7
$\begingroup$

Browsing the specifications for the IC-7300, I noticed it says the receiver system is a "Direct Sampling Superheterodyne" with an IF of 36 kHz.

I was confused by this, since I thought "direct sampling" means the received RF is amplified and filtered, but not mixed; and a superheterodyne employs a mixer. Thus a receiver can not be both direct sampling and superheterodyne. NI seems to corroborate this understanding.

So what is the IC-7300 architecture really, and what is a "Direct Sampling Superheterodyne" receiver?

$\endgroup$
2
  • $\begingroup$ One might ask, "what is ADC sampling rate?". Since the block diagram shows a preceding bandpass filter (15 of them apparently), one could imagine that the ADC consists of a high-speed sampler, followed by a slower sub-sampling ADC. However, the numbers don't quite support 36 kHz I.F. (or psudo-I.F.) The answer may be hidden within the ADC and subsequent FPGA signal processing not shown in the block diagram. $\endgroup$
    – glen_geek
    Sep 8, 2021 at 16:54
  • 1
    $\begingroup$ @glen_geek the 7300's ADC is at 124.032 Msps (deduced based on the schematic and the frequencies where aliases show up). $\endgroup$ Sep 8, 2021 at 20:09

3 Answers 3

1
$\begingroup$

The 7300 is a direct sampling receiver, and you can clearly see it in the block diagram under "Direct Sampling Receiver". Antenna -> BPF -> ADC -> digital processing -> a DAC to drive the speaker. Everyone who's taken it apart is pretty well in agreement about that. The 36kHz is either just a bogus spec, or is a "digital IF" that they use for DSP and Twin PBT filtering before finally mixing down to AF (but it seems impractically low for that). It's not a superhet by any reasonable definition.

At least somewhat related is the "USB IF OUT" option, which is a 16kHz-wide (ish) digital output at a 12kHz IF, which is also confusingly specified and disappointingly implemented, but which gives a tiny bit of insight into what the 7300's digital processing chain looks like.

$\endgroup$
3
  • $\begingroup$ You are looking only at the hardware, which is useless without the software (or firmware). The total functional radio needs to include the software, which could easily be implementing the software equivalent of a single or multi-conversion superhet. $\endgroup$
    – hotpaw2
    Sep 8, 2021 at 19:10
  • $\begingroup$ The whole point of SDR is to move some of the hardware functions into software. Just because it direct samples, doesn't mean it doesn't use superhet principles in software. $\endgroup$
    – user10489
    Sep 11, 2021 at 13:16
  • 1
    $\begingroup$ @user10489 sure, perhaps it does, but A) there's no evidence of that, and B) if it did, to say that makes a superhet receiver would just sow confusion. We have superheterodyne SDRs, as well as direct-conversion SDRs and direct-sampling SDRs, and there are practical differences between the capabilities of each one. How the digital downmixing is accomplished just falls into the "has nice DSP or not" category, which has always been more of a manufacturer-speciffic black box. $\endgroup$ Sep 11, 2021 at 19:56
1
$\begingroup$

The receive chain might have a tunable Local Oscillator, Mixer and BPF, before reducing the sampling rate further for final processing, detection etc. Isn't this a superhetrodyne receiver? The only difference is that these would be in discrete time, after the ADC.

A normal DDC would convert directly to baseband, but you could make the choice to convert to an IF first.

Even the original purpose of the superhet is maintained - the ability to have a nice tight BPF, (very difficult to make adjustable), would be carried into the digital domain. In an analog superhet it's a 10.7 MHz crystal ladder, in this one it'll be a nice tight many-tap digital filter that runs at 100Msps, can be optimised once and doesn't have to be tuned. This is a pre-requisite to decimation anyway. I've never heard of it being called a superhet in DSP though.

$\endgroup$
3
  • $\begingroup$ You mean this tunable LO, mixer, and BPF would be digitally implemented? $\endgroup$ Sep 8, 2021 at 18:29
  • $\begingroup$ IEEE definition of SDR is replacing a radios physical layer with software. That physical layer can include a heterodyne mixer to IF. So if the hardware was a superhet, the total SDR using the same set of layers (not just the ADC) can be called a superhet. $\endgroup$
    – hotpaw2
    Sep 8, 2021 at 19:06
  • $\begingroup$ Yes, I managed to edit that bit out. Hang on. $\endgroup$
    – tomnexus
    Sep 8, 2021 at 19:18
1
$\begingroup$

The 7300 is a direct sampling receiver, since it samples RF at well above twice the the band frequency of interest.

However, there is more than one signal path with more than one bandwidth and center frequency involved. Although one of the signal paths is direct sampling (likely the widest bandscope) other digital signal paths may not be as wide. Those sub-bands can have narrower bandwidths and be offset within one of the wider signal paths (by far more than a BFO tone). Since the frequency of, say, a CW signal can be significantly offset from the center of some signal path, that can make some of the intermediate signal paths into intermediate frequency or IF signal paths, since their center frequency can be different and offset (from both DC and from the CW carrier frequency, by far more than a BFO tone).

This is because a direct sampling receiver can have such a wide bandwidth, that a (usually complex) heterodyning mixer (or two or more!, often inside an FPGA) might be required to pre-process the data so that an affordable DSP or CPU can finish the SDR processing. This means the directly sampled RF result is direct conversion data, but other data paths may be heterodyned or mixed down into intermediate frequency signal paths, before being further heterodyned or mixed and filtering down to a final baseband for demodulation.

Another example: The common RTL-SDR-v3 RTL2832U chip can direct sample MF and low HF by sampling at 28 MHz (in bypass mode from using the front-end mixer such as an R820T2), but then the RTL chip mixes down and ships data over the USB port as an offset IF frequency stream with a lower sample rate of 240k or 2.4Msps (etc.). So the front-end hardware portion (the ADC) can be direct sampling. But the software on the PC only sees a narrower bandwidth IF frequency data stream (and the software possibly uses yet another internal IF frequency data path before demodulation). If an IF frequency signal path is necessarily involved, what would the total system be called other than a superhet?

So, from the hardware point of view, direct sampling. From the SDR software point of view, a superhet. In an integrated receiver or transceiver unit containing both the hardware and the SDR/DSP software, such as the IC-7300, both. (is it a desert topping?)

$\endgroup$
4
  • $\begingroup$ So you're saying the IC-7300 is two or more receivers in parallel, a direct conversion receiver to run the bandscope, and some superhet receiver(s) for tuning individual signals? $\endgroup$ Sep 8, 2021 at 18:16
  • $\begingroup$ Not in parallel. more like some portions serial and some portions parallel. e.g. you can feed the output of one receiver to another receiver. At audio frequencies, that's called audio DSP (or a TNC), at higher IF frequencies into a DSP, that's called SDR (or a subset of). $\endgroup$
    – hotpaw2
    Sep 8, 2021 at 18:26
  • $\begingroup$ In this case, it sounds like "direct sampling" has become the latest technical term to be twisted into a marketing term, like "roofing filter". $\endgroup$
    – rclocher3
    Sep 9, 2021 at 1:15
  • 1
    $\begingroup$ If the ADC is sampling RF signals at well above twice the highest frequency in the band of interest, why might that direct sampling of spectrum be considered just a marketing term? $\endgroup$
    – hotpaw2
    Sep 9, 2021 at 19:30

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .