SB II #112 nearly done / searching the last building mistakes

Sascha Bohnet | DL5SMB

Hi everybody,

Though it took nearly a year now, I am finally almost done with my  SBI kit. Thanks you Steven for this really great ( or better small ) highly portable and multifunctional tranceiver.  I will use it a lot.

I first had some problems receiving anyhing but a rattling noise, but noticed only shortly after that I had wrongly turned X7 around 90 degrees. No wonder  After changing this the  RX worked immediately and I was happy.

The transmitter still has a problem, which I will try to pin down with the voltage chart shortly.

With the bias poti turned fully Acc , the current is only 120m  during transmitting. If I adjust it , I can turn it up quickly, but that does not solve the problem. Somewhere 30 mA are missing, I could imagine in the driver stage or so. I will measure  power tomorrow, when I got more time - the MOD LED does not flicker either, whatever position the mod poti is

I hope I will have sorted it out soon - I will report in again:-)


Don, ND6T

It doesn't seem that it is that far off. The adjustment of the bias trimmer is very touchy. A very careful nudge should bring it up another 30 milliamps. As for the lack of modulation, that is something to trace down.
Congratulations on your progress! Yes, a wonderful rig! 73, Don

Sascha Bohnet | DL5SMB

As I am working myself through the voltage chart now... might it be that not everything there is correct?

In the RX chart at Q20 I am measuring 3,3V instead of 5V Gate voltage - without a microphone connected. I see the refernce
about audio mic input level, but this seems just a bit weird.

Nevertheless the next big difference I noticed - at least compared with my rig - is that at Q3 Collector I am always measuring 13.2 Volt.
RX or TX. Is that right? If I am interpreting the schematics the right way, I think this is as it should be, or am I reading it wrong?

Don, ND6T

You are correct about the voltage on Q3 collector. It's being fed constantly through T1, R12, L10, and F1.
The chart shows 5 V on the drain, not the gate. That voltage is wrong since it is being supplied through the PTT port of the controller which, in turn, is supplied by 3.3 volts. Therefore it should measure 3.3 volts. That same line feeds the PTT and key connections through R52 so you should also get 3.3 volts on those terminals in the initial idle condition.

The gate should measure at nearly open, perhaps 40 millivolts or so if anything. When the PTT (or key) is closed that will measure about half a volt without audio.

No, the note is cogent. The voltage on the gate will change with audio level or state of the VOX_EN bus.

When you close the key or press the push-to-talk button on the microphone does the TX (D3) indicator light? What are you using for a microphone?

Now that you have concluded the voltage checks it might be the time to begin signal tracing. Do you have an oscilloscope? Or perhaps an audio amplifier for signal tracing? A signal injector of some sort? An oscilloscope is handy but not necessary since simple gear will get you through this sort of thing.

73, Don

Sascha Bohnet | DL5SMB

Sorry for the really late reply. I had been sick and was really busy getting well again (you know what they say about the "man flu").

The TX led does light, if I switch to transmitting and I am using a electret capsule. But to try to rule out microphone problems, I use CW for testing.

I have an oscilloscope and tried to get an understanding of what might be wrong, but was not really successful.

But first again to the voltage chart, where I noticed some other thing in TX mode:

The Drain of Q2 should be 2,4 Volt  i think (instead of 5 Volt) , like the Base of Q1, right? The source of Q2 is 5 Volts.
I think there is a mixup in the chart, probably due to the pinout of the MMBF4416.

However i noticed that if i try to measure the voltage of the Gate of G2, this transistors seems to simply turn on and my power supply
needs to limit the current. Is this a usual behaviour?

I know that this is a special amplifier, but I wonder if simply applying the measurement tip, should trigger this. Seems a bit odd.

Now to the oscilloscope
First thing to check should be the final I think.

The voltage of the Drain of Q23 is 13,27 Volt in RX and 13,17 Volt in TX (both Voltages taken with the multimeter).
What is curious is, that the voltage seems to drop while transmitting. In the chart it seems to increase.

This is the picture of the signal at the Gate Q23. 
and this at the Drain Q23

So it looks at least as if the Mosfet is amplifying. However after the second toroid (L9) the signal gets attenuated again
It looks like this at the BNC jack

L9 seems to be okay, I desoldered it and checked its inductance which is as it should be (I don't remember the value anymore, i calculated it and compared).

Any ideas what might be wrong?

