Allison,

I'm afraid if you tested the way you described in your original post on this thread, you weren't testing for crosstalk.? I believe what you measured wasn't cross-talk at all, but rather ground bounce due to current starvation of the Si5351 output buffers.

I've attached the Si5351 datasheet.? On page 5, Silicon Labs gives the following specs:

Vddox, output buffer voltage => 3 ranges, 1.8V +/- 0.9V (5%), 2.5V +/- 0.25V (10%), and 3.3V +/- 0.3V (9.09%).? ?The Raduino uses a 3.3V supply for Vddox.

Iddox -? output buffer supply current per output => 2.2mA typical, 5.6mA max

Zo, output impedence @ 3.3V => 50 ohms

Note this is output impedence, which is very misleading.? The spec is not saying to use an output load of 50 ohms (Rload).? It is saying we need to drive thru a 50-ohm impedence to reach the destination Rload.? So we need to calculate the mimimum output load (Rload) that the output driver can drive.

Rload = Vddox/Iddox = 3.3 / 0.0056 = 589.3 ohms.? This is the minimum load the output can drive, as it gives the maximum allowed output supply current.? Exceed this spec and you enter current starvation on the power supply to the output buffer.

You stated that you used both a 50-ohm load (not impedence) and a 25-ohm load (not impedence) for your cross-talk testing.

With a 50-ohm load, Iddox = V/R = 3.3 / 50 = 66mA, which is more than 10x the maximum allowed output buffer supply current.

With a 25-ohm load, you subjected the part to 132mA Iddox!?

In both cases, the output buffers are so power starved that the chip internally can't supply enough Vddox to control the output buffers.? The result is ground bounce on all the outputs, not cross-talk.? This is typical behavior for digital chip output drivers whose output specs are violated in such a manner.? It also explains why you didn't see as much signal on the Si5351 outputs as you expected.?

The Zo spec was meant to say that the chip was designed to drive thru a 50-ohm impedence (either a 50-ohm coax or a 50-ohm micro-strip line on a PCB) to its load (Rload), not that you could use a 50-ohm or 25-ohm termination as the Rload.? An example of this appears on page 22 of the Si5351 datasheet.? Here the datasheet describes using the part to drive 2 clock outputs with 180 degree phase difference, but the example circuit shows driving the Si5341 clock outputs thru a 50-ohm coax to a voltage divider for the receiving circuit that presents an Rload of 511 ohms + 240 ohms = 751 ohms to each? Si5351 output buffer.? This gives an Iddox = 3.3 / 751 = 4.4mA, well within the 5.6mA max Iddox spec.

I suggest you repeat your cross-talk tests using an appropriate Rload value at the end of a 50-ohm line for each clock output and see how much cross-talk you see.

73,

Carl,? K0MWC

Hi;
The paragraph description on modifying the Beofeng mic connections to the uBITX is vague. Can a schematic please posted a schematic diagram as to how the mods actually are done. 'Thank you '73?

Great data Warren. Are you also going to replace the other 45MHz filter with the matching one you received ? Sure shows what all working together can accomplish........ Good Soup for everyone.?

73 Kees K5BCQ

Raj,

Doesn't leaving R27 in place give you better impedance matching ? Did you try cutting the trace in between and just putting the filter between C22 and R27 ?

73 Kees K5BCQ?

Sorry Bill.? I didn't mention that I do set it with the PTT switch, then exit the setup mode.? This set has exhibited this symptom since day 1.? I thought changing out the Arduino would have helped.? Before that I was adding caps to the reset.? Thanks anyway.

Bob

Hi Bob,

I do not see the part where you save the setting after you have set it. Press the push to talk button to save it while you are still in the calibration mode. Otherwise it does exactly as you describe and will be back to the original next time you turn it on.

Before you leave the calibration and after you have it where you want it, press push to talk. Not your CW key. Not later after you have left the calibration mode. Before you leave calibration.

73,

Bill KU8H

--
bark less - wag more

When the set is first powered on, there is a tinny, high frequency audio.? When you enter the setup mode and enter calibration, pressing the menu button, the audio is normal tone. Then I adjust the calibration level and set it, then exit, all is well.? Until I turn it off. I have tried other software with no luck. So I re-loaded the original factory software. I thought with all the issues you have seen with the various ubitx versions that you people may have come across it. Thanks anyway.

Bob

开云体育

Wow! ?Very nice work, Warren!!!! ?Extremely helpful for many, many people!!

