¿ªÔÆÌåÓý

  1. BITX20
  2. Topics

spur & harmonics fix, v2

 

i have further simplified the fixes:

1. change the resistors R26 and R46 to 220 ohms (from 470 ohms). this increases the darlington pair's standing current and decreases the Harmonic distortions.

2. the 90 MHz trap was difficult to tune unless you had a spectrum analyzer. Instead, we can use a Low Pass Filter instead. A 'T', rather than a Pi was used. This helps better in isolation or RX/TX path.?

3. The LPF consists of an L-C-L of 300nh, 47pf, 300nh.?

4. With the relays that are now used for ubitx,? the harmonics at 3.5, 7 and 10 are in under -45 dbc. we don't need the axicom relays.

I am attaching a hand-drawn circuit to explain this better.

- f

90MHz_LPF.jpg

 

Farhan, thanks a lot for the schematics and great for the minimal hardware upgrade you suggest. Can we use through hole components or SMD are mandatory? Can inductances also be made in through hole discrete components too? I am asking this as the local store where I usually get the components has a very limited supply of SMD components and no toroids, while they have a full set of through hole components.


Il 12/nov/2018 09:01, "Ashhar Farhan" <farhanbox@...> ha scritto:
i have further simplified the fixes:

1. change the resistors R26 and R46 to 220 ohms (from 470 ohms). this increases the darlington pair's standing current and decreases the Harmonic distortions.

2. the 90 MHz trap was difficult to tune unless you had a spectrum analyzer. Instead, we can use a Low Pass Filter instead. A 'T', rather than a Pi was used. This helps better in isolation or RX/TX path.?

3. The LPF consists of an L-C-L of 300nh, 47pf, 300nh.?

4. With the relays that are now used for ubitx,? the harmonics at 3.5, 7 and 10 are in under -45 dbc. we don't need the axicom relays.

I am attaching a hand-drawn circuit to explain this better.

- f
 

Farhan,

What do you mean by "the relays that are now used for ubitx... we dont need the axicom relays"?

Point 5 of the attachment still calls for replacing the relays with axicom relays.

I have a v3 board. Do I need to replace the lpf relays with axicom relays or no?
Regards?


Simon
VK3ELH
 

simon,?
our new relays have less leakage. they are within have sufficiently low leakage to keep the harmonics within the specs. the earlier boards will need to replace the relays.

- f

On Mon, Nov 12, 2018 at 2:22 PM Kelly Jack <kellyjack1968@...> wrote:
Farhan,

What do you mean by "the relays that are now used for ubitx... we dont need the axicom relays"?

Point 5 of the attachment still calls for replacing the relays with axicom relays.

I have a v3 board. Do I need to replace the lpf relays with axicom relays or no?
Regards?


Simon
VK3ELH

 

i have used 47pf standard leaded disc ceramic and two T30-6 coils. A 0.3uh can also be air wound but i am not sure if the lack of shielding will be a problem.

- f

On Mon, Nov 12, 2018 at 2:03 PM iz oos <and2oosiz2@...> wrote:

Farhan, thanks a lot for the schematics and great for the minimal hardware upgrade you suggest. Can we use through hole components or SMD are mandatory? Can inductances also be made in through hole discrete components too? I am asking this as the local store where I usually get the components has a very limited supply of SMD components and no toroids, while they have a full set of through hole components.


Il 12/nov/2018 09:01, "Ashhar Farhan" <farhanbox@...> ha scritto:
i have further simplified the fixes:

1. change the resistors R26 and R46 to 220 ohms (from 470 ohms). this increases the darlington pair's standing current and decreases the Harmonic distortions.

2. the 90 MHz trap was difficult to tune unless you had a spectrum analyzer. Instead, we can use a Low Pass Filter instead. A 'T', rather than a Pi was used. This helps better in isolation or RX/TX path.?

3. The LPF consists of an L-C-L of 300nh, 47pf, 300nh.?

4. With the relays that are now used for ubitx,? the harmonics at 3.5, 7 and 10 are in under -45 dbc. we don't need the axicom relays.

I am attaching a hand-drawn circuit to explain this better.

- f
 

Is there no 12mhz trap now?

uBitx v4 boards were shipping with the old relays, they now have new relays?
When did this change?
How can customers know if they have old or new relays?

