Second build has shipped. Message will appear here when store is ready for third build ordering.

KiwiSDR 2

jksjks
edited December 2023 in KiwiSDR Discussion

[Please see: latest production status info]

I don't like to announce something before it's ready. But the "cat is out of the bag" so to speak.

So please consider this a pre-announcement. There are no guarantees.

KiwiSDR 2 prototypes are now under construction. The problem is I don't know WHO is going to manufacture this or HOW. So I don't know WHEN it might be available. And most importantly I don't know what it might COST.

If you have specific ideas about these questions please email me (support@kiwisdr.com). I'm especially interested in full-service manufacturing and distribution outside of China.


KiwiSDR 2 Goals:

Minimal changes. Fastest time-to-market with lowest possible risk. BUT since the PCB is going to be re-spun fix some of the known limitations that don't add too much risk:

  • New RF front-end:
    •   Balanced input via balun transformer
    •   Digital attenuator (per the advisory group: pSemi PE4312, 0 - 31.5 dB, 0.5 dB steps)
    •   Gas discharge tube (GDT) across input in addition to TVS diodes
    •   Static drain resistors (100K) from input connections to ground
  • External ADC clock brought out on third SMA connector
    •   Self test loopback mode using a short cable between this SMA and antenna input
  • New GPS chip to replace current one which is now EOL
  • Reverse polarity protection (via P-FET) on 5V DC input
    •   TVS diode across 5V input

I have an email advisory group. Many, many other changes and additions were considered and deferred. They have been put onto a list for the future.

Additional info as things progress.

Thank you.



«134

Comments

  • Really Nice!! Hopefully it will be built soon.

  • edited August 2023

    Good to hear!

    Which input transformer do you use?

    According to the data sheet, the PE4312 has a lower frequency limit of 1 MHz. What are the implications for frequencies below 1 MHz?

    Maybe I read it over, but I can't find anything about it in the data sheet. Except maybe low frequency spurs from the internal negative voltage generator.

  • Wow! Although I'm currently not interested in another Kiwi (as long as my existing one is working as perfectly as over the last years), I'm excited that this fantastic project seems to get a new future perspective. Thanks, John!!!

  • This is great news, John!

    I sent you an email regarding a manufacturer possibility.

  • There has been extensive discussion and even some measurements by the advisory group about both the digital attenuator and input transformer (MiniCircuits ADT1-1+)

    There are cost and space tradeoffs. The attenuator used is the same as in the OpenHPSDR Hermes (they use the MiniCircuits version, not pSemi) and similar to that used in the HiQSDR.

  • This is great to hear!

    I'll be watching this progress closely. Hopefully everything goes as planned.

    I'd potentially be in the market for another Kiwi.

  • Does anybody have insertion loss measurements for the ADT1-1+ at VLF/LF range (10-150 kHz), the specs in the datasheet are only for above 150 kHz and it seems to be increasing towards the lower end.

  • jksjks
    edited August 2023

    Yes, Glenn has done some measurements and I should be receiving parts tomorrow that I can measure.

    I would have preferred to use the T1-1T-KK81+ which I've used successfully at 10 kHz with an active antenna. It's spec'd at 2.8 dB loss at 50 kHz but there is no room for it on the board.

  • OK, I just found out about this in the AM Stereo discussion Group on Facebook. Is this basically just a more modern remake of the existing Kiwi? Or will it have increased capabilities, like more wideband receivers, etc? :)

  • Minimal changes. Fastest time-to-market with lowest possible risk. BUT since the PCB is going to be re-spun fix some of the known limitations that don't add too much risk: (short list follows)

    So "modern remake". Anything major (more channels via larger FPGA/host, extended frequency coverage, etc) is something for the future.

  • ADT-1-6 has 0.08 dB at 50 kHz. I think it is better suited for the kiwi than the ADT-1-1 and it has the same footprint.

    I think there was ambiguity whether they are galvanically isolated, but mine are. (And from the datasheet they are as well)

  • John, are you planning to use again a Beaglebone for this Kiwi SDR ?

  • Part of the discussion regarding the choice of input transformer is its performance WRT common mode rejection, which is pretty much defined by its interwinding capacitance (the lower the better).

