DIY "Junkbox-Style" 100khz - 225Mhz Spectrum Analyzer

[SaabFAN]

Am I reading correctly that you've got a 500uH inductor in series with a 700fF capacitor? if so, that's not realizable in practice as the inductor will have far more stray capacitance in it.

Oops. Missed that one. Inductance needs to be much lower while capacitance needs to be much higher, obviously. The 3MHz and the Filter for the second IF had been corrected accordingly.

[SaabFAN]

After struggling a bit with the different filter-types, I have now opted for the same approach HP used: Several Parallel Resonators that can be connected in Series or bypassed.

I also found a much more usable software than Filsyn: Elsie by Tonne Software. It has a free student-version that allows up to 7 stages of a filter to be chained together. The Pro-Version can do a lot more.

With that software, I was able to design someting I should be able to build over the long weekend (we have a holiday on monday in Germany) and test it.


I have also had some thoughts about the Video-Filter: I would like to avoid a complicated network of analog Filters and a solution that I had in mind involves a 3MS/s 12bit ADC connected to a SAMD21 Micro (which I have somewhere in my Parts-Drawer^^) that receives the data from the ADC, integrates the readings according to the setting selected and puts out the current value on 12pins. Running at 48MHz, I think it should have enough computing-power to do the necessary calculations during the short time period the ADC needs to get a reading.
Not very elegant, I know, but it is easy to implement and there's nothing else for this chip to do here

[SaabFAN]

Circuit-design of the analog aquisition board is basically done. I reduced the amount of selectable resolution bandwidths to 3 (1MHz, 100kHz, 1kHz) and replaced the SA602 with an ADE-11X Mixer to reduce noise.

As you can see in the diagram, PIN-Diodes switch between the RBW-Filters and are also used to steer the signal to the SMA-Connector that can send the 3rd IF to another board, as well as to provide the possibility to inject a signal directly into the final IF-Amplifier.

As the 3rd LO I want to use an Si5153, because it allows fine tuning of the 3rd IF, as well as a wide range of frequencies, which makes the requirements for the 2nd IF and LO less demanding - 2nd IF can be between 20MHz and 150MHz basically.
The chip can also be used to provide the reference-frequency for the 2nd LO.
I haven't checked how strong crosstalk between the outputs is though and even though only Output 0 and 2 are used, with Output 1 being terminated into 50Ohms and deactivated by software, I need to perform some tests with the Adafruit-module I have here to verify that the chip generates a stable waveform.
If that's not the case, the reference-frequency for the 2nd LO has to be generated by other means.

Reference-frequency for the Si5153 will be provided by the processor-board (The 25MHz VCXO in the lower right corner), where I have space for a PLL to lock the VCXO to an external reference if available.

[AF6LJ]

Circuit-design of the analog aquisition board is basically done. I reduced the amount of selectable resolution bandwidths to 3 (1MHz, 100kHz, 1kHz) and replaced the SA602 with an ADE-11X Mixer to reduce noise.

As you can see in the diagram, PIN-Diodes switch between the RBW-Filters and are also used to steer the signal to the SMA-Connector that can send the 3rd IF to another board, as well as to provide the possibility to inject a signal directly into the final IF-Amplifier.

As the 3rd LO I want to use an Si5153, because it allows fine tuning of the 3rd IF, as well as a wide range of frequencies, which makes the requirements for the 2nd IF and LO less demanding - 2nd IF can be between 20MHz and 150MHz basically.
The chip can also be used to provide the reference-frequency for the 2nd LO.
I haven't checked how strong crosstalk between the outputs is though and even though only Output 0 and 2 are used, with Output 1 being terminated into 50Ohms and deactivated by software, I need to perform some tests with the Adafruit-module I have here to verify that the chip generates a stable waveform.
If that's not the case, the reference-frequency for the 2nd LO has to be generated by other means.

Reference-frequency for the Si5153 will be provided by the processor-board (The 25MHz VCXO in the lower right corner), where I have space for a PLL to lock the VCXO to an external reference if available.

Very Cool.
You can always add more selectivity positions later on as (or if) you need them.
I was wondering how your project was going.

[SaabFAN]

It is currently the project with the lowest priority (I have about 5 of which two include RF-Stuff).

Btw. I just realized that I haven't put any aliasing filter between the log detector and the ADC. That will be corrected later this day^^

At the moment I am not entirely sure how to build the analog aquisition board: except for the input to the Mixer, the board has to handle "only" 4MHz, so I am thinking about using veroboard and the Si5153 Module from Adafruit I have here, as well as SMD to through hole adaptors for the other chips. Would save me about 40€ for the boards

[SaabFAN]

I have finished an early layout of the aquisition board. Just placed the components in logical groups and took care to have straight lines for the RF-path. Nothing done to make it look more beautiful

I also realized if I want to build it within the Eurocard-Formfactor, I have to order the PCBs. With the adaptorboard, the veroboard would be too small. Putting the power supply and the control for the Filter on a different board would probably give me enough space on the boards for everything.

