|
Greetings from North Carolina.? I'm a uBITX owner and have lurked on this group for some time.? Lots of good discussion.
I have built a fairly close copy of Farhan's Specan analyzer, plus the Two-Tone generator adapted to 40m.? The calculated IIP3 with the generator tones at -20 dBm is around 5 dB (distortion products down around 50 dB), much worse than the expected 15 dB.? Experimentation has shown the culprit is the U2 amplifier which is in the signal chain following the first mixer.? I couldn't find a reliable source of the MAV11 so substituted a GALI 6+.? The datasheet claims a low frequency gain of 12.2 dB and an OIP3 of 35.5 dB at 1 GHz.? Removing/bypassing U2 improved the measured IIP3 to 14 dB which is near the capability of the ADE-1 mixers.
Does anyone have a suggestion of where I went wrong the GALI-6+???
--
David KV4FP
|
David
I cannot easily see from here - but possibly you are only looking at the amplifier ITOI by itself, and not its impact on the chain?
Farhan had placed a cumulative calculation tool at his website (and I can also assist if you email me).? The presence of gain between stages does impact the overall calculation.?
BTW I don't have any linearity issues with my uBitx, but I don't have a huge many element beam high in the air.? Its fine for many of us with normal one and two element antennas.?
On you nice spectrum analyzer - keep the trick in mind of adding 3 dB attenuation (or whatever you have) in various spots to insure the instrument itself isn't limiting your measurement.? Nice work all around and thanks for sharing
Curt wb8yyy
|
Curt, Thanks for the quick reply.? To test the Two-Tone generator, I fed its output through an attenuator to my uBITX and tuned to the expected distortion product frequency.? I couldn't hear anything so believe the generator and uBITX are behaving well.? I'll take a look at Farhan's calculation tool. Here is a scan with the GALI-6+ in the chain.? I wonder if the GALI-6+ OIP3 is worse at lower frequencies (since it was specified at 1 GHz).  -- David KV4FP
|
David
when you put this 12 dB gain in front of that mixer, the effect is to reduce the IIP3 of the mixer by 12 dB.? if you compare the input of the mixer (reduced by 12 dB) to the output IP3 of the galli 6 - you should see that the mixer is dominating the linearity of the network. (which is not surprising).? the goodness of this amplifier is that you are adding some gain (that I suspect will be nice on 28 MHz) without degrading the front end linearity, except for the 12 dB gain that is added.? in effect you are making the mixer handle 12 dB larger signals, so this is why your IIP3 measurement is coming out in this way.? see if it represents a 12 dB reduced IIP3 when adding the amplifier?
cool stuff this RF - you have constructed a wonderful instrument.
Curt
|
I would suggest that you put in a 6db pad in place of the mav11. It will work all the same. The noise figure will be much less but the iip3 will be much better.
David
when you put this 12 dB gain in front of that mixer, the effect is to reduce the IIP3 of the mixer by 12 dB.? if you compare the input of the mixer (reduced by 12 dB) to the output IP3 of the galli 6 - you should see that the mixer is dominating the linearity of the network. (which is not surprising).? the goodness of this amplifier is that you are adding some gain (that I suspect will be nice on 28 MHz) without degrading the front end linearity, except for the 12 dB gain that is added.? in effect you are making the mixer handle 12 dB larger signals, so this is why your IIP3 measurement is coming out in this way.? see if it represents a 12 dB reduced IIP3 when adding the amplifier?
cool stuff this RF - you have constructed a wonderful instrument.
Curt
|
David, It is also possible that the IMD is being generated in the two tone generator! To know this, insert an attenuator pad between the generator and the specan. You should see the imd go down by twice the attenator value (3db of imd down for each db of attenuation). I discovered this problem with my early two tone generators.?
I would suggest that you put in a 6db pad in place of the mav11. It will work all the same. The noise figure will be much less but the iip3 will be much better.
David
when you put this 12 dB gain in front of that mixer, the effect is to reduce the IIP3 of the mixer by 12 dB.? if you compare the input of the mixer (reduced by 12 dB) to the output IP3 of the galli 6 - you should see that the mixer is dominating the linearity of the network. (which is not surprising).? the goodness of this amplifier is that you are adding some gain (that I suspect will be nice on 28 MHz) without degrading the front end linearity, except for the 12 dB gain that is added.? in effect you are making the mixer handle 12 dB larger signals, so this is why your IIP3 measurement is coming out in this way.? see if it represents a 12 dB reduced IIP3 when adding the amplifier?
cool stuff this RF - you have constructed a wonderful instrument.