    Don't become fixated by the insertion loss at very low frequencies, the noise floor is generally pretty high on those frequencies and it's the Signal to Noise ratio that is important rather than actual signal level, or overall gain flatness from 0 to 30MHz.

    Regards,

    Martin

  • I use every tenth of a dB in that frequency range, that's why I mention it.

    But agreed, CM rejection isn't as good, particularly on the higher frequencies. Here's a rough measurement of my FTB-6-1


  • edited August 2023

    ❓️ Also, will KiwiSDR2 fit in the existing aluminum box ?

    73 Costas SV1XV

  • I don't have any technical support ability, but I would definitely be the first batch of purchase participants. Btw, will the frequency range of the 2nd version change (FM/PSB/UHF) in your plan?

  • Although the specs don't indicate this It appears that the Mini-Circuits '6' series transformers are NOT simply flux coupled. They would seem to use a higher permeability core material to achieve increased low end and twisted pair transmission line to get high frequency operation. This achieves wider toatl bandwidth but at a severe increase in inter-winding capacitance. For reducing CM I think they are likely a non-starter.

    Fortunately even if LF loss is increased slightly, this is a region of spectrum where propagated noise is the highest, maximum T-storm energy worldwide, and only the most inefficient of antennas should have difficulty.

  • edited August 2023

    This is good news.

    Not sure if it was an oversight, but will there be a selctable AM BCB filter?

    This was always a problem on version 1.

    Is there possibly the abiltiy to still view the full water fall/ spectrum when used in 14RX mode if not using the BB to decode?

  • edited August 2023

    Yes, field strengths are high on VLF/LF, but most broadband antennas are either severely mismatched or have low gain in this frequency range.

    For example the noise floor on my 50m endfed wire antenna is just -150 dBm/Hz at 50 Ohm on the 136 kHz ham band, after the clifton 23 dB LNA.

    So if the (T1-1) transformer could be bypassed with a jumper, that would be good. But maybe VLF/LF is a niche anyway these days.

  • jksjks
    edited September 2023

    Are you planning to use again a Beaglebone for this Kiwi SDR ?

    Yes. The current plan is to sell bundled with a BBB and (possibly) a newly-designed enclosure. But also (as before) offer the board-only so you could purchase a more powerful BBAI or BBAI-64 separately yourself. Also, BeagleV might be a possibility in the future.

    For RPi a completely different board layout is needed and there is no time for that. Can you even purchase RPi's these days?

    Will KiwiSDR2 fit in the existing aluminum box ?

    Yes. Same dimensions.

    Will the frequency range of the 2nd version change (FM/PSB/UHF) in your plan?

    No. KiwiSDR 2 has no major changes so it can get back into production as quickly as possible. Significant changes will be left for a future device.

    Not sure if it was an oversight, but will there be a selctable AM BCB filter?

    No. There was no room on the board for such a thing. Use the new digital attenuator or solve the problem externally with a BCB-only filter/attenuator. This problem will be more properly addressed in the future.

    There is a larger problem of how you would switch in this attenuation (the BCB-DX crowd would demand that). Using solid state switches (e.g. pSemi PE42421) is about the only answer because of the impossibility of using relays (no space, high cost). But then people will be screaming about the sentence in the data sheet that says: "Note 1. Device linearity will begin to degrade below 10 MHz".

  • jksjks
    edited August 2023

    So if the (ADT1-1) transformer could be bypassed with a jumper, that would be good.

    No one is a bigger fan of LF/VLF reception than I am. So let me see about adding some DNP pads to bypass the balun.

    Also, today I am doing the first response measurements on the ADT1-1 part. I will report back the results. I tested a TT1-6-KK81+ yesterday and it's almost indistinguishable from a direct connection over the 15 to 300 kHz band. But that part is spec'd for -3 dB @ 4 kHz, so no surprise I guess.

  • It might be worth emphasizing that all electrically short/small antennas have the same gain - about .4 dB less than that of a dipole. They have the same pattern and aperture so this must be so. The problem is that the radiation resistance falls as electrical length^2. Thus the available Voltage can get very small at the same time that the Q goes out of sight.

    It is for this reason that probe rather than matched rx antennas need to be used at LF and below. Short antennas need to be either dipoles/whips into very high-Z or else loops into very low Z preamps. Trying to use even what might seem to be a large/long antenna at 25 kHz doesn't get around this. Automotive antennas can be small only because they are probes rather than matched devices. Even a 1-2m whip is essentially unusable below HF if it is directly driving a 50 ohm input. Mismatch loss is huge.