To shield the filters and the detector-stage, I thought about using copper foil and lego bricks as spacers to keep the foil from shorting out anything.
Using small pieces of copper soldered to the board and then putting a copper foil lid on it should work too.

To keep digital noise out of the analog system, I have separated the grounds for the analog and digital parts as much as possible. Unfortunately, one of the DACs needs to sit right in the middle of the analog area, so 2 rather long traces carrying the i2c-bus signals will run along the edge of the board to that particular chip. Fortunately, the i2c-bus is only used to configure devices that can be left alone for long periods of time (1kHz step scanning @ 1kHz RBW excluded - the Si5153 is being tuned about 1Mhz up and down in this mode while the 1st LO is tuned in 1MHz steps all other modes require only the 1st LO to be swept).
The only Bus that is constantly accessed is the SPI-Bus. To get the data out of the ADC fast enough, I need to push the clock frequency as high as possible. I hope signal integrity doesn't become an issue here. Ordering these boards isn't exactly cheap anymore: 40$ + Shipping for 5 pieces.

Any suggestions or advice about what to do and what not to do when doing the layout for ca. 30MHz fast digital signals? Do I need termination at the receiving end of the signals? Can I transport this signal via ribbon cable?

[SaabFAN]

Revision A of the Aquisition Board.
The vias along the coplanar Waveguide-trace for the RF-Signal haven't been added, but everything else is done.
To avoid angles in the traces that carry RF I have used the Arc-Tool and made nice round corners.

As you can see, I have basically kept the layout and just made minor adjustments to the position of the components.
The Ground-Planes are connected at one location: Next to C56, which is in the Power Supply-Section of the board.

By my estimates, the +5V Rail will require about 650mA worst case. This results in about 2,6W power dissipation in the +5V Regulator. With a safety margin, this amounts to 3W power dissipation in the regulator. The same goes for the the -5V Regulator. So these two will be equipped with heatsinks. The other regulators should be fine without heatsinks.

[T3sl4co1l]

Is that one crystal filter really going to do 100kHz width?  That seems unbelievably wide for crystals...

Tim

[SaabFAN]

Now that I think about it, you might be right about that^^

The calculators on Changpuak suggest a shunt capacitor of about 5pF, which is probably too low to implement on FR-4.

I'll build a Filter on the breadboard tonight and will look how well it performs.
If it doesn't work or the loss is too high, I'll switch to a LC-Filter for the 100kHz RBW.

EDIT: Just measured the Crystals and punched in the values I got to get the LCR-Values of the crystals. Now the calculator shows 4pF. No way I can replicate that on the breadboard or veroboard.
So, it's going to be repurposed as a 10kHz filter

[SaabFAN]

I've run into a little problem: The analog frontend won't fit on a euro card (15x10cm). There is no way to fit everything on the board.

So at the moment I'm thinking of reworking the design - Removing some parts, putting other stuff on daughter-boards, or simply switch to a large board where I can fit everything.
At the moment I'm favoring the last option, which would mean that the boards would be oriented horizontally inside the case. With shielding on the top and bottom, I think I can fit 2 cards into a Schroff-Case, which will contain all the necessary electronics. On top of that case the processing-board as well as the display and the keyboard could be mounted inside a custom superstructure.

Another option would be to put everything on an ever bigger board and put it all inside a custom-made case. I'll have to run the numbers on the different options, to see which one is cheaper.

[SaabFAN]

I'm sorry there hasn't been any updates on this project.
Back in November I managed to buy a Spectrum Analyzer for 650€ that includes basically all the functionality I wanted and more (its a 3Ghz unit and has all the features required for EMC-Measurement: R3131A by Advantest - Needs some adjustment and maybe new DACs in the IF-Amp, but other than that it seems to be working ok) so I've now shifted my focus towards the last item that was missing in my test-setup, an RF signal Generator, and also to the project that I want to be testing with all this fancy gear: DC-DC Converters for pleasure craft

This doesn't mean I'm abandoning this project entirely, but currently I don't have the time available to do this project too and get stuff for University done - After all, electronics is more of a hobby to me, while my studies and work is focused on 3D-Art for computer games.
I am considering to combine both for my bachelor thesis though 

If anyone feels like picking it up where I left, the schematics, as well as the layout for the aquisition board are available in circuit maker.
Basically the two major obstacles I can make out so far are the layout of the analog frontend that needs to be divided across at least 2 euro-cards and the software. The aquisition board takes care of bringing the analog signal into the digital domain, so at least that part of the signal-chain is fully done (unless there are flaws in he layout, which I suspect isn't unlikely^^).

[AF6LJ]

This is a complex project and can be tough.
I have thought about building an an analog spectrum analyzer however that is a bit over my head and finding the time is even more of a challenge.
You have done well to get this far.