Si5351 Noise Filter as in Progrock.

In the ProRock Hans uses a noise filter to smooth the power to a modified Si5351 Synth board. This is because some regulators may produce LF noise which could then be added to the output of the PLL in the Si5351.

To maintain the Synth unit's compatability between modules the power filtering as used in the ProgRock may be installed on the Synth board. Page 7 of the assembly manual suggests this.

This filtered Synth unit can then be used with the U3 and VFO.

It looks as if the OCXO could be modified in a similar way. But a very neat and compact method would be required to get it inside the cover. Maybe the filter components could be fitted somewhere off the synthesiser.

The ProgRock is made without the smoothing components similar to its use for the OCXO. On page 6/7 Hans discusses power options, I connected the 5V from IC2 of the ProgRock to pin 9 to the Synth. Note more consideration and maybe modification is required to make an OCXO and the basic Synth unit compatible. It will work my way provided the PLL and Heater supplies are NOT connected when a Synth unit is used.

The filtered Synth board can be built without greatly increasing it's overall dimensions but this would be very difficult with the OCXO.

I do not recommend modifying a built unit. While it is possible there are risks of damaging the board and creating more problems.

It is my opinion that the filter will normally make very little difference. I think only the most demanding application and careful measurment would show any improvement. But, here goes!

Warning:- I believe these steps are correct but you must understand what I have done and verify for yourself.

Synth Board:-

C1, R5 and R6 are not used. (Later I put a surface mount C1 under the board because I thought it would do no harm, possibly good?)

1) Q3 is inserted in IC2 position. The track from it's base is cut above the board. (Was IC2's output to R4/Q2 source and the rest of the circuit.) The later board with the TCXO pads will be different, you will need to find this track.

2) Q1 emitter will fit in the same hole as Q3 emitter. The other leads need to be carefully bent up.

3) Q2 emitter and collector fit in Q1 collector and base holes the base lead is bent up to join Q1 collector.

4) Install the crystal, carefully arrange the transistors making them fit as closely to the board as possible avoiding any contact with the crystal case. Solder the leads under the board.Q2 base may be connected to Q1 collector.

5) Install the 10 pin strip on the top edge of the board.

6) Mount R3 vertically on pin 10. R5 vertically on pin 1.

7) C2 + goes in "C1" hole. -ve to the corner ground pad. C4 -ve also goes here. C4 + and R4 connect to the top of R3.

8) The other end of R4 goes to the top of R5 and C5 + with C5 -ve to pin 5

9) Test. Connect a short wire to the joined emitters of Q1,3. This is the output. Connect a load of about 100 ohms to this and ground. This is to load the circuit as the Si5351 is not yet connected. Connect a voltmeter to this point. Apply 5V to pin 10, ground to one of the grounded pins, The voltage at the emitters of Q1,3 should be about 3V5 if R5 is the specified 39K.

(I reduced this voltage to about 3V2 by making R5 27K. This is because I used a TCXO with a maximum rating of 3V3.)

Fit the other components and 10 pin strip.

10) Connect the wire from the emitters of Q1,3 to the junction R4/Q2 gate. This point also joins to the other 3V power connections.

If a crystal is fitted care must be taken to avoid shorts. The track is cut just in front of the R6 hole.

I used a 27K for R5 on the left

The orange wire sends filtered power to the junction of R4/Q2 gate.(Marked red.) (Also in my application I connected the isolated crystal connection at bottom left to Q1,3 emitters, same track as the orange wire to power my TCXO.) For the photo I left the transistor leads long to show two leads in each hole.

The minute Epsom TCXO is just visible. I cut the track from IC1 pin 3, used a SM capacitor for decoupling. The filtered power is connected to the now isolated crystal hole. The later board would make this easier. This 2x3mm 26 MHzdevice is the only one I know of that is really suitable for use for WSPR at the higher frequencies.

Here is an SMD filter, C2 on the left. Red 5V. Orange 3V5 out with the 39K for R5. I can just fit in a SM R5 but left it until later as if I wish I can trim the output voltage by altering it. As above 27K would give about 3V2 output at 30mA.

Transistors just about any low voltage and current ones should do. I do not think they are in any way critical.

At 25mm, 1 inch wide I made it as small as I could. This might be used with the OCXO when using the U3 or VFO.

Notes on my board construction methods, some mouse-derived rotary encoders at the bottom of the page.

Has anyone seen a transmission like this? 50% duty cycle + extension. Decodes every 2 minutes.

Now it looks like 100% with short TX periods....

Joe, VK5EI suggests 100% and cutting off when the over-temperature limit is reached. Fits!

The decodes show just how much error correction is built in to WSPR, only partial transmissions decodes like at 10.06, 10.10.

And that's on the old WSPR I use on 60m.