开云体育

Maybe the series resistances of the crystals you have is just too high to oscillate in that circuit.? Crystals have much higher equivalent series resistance at their overtones than they do at their fundamentals, and greater gain will be required to overcome that.? What transistor are you using in the circuit?? Have you tried more than one crystal in the circuit?? Same results with all of them?? How high a resistance value can you use in place of the crystal and still get it to oscillate?? Have you tried putting an actual overtone crystal in the circuit and see if it will oscillate with the intended device in the circuit?? Are the crystals you have tried of recent manufacture or older parts?? What case size are they?? There is a big difference in, for example, the tall case HC49 and short case HC49s crystals.

Jim, Thanks for the reply.? I understand your reasoning, but I'm just trying to understand why it doesnt work at all. Then I could worry about whether it was on frequency.? I already have a tiny SA, so I don't really need to build this, I'm just trying to get components to work that others have done before me. When Ashar Farhan builds one out of a junk box and makes it work, I wonder why I can't.? Also Wes Hayward sent me a schematic for his (similar) butler oscillator (5th overtone) with junkbox crystals and showed how the oscillator worked slightly off with just a resistor. (I built that and acheived 100 mhz).? Then he inserted the 20 MHz crystal and forced it into the correct frequency SHAPE as shown on his spectrum analyzer with the variable cap.? Again a fairly simple circuit.? However I could not get mine to oscillate with the crystals I have even though the thing will oscillate at 100 Mhz with a resistor.
Andy

Andrew,

Are you still trying to make that work with a 20 MHz fundamental crystal?? It can probably be made to oscillate in an overtone oscillator circuit, but it won't be on 100 MHz.? It will probably be near 100 MHz, but not close enough to be usable.? One reference suggested frequency errors in the range of (overtone number times 25 kHz), so for 5th overtone operation, expect it to be off by 125 kHz.? That's a "rule of thumb" and highly approximate, of course, but with the potential for that kind of frequency error, why even bother with a crystal?? Stop beating your head against the wall: either buy a 100 MHz 5th overtone crystal, or come up with another solution.? Use your 20 MHz crystal in a fundamental oscillator with a multiplier circuit, or use a packaged 100 MHz oscillator, or something else.? Plenty of other ways to reliably generate a 100 MHz signal.

On Sat, Feb 11, 2023 at 12:50 PM, Gerard wrote:

DL1640

Hi,?
This url from Yokogawa

enjoy

Nigel, Did you use just any 20 MHz crystal, or did you purchase special "overtone" crystals.
I emailed Wes Hayward who has written books on the subject and he said that regular crystals will work,?
but so far I haven't been able to get Ashar's design to work, with the crystal or using the resistor and capacitors.
Hayward sent me another butler design that he uses and that I could get to oscillate at 100 mhz without the crystal
(just a resistor), but no luck with the xtal.
I built Ashar's design using (1)a printed circuit board, (2)A perfboard with eyelets and neither will oscillate.

As far as the Sweeperino, a friend designed a gui that addressed the Si570 and we were closing in on it, but I somehow burnt out the 570s that I had .? I found another supply, but so far haven't go back to that.

I had started the Specan project back in 2019, and have just now got back into the project. ?My last activity had been to build the 100 MHz oscillator. ?I used ugly style construction with Me Pads. ?The oscillator works fine for me. ?I built the module using a 9 volt battery as its supply, and MPF102 transistors. ?I will check my notes to recall if I had any issues. ?I just checked the signal output on a spectrum analyzer and it looks good. ?At the time I may have tuned it using the FM radio method that Ashar Farhan described. I am about 60% completed on this project. ?I had issues with the Sweeperino software not working as well as it was described, and the documentation on the Arduino wiring to the LCD display was wrong.
regards, Nigel, VA3NMA.

Hello,

The RF signal is not getting to the receiving Si4432, so the calibration messes things up.? There is no code in my version to detect this (there is in Erik's TinySA).? We need to find out where the problem is.

Lets start from the beginning with no calibrations done.

The B3555 SAW filters have a centre frequency of 433.96MHz.? Please check the frequency setting in Config/IF freq matches the SAW filter you have used.

