TinySA miniature spectrum analyser

click the picture to read its spec

 

 I bought this little instrument to help look for interference sources. A portable radio works OK but it doesn't give you a complete picture of the noise problem. I've used one to look at our washing machine lately and I heard all sorts of jingly tunes emanating from it, but until now hadn't appreciated the size of the noise problem. I'm guessing the noise is not only radiated but also gets into the house mains wiring. I also found that it doesn't matter whether the machine is actually in operation or in an ostensibly off state. The interference is much the same.

As my amateur band of chief interest is 80m I set the TinySA to scan from 3MHz to 4MHz and the picture above shows the results. The 10 graticules cover 1MHz so the pulse widths indicated are about 100KHz wide with an amplitude ranging over 30dB.

What else can the thing do.. I'd read it works as a signal generator so connected it to my SDRPlay and fiddled around with the settings and you can see the result below. The frequency is 70KHz. Initially I saw a huge amount of overloading because the default was +7dBm output, but cranking it down to a sensible -40dBm the result was pretty good.

 

 

 

 Here are the settings.

Selecting these is dead easy because the screen is touch sensitive and poking it with the plastic stylus that came with it you can choose whatever you want

 

 

 

 At the other end of the spectrum, the highest frequency reached is 960MHz and here the modulation is NBFM. Some overloading is present and the best attenuation accepted is -38dB so an external attenuator would be advisable. I guess the TinySA would make a good beacon for VHF/UHF aerial testing.

The official spec refers to low band output as "sinus" but the high band as square wave. Output level is said to be +/-2dB and whether that includes the usually understood 3dB overall bandwidth figure needs to be checked.

 Next, I decided to see how the TinySA worked with my homebrew noise source. After all, that was the reason for building one a few months back. The results were very surprising because in the process I discovered my favourite BNC to BNC cable was faulty. Below are the results and I've included the shots where the cable was playing up (I think there's a bad screen ground at one end of the cable)..

 

 

 

 

 

 

 Top left. Scan zero Hz to 200MHz with noise base at -90dBm connected to a 50 ohm terminator. Top right as before but with cable fault evident and broadcast signals present (eg FM stations around 100MHz).

 

Centre left. Scan as before but connected to a 30MHz low pass filter and noise source turned off. Centre right with noise source turned on. Noise level about -50dBm.

 

Bottom left as above but displaying cable fault. In many applications the faulty cable may not have been apparent and indicates any cables used in an experiment should be carefully tested. Ref level was set to Auto.

 Next some pictures showing a similar test using a different cable (not perfect). Scan reduced to to 50MHz and reference level set to manual 0dBm throughout.

 

 Scan 50MHz with 50 ohm termination

 As before with noise source on.

 As before but with filter connected, noise source off.

 Noise source on but with RBW reduced to 3KHz hence baseline 10dB down.

 Finally a word on cables. I used really ancient BNC cables and BNC to PL259 adaptors to fit the old 30MHz low pass filter with a BNC to SMA adaptor to fit the TinySA. The larger adaptors showed signs of losses no doubt because they weren't tight fitting. Photos are pretty poor because I used a handheld camera without flash. I think the TinySA can produce jpeg files if I used the software offered by the manufacturer and if so I'll repeat the tests later (see below).

 

 If you're interested in seeing a scan of the same low pass filter using my DSA815TG, then here it is. This uses a tracking generator rather than a noise source but results are pretty close with attenuation showing up at around 50dB with the TinySA and 60dB with the Rigol but with overall shape very similar. Click the picture to see more, then read on below...

 

 Now a couple of interesting pictures. I'd found problems with the cables I was using so instead connected the noise source, low pass filter and TinySA together with a series of adaptors . I also used software (TinySA-App.exe) to save the scans. The results were much better as you can see below. Up to now I haven't found out about averaging the trace, but having seen a u-tube video I suspect the TinySA either has an automatic averaging method or possibly over 10 samples?

 

 

 

 Something rather odd, but consistent with physics (a relationship between noise and bandwidth) can be seen. The output power recorded above right in the shortwave band is registering around -35dBm which is very much higher than the figure without the low pass filter. Below are pictures showing a direct connection between the noise source and the TinySA without the low pass filter.. first with noise off and second with noise on and clearly the noise power is registering around -55dBm or 20dB lower than can be seen using the low pass filter. The effect of using a noise source instead of a tracking generator results in quantitative results which at first are puzzling but nevertheless, for filter experiments and receiver alignment, the use of a noise source is OK as long as qualitative results are acceptable (eg. the shape of the filter response). In fact these two pictures show a 60dB attenuation at 40MHz which is consistent with the Rigol result shown above.

 

 

 

 The scans above were made with my homebrew noise source and I was interested to see how this compared with the Nooelec noise source, albeit using a 10dB attenuator because the total noise power output from this is higher than the maximum rating of the TinySA.

The first two were made with the scan from zero to 50MHz with noise OFF then ON. As you can see the baseline is around -95dBm rising to -40dBm (with a 10dB attenuator fitted)

 
   

 I then repeated the tests with the scan set from zero to 500KHz. The baseline varied and the noise level was arounf -40dB once the frequency had risen to 100KHz which is the lower spec for the TinySA. I also tried the scan in the UHF range but it seemed that the noise source was oscillating somewhere around 500MHz so I didn't bother.

 
   

 In summary, the TinySA together with a noise source would be ideal for testing and adjusting filters, producing similar results to a spectrum analyser and tracking generator. The smoothness of the curves is dictated by the built-in averaging circuitry and the bandwidth setting. This latter is normally automatic but it can be set manually bearing in mind the lowest settings will be pretty slow. The end result will not match that of an expensive spectrum analyser with its much better averaging circuitry.

