New KiwiSDR time to think about the front end
Hi Guys
I'm new to this forum but long time user of a private KiwiSDR set up by the MW Circle in the UK. I'm a huge fan of the collaborative effort behind the KiwiSDR network (and of course kudos to jks) and i really appreciate the TDoA facility..
I'm not the first to bemoan the fact that perhaps half of the publicly accessible Kiwis are performing very poorly from an RF point of view. But what can be done? Not everyone is au fait with the word of microvolts and differential/common mode signals. So how can a receiver like the Kiwi be made as close to plug and play at the RF socket at it is at the data socket? It's sad to see the investment made in receivers that are either deaf or swamped by external electrical noise and digital hash.
Perhaps if a second generation Kiwi is born it would be time to think about the optimum RF front end? So that's why I've started this thread for discussion.
I was triggered by the comment regarding the possible inclusion of a balun on the main PCB . The balun is a mixed blessing as a concept. On one hand it breaks the galvanic connection. But it introduces a frequency limit that wasn't there before. Finding a cheap small balun effective from below 100kHz to above 30Mhz is likely an non-starter.
Many live KiwSDRs could benefit from have a notch filter to remove a high power local "pest". Others would benefit from decent management of earth loops and so on.
I've a few ideas on what a new front end might look like but I'm sure that I can't be the only one interested in this topic?
What do you think?
rgds
Steve
Comments
A discussion is fine. But it's extremely important to understand that "KiwiSDR 2" has very strict limits with what can be done and still meet the "prime directive" of getting back into production as quickly and efficiently as possible. Anything beyond that will be considered in the future.
It's essential to read the KiwiSDR 2 thread carefully: https://forum.kiwisdr.com/index.php?p=/discussion/2955/kiwisdr-2/p1
In particular the PCB dimensions cannot be increased. So any changes have to fit the existing space.
The email advisory group has considered the question of the front end and input balun carefully. Some measurements have been done, but other measurement await delivery of the prototype boards.
Although I'm reluctant to do so, I'll go ahead and post the current front end design below. The DC blocking caps (100V) and 30 MHz LPF are the same as Kiwi-1. The bleed resistors are in 0805 packages. There was only room for a single GDT, so it went across the input SMA. The TVS package size was increased to the largest I could find (0603) to facilitate easier replacement. Everything after the
RFOUTof the digital attenuator is the same, i.e. single-ended input ADC preamp.The balun dual footprint supports both the MiniCircuits T1-1T (better LF response) and ADT1-1 (smaller package size). We willl most likely use the T1-1T unless there are DFM problems using that package.
Hello, @jks! Many use an active antenna powered by a coaxial cable. I use a single 5 and 12 volt power supply with a common ground for this. In this case, it turns out that the SMA housing of the KiwiSDR connector will be connected to the housing of the KiwiSDR itself through a piece of coaxial cable. I don't really understand this, I think it's called "common-mode currents". Won't it cause problems?
The inclusion of a balun is an attempt to resolve some of the more common issues that have been encountered with KiWi installations, that are associated with antenna and KiWi ground paths.
The balun will not significantly affect the KiWi's performance or frequency range. Several of us have already done pretty much the same with external baluns of a similar type and have not observed any issues.
The problem with widespread poor KiWi RF reception, is primarily due to poor antennas and local interference. Given a decent antenna and low noise location it can perform extremely well, and there is little that could be done to the KiWi hardware that would improve the overall receive performance any further.
The main issue would seem to be that many KiWI owners are not 'traditional' amateurs or short wave listeners, but are typically software / computer enthusiasts , with some interest in radio. As a result they are less familiar with antennas and noise reduction and often either just throw out a random length of wire, use an E-Probe (PA0RDT) or cheap loop such as the MLA-30, all of which tend to be 'noise magnets'.
I have helped some KiWi owners to improve their reception when I thought it was worthwhile, either because they had a useful location for TDoA or some other factor. Very often the inclusion of a simple notch filter or balun has made an enormous difference at very little cost, and on occasion I have provided them free.
What is really required is some additional guidance for folks like this, and I'm thinking of writing something to try and improve the situation.
In the meantime, if you have a particular KiWi in mind that is performing poorly, why not message the owner and see if you can help them to improve reception. I have found that folks are generally quite receptive and appreciate your interest in their KiWi, and it helps to develop a further sense of 'community' among KiWi owners.
Regards,
Martin
Martin, it would be very interesting to know your opinion about my KiwiSDR. However, it is only available for my country's IP addresses.
To be honest it's quite difficult to tell from the SNR scores alone, however the day / night variation is a good indicator especially on the upper HF bands.
This is my equivalent plot to yours. I've used Black Cat Systems KiWi SDR monitor to produce the graph. The most useful plots are probably of 0-1800kHz and 20000 to 30000kHz as these tend to show up problems with interference (very little variation between day and night) more than the others.
I'm typically seeing 15 to 20dB day / night variation on http://wessex.zapto.org:8073 which is pretty similar you yours, and I would say yours is pretty good, especially if you are using an active loop.
For comparison here's my home KiWi SNR graph. As you can see there is a lot of interference in the 20 to 30MHz spectrum, so there is minimal SNR variation in the light blue trace.
Anyway, is that you at number one of the list just above me at number 2 ? How dare you :-)
Can you post a daytime screen grab of your waterfall and spectrum display ?
Ideally zoomed fully out with manual waterfall scaling Max -30dB / Min -110dB / waterfall no averaging / spectrum EMA avgs = 4 These are the settings I normally use for comparison purposes between KiWi's.
Regards,
Martin
Here is a screenshot. It's 13:40 local time.
Hi John thanks for sharing the front end schematic.
