On Apr 16, 2025, at 10:41?AM, Ashhar Farhan via groups.io <farhanbox@...> wrote:

You could probe the input to the 3.3v regulator. However, this is a little more complicated than what it appears to be.

First, I spent weeks measuring the spurs. They are all below -43dbc except on 21 mhz. To quote a politician: "those are my alternate facts".

But, we must first be a little careful in trusting the tinysa. About 10 years ago, I built a spectrum analyser. The amount of shielding it needed was insanely high. See?. The spectrum analyzer needs to be carefully shielded at all stages to be useful, esp with a transmitter testng. The signal from the antenna terminal is reduced at least 60 dB before being applied to the spectrum analyzer. On the other hand, there are many other ways that the RFI can leak from the radio into the spectrum analyzer at any stage and produce these heterodyne beats. The TinySA is a single PCB with mixed signals without any shielding.?

The zbitx is a superhet radio with a number of oscillators apart from just the Si5351. There is a 12 MHz crystal on the WM8731, a TCXO at 25 MHz, many spurs coming out of the RPI, the display and the Pico. However, the diode mixer of the zbitx was chosen as a doubly balanced mixer that has far fewer spurs than the Kiss Mixer of the sbitx. This is borne by direct measurements?as well. The uBitx too uses exactly the same front end (of diode mixers driven by an IF amplifier) in transmit, it has acceptable spurious emissions.

For a proper testing, you need a very well shielded or a professional quality instrument. Both my spectrum analyzers, the specan and the Rigol DSA-815 are those. My tests were done on both.?

I have also noted a correlation?between the bias current and the spurs. Here are my settings on the on the test bench with the DSA-815:

The output of the TX is connected to a dummy load made from 20 x 2 W, 1K carbon resistors. A sniffer consisting of a 100K in series with a 50 ohms is used to tap the input into the DSA-815.?

The Spectrum Analyzer is set to 30 KHz resolution bandwidth, span of 70 MH, the zbitx mode is set to CW, straight key. These are the recommendations of the ARRL's testing lab.??

I will repeat these tests and post the pictures tomorrow.

On Wed, 16 Apr, 2025, 6:57?pm Lee via?, <kx4tt=[email protected]> wrote:

GM Gordon,?

I still believe it could be a power issue - the 3.3v regulator does not appear to be properly filtered per the manufacturer Datasheet. The other possibility is that the output of the 5351 is loaded too much under transmit conditions.?The effect could even be combined.?One or both of these problems could occur only when the circuit is loaded (transmitting).? I'd measure the 3.3V regulator output under both receive and transmit conditions. If that varies more than a little bit, there may be an issue on the regulator side. Substituting the 5351 with a signal generator could give an indication if the 5351 might be involved.?

I suspect a combination effect.?

Lee KX4TT

On Wednesday, April 16, 2025 at 08:27:27 AM EDT, Gordon Gibby KX4Z via??<docvacuumtubes=[email protected]> wrote:

Well that tells us even more..

Not due to the power supply.

Unlikely to be produced BEFORE the crystal filter (it would screen it out)

Has to be coming in AFTER the filter

Dependent on the existence of some desired signal level for existence.? Not just being picked up "as is" from another wire.? ? ? Hence, possibly an IMD product with some other signal in the box.? ?

Gordon KX4Z