Dave,?
I bought 5 ADE-25MH from huyai-components on Sept. 22, 2019 @ EUR 4.29 and they work fine. Tried 2 of them so far: one working in the TinySA and one in a modified SPECAN.
--
Joerg, ex-DB2OO

Hi group

I have finally got round to building the Tiny-SA this week.

My version uses B3555 filters, kindly supplied by Tony G4WIF, and I am using an ESP32 as the processor (another processor with impressive ADC specs but almost useless real-life performance).
I tested the filters with/without matching components using my NanoVNA and decided to not include the matching elements.? Without there was a sharp peak at 433.9MHz, and two other peaks at the edges of the passband.? I may be wrong but I thought the narrowing of the passband was better than having a flat top and slightly less loss.
When testing with the Arduino terminal commands the ESP32 responds as you would expect, but when driven by the TinySA.exe TinySA.exe hung after around 100 samples.? It took me some time, but this was partly resolved by setting the speed value to 90.? The scan is really slow but it works.? With faster speed settings the output in the Arduino terminal looks fine.? I could be wrong but I think there is an issue with the TinySA.exe hanging if the samples come too fast.
I didn't buy a mixer module, but bought 5 ADE25MH from a Chinese Ebay seller.? Unfortunately these do not work, but it took me some time to realise this as I was thinking it was an issue with the port to ESP32 with different pin-outs etc.? Avoid these if anyone is tempted.? I tried to have a look to see if they were wired up wrong, but they are so small it is difficult to tell.? No idea how they manage to make these in the factory!? I wonder if these have been manufactured with different pin-outs to the datasheet - it would be interesting to see a picture of the layout and pin-out of the mixers in the module you are using.
There was a ADE-1 lurking in my bits box, so with that in place and some software mods to limit the LO drive to 8dBm (I guessed there would be 1dBm loss in the antenna switch and filters) the unit started to work.
When taking readings I found it important to turn off the pulse output on GPOI2 by setting to a negative value, otherwise the 10Mhz and harmonics show clearly.
No calibration attempted yet, but here is a capture of the output measuring the signal from an AD9850 DDS.




No display yet - that should arrive towards the end of the week.? I already have an attenuator module, but different model so the code will need to be adjusted for that as well .

Great project !


73

Dave M0WID

Joerg,

You have to ensure the inductors can not "see" each other otherwise the magnetic fields will destroy the stopband. At least put consecutive inductors on 90 degrees angle?
The SMD capacitors interact with the inductor windings when mounted on top of each other. Difficult to get a bit distance....
--
HBTE Files section:?/g/HBTE/files
Erik, PD0EK

Touch interface working!
Not all functions done yet but already much nicer to use


--
HBTE Files section:?/g/HBTE/files
Erik, PD0EK

I have built a 9th order Bessel 300MHz lowpass, according to the parameters from?
I set the minimum capacity to 2.2pF and minimum inductivity to 10nH:

I have built this filter with SMD parts in "ugly construction":

and measured the S21 with a NanoVNA with these results:

The -3dB cutoff frequency is a bit lower, most probably due to parasitic capacities. Moreover the stopband attenuation is MUCH worse than calculated, which is most probably due to the construction technique: no shielding, potential issues with undefined currents on the ground plane.
To validate, that this is not a measurement deficiency of the NanoVNA, I replaced the filter with a 60dB (20dB+40dB) attenuator and got a pretty clean result:

The peak at ~300MHz is an artifact of the NanoVNA, that is switching output power of the SI5351 at 300MHz.

I will further investigate, how to improve the stopband attenuation of the 300MHz lowpass using a different construction.
--
Joerg, ex-DB2OO

As the code it multi threaded things get "out of order" in the logging.
The logging is wrong, should be {}

--
HBTE Files section:?/g/HBTE/files
Erik, PD0EK

It should be at the front.
The 350MHz filter has not been tested, its an elliptical filter and maybe too critical in component values.
--
HBTE Files section:?/g/HBTE/files
Erik, PD0EK

Erik,
You originally put the attenuator at the very front end.? Your latest schematic shows it after
the low pass filter.
Which do you reccommend?
Also, I see you redesigned the filter for 350MHz?
73,
Gary
WB6OGD

I'm pretty far along with the Jan 17 tinySA.ino file.
Will probably stick with it unless there's a significant reason to go with newer code.
Hope to eventually be using the native RPi display, not the 2.8" TFT.

Jerry

Erik,

If that is not from the Jan 17 tinySA.ino currently in the files section, could you post the code it corresponds to?
Though I may have all I need if the Jan 17 tinySA.ino file? was working well enough..?