If you connect the ref port to the SA port and turn on the Ref output at 30Mhz you should see something similar to the image in my last post - strong peaks at 30MHz, 60MHz, 90 MHz and so on.? If you do not then something is wrong.

From previous photos you have a libreVNA, with spectrum analyser mode.? The LibreVNA will almost certainly perform better then the SimpleSA.? Using this, an oscilloscope with >500MHz bandwidth or an SDR you can probe around to verify the signals at various points.? The local oscillator frequency will be varying as it goes through the sweep.? To fix the frequency one way is to go to signal generator mode, set a frequency, say 50Mhz and turn the output on.? This will ask the receiving SI4432 to output the IF frequency, 433.96MHz and the local oscillator Si4432 will be set to output the IF frequency plus 50MHz = 483.96MHz.? These two frequencies mix in the mixer, producing 483.96-433.96Mhz = 50mhz and 433.96+483.96Mhz = 917.92Mhz (sum and difference of the two inputs).? The 350MHz low pass filter will remove the higher frequency, leaving the 50MHz signal heading off to the attenuator and then the output port.??
The output signal should be quite strong, and the signals from the Si4432 can be strong enough to damage some equipment so check the limits of your equipment first and? use an attenuator if you are not sure.? The libreVNA specs say -5dBm max.? The Si4432 can output +20dBm so take care.

If you don't get a 50MHz output signal, then trace back to find the problem.? You should be able to change the strength of the output signal using the slider.? I would start by bypassing the attenuator with a 100nF capacitor, and then if still no luck the 350MHz filter.? If still nothing then check each SI4432 output, the output after the SAW filters, and then the mixer.? You could even bypass the SAW filters for this test.

You can do a similar test of the TG output.? In SA Low mode, OUTPUT/TRACKING set the output frequency and then Generate.? For this to work you will need to make sure the TG IF frequency matches the SAW filters you have fitted.? If these are the B3555 then 433.96MHz.? To remind this is set in my_SA.h? #define TG_IF.? You can also just comment out the #define TG_IF.? The code will then use the same IF value as the main IF and apply an offset, which defaults to zero but can be adjusted via a serial port command TGOFFSET.? Leave at zero for now until we have things working properly.

I think there is a hardware issue of some sort.

Dave

I should not get the calibration step. As long as I enter - 30 in the calibration menu, the bottom noise will jump from the bottom to the top - 30 db, just like I calibrate the bottom noise, and every time I start up, there is a prompt of No Calibrations found;



I was also confused about the calibration steps described in Tiny SA Operator's Manual (V2.8). pdf. Click, after connecting REF to LF, set the output frequency, for example, 15M. At this time, there should be at least three peak points, 15M, 30M, 45M (I found no harmonics after erasing the flash, only 15M frequency points can be seen), then switch to CONFIG ->CALIBRATION ->ACTUAL POWER, enter - 30, click the menu switch status and return to the display interface, It still moves to the top as a whole.



I think in this calibration step, unless the calibration code is the frequency corresponding to the peak point in all the frequencies of the calibration scan, of course, the normal situation should be the frequency of the REF output, which is theoretically the largest, but I calibrated it from According to the phenomenon, it seems that the power of all frequency bands has been corrected?

That does not look good.

The maximum signal input to the device must be < 0dBm, ideally < -30dBm.

With nothing connected to the input and frequency span as you have, the level should be around -90dBm.? Do not panic though - it could be just calibration.
To calibrate set the frequency span to be say 1-100MHz, RBW in Auto, Attn and Ext to 0, spur off, connect a -30dBm signal to the input - you can use the ref output for this, or some more accurate signal generator if available.? Then using the menus, select config/calibrate/Actual power and enter -30 on the keypad( or the actual known power of the applied signal) and then X when done.? I would recommend changing the Gain to 37 for a -30dBm signal.? The simpleSA will then record the value of the largest peak signal and work out a correction factor to make this read -30dBm on the trace and store the value- (provided the simpleSA is not turned off within around 10 seconds (#define SAVE_SETTINGS_MILLIS 10000 in my_SA.h).

