A 12-bit ADC has a theoretical maximum of dynamic range like 72 dB. This would be somewhere S9+10dB to S9+20dB above the noise if you have no too much QRN around, but implementations are different. Radios today mostly have only whole ham bands wide frontend filters, also a usual antenna do not change too much on the unattenuated bandwith of signals going to the ADC. You can say that the typical radios 40 years ago had about the same narrow dynamic range, no more than this 70 dB's and we could use them everyday, but sooner or later everybody must learn that overloading an ADC is really different from overloading an analog radio even if they have the same numerical value of dynamics. ADC's have a very different IM curve from the analog stages: they maintain their really good parameters (you can't even calculate IP3) until the sum of the power of signals they are getting exceeds the maximum input power, but if you cross the line... a lot of problems begin to happen. Have you heard what an overloaded sound card abruptly produces? Problems are not limited to getting a weak station wiped out by a nearby strong one. It is enough to have one "too strong" station, e.g. over that S9+20dB power (dBm value the ADC gets) anywhere in the band to overload the ADC, or ten of a bit less strong ones, etc..., and you can't do anything against it - except to attenuate all the signals on the input. So, you need reserve... It is sure that there are conditions where the modest SDR receives stations well but stating (even aggressively!) that the big difference in parameters does not mean, and making up a virtual reality where Hermes Lite (!) 2 only differs from a higher spec. SDR in the owners' fear of their self-integrity and some meaningless numbers is pure ignorance and lack of knowledge. It is sure that there are conditions where the modest SDR receives stations well, they are not broken just limited, but we don't need such "nothing to envy" claptraps here, this is intended to be a technical forum. I hope I'm corrected too if my numbers are bad.
The Lite is a lite version of the Hermes that TAPR originally designed. This turned into the excellent Apache Labs Anan radio. I have the Hermes, Hermes Lite 2 and the Apache Labs 7000 DLE. The quality of the Hermes Lite vs the 7000 is not that much. Much of the work of the system is done on the PC and so a good PC is a requirement. BTW, I also own a Flex 6600. All are excellent radios.
The HL2's ADC IC includes a built-in and configurable front-end attenuator. Often proper use of it is enough to prevent moderately strong signals from often overloading the ADC. And, perhaps surprisingly to some, wideband RF noise plus the processing gain from the FPGA and SDR software filtering often allows seeing and hearing weak signals more than 72 dB below the 12-bit ADC's overload level. See this Analog Devices white paper on S/N improvements via processing gain: https://www.analog.com/media/en/training-seminars/tutorials/MT-001.pdf
The design of the WebSDR protocol and interface includes enough buffering to allow usage over WAN and WiFi. This buffering adds a fair amount of latency. One incurs that latency cost even when using that same interface over a wired connection. The HL2 has very little buffering in the transmit direction (10 to 20 mS in the FPGA), so almost requires a (low-latency-jitter) wired connection when used for Tx via the native UDP Protocol 1. The amount of buffering in the receive direction depends mostly on the buffering required for the DSP filtering algorithms and the computer's audio output driver and OS subsystem, although some of the SDR software applications might add increased buffering latency when used over lower priority network connections. As for envy: If you need a system with certain specification for successful operation, then a system with that specification isn't enviable, but necessary. But if you don't need that a system with that specification for your actual operating conditions, then any significant extra cost might be seen as more a form of conspicuous consumption rather than enviable. I suppose paying more for a system with higher specifications to cover the risk of a potential need in the future might be considered preparedness, up to some point. But a lot of hams might be buying more expensive rigs than they currently really need because they don't know how to operate their stations more optimally. Keeps the ham radio vendors profitable and in business I guess. The goal of the HL2 project was to produce a low cost network connectable direct sampling transceiver. I currently fail to find any other system that is enviable for meeting those goals at an even lower cost. Maybe the Radioberry project, which costs less if you can find one or build one, but that's basically an HL2 without the case, 5 Watt amplifier, filters, T/R relay, or network jack.
I agree that HL2 offers a lot of bang for the buck. I take your point about preparedness for future potential operating situations that might need the higher dynamic range but those situations are few and very specific. You would have to have a pretty strong sense that you are going to one day operate in some of the bigger contest situations to justify the thousands of dollars those rigs demand. Chances are if that's the case you would be investing a lot of money in other aspects of the station anyways so a low cost xcvr doesn't really fit anyways. The Kiwi with its 14 bits is a pretty impressive receiver in most situations. The HL2 should be all that is needed in most casual operation scenarios.
If we want to see a less simplified picture: - The number of usable bits of a 12-bits ADC, the ENOB is between 10-11 bits usually, this degrades the theoretical dynamic range value to the 6.. dB zone. - Depending on the sample rate and the bandwidth we want to observe, the decimation is possible, and it will increase the dynamic range a lot at the expense of needing much more CPU power, so power consumption will increase as well - besides narrowing down the bandwidth a lot (but we would need it to be fairly high e.g. for an effective noise blanking). - Dithering adds a lot of complexity but also helps increasing the dynamics, but AFAIK, this receiver does not have dithering. By they way, it is very hard to find the optimum of several parameters, not an easy development. - And solely the random noise may increase the DNR but it also has limitations, and the amount of changes, number of real dB's highly depends on the ADC parameters (e.g. the details of SFDR), the architecture and the goals of the actual receiver, etc. etc. So the sum of these factors will not make a similar receiver something you can say of that it does not handle only an extremely strong station in an extreme situation... It's simply not true. And again, we are not speaking of the effect of _one_ strong station but the total power of signals in a band.