Here's few inconsistencies between that recent trace and the tinySA.ino code in the files section"

The history you sent appears to be much less verbose than what should come out of tinySA.ino.

Now using 'b' and 'e' instead of '{' and '}' around the sweep data.

A bit of weirdness near the bottom, the "VFO:1" looks like the tinySA acknowledges a "V1" command
between the "p6 0" and the "o433700000", but that "V1" command? was apparently never sent.

Erik,
I really appreciate, that you seem to read my mind: I am eagerly looking for the code for the ILI9341 user interface. I have two of those: one with parallel interface and the other with SPI interface. Up to now, I used those only with the "blue pill" and would switch to that from the Nano, which I am currently using.?
--
Joerg, ex-DB2OO

Excellent, that's perfect!!!
Jerry

As requested, inverse order (latest on top), generated by enabling the logging in tinySA.exe
Scan of 100 points from 0 till 100MHz
Send by tinySA: "<<"
Received by tinySA: ">>

"__" means CR/LF as sometimes multiple responses on one line

--
HBTE Files section:?/g/HBTE/files
Erik, PD0EK

Erik,

I did find your list, it was a big help in getting a handle on tinySA.ino
Have you verified that the current MSWindows package does not use any commands other than listed?

Better yet, could you capture the commands as they are executed from a complete session
from initialization through a sweep, perhaps from 1 to 100 mhz?
With that I might be able to run this thing on an RPi with real hardware,
and then plot the resulting data using pyplot.

Yes, the code looks like it has grown organically over the years.
Similar to the debuggers I have whacked together over the years when doing hardware design.
Exactly the same style as many of my hardware debuggers, straight up K&R with zero attempt
at being trendy with object oriented code and ridiculous levels of indirection using #defines.
I find it more readable than much of the code found on the web for this sort of thing,
will prove useful if all I do is make a radio collar for the cat.

I'm removing any cruft that gets in my way of figuring out the tinySA
without major changes to style or naming conventions.
Though I seldom prefer working with somebody else's idea of how my code should look,
let me know if interested in seeing it once I get it to compile.

Jerry, KE7ER

开云体育

Yes,

The tinySA.ino is derived from my generic VNA/SA sweeper that includes the commands for various tracking generators.
Inside the build instructions (see file area) there is a list of used commands with some explanation
--
HBTE Files section:?/g/HBTE/files
Erik, PD0EK

Looks good!
That 2.8" screen makes it a good complement to a nanoVNA.
And the nanoVNA is certainly popular.
Though the the newer nanoVNA versions are going to a bigger screen.

I'd prefer to have both the nanoVNA and the tinySA hooked up to an RPi via SPI.
No MSWindows host, no tinySA Arduino, no serial protocol between the two.
Screen could be anything from your 2.8" to a 1080p hdmi monitor.

For now, I'm trying figure out the tinySA.ino stuff.
My assumption is that the VFO variable gets set to a 1 using a "V 1" command
(so points to the LO Si4432) before the "M" command is issued to make a sweep.
And that extraVFO and extraVFO2 are never used.
Is that correct?

Jerry, KE7ER

The first display was really tiny, the TFT 2.8" ILI9341 with touch is still cheap but more eye friendly and not too difficult to interface.
With the touch I can drop the rotary and have a clean setup. This all will fit nicely in a small box.
Would this be a useful option for the tinySA?



--
HBTE Files section:?/g/HBTE/files
Erik, PD0EK

Don't know a lot about the Si4432, but on the Si5351 these can improve spurs and noise:

Shut down the other two outputs or there will be crosstalk.
Load the single remaining output lightly, capacitively coupled into perhaps 200 ohms.? ?Not 50 ohms.
Bring the frequency down to something more like 50 or 100 mhz,? At least down within the 225mhz spec.
A clean power supply.
Use an integer output divider, a fractional pll divide if you must? (though not as much of a factor as you might think)

Spurs at 40dB down strike me as excessive.
I would want to know a lot more about the test setup (and understand the parts better) before coming to any conclusions.
Regarding the Si4432, the tinySA will still be very useful even if there is some cruft within 100khz of the fundamental.


On another note:
I've never played with the RPi much.??
However, poking around some, I see that the 435mbyte Raspian Buster Lite image (based on Debian linux)
comes with gcc 8.3 and python 3.7.2.? ?I believe it also includes the wiringPi C library to access the header pins.
? ??
Accessing an SPI device on the header pins from a C program is quite easy:
? ??

Jerry, KE7ER