There is a bug as this new calibration value is not applied until a change of some sort takes place, eg changing start frequency.
You may have to change the ref level (32 in your photo) to something more normal such as -10.

Then you should end up with something like the attached photo, with the ref output still on and connected.

Dave

Do you get the same effect when scanning the lower frequency range?

I think also you may have found a bug.? Tracking and spur reduction should not work in the high range, only the low range.? It is quite possible there is some effect going on here - what happens if tracking is turned off?? I think you may have to go to low range, turn it off, then back to high range.

It could also be harmonics from the 30MHz oscillator signals if they are not carefully routed away from signal paths and kept short.? Unfortunately I am away from home with just company laptop, which is locked down to some extent, so I am not able to install kiCAD and view your layout at the moment.

The mixer, LO and SAW filters are not used at all in the High range - it is just a signal into the SI5531, filtering is mainly done in the SI5531 internal DSP.

My guess is some of the 30MHz clock is getting into the signal path, or possibly there is a bug and even though the Ref output is shown as off maybe it is not.? That would need some testing.
Your noise floor is around 10dB lower than my one, which is good.


Dave

The attachment is here.

Do you know why? Is it where the code needs to be modified?

The attachment is the PCB version that I have modified at present, and gerber is the original version that I have welded at present

Interesting to hear of different techniques.? The first one I built by hand and used a heat gun and normal solder, the second in someone else's reflow oven using a stencil and paste.? Both are possible but the oven is much easier and gives a better looking result.? When building by hand I first applied solder to each pad using a normal iron, not too much.? Use solder wick to remove some if too much, or level off with the heat gun.? I then placed the filter in position carefully aligning and applying some pressue on top while heating with the hot air gun until I felt the filter sink slightly as the solder melted.? I have to admit it took three goes to get it right, however the PCB was also home etched so did not have the solder resist of your boards, which makes a huge difference.? With solder resist the part will tend to find the correct position from surface tension of the liquid solder.? I have to admit to subjecting the filter to much more than 215C each time as I heated from above.? Heating from below as Blackberryer suggests would be much better.? I found the attenuator however much harder to do by hand, again no solder resist did not help.?

The Tracking Generator(TG) function provides an output signal that tracks the frequency being "listened" to by the SA port during a sweep.? This is a really handy feature for testing the response of filters and amplifiers.? You can also use a VNA for that of course, but the TG adds versatility to the analyzer and compared to my old Nano-VNA at least has a much faster sweep making adjustment of filters easier.? You are also outputting a proper sine wave while the Nano Vna outputs a square wave.
The TG output is a low pass filtered mix of the outputs of the TGIF Si5531 and the TGLOSi5531, while the analyzer part is a mix of the signal being measured and the output from the Si5531 SALO.? If both SI5531 IF frequencies are the same then the strong signal from the TGIF Si5531 bleeds into the receiving SI5531 and reduces the dynamic range quite significantly.
In my second build I actually used 447MHz (B3767 - datasheet attached) filters as that was what was available at the time to give a larger frequency separation and the narrow 1Mhz or less filter bandwidth needed.? In the first prototype I used the same B3555 SAW filters for both SA and TG, and this could be made to work very well by setting the IF frequencies at opposite ends of the 1MHz or so pass band of the filters.? Not too close to the ends or the results with wide RBW settings will not be ideal.? I think I used B3760 433.93MHz filters for the SA in the second version, again because that is what was available at the time.
In short different SAW filter frequencies is a benefit, but by no means essential.

If all four SAW filters are of the same frequency then the offset between the two can be be adjusted using a terminal command using the USB serial port? (TGOFFSET).? You can see a list of available console commands in the cmd.cpp file.? If different IF then the offset can be left at zero.
Whatever filter is used for the TGIF change the #define TG_IF value in my_SA.h to suit, and the initial value for the SA IF frequency is set in the initial values of the settings_t type in simpleSA.h (line 272).
I found it quite educational to connect a bandpass filter for 20Mhz or so, whatever you have or can make (I used one from QRP-Labs), between the SA and TG and then try the results with different IF frequency settings and offsets to see what works best.


Hope that helps

Dave M0WID.

re,

?Dave, m0wid
Can you tell us more about that please?