ad: HamInsurance-1

A technical point about analog Vs SDR radios

Discussion in 'General Technical Questions and Answers' started by KC6ETE, Nov 24, 2021.

Tags:
ad: L-HROutlet
ad: l-rl
ad: Left-2
ad: Left-3
ad: L-Geochron
ad: MessiPaoloni-1
ad: HRDLLC-2
ad: L-MFJ
ad: abrind-2
  1. KC6ETE

    KC6ETE Ham Member QRZ Page

    For my purposes, the peak to peak level of a recovered baseband signal for an input signal of a given level.
    The SA636 RSSI I only used as an example, having more to do with the variability of the RSSI chip to chip under identical conditions, than how it's derived.
     
  2. KC6ETE

    KC6ETE Ham Member QRZ Page

     
  3. KC6ETE

    KC6ETE Ham Member QRZ Page

    Working definition: For a given mid level signal, the peak-peak output voltage of a Tayloe detector.

    If I supply a mid-level signal to a Tayloe detector, ignoring the effects of any components before or after the detector, how consistent would the P-P output voltage be if I built say 100 or 1000 units?

    Looking at it more pragmatically, assume high quality opamps on both ends, and 1% or 0.1% resistors. The detector capacitors can be matched better than 1% without too much trouble.
     
  4. N5HXR

    N5HXR Ham Member QRZ Page

    I think you'll still see measurable variation. You can control matching on your sampling caps, and you can be very precise on whatever filtering goes in front. But you still have the switch to contend with. I've built a few with the FST3253 and SN74CBT3253, for example, which will exhibit variation in on-resistance and switch timing. I suppose you could develop some process for matching the switch chips :).

    If you read Dan Tayloe's articles, you'll see that his suggestion for maximum sensitivity is to design the negative feedback on the first amplifier stage after the switch over the system impedance of the input (you can see this in this article). This avoids any resistive elements contributing their thermal noise in the signal path. But it also means that the gain of that first stage is particularly dependent on any front-end elements. Even small variations in that input system impedance will produce variations in level at that first stage, as they will be amplified by the gain of the first amp (!).

    Even Dan Tayloe added inline resistors to stabilize the gain in the NorCal 2030, and most Tayloe detectors that I've seen do the same. I've read of some other designs that terminate the input differently and use instrumentation amplifiers that are supposed to produce more independence from the input impedance. I haven't built the different designs and played with them myself, just read about them.

    Maybe that's an unimportant digression... Anyway, you can't avoid the fact that from one device to the next, and from one frequency to another, different parasitics in a filter element, etc., there is going to be variation, etc. It doesn't matter from a sensitivity perspective... but still...

    If you want a reliable "RSSI" value, I think you'd get a lot more benefit from an adjustable element that you can calibrate than trying to whack all of the variable moles in all the parts to make it somehow inherently repeatable.
     
    SWL37632 likes this.
  5. SWL37632

    SWL37632 QRZ Member

    Sounds like you are attempting to ask and/or determine the statistical tolerance performance of a Tayloe detector.

    Is this correct?

    Bill

    20+ WPM 1970's Extra
     
  6. SWL37632

    SWL37632 QRZ Member

    N5HXR....looks like we 'doubled' posts.

    To add to your cogent response to the OP, typically, any 'signal level' measurements (tolerances), whether, dBm, dBW, uV, 'S' units, RSSI, Volts per meter, et. al., are 'calibrated'/'adjusted' in the SDR software/GUI to whatever reference level that is selected.

    Hardware tolerances, whether it's 1%, .01% or 50% are easily compensated for.

    Bill
     
  7. KC6ETE

    KC6ETE Ham Member QRZ Page

    That might be another way to put it.
     
  8. SP4IT

    SP4IT Ham Member QRZ Page

    May I ask what is the use case and why does it matter? 3 to 1 variation is a bit of under 5db. If your use case is to use multiple receivers at the same time perhaps you can calibrate the whole system by sweeping through the frequencies of use with a signal of known amplitude. Then you set the gains accordingly so all receivers get similar results.

    If you want to design a single receiver with very predictable characteristics I would use a calibration method too.
     
  9. KC6ETE

    KC6ETE Ham Member QRZ Page

    5dB of error matters to me. I know how to do it with calibration, but I'd prefer not to have to measure every device. I'm wondering how much of an improvement I should expect.
     

Share This Page