The experience of cleaning up spurs and harmonics in a V3 uBITx

v4 board at 10w on 21 MHz (overdriven)

Mike KU4QO has shared his results in trying to clean up the output of his V3 uBitx.

He thanks all the folks who spent their time on tracking down the causes and cures and shared his results with the group for others to reflect on.

His setup comprised a signal generator at 1kHz feeding the microphone input, and the radio set to SSB mode.  He applied 12.0 VDC for power. Output of the radio was fed through a 30db attenuator into a borrowed spectrum analyzer. He adjusted the signal generator’s output to vary output power in the transmitter.

Power output of the fundamental was just below the point where more microphone signal did not appreciably increase the output power:

80m  10 watts

40m    8 watts

20m    5 watts

17m    2 watts

15m    2 watts

12m    1.5 watts

10m    1 watt

As built, his V3 uBitx was marginally compliant to US FCC rules on 80, 40 and 20 meters.  Above 20m the output was messy as a result of lower frequency spurs. For example, on 15 meters there was a spur at 2.6 MHz that was down only 23dB from the fundamental.

He first completed the L5/L7 mod using surface mount 0.68uH inductors (from the Ebay seller in Poland), marked “T R68K”.  He re-ran the tests and saw no appreciable change in any of the results.

Next, he replaced the 5 relays with Axicom relays from Arrow.com. This change showed a significant change in the harmonics on the low bands. 80m, 40m and 20m went from marginally acceptable (barely -43dB from fundamental) to easily meeting FCC specs on those bands, with unwanted harmonic outputs dropping an additional 10dB or so. However (as expected), the relay change did nothing for the spurs resulting from operation on the higher bands.

Most interesting was watching the output spectrum while adjusting the microphone drive. For example, the nasty spur on 15 meters at 2.6 MHz went from 23dB down at a fundamental power of 2 watts, to below the noise at a fundamental power of 1 watt. At significantly reduced drive (power) levels, the high bands are all in compliance with FCC specs. Here are the reduced powers that provide a clean output signal:

17m   1.5 watts

15m   1 watt

12m   0.6 watts

10m   0.25 watts

Mike followed Raj’s directions for the L5/L7 mod. What should the affect of that mod have been? Should it have impacted those spurs on the high bands? He is questioning whether the inductors were actually not the correct type.

Jerry KE7ER thinks the consensus is that replacing L5,L7 should reduce coupling from the power amp back into the 45mhz IF amp where it takes another trip through the D1,D2 mixer.

Raj VU2ZAP, the originator of this mode, had a specific manufacturer in mind.  They have to be well shielded, commonly available surface mount inductors did not help at all.

Curt WB8YYY is pleased that the results show that the Axicom relay replacement provided around 10 dB improvement.   He has Gordon’s relay board installed, and he imagined his set of new relays will do at least that good.

He notes that some form of added 45 MHz filtering should help in most rigs. He first tried a xtal filter stage – but it disrupted his 40m CW output (He could restore output only by touching L5 – hence the reason for considering shielded inductors). He instead added an LC filter similar to what now appears in the stock v5 board, but he is not sure of the inductor’s Q.  He thinks it may be around 70.  According to simulation it provides more than 10 dB.  He is still waiting for verification in real life.

Curt notes that in the club’s initial measurement of many ubitx rigs, there is appreciable variation even though these are all the same board design (v4) and they were all received at the same time.

John Sutter W1JDS uploaded a spreadsheet of measurements for his µBITx for 80, 40, 30 ,20, and 10m:

1. Stock
2. Replaced L5/L7 with SMD inductors I found locally, kept C211 in circuit
3. Replaced the relays
4. Replaced the inductors with the specified Murata parts
5. Bypassed C211 per the mod

The spreadsheet can be accessed here: https://docs.google.com/spreadsheets/d/1eUFuHcprJhk_HuINxXJLhBuZ7-0DHPfaGPV-RoN8wBY/edit?usp=sharing

The mods he used are those from this website:  http://ubitx.net/spectral-purity/

Both mods helped.

To desolder the relays, he used a piece of desoldering stick,
a low temp alloy along with solder wick and a hot tip on the iron,  and this made easy work of it.  Cleanup was the worst bit.

Probably about the same as: http://www.zeph.com/lowmelt.htm

He’ll probably put the cap back in the circuit in order to be able to use 30m.

He is still needing to find a solution for 80m in order to ensure compliance.

His testing was done in CW mode with the following configuration:

  • 12V LiPo Battery pack  (link)
  • 40dB homebrew attenuator which I’ve characterized as being pretty flat
  • Agilent E4404B SA & HP 8595E SA depending on which one I could use at the time
  • 3D printed key
  • Misc cables…

    This is a v3 radio.  Right now it’s mounted in the plastic box it came in until it decides what it wants to be when it grows up 🙂   John has a JackAl board ready to go when he is happy with the harmonics.