The hand drawn schematic looks good enough for me, but the handwritten text is a bit botched.
1)? "Change R26 and R46 from 220 to 470"? ? ?No, from 470 to 220
3)? "Solder a 47 ohms leaded"? ? ?Some will need a bit more guidance here
5)? "Change the relays to AXI"? ? ?Change to Axicom?? What relays do you now ship?

For the two new 0.3uH inductors, I'd prefer to use 1206 surface mount parts.
Perhaps mounted to the back of the board manhatten style.

Anyways, looks very promising, and I appreciate you getting this out
as quickly as possible for others to test.??
Those that find the instructions confusing and/or don't have a spectrum analyzer
should hold off a bit longer.

Jerry, KE7ER



On Mon, Nov 12, 2018 at 01:06 AM, Ashhar Farhan wrote:

i have used 47pf standard leaded disc ceramic and two T30-6 coils. A 0.3uh can also be air wound but i am not sure if the lack of shielding will be a problem.
?
?
 

Ok to use axial inductors the ones that look like resistors instead of winding bulky toroids?

Their size would be more convenient when modifying existing UBITX PCBs..



73 Steve
 

I am a little wary of axial inductors. I would have easily recommended air wound coils. However, given all the RF floating around, i would recommend a toroid.

The 470 ohms to 220 ohms is jacking up the gain by a few notches to drown out the 12 mhz spur. Remember that the 12? mhz spur is not from the IF amp or the mixer, but from the Si5351's internal cross talk

On Mon, 12 Nov 2018, 17:12 G1KQH via Groups.Io <g1kqh=[email protected] wrote:
Ok to use axial inductors the ones that look like resistors instead of winding bulky toroids?

Their size would be more convenient when modifying existing UBITX PCBs..



73 Steve

 

I'm very happy to see that you are digging into this so deeply.

Raj found that replacing the toroids at L5,L7 with sm1210 shielded inductors reduced coupling,
so they may work well here.? But at least we have a baseline fix, others can try variants.

>? The 470 ohms to 220 ohms is jacking up the gain by a few notches to drown out the 12 mhz spur.
>? Remember that the 12 mhz spur is not from the IF amp or the mixer, but from the Si5351's internal cross talk

If jacking up that gain at Q22 and Q42 fixes it, then the 12mhz we're worried about
must be riding along on clk2.? But jacking up that gain may further overload the D1,D2 mixer.
Since clk2 is always above 45mhz, seems a low pass filter on clk2 would be the better fix here.

I have not made measurements of my on on this stuff.
Just basing this on Allison's reports that the signals going into the mixers on transmit
are already too high (and that the LO injection levels are too low).

If we could use the BFO to select USB vs LSB, then clk1 could be parked high
at 45+12=57mhz, not flipping between 57mhz for USB and 45-12=33mhz for LSB.
Any residual 57mhz is easily filtered by the 30mhz LPF if clk2 is showing some crosstalk from clk1 also.
However, the current scheme of BFO always low is probably good enough so long as LSB is not used
on 15,12,10m bands since the transmit LPF's should take out the 33mhz.?
Assuming the transmit LPF's are working properly.

Flipping the BFO like that doubles our chances of the 12mhz Nano resonator harmonics
beating with BFO harmonics, creating audio tones.? ?Assuming that's where the tones come from.
But that issue needs to be fixed anyway, we have enough reports of audio tones already.??
The newer Raduinos now have a socketed Nano, and with Nano's so cheap
that at least allows us to try a bunch of different Nano's out if audio tones are a problem.

Just pushing some thoughts out.
I'm good with what works.?

Jerry, KE7ER




On Mon, Nov 12, 2018 at 05:56 AM, Ashhar Farhan wrote:

I am a little wary of axial inductors. I would have easily recommended air wound coils. However, given all the RF floating around, i would recommend a toroid.
?
The 470 ohms to 220 ohms is jacking up the gain by a few notches to drown out the 12 mhz spur. Remember that the 12? mhz spur is not from the IF amp or the mixer, but from the Si5351's internal cross talk
 

jerry,

we should have used 11.052 MHz IF. if we were to start it all over again, i'd give my rigol's channel 1 to do that. unfortunately, there are others who share our misery, we must bear this with them for the sake of backward compatibility of the software. do you feel bill gates pain now?

- f

On Mon, Nov 12, 2018 at 9:49 PM Jerry Gaffke via Groups.Io <jgaffke=[email protected]> wrote:

I'm very happy to see that you are digging into this so deeply.

