New expansion - UPD858 SSB chassis

[Snow Bird]

Hi Sue Glad To hear You are Ok Down Under Hope you all stay that way Later Sweet Lady. Rich

[Richk]

Hi would it be posible to get the windows tune up APP,
Thanks
73's
Richard G

[nomadradio]

Something I want to try with the Si5351 is a simple divide-by-two on its output. Seems to me this would serve to clean up oddball distortions on the waveform. Filtering a square wave to use as a LO for transmit and receive should not be so tough.

Maybe.

73

[lbcomms]

In this application, where it's only being used to feed a downmixer that it turn only feeds a divider in the PLL chip, it doesn't need to be super clean. The low pass filter between the PLL and VCO (essential to remove the 5KHz / 10KHz phase detector waveform that would otherwise cause unwanted sidebands to appear) will remove any nasties. The Cobra 138 radio we used it in is getting IMD figures of -45dB to -50dB and sounds really clean on air (after we upped the final bias to 90mA, but that's another subject altogether!).

If the output of the '5351 is either mixed to the LO (i.e. PLL02A Cybernet SSB radios) or becomes the LO itself (replacing the VCO in an "unexpandable" PLL circuit with no downmixing, such as the LC7131 / TC9106) then additional filtering would be needed to keep it clean.

A better approach for PLLs with no downmixing is to replace the 10.240 reference on the PLL with the output of the '5351. The PLL chip is wired for one channel only (we hard wired our LC7131 to channel 40) and the channel switch codes sent to the Mini. The divider math is easy to reverse in code so the output of the VCO is the exact correct frequency. It's clean because the low pass filter between the PLL and VCO remains in place. BUT - this will only work for single conversion SSB radios as these don't need the 10.240 as a second local oscillator (10.695 - 10.240 = 455 KHz), not an issue locally though as 27MHz is SSB only down here.

The output can also be used on the 23 channel radios (at least the ones common down under) though, as these have plenty of filtering on the local oscillator output. Downside is it limits the bandwidth to about 60 - 80 channels, but not really an issue as the 23 channel radio owners we've dealt with are only interested in hitting the two SSB call channels (27.355 and 27.555) and a few channels either side of that.

PS: Are you the nomadradio from the old CBT forum? Is so, hi...

[nomadradio]

Yes ma'am, same one from CBT. Howdy yourself!

Using the radio's PLL to clean up the Si5351 takes care of the problem quite neatly. Gets you extra channels with the radio's LO still as clean as the stock setup. Kudos for taking that approach. I'm pondering a way to use the Si5351's output as the LO source.

One application I want to try is a VFO/synthesizer for old two-crystal-per-channel radios. One output from the 5351 for receiver LO, one to generate the transmit frequency directly. Spectral purity for AM-only radios probably means more in the receiver than the transmitter, but nobody else has offered a packaged solution of this sort to date.

One far-fetched notion still not even half-defrosted, let alone half-baked. Use one output for VFO, one for HFO and the third outut for carrier in old analog three-oscillator ham HF transceivers. Spectral purity becomes a Really Big Deal in this context.

Just the 'baby step' of replacing my old MC145106 synthesizer we no longer sell seems like a good place to start, at least.

The allure of selling a "nobody else" gadget is all very good, even though it may yield little more than bragging rights at best. Deciding to come up with something nobody else offers raises an inevitable question, "Does this mean that anyone sharp enough to pull it off already knows there is no money in it?" Sometimes that's the real reason that "nobody sells" (fill in the blank).

The complexity of program code to jam into an embedded controller has me more than a bit intimidated for now. Coming up with nine-bit divisors for a MC145106 was not so complex to do wtih a PIC16C57, but the 5351 is a horse of a different color.

Haven't got my big toe wet with this technology, let alone my feet. Plan is to use Arduino pro-mini with the Adafruit 5351 breakout and a 0.96-in OLED display. Still trying to wrap my mind around feasibility so far.

I do appreciate your generosity with your design efforts. I've always been stingy with my firmware. If I use open-source stuff to build this kind of toy, I'll be obligated to publish the whole thing as I understand it. Just gotta keep reminding myself it's not the nineties any more.

73

[lbcomms]

Yes ma'am, same one from CBT. Howdy yourself!

Hi after the last few years!

I'm pondering a way to use the Si5351's output as the LO source

There are two ways to do it if you want to keep spectrally clean.