 A "Faulty" TinySA

Old TinySA version V0.3.1_E, S/No 21041937

 

Faulty TinySA version V0.3.1_E, S/No 25082009

 I spotted a faulty TinySA (made in 2025 right) on Ebay and bought it for £19.95 after the price had been reduced. The fault described by the owner is reported from the internal Self Test and is related to the attenuator chip mounted at the RF Input socket. I switched it on and it seems to work normally as far as basic functions are concerned but there are clues to its fault. As my older TinySA (made in 2021 left) is working OK (self-test good) I used it to check the faulty one (Test 3 bad).

Using a frequency of 300MHz as Low Output at any level down to about -25dBm the new one registered around -24dBm. Beyond this at say -35dBm the new one registered -38dBm. In the transmit mode the output was below -25dBm. These tables give the responses in the low band to each others output signal from the low band and show the faulty Tiny low band input and output are both bad.

 dBm

 dBm

 dBm

 dBm
  dBm

 OLD

 0

 -10

 -15

 -25

 -35

 NEW

 -23.9

 -23.9

 -24.4

 -24.4

 -38.9

 dBm

 dBm

 dBm

 dBm

 dBm

 dBm

 NEW

 0

 -5

 -25

 -28

 -34

 -36

OLD

 -24.9

 -24.9

 -25

 -36.4

 -36.4

 -34.4

The lower band shows consistent losses between the two devices when either is transmitting

The first of the two pictures below shows measurement ** and the second ***.

 dBm

 dBm

 dBm

 OLD

 -7 **

 -17

 -27

 NEW

 -84

 -94

 -103

 dBm

 dBm

 dBm

 dBm

 dBm

 NEW

-7

 -17

 -27 ***

 -37

 -47

OLD

 -84

 -93

 -88

 -96

 -111

 Now a test at 10MHz as shown below

The picture shows a loss of 84.2-7.0 = 77.2dB in the input circuit of the faulty Tiny.

 

 Now reversing the two

The picture shows a loss of 88.2 -27 = 61.2dB in the output of the faulty Tiny.

 

 As both the input and output are governed by the same components and other parts around the SMA connector seem to be OK I've ordered a new attenuator chip type PE4302 and with luck this should fix the fault.

In the meantime I checked the two TinySAs on my Rigol using my old TinySA initially, as this had passed all its tests, and much to my surprise the first (nearly new) test cable carrying an N-type to a BNC to the SMA-BNC adator resulted in hardly any signal being displayed. Eventually, having switched in the Rigol 20dB amplifier and setting the vertical scale from 10dB to 1dB I saw a 10MHz signal but really really weak. Now, this was very odd, so I swapped the cable and the Rigol strongly objected! How long had I been using that faulty cable? Anyway, with a working cable I was able to complete the tests on the faulty TinySA.

 First the old TinySA set to -7dBm output

 

 Now the faulty TinySA, also set to -7dBm output

Clearly the signal is being attenuated 77dB by the faulty PE4302. Much to my surprise this figure of 77dB matches that when testing with my old TinySA.

 

 The old one set to 300MHz High Band output at +9dBm

 

 The faulty one also set to 300MHz High Band at +9dBm

This is a little low but suggests that the components around the High Band SMA connector are probably OK and the output is unaffected by the Low Band fault.

 

 This shows the old TinySA High Band set to its maximum output of +16dBm. Interestingly this is way above the maximum setting of +9dBm for the faulty one.

My guess is the firmware of the later example no longer permits a signal of greater than +9dBm.

The maximum input of the TinySA is specified at +10dBm ***

Another firmware change is in the output level setting. The old firmware only allows the minus sign to be selected ahead of the numbers whilst the newer firmware allows for the minus sign to be added either before or after the numbers are selected.

 

 *** Here's something quite important. The maximum input rating for the TinySA is given as +10dBm. However the spec does state that the output of the High Band is a square wave so does this mean that if one sets the High Band output to say 300MHz at +9dBm and connect this to a second TinySA will this be OK? No.. because there will be harmonics of that 300MHz signal whose power must be added to the power at 300MHz. I didn't measure the harmonic levels in the tests above but I shall carry out further measurements later.

 

Now for the repair

 

I was slightly concerned about the chip change because the display is glued directly underneath the faulty PE4302 and I didn't want to detach it in case I damaged it.. I did however remove the lithium battery because I've had experience of one catching on fire whilst repairing a mobile phone.

My hot air gun has a nozzle diameter of 5mm so is fine for the chip removal. I set the temperature to 350C with an air flow setting of "20" whatever that means. To aid heat distribution I smeared a flux solution on the 20 pins. Using tweezers with ceramic tips the chip lifted off within around 20 seconds.

The picture on the left below was taken after the chip had been lifted off. Residual solder looked fine so I just balanced the new chip carefully in place and used the heat gun again. After about 20 seconds the chip settled in place and the soldering looked perfect.

The display glass felt very hot to touch but there were no signs of damage.

 After refitting the metal screens (I'd removed both so I could check the High Band components) I put back the battery and crried out tests. Note the the minimal clearance between the outer case and the circuit board assembly means the battery needs to be positioned as it was initially.

The tests looked OK to me so I put the assembly back in its case.

 

 Now the test that really matters! The seller had been kind enough to show the test results in his Ebay description.. these are shown below left. On the right are the test results for the repaired TinySA. I guess lots of TinySA's will be damaged like this one.

My tests are always done after careful consideration because I regularly use a Rigol spectrum analyser. Although its spec leaves a little more latitude than the TinySA I'd be more than a little upset to damage it.

 

 

 

 Bearing in mind my comments on looking after the TinySA I also bought this attenuator. It cost me £2.85 including postage and, although its very simple in design, it seems to do the job. It's spec tells me it'll be good up to 4GHz.
 

 Return to Reception