73 Sascha

Don, ND6T

I'm glad that you're getting back to health!
Yes, looks like you found another anomaly in voltage chart. The odd gate condition (Q2) is probably an artifact of the probing. But do check the DC resistance to ground and insure that it is close to zero ohms (a check of the connections through L1). The drain of Q23 should pull down slightly during transmit because of the ohmic losses of the supply chain. The photos of the signals seem like they might be OK but the frequency response of the probe and 'scope make it look cleaner than I would expect. Also, it would be more instructive if the voltage range (or estimated peak-to-peak voltage) was included. I assume that you were measuring with a 50 ohm load right at the BNC (not more than a few cm distant)? May I also assume that you are using a X10 probe? What voltage is your supply? What current is being drawn while transmitting a CW carrier? This would give us an idea of what power is being presented at the output load and the efficiency of the transmitter. What band are you on?
From what I can see, it appears to be working a bit at least.
73, Don

Steven Weber


Be sure to calibrate. Otherwise the BFO can be off enough to affect power output since It’s not lined up well with the pass band of the IF filter until that is done.





Sascha Bohnet | DL5SMB

I had not too much time yesterday - but I can at least answer some questions.

I built the SBII for 20M, so every test is run on 14.160 Mhz (start frequency). I run it with 13,6 Volts and the current drawn is 160mA when sending a CW signal. Like I wrote earlier, if the BIAS poti is turned fully ACC its 130mA. So I adjusted it to 160mA which is the same value for CW and SSB. However SSB can't be seen on the scope at all.
The RX draws only 40mA. And yes I am using a X10 probe with a dummy load (10 cm cable).

The VFO is calibrated , the BFO not yet. Need to figure out how I do this best......... I read somewhere for filter adjusting using an audio spectrum analyzer is one of the best ways.

Sascha Bohnet | DL5SMB

I am just coming back from playing with the scope, trying to measure the peak-peak voltage.....

I am a bit confused as it seems as if the probe does not attenuate by factor 10 but amplify by 10.

I used  the  5V DC voltage from U1 as reference, switched the scope to "DC" and "0.1 V per Division" and saw the line go up 5 boxes - so one box should be 1 V then

So after switching back to "AC" I measured the output of the final again - about 1.6 boxes peak-peak  - so I assume that means 1.6 Volts.
This would also not contradict the pictures I sent yesterday - there you can the that the scope channel A was set to 20  mV (which makes magically 200mV) per division. There were about 8 boxes Peak-Peak visible, which would also make 1.6V.

This seems all way too low in my eyes.

Maybe the input to the final is just too small. However I can't see much of a signal at the drivers /  bases of Q3 and Q4.

By the way - I did measure the DC restitance of the gate of Q2 to Ground through L1. It's about 0.77 Ohm.

And another point: Is it possible that Pin 4 and 5 of U3 got mixed up in the schematic? I first thought the mixer was soldered in the wrong way, but noticed then that turning it would not make it like in the schematic.


Don, ND6T

Yes, U3 pins 4 and 5 are swapped in the schematic. See my message posting #610.

Indeed it would be best to resolve the confusion before attempting to repair the circuit problem. I am no help there! I'm not familiar with your oscilloscope. If it does not automatically compensate for the probe change from different scales (some do that) then you need to spend some time experimenting with it and becoming comfortable. Check to ensure that the scales track (use your calibration output to see if the measurement at 1 v/div is the same at .1 v/div, etc.). Be sure to frequency compensate your probe.
Looking at your photo of the voltage across the BNC jack it looks like the positive peaks reach 1.6 divisions and the negative peaks 1.8 divisions. That would be about 3.4 divisions peak-to-peak, right? At 20 millivolts/div and a 10X probe then that would be about .68 VPP (.2 X 3.4 ). That would be about a milliwatt output. We would rather see 50 VPP there.
Also at issue is the bandwidth of the instrument. Is it accurate at 14 MHz? Additionally, there is a bit of noise that widens the trace, so these are approximate values, good enough for comparison level checks. That is what is important.

So your levels are really low where you measured. Check at U3 pin 4. Check the level at pin 6. Work your way back.

Ensure that the signal that you are measuring is the expected frequency. Measure the period. That makes an oscilloscope very useful.

The most likely problem is a solder connection. They can be deceiving. With a continuity tester (or ohmeter) probe the pin where it just leaves the package and place the other probe at a point removed from that area somewhere (like a component, via, or pad) that should be directly connected. Try to test the continuity without accidentally pressing the pin down onto the pad while testing, therefore closing the connection that you are testing for an open. This is what happens to me on occasion. The connection looks good but isn't quite there.

Keep us posted! 73, Don

Sascha Bohnet | DL5SMB

Hi everybody and especially Don,

I am still working on this - I did some experiments in the meantime, completley rebuild the output LPF (which did nothing at all)
and then used another scope to check the frequencies.

I think I have a frequency problem, though I do not understand the schematic completly. I assume that my signal is only the VFO frequency which
gets filtered out again by the band / low pass filters so that nothing is left anymore.

