Take a look at this video for an explanation why square signals are better for mixers:?https://youtu.be/WwJKxvz7qbs

Thanks Dave.

I am going to try controlling the PTT through the nano to control the clock better before the PTT off is triggered. So a new nano input from the mic PTT and a new output from the nano to a 2n3904 to switch the PTT line on the board.

I dont have AGC or CW on my rig so less complications.

I have a transistor across the lm386 input that switches to ground when PTT is pressed at the moment and I have no audio pop. I cant remember how I triggered the transistor though so Ill have to work through that and hopefully it will be compatible with this new plan above.

73
Simon VK3ELH

The ARRL Handbook says square waves are better,
though does not say anything about how huge the difference is.

My understanding is that the more a diode ring mixer departs from being a perfect switch
driven by a square wave, the greater the intermodulation distortion.
Slowing down the switching edges by driving with a sine wave does not help
in most measures of performance.

The action of an ideal switching mixer is that of doing an analog multiply of the incoming signal
with a square wave, as described in post 22538 (one beyond the post previously cited).
You could break that square wave down into odd harmonics of specific amplitude through
its Fourier expansion ? and multiply those by
the incoming signal to calculate?hard numbers for what comes out of the IF port. ?
No need for any small-signal handwaving.

Jerry

I wanted to get this project off the desk as I don't have time to work on it and no cash to get a 2nd bitx for it.

I have 2 x W8TEE boards for the tft display 2x regulators one AD9850 2x encoders 5 x AO3's one mega2560 pro mini (purchased from the same place Jack has gotten his in the past)

?

I am looking to sell the lot 35$ shipped to your door.

?

?Any questions please e-mail me direct

d herron at live dot com

?

73

?? David

????? N8DAH

Glad to hear that this idea is working out for you Simon.
I?am now?on holiday?so unable to play/experiment?for a few weeks, and have not had?chance to load Allard's new release.?
I need to experiment more but I think the PTT sense is not required at all if the mic PTT is connected directly to the Raduino (A0).? For CW I think it is it is not needed, the Raduino keys the transmitter itself via D7?.? A delay from keying to operating the relays will be required in CW also o stop the carrier burst - I have not looked in the code to see if that is implemented, I did do something in my version but cannot look right now.? If the relay?switching or cap charging/discharging can be made faster then that could help reduce the delay.? Not looked at that at all as CW is still a mystery to me.
I did manage to build Farhan's AGC circuit and have tried?connecting the PTT output from the Raduino (output D7) through a diode to charge up the AGC cap and mute the audio on TX.? This unfortunately did not solve the RX-TX pop, and the time constant going from TX-RX was quite long before audio could be heard again.? I injected the 5v between the agc diode and the resistor, so next attempt will be after the resistor to reduce the rise time, as Gary suggested.? Not decided the best approach for reducing the TX-RX decay - options include reducing the bleed resistor, reducing the capacitor value or reducing the voltage the cap ultimately charges to?down from the 5v output by the Raduino D7.? I may add another FET to invert the PTT input from the Mic switch to get the mute going earlier and then try lowering the ?voltage applied to the AGC circuit to reduce the decay time.
This is what its about for me - thinking about a problem, playing, trying things and seeing what does/doesn't work.? Just hurts a bit when the smoke gets let out and some poor component dies, but all part of the learning experience.
Have fun!

Dave M0WID

As I pointed out, I'm using my own Nano. No Raduino is used in this.

Vince.

--
Michigan VHF Corp.

开云体育

?Again I'm not an expert, but even tho this isn't a linear system, for small signals one might be able to make an approximation of it using the superposition principle.

In other words,? ?that square wave has WITHIN IT the sine wave of the desired frequency, and the diode mixer probably responds (to a first order appxoimation) as:

sine wave LO? x modulation frequency --->? ?sum & difference

2nd harmonic of LO? x moduclation frequency? ---> sume and difference

......

nth harmonic of LO x modulation ferequency --->? sum & difference.....

and then the filter next throws out everything except the desired one......hence the book's observation that it doesnt make a huge amount of difference.? ??

In the long run, it is the tiny tiny changes as a result of interaction between the two frequencies on a non-linear transfer function that accomplishes the mixing.? ?The diode isn't a binary device (1 and 0 only) but instead is a non-linear device with a transfer function that has infinite steps between? ?0 volts and 1 volt.? ?