Raj found that replacing the toroids at L5,L7 with sm1210 shielded inductors reduced coupling,
so they may work well here.? But at least we have a baseline fix, others can try variants.

>? The 470 ohms to 220 ohms is jacking up the gain by a few notches to drown out the 12 mhz spur.
>? Remember that the 12 mhz spur is not from the IF amp or the mixer, but from the Si5351's internal cross talk

If jacking up that gain at Q22 and Q42 fixes it, then the 12mhz we're worried about
must be riding along on clk2.? But jacking up that gain may further overload the D1,D2 mixer.
Since clk2 is always above 45mhz, seems a low pass filter on clk2 would be the better fix here.

I have not made measurements of my on on this stuff.
Just basing this on Allison's reports that the signals going into the mixers on transmit
are already too high (and that the LO injection levels are too low).

If we could use the BFO to select USB vs LSB, then clk1 could be parked high
at 45+12=57mhz, not flipping between 57mhz for USB and 45-12=33mhz for LSB.
Any residual 57mhz is easily filtered by the 30mhz LPF if clk2 is showing some crosstalk from clk1 also.
However, the current scheme of BFO always low is probably good enough so long as LSB is not used
on 15,12,10m bands since the transmit LPF's should take out the 33mhz.?
Assuming the transmit LPF's are working properly.

Flipping the BFO like that doubles our chances of the 12mhz Nano resonator harmonics
beating with BFO harmonics, creating audio tones.? ?Assuming that's where the tones come from.
But that issue needs to be fixed anyway, we have enough reports of audio tones already.??
The newer Raduinos now have a socketed Nano, and with Nano's so cheap
that at least allows us to try a bunch of different Nano's out if audio tones are a problem.

Just pushing some thoughts out.
I'm good with what works.?

Jerry, KE7ER




On Mon, Nov 12, 2018 at 05:56 AM, Ashhar Farhan wrote:

I am a little wary of axial inductors. I would have easily recommended air wound coils. However, given all the RF floating around, i would recommend a toroid.
?
The 470 ohms to 220 ohms is jacking up the gain by a few notches to drown out the 12 mhz spur. Remember that the 12? mhz spur is not from the IF amp or the mixer, but from the Si5351's internal cross talk
 

I don't have the audio tone problem, so can't say for sure it is due to the 16mhz resonator.
However, when I put my finger on my 16mhz resonator, it's more than happy to move.
So there might be an easy solution here, perhaps solder a cap or two from resonator to ground,
though the pins are awfully small for most folks in the forum.
An easier solution might be to cover the resonator with insulating tape, then cover that
with some grounded copper tape.? About the same as putting a finger on it.
I'll experiment with that here.?

And while we're at it, that unshielded USB interface does not belong in a radio with a 12mhz IF
if it is ever going to be active while the radio is operating.

Long term, put a $2 processor chip down on some new Raduno
and feed it a different processor clock rate.
Or move to a different filter frequency on v5, and put a #define for it at the top of the source code.




##############? Continue reading at your own risk!!!? #################

Way off topic, we may need to create a new group for this one.? But oh well:

Don't do a Bill Gates and wallow in mistakes made decades previous.

You may make billions, but Bill (and that IBM PC crew) were responsible for an entire generation
of programmers making a career of cooking their shoes.? Just google for "segmented memory advantages".
Here's an especially good one, posted recently, aimed at CS students:
? ?

Those "advantages" were laughable in 1980 when Intel came around to us flogging their new 8086 against
Motorola's 68000 with its flat?4 gigabyte address space.? ?A 40 year old list of bullet items
that students today are still trying to make sense of.?
Try to avoid doing that sort of thing to ham radio.? ?;-)

"Cooking their shoes" is a reference to this old chestnut of a unix fortune cookie:
? ? The day-to-day travails of the IBM programmer are so amusing to most of
? ? us who are fortunate enough never to have been one - like watching
? ? Charlie Chaplin trying to cook a shoe.
?
Of course, that was a reference to the IBM 360 crowd, not the IBM PC,
which did not yet exist when that was written.? But applies nonetheless.

And I won't mention Bill's snuffing out of any competition through any means available,
including that SCO vs linux crap.?

Though I must admit, Bill is trying to do good things with his money these days.