First question ahead: Is it normal, that the VFO (Picture 1 and Picture 2)  and the BFO (Picture 1 and Picture 2) have different signal levels?
At the VFO I am seeing about 1,1V while the BFO seems to have around 0,5 Volt in SSB?

Is this supposed to be this way? I tried to resolder Pin 10 of the SI5351, and the 3,6k Resistor - but it made no difference.
Then I resoldered Pin 27+28 of the microcontroller - but this did not help either.

Also the question: Is there a difference between CW and SSB regarding the generation of the BFO signal?
I am not exactly sure anymore, but I think in CW the signal is so low that it is nearly undetectable - though the scope is able to detect a  frequency (picture)

However if i try to look at the signals at U3 when transmitting in CW it gets strange.

At 14.160 in CW I can see a 9,213 Mhz signal (the VFO I assume) at the output Pin4.  Should this not be already the mixed, final frequency?
(Pin 4 Picture 1 and Picture 2)

At Pin 6 the VFO is visible ( Picture 1 and Picture 2).  However I can nowhere detect the BFO Signal?
At Pin1 I also see see the VFO (sorry, no picture) and at Pin 5 the signal is too low to see anything useful.
Should not Pin 4 and 5 show the same signal, only in phase shift?

I tried to recreate the filter after U3 with LTSpice (never did something like this before). I I did it correctlya this would explain the low signal at Q2.
At 9,213 Mhz there seems to be an attenuation of about -30db (Picture).

Any ideas?

Don, ND6T

Hi Sascha,

We are getting closer.

According to the schematic:
The VFO is CLK0 on the Si5351 pin 10 and it connects to R45.
The BFO is CLK1 and that connects to R46
On 20m the BFO is 9.2169 MHz and the VFO is the sum of the operating frequency and the BFO (ie: @14.16 MHz SSB the VFO=23.3769 MHz).
For CW the BFO frequency is adjusted 600 Hz lower.

I do not have a 20m SB2 here that I can open and measure. I repaired one for a fellow but returned it. I may have one shortly if I am unable to help him on line. I can be more definitive if that happens.

I would not be concerned with the levels that you measure at the Si5351 outputs but you should have the 14 MHz signal on pin 4 but instead you are seeing the VFO frequency as it has passed through the crystal filter and into the mixer. The signal that you see on pin 6 of U3 is, indeed, the BFO.

You stated that the receive is working properly. Since U3 is the product detector in receive mode then it must be mixing at that time. So let's pursue the transmit path some more: While transmitting CW, check the signal at pin 1 of U3. That's the input. If it is very low then check at the junction of Q12 emitter, C47, and R28. That is the input to the crystal filter. You measured 9.213 MHz for that signal and that is 3.3 KHz lower than expected and would place it out of the filter band pass. If the SB2 has not been calibrated then that would explain it. If your oscilloscope can be trusted to be properly calibrated then that, too, should be considered.

If the receiver is NOT working properly then check that U3 pin 8 is +5 volts, pin 3 is well grounded, the bias on pin 1 and 2 are about equal, and that very little or no signal appears on pin 2 (it's a differential input, as I recall, and without C37 the signal will cancel itself).

The output of U3 pin 5 is bypassed to ground by C59. So, on receive when that output is audio, you will see detected audio (which that capacitor ignores). On transmit, that capacitor shorts the RF to ground so you shouldn't see much there.

I am pretty much "talking through my hat" here, Sascha. So forgive my inevitable mistakes. I am only guessing at things so accept my advice for what little it is worth ;-)

Don, ND6T


My apologies! U3 pin 6 should be the VFO! Looking at your photo I misread the finer trace as being high frequency, like the 23 MHz expected there, but reading the tool bar I see that it is the BFO frequency. Then I "caught" myself and corrected the text like the fool that I am.

This is probably the problem. The signals into pin 1 and pin 6 are supposed to mix and the results output from pin 4. Since both inputs are the same then that same frequency is output instead. I believe that the input to pin 6 should come from U4 pin 15 which is switched from (I believe) pins 1 and 2 of U4 by the TXEN (transmit enable) pins which are pins 9 and 10.

Check to see if that is happening. Check the solder connections on pins 9 and 10. Does the voltage change there when you key it? If not, follow the trace back to u8 pin 4 and check that connection.

There it is, my "inevitable mistake". :-)

Sascha Bohnet | DL5SMB

Hi Don - no problem.

As it was me who first confused BFO and VFO  I most probably caused your mixing up in the first place.

Once again its gets interesting now:

In SSB I can see the switching of the signals / frequencies at Pin 6 of both SA612 mixers and U4.
TXEN seems also to work in both modes, SSB and CW.