    Even so, ITU R.372 shows typical 'quiet rural' noise at 50 kHz to be ~100 dB above KTB, so around -74 dBm/Hz. One can sometimes 'get away' with not using an active preamp as a suitable probe follower but I wouldn't recommend it.

    Trying to fix this issue in the Kiwi seems to me to be misquided. Even so, if the noise comes up from a ~-154 dBm/Hz no-antenna sort of typical Kiwi number, even a few dB, the degradation to SNR is fairly small.

  • Well, a mixed bag for the ADT1-1+

    -3 dB at 30 kHz which is much better than the 150 kHz spec. But the drop-off is pretty severe below that:

    Flipped over you can see it's a binocular core. That makes it bifilar wound TLT, right?



  • The binocularism doesn't make it so but the twisted/paired primary/secondary do. No doubt the mu of the core material is pretty high but perhaps not as high as the -6 version. Probably they've reduced the total length of the winding(s) by changing core but not as much as on the -6's. This is probably why it shows higher inter-winding C than the T1-1 but not as much as the -6.

    As I wrote elsewhere, I don't feel that even the drop at 10 kHz is prohibitive. ITU says

    which shows propagated noise floor at 'quiet site' so far above KTB that anyone serious about ELF/LF will not be trying to match to the radiation resistance of their antenna but will be using an active element to create a probe. This will certainly push propagated noise above the Kiwi's floor.

    But that's just my opinion.

  • jksjks
    edited August 2023

    I looked at the mechanical aspects again. And I think I can actually fit a KK81 package inside the Kiwi's RF shield. So using a T1-1T-KK81+ is a possibility.

    When I looked at this some years earlier the issue was the package height versus the vertical clearance of the shield cover. There's about 0.4mm to spare according to the shield specs. But I tried placing one inside the existing Kiwi shield, put the shield cover on, and it seemed to have acceptable clearance.

    So let me do a layout that supports both the KK81 and CD542 pads (for the ADT1-1+). After the prototypes are built, and we do some testing with real signals, we can decide which one to choose.

  • That's great news! For some time I was looking for the first version, but it is not available anymore.

    73, Albert PA3GUK

  • edited August 2023

    For those of us utilizing the BBAI, BBAI64 or future BB, cooling will likely be an issue.

    Since the updated board is still in progress, would it be of consideration to add a pair of 5v output

    JST style 2 pin connectors for fans?

    Currently I solder the fan power leads to the bottom side of the board, under the barrel connector.

    Doesn't make for easy replacement and it can be improved upon.


    I currently have double fans on 7 Kiwi/BBAI. Configuration is as seen below.

    Fans being utilized draw .38w each. Keeps BBAI below 45c in a 24c ambient environment.


  • jksjks
    edited August 2023

    No, something like that violates the low-risk principle. And there's no space for them. Besides, you have better alternatives.

    In the case of BBB/AI/AI-64 you can connect to the 5V (outer two pins) of the USB-A host connector. In addition to USB-A the BBG also has 3.3V on the Grove connectors.

  • Since I am still powering my KIWI via the barrel connector, I had purposely placed the fans connection before that CM filter on the KIWI due to the current rating concerns that had come up earlier this year; about that part not being suitable to supply the current needed for the BBAI.

    So, if use the USB-A, then I am pulling more current through that already questionable part.

    I guess I could re-do power distribution and power via the BBAI USB-C.

    Glenn, that does bring up a good point: is the CM filter on the KIWI being reevaluated with a beefier component?

  • Some time ago we changed the recommendation that BBAI not use the Kiwi's DC input jack, but instead use the BBAI USB-C as you mention. Kiwi + BBAI current through the Kiwi CMC was much too high and there were some reported failures.

    On KiwiSDR 2 the CMC has been removed entirely. It never had enough inductance to have any beneficial effect. And when more powerful Beagles came along it became a liability.

    The entire DC input situation will have to be revisited in the future. What's really needed are large filtering components and a quiet, shielded SMPS buck/boost solution that accepts a wide range voltage input. Mainly so we don't have so many brown-out problems that first time users currently experience (i.e. DC cable Vdrop).

Sign In or Register to comment.