School spectrum analyzer project.

[Smokey]

Learn how to make a real circuit board and have them made at JLCPCB or something.  However your project turns out, you will at least be able to make circuit boards at the end of it.

[BlownUpCapacitor]

Learn how to make a real circuit board

Already do! https://youtu.be/02W6lhY8H5M?si=VaUy5YmIQaYIvUBl

Jokes aside, I will. I have already designed a few PCBs on easyeda and ordered from jlcpcb. I get free pcbs from jlcpcb essentially turns out. If you use easyeda you get a $8 coupon everymonth that covers the pcb cost. If I order say a $7 pcb, and use the $8 coupon, all I have to pay for is shipping.

[Sensorcat]

A lot of stuff for a 15 year old eh? My local e-waste center sells this kind of stuff for very very very cheap. The 7904 was $50, the 5110 with all plug-ins as $75, and the BK 4040 was $20. Just some small repairs and cleaning were all that was needed.

OK, your situation is completely different to what I imagined. I don't think anything important is missing. That's good, because building an instrument without an instrument is almost impossible.

So you are 15? It could be that you are the youngest person to ever build a spectrum analyzer. And once you succeed, the only problem is that many people will not be able estimate the achievement.

[BlownUpCapacitor]

Update: Don't expect anything new until June 10th. I have finals coming up and my grades are falling behind. A series of screw-ups and simple mistakes cost me a bit.

[zapta]

Have you considered to use a fast ADC and FFT algorithm, e.g. on a raspberry Pi or a PC?  It may be easier to design and more flexible to use. Just sample a short burst of samples at 20 Mhz, and use a python script to compute the FFT and display the graph.  The python script is trivial once you read the burst of ADC samples.

[BlownUpCapacitor]

Have you considered to use a fast ADC and FFT algorithm, e.g. on a raspberry Pi or a PC?  It may be easier to design and more flexible to use. Just sample a short burst of samples at 20 Mhz, and use a python script to compute the FFT and display the graph.  The python script is trivial once you read the burst of ADC samples.

I could, and that would be easier, however, the school requires that I learn something very significant from this. I think if I do that, I'll learn how to code and use uCs, and some RF design, but not much else in terms of electronics I think.

A swept front end would give me lots of experience in RF and electronics design. Plus I have until January to complete this. That's nearly 6 months. Really the project is due near end of December, as I have to write up a 25 page report on this project, but still almost, or maybe more than half a year's time for this project. So it can't be too simple.

[zapta]

Makes sense. I don't know what the requirements at your school are, just speculating.

If you want to add complexity, once the fft analyzer works, you can add to it a frontend mixer that will shift a narrow band from high frequency to lower frequncy. This will give your analyzer high resolution and at high frequency bands. Probably much better than you can do with mixer only. And then you can record it over time and generate a spectrum over time graphs. It's a good segue to SDR and signal processing.

https://training.dewesoft.com/images/uploads/29/fft_3d_measurement_1587721452.png

[BlownUpCapacitor]

Update: Don't expect anything new until June 10th. I have finals coming up and my grades are falling behind. A series of screw-ups and simple mistakes cost me a bit.

Yea, uhmm, I got bored from studying Chemistry after an hour and decided to create a 3D model of the Tektronix 5000 series plug-in chassis. https://www.tinkercad.com/things/iGbtzBgoZeZ-tektronix-5000-series-plug-in-shassee

I plan to 3D print this and use it to make my project work in a Tek 5110. I just need to figure out how to do that in the first place.

Stoichiometry hurts my brain sometimes.

Anyone know some good pointers for Lewis structures too?

[erikka]

In case you have still time to consider options you could search ebay.com for these modules "rf oscillator voltage controlled"
They come in different frequency ranges.
Using such a module, a mixer, a band pass filter at the high if frequency, a second module and mixer for the down conversion to the low if where you have the resolution filters, your most difficult components become the high if filter and the resolution filter(s)
For the log power detector you can use a AD8307 module, also from ebay
The log power detector doubles as a generic rf power meter that you can use, together with one of the vco modules, to test your if filter.
You can use a 4 opamp ic and some potmeters and other passive components to build a voltage sweeper where you have one potmeter for the center frequency and one for the span.
In this way you can build an sa without microcontroller, using a xy scope for display,  and I know it will work as I did build it.

[BlownUpCapacitor]

So I made the 3D model easier to 3D print: https://www.tinkercad.com/things/82Jht3C4EZv-3d-printable

Also, my finals start tomorrow... and my first one is chemistry. Wish me luck on figuring out Lewis structures, stoichiometry, atomic theory, and literally everything else as a freshman with 45 seconds per question.