Warren,

Try with both matched filters.. that should get further reduction.
I don't have a pair right now..

Raj

At 08-09-18, you wrote:

Hi Sunil,

Transformer isolation is VERY useful for eliminating ground loops and Galvanic isolation for the audio circuits. should you design a revised digital interface board provisions for installing audio isolation transformers would be much appreciated.??

-Justin N2TOH?

And finally, finally (for real this time) a pic of the filter in place. The center ground lead of the filter is not visible but it bend under the case and picks up the R13 ground.

Good work !

I have been using 3 to 31MHz as my span as you will see these dancing deverishes darting in and and out of the lower end
of the spectrum.

I got better results by replacing C22 with filter and leavening R27 intact!

Have a nice meeting and catch you later!

Raj

At 08-09-18, you wrote:

And finally, here is a wider span showing 15 through 10 meters harmonic performance. My unit has the onboard filters completely removed and this scan was made with an external 30 MHz LPF plus the R27 filter.?

A clarification on the IMD comment in the last post. IMD was unacceptable at 100 mVrms audio drive before the filter was in place. It was not clear to me at that time whether the IMD was being introduced in the stages prior to the filter or after. If it was after, then increasing the audio input may be ok because it simply compensates for the insertion loss of the filter. If, though, the IMD was introduced in the SSB generator or the stages before that, then these pix may be a cure for the spurs but still have unacceptable IMD. Further testing is required to figure that out.

It turns out I did have time for one pass of spur measurement. Here are screens for each of 17 through 10 meters. Conditions of the test: Audio input: 100 mVrms, 1.5 kHz tone. RV1 adjusted in each case for 2 watts output.

So far it looks really good with the eBay filter in place of R27. No other changes.

Yet to be verified: 100 mV audio drive, without the filter in place, gave very unacceptable IMD performance. It may well be in the case of the added filter that the stages preceding the filter have enough dynamic range to work at that level and it is simply compensating for the insertion loss of the filter itself. That can and will be confirmed with two tone IMD testing that will indeed have to wait. Also, we simply have to listen to the recovered voice quality and decide if it is adequate. Adding the filter has introduced low frequency rolloff for LSB and high frequency rolloff for USB. The change is less than 6 dB and may not be objectionable but that will be a subjective judgement.

Overall, very promising as a cure for the spurs.

Raj

We may be talking "past" each other or I may not understand the intent. But as I see it the added filter is indeed a passband filter for the 45 MHz LSB/USB signal.

Referring to the attached scan: Blue trace is the filter passband measured out of the circuit with a 50 ohm in/out. Purple is measured with a 10x scope probe on the S/A input and attached to the same filter in place in the circuit. Purple is the input side of the filter. Yellow is the output side of the filter. So the screen reads top to bottom: Purple input, Blue filter, yellow filter output.

The two "humps" on the purple and yellow traces are the LSB and USB signals respectively with the audio input being swept from 0 - 5 kHz. Due to the way the 45 MHz SSB signal is generated and later heterodyned to operating frequency, the right side of both humps is the low audio frequency and the left is the high frequency. While the sweep was done from 0 - 5 kHz, the humps themselves represent only about 300 Hz (on the right side) to 2.7 kHz on the left. The 'scalloping' evident on the purple trace comes from the response curve of the 12 MHz crystal lattice filter in the SSB generator.

My comment about the shape of the newly added filter is evident by comparing the yellow output humps to the purple input. The right side of the leftmost hump and the left side of the rightmost hump are attenuated by the filter shape. This will have the effect of attenuating the low audio frequencies on LSB and the high frequencies on USB.

Some good news is the eBay filters appear to be a good match for the SSB generator as both humps are centered in the filter passband. So I think I am in good shape to do some testing on the effect the added filter has on the spurs. That will have to wait until the afternoon as our club technical breakfast is in about an hour.

Glenn,

The intention of the filter is not the passband. I wanted to get rid of all the HF 3-30MHz that was
being picked up from messing with the mixer and BiDi amp. So a 45M30 say 30KHz filter may be
fine or better if any center freq is off.

The need as I see it is the ultimate rejection beyond 1MHz is -90db which is great!

Raj

Gosh, I have received now the comment by Ray, so mine is a kind of repetition of it.

You measured the filter at 50ohm but the terminating impedances should have been 650ohm/3pf. I am not sure whether the impedance of the stock filter is similar and the matching in the Ubitx might be adequate as is.