Jerry,

Only having a 120kHz resolution filter will make certain SA applications impossible.
Next of being able to resolve two closely spaced signals a 3kHz resolution filter also gives substantial less noise.
Further the AD8310 has only 95dB (at best) range, in reality you will lose about 5dB due to noise ingress as the AD8310 is a extremely wideband logarithmic detector
You have to study the maximum level acceptable for the SAW filter (+5dBm?) and the input range of the AD8310 at 400MHz (-95dBm till -5dBm?) and the loss in the mixer (8dB?)
You may find you will need some amplification to make everything match which may introduce additional noise.
Everything gets more complex and soon you will be building something like my 2GHz SA. Which is perfectly OK but it is good to have clear objectives from the start:
Either the most simple, cheap acceptable performance SA or go for a best obtainable performance and much more complex design (in terms of matching all the parts)
Suggest you download ADIsymRF from here:?
and do the signal path calculation.


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

There are other LO's which can move much faster than the Si4432.
Would be tough to find something at that price.

No idea how many ADC samples from the AD8310 we need for good results.?
But given my previous assumptions, 100 samples is still plenty fast.
Having the very fast AD8310 and a 1mhz ADC would allow reasonably fast scans
with the 120khz bandwidth filter.

Regardless, I have plenty of time.
And I like the looks of the B3790 SAW filter (B3780 in post #609 was incorrect).

Jerry

SA.exe scans all COM ports and tries to open each port and if open succeeds it gives the "v0" command and check if the response starts with "VFO". Only then will the COM port be shown.
You can use the monitor function of the Arduino IDE or any other terminal emulation SW to check.
Open the monitor on the correct COM port and give the "h" command
This should list all available commands in the HYPERVFO SW
Then give the "v0" command and check if the reply starts with "VFO"
--
HBTE Files section:?/g/HBTE/files
Erik, PD0EK

I could not resist to make the tinySA with its tiny display operational.
All local control goes through one rotary encoder with push button.
Local settable are from (MHz), to (MHz), max display (dB), min display (dB) and attenuator (dB). Resolution width is automatic.
First a picture of a 150MHz wide scan to see if there is any signal. Vertical scale is from -100 till -20 dB



Next is after zooming in to 5MHz till 10MHz range



The reduction in noise level due to the reduce resolution bandwidth is clearly visible.
Performance is almost identical as when using with PC control.
Standalone operation is nice when you quickly want to check for presence of certain signals or harmonics
For serious measuring the PC application is a must

Arduino source in the files area is updated to this latest version. It has not been tested if the PC application still works.....
--
HBTE Files section:?/g/HBTE/files
Erik, PD0EK

Hello Erik,
I tried to connect with the SA.EXE program from the FILES section to an Arduino running your HyperVFO software, which I pulled from a different place.
Unfortunately the SA.EXE does not provide any choices for the COM-Port, but only the MOCKUP option!?
In an earlier version of the SPECAN software, one could select the COM-Port and using that, I CAN connect to the Arduino HyperVFO, but that old version has very limited functionality.

What do I need to do, to make the the SA.EXE recognize the Arduino on my COM8: COM-Port?
Any chance, to get a "Connect" dialog with a list of COM-Ports?
--
Joerg, ex-DB2OO

Scanning 150MHz with 120kHz bandwidth requires 1200 measurements.
Each measurement needs a new frequency of the LO SI4432 PLL. This could take up to 1 ms to set and settle.
Then you have to average the signal strength to get rid of noise. Possibly 2-5 ms.
So a full scan may take 2 to 7 seconds
Scanning 400MHz will be even worse

But if you are not in a hurry this is no problem.
--
HBTE Files section:?/g/HBTE/files
Erik, PD0EK

Could be that delay through the SAW filter limits our scan rate.
Must wait that delay time after a change to the LO before sampling the 415mhz RF.
But I'm betting we are typically limited by how fast we can move the LO, not SAW filter prop delay.