Curt
|
Farhan and Chris, With the GALI-6+ still in place (it's U2, after the first mixer), adding a 6 dB attenuator between the generator and Specan caused the distortion product to fall 17.9 dB.? But the tone level only dropped 0.8 dB.? I could see some harmonic distortion in the IF amplifier output with a scope, suggesting significant non-linearity in the signal chain. Then completely removed the U2 GALI-6+ (and re-calibrated the firmware power meter constants).? There was already a 6 dB pad after the first mixer, so took the LO module output from that pad.? Results are shown below and are much improved.? With the generator directly connected to the Specan, the IIP3 is now -20 + 68/2 = 14 dBm (I think?).? When a 6 dB attenuator is inserted at the Specan input, the measured tones fall 5 dB and the distortion products fall below the noise floor. This seems to vindicate the two-tone generator though don't have enough dynamic range for the 6 dB attenuator test.? This seems like usable performance, though I may search for a replacement for the MAV/GALI to improve the noise figure. Thank you for your help, -- David KV4FP 
|
Yay!? This also means that we cant do away with the post mix amp altogether... Maybe. We need 80 db of usable, distortion free range for good measurements. Let's see how we will get there. With 14 dbm iip3 of the specan, this would mean, the most powerful signal you can feed into it is 14dbm - 80d /2 = 14 - 40 = -26 dbm. The noise floor will have to be -26 dbm -80db = - 106 dbm. At 20 Khz bandwidth and 20db noise figure, the projected noise floor will be -174 + 43 + 20 = -111 dbm. You should be there. The scan showa that, instead, your floor is -90dbm. You should shoot for that extra 10 db dynamic range. The mav11 could have given you that extra sensitivity without compromising on the iip3. Infinion has some mmics you could use. Alternatively, a discrete BFR106 feedback amplifier could do the trick (with 20mA collector current). It is my favourite transistor now. Allison pointed it out to me. The original software is very dated. A better written software or a nextion display will make it much more useful. - f .
On Wed 9 Jun, 2021, 3:49 AM David Cooper KV4FP, < coopers04@...> wrote: Farhan and Chris, With the GALI-6+ still in place (it's U2, after the first mixer), adding a 6 dB attenuator between the generator and Specan caused the distortion product to fall 17.9 dB.? But the tone level only dropped 0.8 dB.? I could see some harmonic distortion in the IF amplifier output with a scope, suggesting significant non-linearity in the signal chain. Then completely removed the U2 GALI-6+ (and re-calibrated the firmware power meter constants).? There was already a 6 dB pad after the first mixer, so took the LO module output from that pad.? Results are shown below and are much improved.? With the generator directly connected to the Specan, the IIP3 is now -20 + 68/2 = 14 dBm (I think?).? When a 6 dB attenuator is inserted at the Specan input, the measured tones fall 5 dB and the distortion products fall below the noise floor. This seems to vindicate the two-tone generator though don't have enough dynamic range for the 6 dB attenuator test.? This seems like usable performance, though I may search for a replacement for the MAV/GALI to improve the noise figure. Thank you for your help, -- David KV4FP 
|
It's interesting that the noise floor has been around -93 dBm, with and without the first mixer post amplifier.? Will explore this some more.? I wonder if the log amp is setting the floor now.
The BFR106 seems like a good candidate.? The only concern might be power supply (Vceo is 15-16V and specs are at 8V).? What voltage have you been using?
I'm mostly a HW guy but an improved user interface would be nice.? I had thought of changing the pot to an encoder and some other improvements.? More computing power and a graphic display would be nice.? I've thought of pairing the Specan with a Rasberry Pi.
I'll order some BFR106's and let you know how I make out.? Thank you for helping,
--
David KV4FP
|
By the way, did I buy the proper version of the Si570?? I ordered the?XTAL OSC XO 100.0000MHZ LVDS SMD.? Output drive seems low.
Thanks,
--
David KV4FP
|
If I were doing it all over again, I would use an Si5351. That way, both the oscillators (plus the tracking generator too) would be taken care of. The low noise figure could be due to low gain in the front-end More gain in the later stages just amplifies the noise of the front-end along with?the signal. - f
On Wed, Jun 9, 2021 at 10:58 PM David Cooper KV4FP < coopers04@...> wrote: By the way, did I buy the proper version of the Si570?? I ordered the?XTAL OSC XO 100.0000MHZ LVDS SMD.? Output drive seems low.
Thanks,
--
David KV4FP
|
Farhan,
This the same transistor BFR106 I used when improving the earlier
BITX!
David this transistor work very well!
Raj
At 09/06/2021, you wrote:
toggle quoted message
Show quoted text
It's interesting that the noise
floor has been around -93 dBm, with and without the first mixer post
amplifier.? Will explore this some more.? I wonder if the log
amp is setting the floor now.