Not built in, but many stations get a lot of incidental dithering for free, from their significant and wideband local RF noise floor. Test setups that don't include this wideband RF noise produce misleading dynamic range results, compare to actual use. The total RF power is rarely more than that of the one strongest signal. In practice, multiple signals, even in a crowded band, don't sum statistically often enough to affect the Rx of narrow band signals (CW or FT8, etc.) IIRC, one of the founders of Flex Radio put out a white paper (or maybe presented a video seminar) on this topic. Anyone have a reference link?
In fact this video shows many strong signals reaching S9+10dB within the passband of the receiver's front end and yet the audio shows no sign of the claimed nastiness.
That's great, but this is only the upper side of the available range. The Overload sign was flickering for a little - listening to an empty frequency between strong stations, hi. We can shift this window by changing the frontend amplification or attenuation just like in an analog radio but the weakest stations will certainly diminish if the window is smaller than the distance between the noise and the max. signal the radio can process, so you can't tell only from this example what the receiver is capable of. I did not see the whole video yet but I will.
The point of my post was that the signal of interest sounded Q5 while several adjacent signals were peaking above S9. This suggested to me that your claim about total wideband signal and noise power seen by the ADC would be a big problem for 12 bits. It didn't sound too shabby to my ears. This receiver may desense or misbehave when trying to pull a weak signal in the presence of monsters but it was not intended to be on the top of sherwoods list eh? For the entry level who aspires to higher goals like that, I suggest it is part of the learning curve of amateur radio and lofty ambitions requires a heavy wallet to get much lighter. How many newly licensed ops would be willing to drop thousands of dollars on a rig which has that capability when they will probably be content with 70dB of dynamic range for a good long while? The intended market for this radio probably doesn't have the array of TX and RX antennas and other station gear to go hunting with a little pistol among the lions anyways. Many people built DC receivers with no image rejection and no AGC and still had a lot of fun until they wanted more. I agree that people should not compare to high end equipment, but this little rig stilll has enough performance per buck to be recognised as a great achievement (just a personal opinion of course).
I don't consider the Hermes Lite 2 to be just an "entry level" rig. I consider it to be an very low cost implementation of an advanced, near state-of-the-art (given its cost) transceiver. The HL2 includes many of the same capabilities as a few rigs near the top of Sherwood's list, direct sampling of the full HF band, simultaneous multi-band receive, wide bandwidth display support, Tx adaptive pre-distortion support, low latency networking built-in, supported by multiple SDR client applications. The Tx output power, and a few other specifications, did have to be relaxed to meet its highly affordable cost goal. As an example of its dynamic range in actual use, note that the top of the list of systems reporting the most spots per minute of FT8 signals (a mix of both weak and strong ones in crowded 3 kHz narrow band slices) to pskreporter includes HL2s. In terms of weight, it has a better spec than most other multi-band transceivers: at only 440 grams, with low idle power requirements, and robust Tx finals, the HL2 quite suitable for multi-mode SOTA lightweight-carry ops (when paired with a suitable lightweight mobile device for the SDR UI).
@ N6YWU I meant entry level in terms of the purchase price. Certainly the performance has set a new standard for what is available at that price point.
What this gentleman said is golden. In a few paragraphs, he explains the major shortcomings of a "pure" SDR, a receiver without a superhet front end and its narrow IF filter which allows only the signal of interest to reach the FPGA, and to greatly attenuate the other signals that made it through the band pass filter. AFAIK, current test methods use only 2 or 3 signals into the receiver, a far cry from simulating a crowded ham band. This is a valid test method for a superhet receiver, but it makes the "pure" SDR look far better than it actually is in the real world.
Actually, the exact opposite. This testing method benefits an analog superhet or superset SDR more than a direct sampling SDR. A direct sampling SDR often works better with all the noise from several crowded hams bands present, as long as they don't constantly overload the ADC (typical for some number of users). Using only 2 or 3 pure signals for testing reduces the amount of processing gain possible when using a high sampling rate ADC to detect a narrow band signal (CW or FT8 for instance). Which is why some currently competitive contest stations do quite well using direct sampling HF transceivers (Flex Radio's). The HL2 has more modest specifications, but it wasn't designed for contest stations with large budgets, but more for higher affordability given the use of the same basic technology.
As you said, "as long as they don't constantly overload the ADC (typical for some number of users)". THAT is the point that HA3FLT made, and it is the major shortcomming of SDRs that is "typical for some number of users" (your words). A scenario where this might not apply very often, would be if someone was using a poor antenna system that doesn't deliver a lot of cumulative signal strength to the FPGA. Current test methods don't simulate crowded band conditions, maybe you can explain to me how this benefits superhets over SDRs. From what I see of Rob Sherwood's numbers, SDRs achieve their DR3 at the expense of noise floor and sensitivity compared to superhets. Also the 100kHz blocking is orders of magnitude worse for SDRs than that of superhets. Again, current test methodology makes SDRs look better than they are under real conditions when using a decent antenna system.