The AD8310 should settle out in under a microsecond or so.
And should be possible to take 16 bit ADC samples from the AD8310 once very microsecond.
Advancing the LO by 100khz every microsecond, it would take 2500 samples to cover 0 to 250mhz.
So 2.5ms per scan, which seems about 100x faster than I need.

Do I have that right?

Jerry, KE7ER

Will not help when the saw filter bandwidth is the limiting factor.
For a wide scan you can opt to remove the saw filter as the mirrors will not be visible
--
HBTE Files section:?/g/HBTE/files
Erik, PD0EK

So, maybe your tinySA plus an AD8310/AD8307 to improve response time for wide sweeps
if the SAW filter is too narrow.

Disadvantage of the ad8310 is the lack of selectable band filter. The SI4432 offers between?1kHz till 600kHz selectable? selectivity.
The saw filter is only used to suppress mirrors and should be best wider than 600kHz but less then 1.5MHz as the IF of the SI4432 is around 900kHz
--
HBTE Files section:?/g/HBTE/files
Erik, PD0EK

Eric,

Thanks for all the answers.

I'm no expert on SAW filters, or anything RF for that matter.
But have read that impedances on a SAW filter can be modeled as a capacitor from input to output,
with a fairly high resistance in parallel.
The series inductors on input and output cancel out that capacitance, making the impedance look resistive.
With two SAW filters in series, there's effectively a very large capacitance between the two.
Adding an inductor in the middle would cancel that out.?
Or perhaps could double the inductance at each end instead?

On the other hand, if it works well, that's good enough!
And I could be wrong.

I like the looks of the 120khz wide B3780 SAW filter, very sharp, good stopband.
Perhaps a 415mhz+(0-250mhz) LO into an ADE-1, two B3780's feeding an AD8310.
An MMIC for gain if needed.? The AD8310 can respond very quickly for fast sweeps.
When a narrow detection bandwidth is needed, inspect the 415mhz using an Si4432?

Though your tinySA would likely be all I need,
Being retired, should have plenty of time available for long sweeps if using the B3780.
?
I've been watching the various SA projects you post here with great interest.
All of them very cool!

Jerry, KE7ER

Hi,

Will there be any chance of a Linux version of SA.exe ?
It will probably run with Mono/Wine(not tested yet), but I always like to have something native....

Since it is written in C++ (I believe), the only real work would be in converting the
display/graphical routines because this is very different on Linux.
Sadly, but understandably, the source is not available. I'm not a very good programmer (also
my nickname on Stackoverflow.com) but always willing to help.....

73
Hans

Document is updated to PDF
--
HBTE Files section:?/g/HBTE/files
Erik, PD0EK

Hello,

This email message is a notification to let you know that the following files have been updated in the Files area of the [email protected] group.

• /Tiny Spectrum Analyzer/Build instructions for the Tiny Spectrum Analyzer.docx

Uploaded By: erik@...

Description:
A document describing how to build and use the tiny spectrum analyzer

Cheers,
The Groups.io Team

On Wed, Dec 25, 2019 at 05:18 PM, Jerry Gaffke wrote:

Post 516 gives a link to the datasheet, for a? B39431-B3555-U310
That SAW can deal with 10dbm max, so the ADE-25MH does give additional dynamic range.

In real life you never want to exceed -10dBm at input as the RX SI4432 will be overdriven and the mirror suppression starts to fail.

?I would like to see more about the protocol between the host and the Arduino,
would make that Arduino code easier to read.?

Will add to the " how to build" document
?
--
HBTE Files section:?/g/HBTE/files
Erik, PD0EK

On Wed, Dec 25, 2019 at 09:42 AM, Jerry Gaffke wrote:

I assume "the switch" is the small IC on the Si4432 module (a PCB assembly that we buy)
that selects either the TX or RX port from the Si4432 to be connected to the module antenna port.
And "set the first IF to zero" is done in the software, as we no longer have a front end diode ring mixer.