However I noticed now, that in CW the VFO does not seem to work correctly. Nearly no signal strength in CW (maybe 1 or 2 mV at the testpoint)
and the frequency seems to be the same or similar as the BFO.

Pictures CW : BFO 1, BFO2, VFO1 VFO2
Pictures SSB (USB) for reference: VFO1  VFO2

Seems like a firmware problem (why else should the VFO signal be present at the Testpoint in USB and not in CW?)
- though this should be impossible as it works in everybody else's rig.

I have flashed the latest firmware 1.2 two times and used two different microcontrollers, but it helped nothing.

Tomorrow I will check if the VFO also works in LSB. Maybe this gives some additional clues.

I read earlier that the SB 1 had earlier an issue with the BFO becoming the VFO frequency, 
induced by the BFO adjustment which called the wrong clock of the SI5351 - but this was solved.....

Or could this be a compiling issue - that something changed in the way the source file is interpreted?
I am using Arduino 1.8.9. with the libraries in the file section.

Maybe then my SSB mode is okay then and I have only an audio and CW problem? I am puzzled.

Clueless regards


Sascha Bohnet | DL5SMB

So I just checked LSB - no VFO signal either.
Then I took a look at the code , line 600.

There is:
VFOfreq=  IFfreq - OPfreq
Is this really correct?
I think this works for 80, 60 and 40 - but on 20 Meters the VFO frequency gets negative. I don't think it is supposed to be this way.
I tried to put it in an abs() clause but this did not work out.

So I just changed it to OPfrequ - IFfrequ.
And this way I finally get a high level  VFO frequency at the test point.

I did not have a look at the mixers yet - but this should be right I think. Or not?

Don, ND6T


I believe that you found it!
You might just add a line to the sketch to patch it to still work with the other bands like this third line:

 void PLLwrite() {
         VFOfreq = IFfreq - OPfreq;
           if((IFfreq - OPfreq)<1) VFOfreq = OPfreq - IFfreq; //Bug correction
         if (sideBand == 1) {VFOfreq =OPfreq + IFfreq;}
      if (Tx_mode == 1){VFOfreq = VFOfreq - 600;} 
I don't have a rig to test it at this time so please let us know if it works.

Sascha Bohnet | DL5SMB

Hi Don

- that really was the problem.

Now I get about 10W Output on CW .

Still no SSB modulation but that is another problem I have to investigate. I will look into this matter tomorrow.

About your code snippet:

I tried it, but it did not work. Switched back to my other MC with the "hard wired change" and it was okay again.
Most probably there is only a small issue with it ( just noticed a missing curly bracket). I'll test tomorrow again.

Thanks  a lot for your guidance - I had a steep learning curve on this.


Don, ND6T

Hi Sascha,

No, the code was a rookie mistake. Although the brackets are optional when used with a single statement like that, I forgot that we are dealing with an unsigned integer. So it was not only an inelegant solution , it wasn't the right approach. I'll ponder it later. Your fix was perfect for the singular solution. Go pursue the audio. You have it in retreat! -Don-

Sascha Bohnet | DL5SMB

Hi Don,

and - I am nearly done :-)  I found a bad solder connection on the capsule.

However I get lots auf audio on LSB und not so much on USB - or at least it looks that way when watching the power meter,
I could not listen to the signal yet with another radio.

Where I get 1W output by blowing in the mic on USB, I get 4-5W output on LSB.
The Mod LED does not flicker either, though I don't pay attention to that yet.

Is this maybe also caused by the redesign of the SB II from an original 20m USB to a 40m LSB radio?
Could this be related to the higher VFO frequency in USB (23,3 Mhz) than in LSB (4,9 MHz)?

I just tried to compare the source codes of the SB I and the SB II but did not find anything which could be causing this.

Or is this maybe a filter problem?

I noticed earlier that the radio is a bit deaf (only checked USB). It works, but signals need to be rather strong (S5) to be heard well.
I have not tried moving/adjusting the BFO signal to the IF FIlter yet - but if I look at the schematic, this could help as well with the transmitter, right?


Don, ND6T


I agree. The higher BFO frequency is considerably weaker. This might be due to the frequency response of U4 but I wouldn't doubt that the capacitive dividers (C60-62, 67) might also contribute. Well worth investigating.

Optimizing the signal frequency within the crystal filter pass band is a good idea. First calibrate everything before trying that, of course. Both of my SB2 did not need it but they are on 80m and 40m. The Upper Sideband setting on 40m results in an appreciable loss of sensitivity but I never use this rig for digital modes.

Good work, Sascha. Thank you for sharing your adventure. Please keep us posted if you find more 20m solutions.