Thanks.

Update: I think I got at least a D- 

[Solder_Junkie]

If you want to view up to 500 MHz with a few Hz resolution bandwidth, this simple idea is hard to beat... not really what you are looking for but a fun and very low cost way to check IMD and keying bandwidth of transmitters.

https://www.qsl.net/g4aon/ssa/

SJ

[BlownUpCapacitor]

Does anyone have any tips on creating LC oscillators?

So while experimenting with a super simple variable resolution filter, I finally decided to ask about this pet peeve.

So whenever I make an LC tank circuit for, well, ANYTHING, there is always something wrong. Either the Q of the system is too low, or the inductance of the coil is calculated wrong somehow and the resonance of the tank circuit is off by a few megahertz. And when I move my hand too close, the thing changes its resonance also. Same thing with an oscilloscope probe when I'm trying to tune the thing with a trimmer cap, and when I remove the probe it oscillates at a few megahertz higher AFTER I installed it in such a way that I can not fix it.

So far, the only thing that could possibly bypass the scope probe thing that I've thought of and tried myself is using a different coil to induce an AC magnetic field in the tank circuit and use another to pick up the resonance from another coil that feeds into a buffer to prevent any capacitive changes that would mess up my resonance. But my hand, even 4 inches away messes with it when trying to tune using this method.

Does anyone here hold some tips they can share that would help?

[Solder_Junkie]

And you want to build a spectrum analyser? Making an RF oscillator is pretty basic stuff, your image shows both a crystal and an inductor, and a physically large inductor at that!

Either you want an LC oscillator, or a crystal one, don’t try to combine them.

There are literally hundreds of circuits on the web for crystal oscillators, one of the most reliable types is the Collpits oscillator. If you want a really clean crystal oscillator, there are techniques to do that too. This set of low noise beat frequency oscillators are based on work by Ulrich Rohde ( of Rohde and Schwarz fame):
https://www.qsl.net/g4aon/pdfs/BFO%20v2_3.pdf

SJ

[BlownUpCapacitor]

The main reason why I'm using a tank circuit is because it offers a low impedance to other frequencies compared to a high impedance for it's resonance frequency. This is something a crystal can not provide as it is inverse. The crystal has a low impedance for a single frequency (ignoring spurious resonances) and a high impedance for other frequencies.

I've also gotten around the LC resonance shift problem by shielding the crap out of it using copper clad boards or some other thing. This has worked okay, but it uses more material than I'd like, thus my question of tips. But if it is the only true way, I guess I'll keep doing that.

[BlownUpCapacitor]

Progress report:

After a while of trying to make my own ring modulator, I gave up as I didn't have any of the correct parts; it would also be more economical to buy a commercial one instead of buying the parts I needed.

So instead of using a ring modulator, I decided to use a 6BE6 tube for mixing as I happened to have some extra after my tube transceiver project. It worked very very well at audio frequencies, but the higher the frequencies got, the worse the tube performed. This was likely me setting up the tube improperly, but I just really didn't want to deal with it anymore as it was getting hot and deforming the plastic on my breadboard.

So I gave in and bought a ring modulator from Digikey. After the parts arrived, I swiftly assembled a very crude prototype/proof that I could solder a piece of metal to another piece of metal.

The thing has no detector or log amplifier. It has no good way to adjust the center frequency nor the frequency span. But it works. All it has is a mixer, a crystal filter, and a very bad and not very reliable VCO. The VCO issue is just me using a very scratchy trimmer pot for R1 I pulled out of some random board. I am probably going to just buy some VCO off ebay instead to ration my sanity well.

Here is a video of me stuttering a lot while struggling to speak normally: https://youtu.be/8rkVdlCWSvI

I have chosen to not use the 5110 anymore, and opt to use my 7613. It is variable persistence storage which makes it an excellent display unit for a spectrum analyzer. The downside is that it will take more effort to properly fabricate a plug-in module for it as it uses that special alignment plastic bit on the rear of the plug-ins, non-similar to the 5000 series or 500 series where they don't require such things.

I also somehow managed to pass chemistry with an A, if anyone was curious.

[BlownUpCapacitor]

I've designed a promising sweep generator design using some simple components.

It operates by using a 555 timer to generate a square wave, when then gets fed into the sweep generator. It consists of a simple capacitor charger, and when the 555 timer outputs a hi signal, the NPN transistor turns on, quickly discharging the capacitor, and when the 555 timer outputs a lo signal, the transistor turns off, allowing the capacitor to charge again. This capacitor charging and discharging is buffered through an OP amp, and will later be amplified by another op amp.