(my 2 cents worth)

I think looking at it in the frequency domain may be more instructive than looking at it in the time domain.? ??

gordon

Hmm. ?I guess opinions vary.

In my 2015 edition of the ARRL Handbook, page 10.19, section 10.5.2 for "Switching Mixers",
the discussion about diode ring mixers assumes a square wave for the local oscillator.
The only reference to a sine wave local oscillator in that section is this:
"Normally, the harmonic outputs are so widely removed from the desired output frequency
that they are easily filtered out, so a reversing-switch mixer is just as good as a sine wave
driven analog multiplier for most practical purposes, and usually better for radio purposes,
in terms of dynamic range and noise."

The si5351 is not very well spec'd, but the si5338 of the same famiily is, and those docs suggest?
the phase noise when using fractional values in the output msynth dividers is well under
a factor of two worse than running the part with integer divides on both pll and output msynths. ?
Many in this forum are very happy with their analog vfo, once they've tamed it's tendency to wander.?
That analog vfo would have much worse phase noise than the si5351.

I suspect many of the issues with mixers and birdies and such revolve around poorly terminated mixers,
especially at the if port, and poorly shielded rigs allowing vfo harmonics to crawl into places they should not. ?
On the si5351, I'd be more concerned about crosstalk when using mulltiple outputs than with phase noise. ?

This post from last February: ?/g/BITX20/message/22530
plus a few of the neighboring posts in that thread describe how a diode ring mixer works,
and how a square wave switching an incoming signal can be considered a multiplier. ?

Jerry

Oops! Hit send mid-edit. Sheesh! ?I’ve shunted the PTT input with 1uF and a reverse biased diode to suppress this exposure, but otherwise did nothing to ensure that debounce is effective at the mic switch. I suspect Allard’s CLK2 blanking code is bypassed when CW mode is initialized.

73

Gary, N3GO

I considered that... at least in the context of the PTT input from the mic switch, and that may well be the source of the intermittent bursts in sideband mode. I’ve shunted the PTT input haven’t attempted to verify this though. 75 ms is a very long time for a pair of 7 ms relays to settle however, and the 6 ms burst in CW mode is consistent and reliable, as is the dropout that follows and then proceeded by the start up of the initial CW element.

Gary, N3GO

El 23/10/17 a las 13:34, Jerry Gaffke via Groups.Io escribió:

Some in this forum have experimented with adding filters between the si5351 and the diode ring mixer.
Some do report fewer birdies, though there are lots of possible reasons for that.

Hi, My two cents on this.

I used a Si5351 for VFO and BFO in another project for a generic radio controller, like the raduino but more broad in capabilities (see ) and I can tell that lower the freq you generate greater the harmonic content (aka. general noise and birdies).

Two samples, and comments.

-> Old marine radio (really oooold... Spilbury... if not wrong), BFO freq is 1.6 Mhz, direct square wave to DBM as a detector yields to a noisy but usable receiver... after placing a back to back red toko If cans (broadcast radio oscillator IF can, via a 47pf cap) and tune to 1.6 Mhz you got a very decent receiver noise, not measured by lab instruments but by ear I say about one or two S unit of band noise was cut after filter placement.

-> Homebrew monoband (40m) 500Khz heterodyne radio based on a old Russian filter/xtal from a KARAT transceiver... Rx was useless without the filter... with the lesson learned from the above example, two back to back 455 khz toko IF cans re-tuned via a 100 pf cap solved the issue and reception is very good now.

So, the lower the freq you generate the dirtier it get's and it make sense as the harmonics are quite close in range and the mumbo-jumbo of strong freqs mixing each other it's a nightmare...

Then you have the Square vs Sine in a DBM...

I doubt it has anything to do with presenting a square wave to the mixer.

I'm not agree, in my experiments I found that driving it with Square Waves is more efficient, but a lot dirtier. ARRL and RSGB Handbooks agree on that, several academic papers on the internet are agree.

Back in the beginning of the year I spent a few days with a borrowed Tektronix oscope and this was one of the issues I studied with it... (Santa take note: I like a Textronix this year for Xmas, hi hi hi)

I'm not a electronic engineer (actually industrial engineer in the ISO certification and IT consulting field)? but I dare to speculate about the root cause: zerocrossing.

In a square wave via a DBM the harmonics (dirt) are created in the zerocrossing of the square wave from one cycle to another.

Think about it: a sine wave has a region that results on both diodes to be off in a certain time (amount) before cycle inversion and that's allows the inductors to collapse it's fields and reset for the next cycle.