Now you got me going.
Don't worry, I'll go have a beer and try to calm down.

Jerry, KE7ER



On Mon, Nov 12, 2018 at 09:49 AM, Ashhar Farhan wrote:
we should have used 11.052 MHz IF. if we were to start it all over again, i'd give my rigol's channel 1 to do that. unfortunately, there are others who share our misery, we must bear this with them for the sake of backward compatibility of the software. do you feel bill gates pain now?
?
- f
 

Is it worth adding a notch to the LPF mods suggested at R27?? For those of us where -43dB is not enough. (-50dB is the spec in VK)

eg reduce the 'input' 300nH to 220nH with a 15pF in parallel. (maybe do it for the other 300nH also)? 220nH is 8t on same toroid.

glenn
vk3pe
 

On Tue, Nov 13, 2018 at 03:49 AM, Ashhar Farhan wrote:
we should have used 11.052 MHz IF.
NOW that's a hack I like the sound of... I have a bag of 50 11.052 xtals for another home brew project... Time to start matching them up.
73 Nick VK4PP
 

My suggestion would be to switch to 11.052mhz and make it an option in the firmware itself. It could even be part of the initial setup sequence, with the frequency written clearly on the board.?
--
Ryan Flowers - W7RLF
 

ryan,

i did think of that, the trouble is that by default, it would switch to one frequency or the other. it is not so easy for most of us to figure out all the convoluted historical logic for doing things in these ways. i can well see people tuning to 11.052 Mhz instead of 12 MHz and wondering why they are singled out by wowbagger for being stupid.

it is still an option. the crystals for 11.052 mhz have suprisingly different characteristics from those of 12 Mhz, we will need to characterise these and build a decent filter.?

- f

On Tue, Nov 13, 2018 at 8:16 AM Ryan Flowers <geocrasher@...> wrote:
My suggestion would be to switch to 11.052mhz and make it an option in the firmware itself. It could even be part of the initial setup sequence, with the frequency written clearly on the board.?
--
Ryan Flowers - W7RLF
 

I don't see the problem.

Anybody upgrading firmware on a v3 or v4 board typically recompiles the source using the Arduino IDE.
The v5 boards could have the same source code, I assume 11.052mhz crystals would be the default going forward.
Those with 12mhz crystals on v3 and v4 boards need only change one line of code before compiling.

Consider the code of post? ?/g/BITX20/message/44278

To update that code fragment for v5 boards, just change this line:
<? uint32_t? f12c? = 11998000;? ? ?// center of 12mhz filter
to this:
>? uint32_t? f12c? = 11050000;? ? ?// center of filter using 11.052 crystals

Probably best done as a #define at the top of the main file where folks can easily find it.
I assume that the filter center frequency is about 2khz down from the parallel resonant frequency marked on the crystals.

The most difficult part of this switch would be to find a more appropriate name for "f12c".

Note that these RAM variables in that code fragment:
uint32_t? f45c? = 44995000;? ? ?// center of 45mhz filter
uint32_t? f12c? = 11998000;? ? ?// center of 12mhz filter
uint32_t? pbt? ?=? ? ?1500;? ? ?// bfo is 1500 hz below f12c

can be usefully changed by the user to set the BFO offset and the center frequencies
of the two crystal filters.
It should be possible to save all those changes to eeprom for the next power up.
So even if f12c did get compiled to be 11050000, users of v3 and v4 boards could
simply scoot that value over to 11998000 using the configuration menu.
But I think it's better to just change a #define for f12c before compiling
so it is close to working at first power up.

If that's some sort of problem, it's way way less of a problem than being plagued
by audio tones from the 12mhz BFO beating with stuff on the Nano.

Jerry, KE7ER


On Mon, Nov 12, 2018 at 07:38 PM, Ashhar Farhan wrote:
i did think of that, the trouble is that by default, it would switch to one frequency or the other.
 

jerry,

i guess the old boards will already have the bfo set to 11.995 or thereabouts, if we sniff it we can shift to 11.052 mhz. will that work?

- f

On Tue, Nov 13, 2018 at 10:23 AM Jerry Gaffke via Groups.Io <jgaffke=[email protected]> wrote:
I don't see the problem.

Anybody upgrading firmware on a v3 or v4 board typically recompiles the source using the Arduino IDE.
The v5 boards could have the same source code, I assume 11.052mhz crystals would be the default going forward.
Those with 12mhz crystals on v3 and v4 boards need only change one line of code before compiling.