The BFR106 seems like a good candidate.? The only concern
might be power supply (Vceo is 15-16V and specs are at 8V).? What
voltage have you been using?
I'm mostly a HW guy but an improved user interface would be nice.? I
had thought of changing the pot to an encoder and some other
improvements.? More computing power and a graphic display would be
nice.? I've thought of pairing the Specan with a Rasberry
Pi.
I'll order some BFR106's and let you know how I make out.? Thank you
for helping,
--
David KV4FP
|
toggle quoted message
Show quoted text
Farhan,
This the same transistor BFR106 I used when improving the earlier
BITX!
David this transistor work very well!
Raj
At 09/06/2021, you wrote:
It's interesting that the noise
floor has been around -93 dBm, with and without the first mixer post
amplifier.? Will explore this some more.? I wonder if the log
amp is setting the floor now.
The BFR106 seems like a good candidate.? The only concern
might be power supply (Vceo is 15-16V and specs are at 8V).? What
voltage have you been using?
I'm mostly a HW guy but an improved user interface would be nice.? I
had thought of changing the pot to an encoder and some other
improvements.? More computing power and a graphic display would be
nice.? I've thought of pairing the Specan with a Rasberry
Pi.
I'll order some BFR106's and let you know how I make out.? Thank you
for helping,
--
David KV4FP
|
I've ordered some BFR106.? Looking forward to building a post-mixer amp.
The suggestion of the Si5351 is interesting too.? In my Specan, I packaged the Si570 in the first mixer module thinking to minimize the distance between oscillator and mixer.? But I forgot about the temperature rise generated by the MAV/GALI buffer amps in a small enclosed box.? I'm thinking about switching to the Si5351, locating it near the Nano/display, and using it to generate the two oscillator clocks similar to the uBITX architecture.? I'm also hoping that this will generate stronger drive to the mixers.
--
David KV4FP
|
I saw a lot of noise from arduino into the signal path. I had to use very aggresive shielding.
On Thu 10 Jun, 2021, 6:53 PM David Cooper KV4FP, < coopers04@...> wrote: I've ordered some BFR106.? Looking forward to building a post-mixer amp.
The suggestion of the Si5351 is interesting too.? In my Specan, I packaged the Si570 in the first mixer module thinking to minimize the distance between oscillator and mixer.? But I forgot about the temperature rise generated by the MAV/GALI buffer amps in a small enclosed box.? I'm thinking about switching to the Si5351, locating it near the Nano/display, and using it to generate the two oscillator clocks similar to the uBITX architecture.? I'm also hoping that this will generate stronger drive to the mixers.
--
David KV4FP
|
Here is a picture of my Specan-- have already begun modifying it to use the Si5351-- thus some loose wires.? All the modules are shielded except the UI/Arduino, though some room is reserved for a shield if needed.? The chassis is 8 7/8 x 11 1/2 inches which turned out a little tight. The notes for the uBITX '5351 routines say that the maximum frequency is 109 MHz but the Specan needs 182.? Any thoughts on what library or settings to use?  -- David KV4FP
|
The better routines are in the Antuino git in?. it works cleanly upto 280 Mhz. Thanks to Hans' help. There is a little trick I used in Antuino. You can use the third harmonic of the local oscillator to extend the range (the usual caveats of images and spurs) to well beyond 500 Mhz. You will have to remove the Low pass filter. Another neat hack would be to use a 3.5 inch or a 2.8 inch tft display in place of the LCD, you will have an fully container specan! To save yourself time, you could order a Raduino that has the entire Arduino+Si5351+TFT built in. - f
On Sat 12 Jun, 2021, 4:33 AM David Cooper KV4FP, < coopers04@...> wrote: Here is a picture of my Specan-- have already begun modifying it to use the Si5351-- thus some loose wires.? All the modules are shielded except the UI/Arduino, though some room is reserved for a shield if needed.? The chassis is 8 7/8 x 11 1/2 inches which turned out a little tight.
The notes for the uBITX '5351 routines say that the maximum frequency is 109 MHz but the Specan needs 182.? Any thoughts on what library or settings to use?
--
David KV4FP
|
Thank you for your help.? Have started installation of the 5351-- have an Adafruit board on hand.? Some other priorities have intervened but will report back when there is some progress.
--
David KV4FP
|
You can reuse the antuino code (v2.2) most of it is there, minus the UI and the second oscillator (which is fixed anyway) - f
On Tue, Jun 15, 2021 at 3:42 AM David Cooper KV4FP < coopers04@...> wrote: Thank you for your help.? Have started installation of the 5351-- have an Adafruit board on hand.? Some other priorities have intervened but will report back when there is some progress.
--
David KV4FP
|
Raj vu2zap