Square wave have virtual no zero off time between cycle inversion and...

With light filtering (top square smooth on the scope but not good like a sine wave, just a low pass filtering) you don't have chance to see it as it's almost square at the bottom of the signal near zero, it still a square wave.

To actually see a improvement you have to place a real band pass filter that get the signal as much as a sine wave and the important part is the lower one near the zero, to let a little gap of both branches of the DBM in the off state by the 0.2 or 0.6 volts of conduction of the diodes.

I'don't have the chance to test the back to back IF cans with the oscope, but in practice they make A LOT of difference, some simulations with ELSIE software shows that with that kind of filtering the signal get a pretty sine wave... (also confirms that a simple LPF don't change the skirt of the Square wave)

Bitx40 uses 12MHz IF so dirt is 12 Mhz away (not taking into account reciprocal mixing and other inter heterodyne products) so you may expect a few birdies and low noise, but beware if you use a low IF frequency...

BTW I saw a youtube video this weekend of a SDR forum presentation in germany on this year (don't have the URL, a friend pass it to me via a fash stick) of the CEO of Flex talking about SDR tech, the reciprocal mixing of phase noise in the VFO make me proud of my decision of make my Si5351 lib with only two outputs at a time and use integer division to avoid jitter and phase noise.

Google it, it's a must see one.

--
73 CO7WT, Pavel.

开云体育

I have made my own version of these.? They work fine on a Nano, but since there is NO Address selection it probably collides if you put it on Raduino with its SI5351.

The 20 Pin version of the SI5351A has ONE address selection pin, so the next ones I make with be with the 20 QFN part.

It is over 2 the price and in the QFN package.

Mike, WA6ISP

Not coming off the Raduino, I'm using one of these:



Much simpler but does the same thing.  My own Nano connects to it without
the need for a display.

Vince.

thanks.   I assume you measured right at the output of the radiuno and not
after the 6db attenuator.

I'm hoping to use this thing as a replacement VFO for some old heathkits
once I learn what THEIR vfo output voltage was.....I'm pretty sure they
were at 2K ohms so I may be able to use a ferrite transformer and step up
voltage through simple impedance matching.     It would be really cool to
have the advantages of the digital VFO on my old (and EMP proof)
SB-102!!!!

Gordon


________________________________________
From: [email protected] <[email protected]> on behalf of Vince Vielhaber
<vev@...>
Sent: Monday, October 23, 2017 1:20 PM
To: [email protected]
Subject: Re: [BITX20] Raduino question.

measured where, across what impedance?
thanks for the information!!!

Typically my RF measurements are all at 50 ohms, but right now I can't say
for certain that last nite's were.  I'll have to double check when I get
home.  Also I need to correct something, my initial output was 4ma, not
2ma.

Vince.
--
   Michigan VHF Corp.



               

-- 
Mike Hagen, WA6ISP
10917 Bryant Street
Yucaipa, Ca. 92399
(909) 918-0058
PayPal ID  "MotDog@..."
Mike@...

Not coming off the Raduino, I'm using one of these:



Much simpler but does the same thing. My own Nano connects to it without
the need for a display.

Vince.

--
Michigan VHF Corp.

Hi Gordon,

...With a diode mixer, yes you get the sum and difference, but you also
get all the harmonic mixing products, too; reduced in amplitude, of
course---it's the "nature of the beast," so to speak.

73, David KB4FXC




On Mon, 23 Oct 2017, Gordon Gibby wrote:

fascinating.


I'm not an expert at all on mixers, so you know a lot more about them

than I do! But the desired frequency is still only a sine wave. The
rest of the squre wave....is not at the desired frquency. If all
those other harmonics etc help you out in some interesting way, that's
fine, but my pea brain thought you mixed one frequency with another and
through the miracle of non-linearity, you get their sum and difference.
I don't quite understand how any other frquency enters into the equation.
but then again, I don't claim to understand them. Maybe I'll better
understand them some day!!!!

cheers!


gordon




________________________________
From: [email protected] <[email protected]> on behalf of Jerry Gaffke via Groups.Io <jgaffke@...>
Sent: Monday, October 23, 2017 1:34 PM
To: [email protected]
Subject: Re: [BITX20] Raduino question.

No. The desired drive into a balanced ring mixer is a square wave.
You are driving diodes, the current through the diodes is an exponential function of the voltage across them.
So they are effectively either on or off, with very little transition between the two.
Which is not necessarily a bad thing.
The diode ring is best thought of as a commutating mixer.
Which means it is best analyzed as switches.