I have chosen a fairly large value for the discharge and charge capacitor as I have discovered that the greater the value, the more linear the sawtooth waveform's rise is. But this comes at the downside of reduced signal amplitude, which can be rectified by using another opamp to amplify this signal.

I chose this method for creating a sweep signal because the 555 timer outputs a square wave that can be used for blanking the z-axis. The squarewave's phase directly aligns with the time the saw tooth wave is not sweeping/ retracing, which fits my needs. This will remove the issue of the SA redrawing everything in reverse, which was an issue in my mockup design. The only possible issue I see with it is that the re-tracing happens exactly when the square wave goes hi. This can mean that the trace doesn't fully blank before the sweep re-traces. It would be nice if there was some way to add a delay to when the 555 timer goes hi, to when the discharge transistor turns on. Maybe I can use an arduino to add a 50ms delay to that. I will be incorporating an Arduino anyway to display center frequency and other stuff on an LCD display. I can't figure out how to interface the readout on the 7000 series. It would be nice if I could figure it out because if I did, I could just use that instead of a separate LCD display.

So this will require only two ICs, as I will be using an LM358 dual op amp, which covers the two op amps: the buffer, and the amplifier. The other IC will be the 555 timer.

Here is my circuit: https://tinyurl.com/2j9pu69m

[Sensorcat]

I've designed a promising sweep generator design using some simple components.

It operates by using a 555 timer to generate a square wave, when then gets fed into the sweep generator. It consists of a simple capacitor charger, and when the 555 timer outputs a hi signal, the NPN transistor turns on, quickly discharging the capacitor, and when the 555 timer outputs a lo signal, the transistor turns off, allowing the capacitor to charge again. This capacitor charging and discharging is buffered through an OP amp, and will later be amplified by another op amp.

Hmm. The 555 is not a very stable source for clock and timing signals. Since you seem to aim at about 20Hz, you can use the Arduino for all timing signals, including the delay signal. It is accurate to generate slow signals with a µC, because the timers/counters reach large values. To get the Arduino stable, add a quartz oscillator, if your module does not already have one. The internal oscillator of the µC will likely drift as a 555 does.

[BlownUpCapacitor]

Hmm. The 555 is not a very stable source for clock and timing signals. Since you seem to aim at about 20Hz, you can use the Arduino for all timing signals, including the delay signal. It is accurate to generate slow signals with a µC, because the timers/counters reach large values. To get the Arduino stable, add a quartz oscillator, if your module does not already have one. The internal oscillator of the µC will likely drift as a 555 does.

I'll take the advice for using a uC for square wave generation. It would also give me more control over the duty cycle of the square wave, which is good as I can adjust the sweep hold-off that way. I don't think clock cycle drift is a huge problem as I know that the sweep frequency can vary by approximately +-10%. I was planning to add a variable control for sweep speed on the front panel anyway. I guess I can add a rotary encoder for my Arduino to use to change the sweep frequency.

[BlownUpCapacitor]

The circuit works well on the breadboard using an arduino to drive the discharge transistor. I noticed that the falling side of the waveform wasn't that clean, but I can just blank that away.

You may notice that the trace seems quite bright, and that's because I have to put the intensity quite high to capture the waveform using a camera.

[BlownUpCapacitor]

So I made a 3D model of the mating part for the 7000 series mainframe plug ins: https://www.tinkercad.com/things/bED8Zr7AeYU-frantic-tekwork

I don't have a 3D printer to I'll have to wait until late August to hopefully print it; maybe the computer science teacher won't even let me print it because of its size. But I have (some) confidence in my tinkercad skills 

So I have everything I need to make a 7000 series plug-in chassis.

Those of you who are more observant and more savvy with tek plug-ins may notice the lack of the PCB mater, and the weird 4 hole thingy on top of the part. That's because through my geniusness I have concluded that the effort to model the PCB mater does not meet my effort-to-product ratio requirements. This is because I am hoping that my future self will be able to design the chassis and the PCB to fit into the mainframe socket without the need for such a PCB mater to hold the PCB secure.

Also, I don't know what the 4 holes thingy on top of mater for tek plug-ins are for. So I didn't model it.

I'm also taking an interest in trying to understand the readout system of the 7000 series mainframes. My hope is to be able to use that to display the center frequency of the spectrum-analyzer and some scale factors. I am, however, struggling to do so and might just use some of my own vector graphics if I can figure that out.