The only bit that causes me a bit of concern is the two 100k resistors linking the "earth" on the coax feedline to the "earth" on the pcb. I think the resistors and stray capacitance will bridge the hopefully "ideal" primary-secondary isolation of the transformer.
One doesn't know how the user of the kiwiSDR will cable their installation and what ground loops might exist. At one extreme the kiwi might be completely isolated from local RF earth apart from various stray capacitances. But at the other the user might have deliberately bonded all earth points together with a low impedance ground plane.
Some Kiwi users will install the receiver in a enclosed screened box hoping to keep interference out, but this may create new problems due to the way the metal is "earthed" - akin to the famous "pin 1" problem affecting audio engineers.
It's a personal choice, but I always prefer to have the flexibility to change the earthing connectivity when dealing with receiver inputs. Using short jumper connections I like to be able to tie "pseudo earth" points together or isolate them so that a set-up can be optimised to reduce extraneous noise and signal ingress mostly caused by common mode signals.
With the two 100k resistors hardwired on the pcb the only way to achieve complete input isolation would be to attack the pcb with a soldering iron.
I like the idea of including some transient protection on the input but it could lead to a false sense of security. First a surge could damage the on board protective element so rendering it useless against future events. Secondly a voltage surge induced on the outer screen of the coax feedline could appear across the 100k resistors at the input and damage them. I have seen 2 watt solid resistors burnt out when used to terminate long MW Beverage antennas despite the use of neon bulb protection. The cause was probably not lightning but long lasting snow static build up. If I wanted to give my Kiwi the best chance of survival I'd add external gas discharge and DC paths from the inner and outer of the input coax feedline directly to a local safety ground point. This would depend on local configuration and whether coax was carrying DC power to the antenna/preamp so should always be external to receiver.
Lastly, the transformer is really not operating as a balun. Both sides are unbalanced but isolated. Does anyone actually want to connect a truly balanced input to their KiwiSDR. Perhaps using balanced feedline rather than coax?
73
Steve
Hi Steve,
Yes you're correct. I should not have used the term "balun" for the input transformer. For some time the advisory group was considering a balanced structure from the first transformer into a second 1:4 transformer feeding the ADC preamp in a balanced manner. Instead of the single-ended DC input it shares with Kiwi-1. But there was hardly room for two transformers, let alone one, and great concern for LF/VLF frequency response (which I personally care about). This issue will be revisited in the future.
You're other points are all valid. It was felt though that something had to be done given the pattern of failures we've seen over the years. It seems as though some Kiwi owners don't take their local conditions into account (T-storm possibility, co-located transmitter signals) at the antenna interface. And improving the level of protection didn't have much downside.
The parasitic capacitance across an 0805 SMD resistor is maybe 0.05-1 pF according to the literature I found. About the same as a PCB pad. And I selected TVS diodes with a Co of 0.05 pF. And we haven't even talked about how big the oversized isolation hole should be around the antenna SMA in the enclosure. I think these questions are much more of an issue at 1 GHz than 30 MHz.
Yes, adding some flexibility via configuration jumpers and a separate antenna ground post rather than dumping the drain resistors and TVS diodes into the system ground would be desirable. But these are incompatible with the KiwiSDR 2 prime directive. They will have to wait for a future design with a larger PCB.
Hi John
re the transient suppression TVS diodes
I'm not sure which device you've selected but does 5VW refer to a 5V reverse stand-off voltage?
If so isn't Z401 at risk of damage from DC inserted into the antenna socket? There have many instances where people have wired bias-T units incorrectly and fed DC into the receiver rather than up to the antenna.
As a rough assumption this erroneous DC is most likely 12v but I guess it could be more and could be either polarity.
I can see you've wisely put 100V ratings on C405/406.
Returning to the question of stray capacitance you are of course correct regarding pads/SMD resistors, however I think some TVS diodes have several hundred pF of capacitance which could affect performance above 20MHz. Certainly Z402 could compromise filter cap C1
regards
Steve
Steve_MWC asks:
"Lastly, the transformer is really not operating as a balun. Both sides are unbalanced but isolated. Does anyone actually want to connect a truly balanced input to their KiwiSDR. Perhaps using balanced feedline rather than coax?"
I very much DO want to use balanced input to my Kiwis and to other receivers. This is extremely useful in determing whether unwanted ingress is from the antenna or from CM due to currents on the feedline and/or poor conversion (asymmetry) at the antenna.
If/when one only has single-ended inputs, shorting the perceived "antenna" shorts both common mode as well as differential signals. OTOH, if one has a balanced system, the antenna can be shorted leaving only the CM portion. IF what remains isn't at least ~10 dB lower it may be dominant in setting a limiting system noise floor. One can then discover that "all that digital noise coming in the antenna nowadays" is ACTUALLY something that can be cured.
From my measurement over the years, using my own data along with the WSPR databases, it's quite clear that the vast majority of amateur HF stations are not limited by propagated, DX noise. Particularly at longer wavelengths this is NOT inverse-square propagated energy but CM ingress or near-field sourced. These are issues that, with effort, can be removed or at least mitigated resulting in greatly improved receive performance rivaling the small percentage of truly good stations/sites. This is really good news for hams, I think, but it takes effort to discover and cure the problems which so often aren't even recognized. Differential/balanced systems make this insight possible.
In the schematic "5VW" means a "working" voltage of 5V. The "continuous operating voltage" in the data sheet. I picked this TVS specifically for its low Co of 0.05 pF. The clamp voltage (Vc) is 25V. There is no "reverse standoff" voltage spec stated.
Would it survive 12-15V continuous DC applied at the antenna input? I don't know. We've certainly had cases over the years where the TVS appeared to short, for whatever reason. And the advice to "knock the device off the board with a small soldering" iron restored operation (Kiwi-1 used a 3.3WV TVS in a tiny SOD-323 package).