Something like a Gilbert cell can be used to truly do an analog multiply of the incoming signal with the local oscillator.
But even these mixers are typically driven hard to where the Gilbert cell acts as switches.

If you really want a sine wave into your diode ring mixer, you will have to choose
between a sine wave in voltage presented or a sine wave in current through the diodes.
You can't have both, as them diodes are far from linear.

Some in this forum have experimented with adding filters between the si5351 and the diode ring mixer.
Some do report fewer birdies, though there are lots of possible reasons for that.
I doubt it has anything to do with presenting a square wave to the mixer.

If using more than one of the si5351 outputs, a filter might help reduce crosstalk.
It would also help prevent vfo harmonics from radiating into off the open cabling into other parts of the radio.

Jerry

On Mon, Oct 23, 2017 at 10:08 am, Gordon Gibby wrote:

but the DESIRED frequency is a sine wave.

Nope, I don't bandwidth limit that scope (Tek 2430A). I'll do another
measurement tonite with the Tek 11402 and see what it says.

Vince.

--
Michigan VHF Corp.

thanks. I assume you measured right at the output of the radiuno and not after the 6db attenuator.

I'm hoping to use this thing as a replacement VFO for some old heathkits once I learn what THEIR vfo output voltage was.....I'm pretty sure they were at 2K ohms so I may be able to use a ferrite transformer and step up voltage through simple impedance matching. It would be really cool to have the advantages of the digital VFO on my old (and EMP proof) SB-102!!!!

Gordon


________________________________________
From: [email protected] <[email protected]> on behalf of Vince Vielhaber <vev@...>
Sent: Monday, October 23, 2017 1:20 PM
To: [email protected]
Subject: Re: [BITX20] Raduino question.

measured where, across what impedance?
thanks for the information!!!

Typically my RF measurements are all at 50 ohms, but right now I can't say
for certain that last nite's were. I'll have to double check when I get
home. Also I need to correct something, my initial output was 4ma, not
2ma.

Vince.
--
Michigan VHF Corp.

开云体育

fascinating.? ?

I'm not an expert at all on mixers, so you know a lot more about them than I do!? ?But the desired frequency is still only a sine wave.? ?The rest of the squre wave....is not at the desired frquency.? ?? If all those other harmonics etc help you out in some interesting way, that's fine, but my pea brain thought you mixed one frequency with another and through the miracle of non-linearity, you get their sum and difference.? ? I don't quite understand how any other frquency enters into the equation.? ? but then again, I don't claim to understand them.? ?Maybe I'll better understand them some day!!!!

cheers!

gordon

Oh, and the si5351 is driving those series schottky diodes through a 6db attenuator, which cuts the voltage in half.
Maybe you had your scope on some low bandwidth setting or something.

Rather surprising you only saw a fraction of a volt from the si5351 at 35mhz..

If so, hard to imagine Farhan would get the ubitx to work at all, as his first local oscillator is at 45mhz + signal
Apparently works up to a 30mhz signal frequency, so 75mhz from the si5351.
And it has to drive something like 0.6v into two series schottky diodes for them to start conducting.

The si5351 outputs should work up to something like 200mhz.
(Though my library is only good to a bit over 100mhz, as it does not mess with varying the vco frequency).

Jerry

No. ?The desired drive into a balanced ring mixer is a square wave.
You are driving diodes, the current through the diodes is an exponential function of the voltage across them.
So they are effectively either on or off, with very little transition between the two.
Which is not necessarily a bad thing.
The diode ring is best thought of as a commutating mixer. ?
Which means it is best analyzed as switches.

Something like a Gilbert cell can be used to truly do an analog multiply of the incoming signal with the local oscillator.
But even these mixers are typically driven hard to where the Gilbert cell acts as switches.

If you really want a sine wave into your diode ring mixer, you will have to choose
between a sine wave in voltage presented or a sine wave in current through the diodes.
You can't have both, as them diodes are far from linear.

Some in this forum have experimented with adding filters between the si5351 and the diode ring mixer.
Some do report fewer birdies, though there are lots of possible reasons for that.
I doubt it has anything to do with presenting a square wave to the mixer.

If using more than one of the si5351 outputs, a filter might help reduce crosstalk.
It would also help prevent vfo harmonics from radiating into off the open cabling into other parts of the radio.

Jerry