Consider the code of post? ?/g/BITX20/message/44278

To update that code fragment for v5 boards, just change this line:
<? uint32_t? f12c? = 11998000;? ? ?// center of 12mhz filter
to this:
>? uint32_t? f12c? = 11050000;? ? ?// center of filter using 11.052 crystals

Probably best done as a #define at the top of the main file where folks can easily find it.
I assume that the filter center frequency is about 2khz down from the parallel resonant frequency marked on the crystals.

The most difficult part of this switch would be to find a more appropriate name for "f12c".

Note that these RAM variables in that code fragment:
uint32_t? f45c? = 44995000;? ? ?// center of 45mhz filter
uint32_t? f12c? = 11998000;? ? ?// center of 12mhz filter
uint32_t? pbt? ?=? ? ?1500;? ? ?// bfo is 1500 hz below f12c

can be usefully changed by the user to set the BFO offset and the center frequencies
of the two crystal filters.
It should be possible to save all those changes to eeprom for the next power up.
So even if f12c did get compiled to be 11050000, users of v3 and v4 boards could
simply scoot that value over to 11998000 using the configuration menu.
But I think it's better to just change a #define for f12c before compiling
so it is close to working at first power up.

If that's some sort of problem, it's way way less of a problem than being plagued
by audio tones from the 12mhz BFO beating with stuff on the Nano.

Jerry, KE7ER


On Mon, Nov 12, 2018 at 07:38 PM, Ashhar Farhan wrote:
i did think of that, the trouble is that by default, it would switch to one frequency or the other.
 

I suppose one could read your USB_CAL parameter from EEPROM,
If it's around 12000000 then assume it's a v3,v4 rig.
If it's around 11050000 then assume it's a v5 rig.
If it's neither, print a warning and assume it's a v5 rig.

Include a menu item to force it one way or the other.?
People may have reason to swap Nano's around between rigs.


But I'm not convinced any of that extra code is necessary, all we need is a? #define?
at the top of a file somewhere to switch it one way or the other.?
If somebody is upgrading firmware on v3 and v4 boards, they are apparently competent enough
to install the IDE, download your code,? and get the host to talk to the Nano.
And thus competent enough to change a? ? #define.
Or choose the correct zip archive that already has that? ?#define? ?set for their v3,v4 board.

You may come up with other reasons to have a? #define? ?in there.
Maybe stir the Nano pinout for cleaner CW keying, an SWR indicator, ...
And if major changes are warranted, fine with me if it becomes a whole new body of code.

Jerry, KE7ER


?

On Mon, Nov 12, 2018 at 09:51 PM, Ashhar Farhan wrote:
i guess the old boards will already have the bfo set to 11.995 or thereabouts, if we sniff it we can shift to 11.052 mhz. will that work?
?
 

Or call it something that makes it obviously different like the uBITX mk2. People are fairly used to having different versions of things. I think this fix is significant enough to warrant such a change, as it really redefines the uBITX in the ham radio world. I hate to say it, but the uBITX has the spur reputation to shake off in some communities, and a "mark 2" or other revision would really help. "v5" might not be enough to set it apart. And, it solves the firmware compatibility issues, because people will generally be smart enough to download the correct firmware for their board, especially if the "MK2" designation is silkscreened in an obvious spot on the board for easy identification.?

--
Ryan Flowers - W7RLF
Martin KM6TCD
 

How do we tell if we have a board that needs relays replaced?
?
Sent:?Monday, November 12, 2018 at 1:05 AM
From:?"Ashhar Farhan" <farhanbox@...>
To:[email protected]
Subject:?Re: [BITX20] spur & harmonics fix, v2
simon,?
our new relays have less leakage. they are within have sufficiently low leakage to keep the harmonics within the specs. the earlier boards will need to replace the relays.
?
- f
?
On Mon, Nov 12, 2018 at 2:22 PM Kelly Jack <kellyjack1968@...> wrote:

Farhan,

What do you mean by "the relays that are now used for ubitx... we dont need the axicom relays"?

Point 5 of the attachment still calls for replacing the relays with axicom relays.

I have a v3 board. Do I need to replace the lpf relays with axicom relays or no?
Regards?


Simon
VK3ELH

?

?

1 - 20 of 70