Today we had a meeting in our local amateur radio club ESSLINGEN P02.The topic for this meeting was WSPR.
Bernd DK9BH was the organizer of this meeting. He has ordered about 10 WSPRlites from SOTABEAMS.
I had the chance, to do same basic measurements on one of the WSPRlite units.
Here the results.
80m
The WSPRlite has only filters for 30m and 20m. But even better knowledge, we did a test for 80m to check, which problems this may cause.
Powered by LENOVO Laptop
You can see that power output for 80m is more or less identical to 30m and 20m - this is remarkable in my opinion.
But the harmonic suppression is not existing. The first four harmonics are suppressed a maximum of 20dBc - this is far too less to meet any regulatory requirements and should also cause some bad feeling for the WSPRlite user!
30m
Scan width 30MHz
Scan width 1MHz 
On 30m the harmonic suppression is about 35dBc. This should not cause any problems but may not meet all regulatory requirements ...
Interesting is, that the emissions below the main carrier are much better suppressed (-60dBc) than on 20m. So I assume, another type of low pass filter is used?!
20m
Scan width 30MHz
Scan width 1MHz 
On 20m all is fine. The harmonic suppression is about 50dBc. This should meet all regulatory requirements!
What makes me a little bit wondering, why below the main carrier a number of harmonics are visible - all about -50dBc. So I assume on 20m the unit uses a low pass filter for suppression of harmonics above the main carrier.
Power filtering
Powered by powerbank A
Powered by powerbank B
If you compare these two scans with the 80m scan powered from a LENOVO consumer laptop, you can see, that the USB power quality has a direct impact to the output signal.
I assume, that powerbank A delivers significantly less output voltage, therefor the output power drops from 226mW to 36mW.
The power amplifier of the WSPRlite is directly fed from the +5VDC from the USB port, noise on the USB port will be modulated (with a little bit of filtering) on the output signal.
Drift
I've done some statistic on about 4.000 WSPR spots from 9 WSPRlite and the frequency drift is acceptable:
General
The power output from the WSPRlite is nearly exactly 23dBm == 200mW as set in the config utility.
If someone has more details on the WSPRlite circuit, I would be happy Diese E-Mail-Adresse ist vor Spambots geschützt! Zur Anzeige muss JavaScript eingeschaltet sein!.
All measurements were done using an HP432A powermeter with a thermal 30mW powerhead, a SIGLENT SSA3021X spectrum analyzer and some SUHNER attenuators.