开云体育

Looks great! ?Not using the modules allows much more layout flexibility. Did you find a 30MHz TCXO? ?Looking forward to hearing how it performs. Nice you have an oven for soldering, that attenuator was tricky by hand.?

Fortunately, all materials arrived before the New Year, and I successfully completed the welding work today

I think that because there is a DC isolation capacitor inside the module, the external capacitor can be omitted

You can see also i have made some modifications because no standard values of capacitors and inductances? and add a solder jumper on the DNF of the filter

Hello

M0wid, you wrote:???At some point very early on I has a problem with a mixer that did not work properly unless DC isolated, so I added dc isolating caps on all mixer ports.

So need to add also a capacitor on output 13 of U8??
We? have also a mixer? U11, it's the same schematic that the RF switche.

RF SWITCHE? ?with added capacitor



The second mixer. Need to add a capacitor??

cdt

Well,I see. It seems that C40 cannot be used to replace the additional DC isolation capacitor you mentioned, because there are 3db components in the middle, and the DC component may cause abnormal operation. In addition, I see your positive reply to using a TCXO to drive the four SI4432s. I am sure I will do this.

Hi.

A response to various posts in multiple threads, which might confuse!

The RF switch must not have a DC path on any of its ports, hence the need for the additional capacitor.? At some point very early on I has a problem with a mixer that did not work properly unless DC isolated, so I added dc isolating caps on all mixer ports.? In hindsight I think that may have been a faulty part but have not checked again.? I just chose a value that I had lots of.? It could be much lower here as the frequency will always be high.? Something to experiment with to see if it makes any difference in practice.
If not using the SI4432 module then make sure a capacitor is present between the RF switch that would be on the SI4432 module (for RX/TX) and other parts outside of the module such as attenuators or other RF switches.

I used a separate SMA port for the HF and reference as I did not want to load the High frequency range with the resistor network for the reference output, I have a bag of SMA PCB connectors, and am only making one or two.? I think it better to keep them separate.

Good to avoid using the pre-made Si4432 modules if you can, but then I would have one resonator or ideally TCXO clocking all four SI4432 rather than four.

Interesting to see how your BPF design works out in reality.

I hope to see your results as soon as possible. It is worth mentioning that when I measured some other modules in WIFI, I found that the current peak value was only 200~400mA. Therefore, I do not plan to put such a large electrolytic capacitor, which makes it difficult for me to miniaturize the assembly. I replaced it with 47uf or smaller MLCC, and replaced it with DCDC+TypeC. In addition, I'm interested in 4inch LCD, and I don't know whether the code is compatible

Hello Blackberryer

Be aware that there is a mistake in my schematic, which I really must fix.? The RF switch U9 needs to have a capacitor between the output RF2 on pin 3 and R36.? I used 100nF but its is not at all critical.

-----The added capacitance between PIN3 and R36 needs to be repaired. Is it a DC isolated capacitor? Can C40 replace it

JLC recommends 0.3424mm (7826-CPWG) with accurate line width. I have added some important network labels for easy identification to the schematic diagram, as well as 50R text label, as shown below. I hope you can identify

Hello Blackberryer,

Another problem is TinySA. It seems that the output of GPIO2 is connected to the output of RF switch and connected to HF-INOUT. I guess that it connects the_ OUT and HF_ INOUT shares a SMA interface. This design requires the software to consider the timely switching operation.



The advantage is that you can save an SMA interface, so I don't know why you should consider adding an SMA separately in your design.

I wonder if it is feasible to modify your original schematic diagram in this way?

All very good observations.

First of all the credit for the original design goes to Erik Kaashoek.? He developed it much further and much more polished than this project, producing the TinySA and recently the TinySA Ultra.? This was started before the TinySA became commercial, several of us took on the task of making our own version of Eriks original using the ESP32.? John WA2FZW contributed a lot of time improving the software, and Glen VK3PE created his own PCB design.? Unless you particularly want to make your own you would be better off buying a TinySA or TinySA Ultra from a reputable source.? Personally I like making things and on the way learning something.? It is rather boring if something just works first time!

I am not an electronic engineer and not trained in RF design - so all suggestions, improvements and even criticism are very welcome.??

The LPF in practice is closer to your second trace with real components, actually a bit worse.? It does affect results at those higher frequencies.? I don't use mine above 150MHz so I am OK with it.? If you can come up with a better design I would really appreciate the input.? The LPF is mainly there to suppress the LO leaking through the mixer when the port is used as signal generator as well as possible alias effects.? No doubt smaller parts would help, but then harder to solder.? I had the idea that this could become a club project so ease of soldering was important.? Even 1206 parts would be too too difficult for many unfortunately.
I have not come across RFSim99 before - will take a look at it - thanks.

The other limitation was using four of the pre-made Si4432 boards.? If starting from scratch it would be far better to not use these modules and fit the SI4432 directly to the PCB, saving space and allowing the use of one clock source as you suggest.? This would also remove the need to modify the modules with the removal of their filtering.? The separate resonators on each module is far from ideal as they all drift slightly differently to each other.
It should be understood however that the frequency resolution of the SI4432 is not that good, and the smallest RBW is 2.8kHz so it is not that critical in most circumstances.

I would say the software of this project is usable but still could do with some work.? For example there are no markers on the web page (they are there on the local display), it would be handy to be able to easily use the two outputs as independent signal generators, amplitude modulation of the signal generator would be nice, but it is useable especially with the encoder and sees frequent use on my test bench.? The hardware allows for these improvements but time available for coding does not.

I do seem to recall using the impedance calculator on the JLPCB web site for the four layer PCB to get the track width but may not have been consistent in applying it.

Which country are you in?

Dave M0WID

Hello,

Recently, I saw the open source design of SimpleSA, and I thought it was great.Thank you very much for such an excellent design, which benefits most people.
I wanted to make one at home, but after looking at the schematic diagram of the source design, I found several questions, which made it difficult for me to make one.

1.How to select components of LPF?
PCB packaging?
Most of the components selected for the source design are 0805. For general RF design, I think the package should be close to the PCB line width. For example, the 50 ohm line width should be 0.3424mm based on the stack of 7826. I would choose the 0402 component instead of the 0805, because it is closer to 0.3424mm.
顿颁搁/蚕？


RFsim99 simulation results are as follows:



When I find the device on JLC and input the parameters into RFsim99 for simulation, the results are as follows, which is slightly worse than the simulation results of ideal components



2.Why is the PCB trace width of RF signal 0.25mm?
Don't you need to ensure 50R impedance matching?

3.The 30MHz crystal does not need to select VCTXO?
I see that the 30MHz crystal in the si4432 module should be an ordinary passive crystal?
No need to select VCTXO with better accuracy and stability?
Is it because the software has RF calibration?

4.How should I choose a 470MHz SAW,JLCSC seems to have no such device?

/g/HBTE/files/WA2FZW%20-%20M0WID%20ESP32%20-%20ILI9341%20Version%20of%20the%20TinySA/Tiny%20SA%20Operator%27s%20Manual%20%28V2.8%29.pdf