrive. In any event, about 3 dB (instead of 6 or 7 dB) of attenuation in the LO would be a good thing.
Fresh off my mixed success with the W1GHZ 1296RSU (which are neatly packaged in Hammond 1590BB-sized cast aluminum boxes), I wanted to give the 902/903 version (which is still loose boards) a try with the spectrum analyzer. So, I hooked up the 756LO board first. It made -5 dBm at 756 MHz. These numbers are more like what W1GHZ was promising. So, I put the MAR-3 mentioned above onto the 902/903 transverter board. With 2 dBm of 147-MHz drive, the output was a very clean 16 dBm on 903. Score! Must be livin’ right at least half the time. My last DigiKey order included a 1590BB for that transverter, but the 756LO board is a bit longer and I haven’t yet found a suitable case for it.
Thought I was done? Me too.
I admit it. I’m a sucker for this sort of thing. Fred, N1DPM, recently posted to VHFcontesting and the “Stanford” VHF lists that he was selling a bunch of spare microwave gear, including a first-generation DEMI 3456 transverter/LO and some amplifiers to get the output up to 4 watts. Cheap. He had some 2304 stuff, but ye old project fund is pretty much depleted since Dayton and I always try to keep a little bit in reserve for just this sort of opportunity. Anyhow, the transverter is set up for 10 watts of drive on 144 MHz. I gave it 250 mW (24 dBm) from my modified IC-290A (post to follow on this—not rocket science, just bypassed the PA) and the output came up at -9 dBm on 3456 MHz. Once I remove the attenuator from the input, I should be able to get it up to about 13 dBm. At least it seems to work on TX. Need an antenna to try RX because the K3UO beacons are not as close as W3APL.
Mirage Active Bias mod, Part 1
After chatting with Terry, W8ZN, about whether to keep or sell the AM-6155s (which I did sell to finance a tower), he suggested that I apply the N1DPM “Active Bias” modification to my elderly “160-watt” Mirage 2-meter amplifier. I contacted Fred, N1DPM, and he sent me a copy of his paper Linearization of Solid State “Brick” Amplifiers from the 21st Eastern VHF/UHF Conference (1995), along with some additional notes from his notebook. It’s pretty eye-opening how non-linear the amp is without the modification! I finally gathered the parts and hacked it together last night. Hope to test it soon with a “dummy” transistor and then live on the amp. The braided wires go to the thermal compensation transistor which is thermally bonded to one of the RF power transistors.
It probably won’t be ready for the ARRL June VHF this weekend, though. So, I’ll only be using the amp on CW. That’s no matter since I haven’t hung the low-loss cable (LMR and Heliax), nor have I received a D1010 432-MHz amp that will be on its way to me soon. Everything seems to take longer than it ought to!