Indeed, in SA.exe

This was even more cryptic to me at first, just now getting my head around it all:
> If you flipped the switch before init, the LO?one would become the RX, the rx ?one will?be the LO one but it’s tx goes nowhere...

By "one", I'd guess that means a Si4432 PCB module.
And the only "one" we need in single Si4432 mode is the
module that is normally used to drive the diode ring mixer LO port.
I had been assuming that the Si4432 only worked in half duplex mode,
not using the transmitter and receiver at the same time.

They can't. You can tune and receive, or tune and TX

Perhaps it's been updated, but the build documentation specifies a matching network
for the SAW filter, but doesn't bother to say exactly what SAW filter was used.

The matching network depends on the actual SAW filter you use, YOu have to find the relevant datasheet

I'm looking at the B3790 in the RF360 series from Qualcomm.
It's cheap, Mouser has it in stock, 40db down at 500khz out and headed for 60db down.
But passband is only about 120khz wide, will this work with the current software?

It will work but severely limit the ability to do quick scan over a large frequency range. The max RBW of the SI4432 is 600kHz, a 1000 point scan covers the whole 400MHz range easily.
With a IF BW of 100kHz you will need to scan 5 times the points otherwise you have gaps and can miss a signal

Build photos appear to show two SAW filters directly connected, would it not be better
to have 2x the recommended matching inductance between the two?

Could be but as the input and output impedance are the same I assumed you could couple them also directly

Appears to me that top end of the 5dbm absolute max spec going into the B3790 SAW filter will be
the limiting factor when using the ADE-25MH.?

The +5dBm for the IIP3 does not mean you can use that large input signal. It means you should stay below -25dBm to have spur-free measurement.
And if you measure above -25dBm you have to be aware the mixer may generate all kind of harmonics so you should use the attenuator to test if something is in the input signal of internally generated

Though I suppose running an ADE-25MH with 7dbm going in may be cleaner than 7dbm into an ADE-1.
And the ADE-25MH is surprisingly cheap and easy to get.

In this application there is not much difference between ADE-25 and ADE-1

Amazing these Si4432 modules can be sold for $2us.
Mouser's best price is $4.20us each, if you get the 2500 piece reel.
So the cheapest way to get the Si4432 chip is to heatgun them off of the modules?

They build millions of these modules.

Anyways, the tinySA looks revolutionary.
An excellent complement to a nanoVNA, and about as useful.
If a board were fabbed to take the 250mhz filter, mixer, SAW filters,
and sockets for the two Si4432 modules plus a Nano, it should be
very easy and cheap to duplicate.

All them SMA connectors and cables in the rig Eric built dominate the cost!

Indeed, just add a 2.8" touch display and you have a nanoSA
?
--
HBTE Files section:?/g/HBTE/files
Erik, PD0EK

Afraid not as the RF and LO range is way to high.
At lower frequencies this mixer will be completely deaf.
It is a very nice microwave mixer though

Frequency range, RF&LO	9.0-15.0			GHz
Frequency range, IF	DC-2.5			GHz

This is a very well usable mixer

Excellent performance!

Or even cheaper the ADE-1 found here



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

HMC412 is specified for 9 to 15GHz with +9 to +13dBm LO drive. So I'd say no.
The IF does go from DC to 2.5GHz, which would make it an interesting 1st mixer candidate for a spectrum analyzer with the first IF around 9-15GHz and say a stellex yig as the first LO, but not really for the neat low cost tinySA project.

Came across this one :


Would this be a suitable mixer for the TinySA ?

Thanks,
Hans

开云体育

Eric,

Might I suggest using a 2.8inch TFT SPI display like those used on the NanoVNA and uBITX V6. they are fairly inexpensive. As far as the microcontroller, maybe a STM32, esp32, or a pi-zero. If the software were cross-platfrom compatible that would be nice.

Jerry

AC9NM