The first is to do what the first gen 23 channel Cybernet SSB - the chassis with 10 crystals on the channel selector board, branded Kraco, Hygain, and a few other names did. Lots of tunable filtering between the synth output and the Tx / Rx mixers, needed in that chassis as the crystal synth was a rather dirty beast:



One of the crystal sets was in the 14 MHz range, which if allowed to enter the Tx mixer would cause spurs at 26.815 (14.91 + 11.275) and 29.82 (second harmonic of 14.91) on channel 1. There were three mixers - each with their own filter stage - before the radios main Tx and Rx mixers to keep it clean, and to save four crystals be using high side injection on AM / USB and low side injection on LSB.

By removing C3 (high side LO coupling) and Q24, and injecting a 15 MHz LO in all modes at Q24, the filtering between C131 and C147 can be used to nicely clean up the output of the '5351. Two minor downsides - the filters will limit you to about 600KHz coverage (nobody asked for more than this) and you'll need to generate a 11.270 MHz carrier oscillator signal (used the other 5351 output to do this, through a very narrow filter - it only needs 5KHz bandwidth - to keep that side clean).

The other way is to use mini fixed ratio PLL circuits between the output of the '5351 and the radios LO connection/connections.

Something like the little 8 pin Analog Devices HMC1031 PLL chip. This is a fixed ratio (divide by 1, 5, or 10 depending on how its connected) PLL synth, which appears as a X1, X5, or X10 multiplier mathematically but is a proper PLL that controls a VCO and needs a DC low pass filter between its output pin and the VCO. Like using the 5351 to generate the downmix signal on the uPD858, the filter only lets the single strongest signal through and leaves you with a perfectly clean LO to feed the radio with.

Again, two minor downsides - the HMC1031 chips cost $15 each (Mouser Part# 584-HMC1031MS8ETR), and you are limited to the lock range of your VCOs. Bigger range means more channels covered, but also adds more phase noise to degrade your spur / selectivity numbers. Limit the range of the VCO to 3 MHz and the rest of the radio to 2 MHz (to allow for DC control voltage drift caused by temperature changes) and they are squeaky clean.

One application I want to try is a VFO/synthesizer for old two-crystal-per-channel radios

Might be a bit before my time, are you talking one crystal for Rx and the other for Tx? If so, feeding the one signal to both and sensing Rx / Tx condition at the micro is going to save you a 15 dollar IC and $5 for the VCO / filter components. If the difference is only 455 KHz you should be able to keep the lock range to 3 MHz easily, most folks only want 80 to 120 channels (under 1.5MHz LO range).

The allure of selling a "nobody else" gadget is all very good, even though it may yield little more than bragging rights at best

We don't go big on selling ours, only have maybe 30 or 40 regulars who are either customers of our core business (business band 2 way and point to point data links) or friends of those that are. Never made any serious money from hobby radio, we do it for the technical challenge and to break the monotony of the 100th DMR repair we've done for the year instead. Happy to help others on the forum to duplicate our results, but no time for certain individuals who have copied out stuff and sell it as their own product without even giving us a mention.

Coming up with nine-bit divisors for a MC145106 was not so complex to do wtih a PIC16C57, but the 5351 is a horse of a different color.

There are only two calls needed, one to initialize it (called once when the radio is turned on) and another to set the output frequency.
Have a look at https://github.com/etherkit/Si5351Arduino for details, load their lib and it's a one line command to set the frequency of an output pin:

Code: Select all

si5351.set_freq(2696500000ULL, SI5351_CLK0);

sets the first output (CLK0) pin to 26.965 MHz. Frequencies are longint units of 0.01 Hz, so as an example "2696500000" will make it onto a signal generator for channel 1...

[dt307]

Been AWOL and so busy with repairs ect no time for my own fun but slowly getting there .....bagged an Adams so made some good head way and have your expander all in and working...just the alignment and band wires left to do will upload some pics asap ....hope all you guys doing ok

[lbcomms]

Been AWOL myself - well not quite, but that pesky virus thing got in the way for a while. I got a uPD2824 and an MB8719/34 radio to recode it for those PLLs, but the whole thing got pushed to the back while we learned our way of working with the NSW govt. COVID restrictions in place.

Things are finally returning to normal here with some restrictions removed or watered down, so after we've cleared a bit of a backlog we couldn't before (such as site visits where there are lots of other people present, like schools and office blocks) we'll get back to the 2824 / 87xx versions.

Love to see some pics of your install and hear any other comments good or bad, this thing is a work in progress and can hopefully be tweaked if you get any issues with the instructions or the software...

[dt307]

yes same situation here england is more open than wales so a little harder to get things done with the virus....i did notice under testing the lsb icon dont light on the debugging app so i may have a wiring issue i need to look at as i may have done something wrong. also i think i have said before that the option for the coarse tune would be perfect if a -/+ 5kc was added and remove the 15k 1 as then when the coarse turned fully each way would be on a zero and would then be no need for a -/+ 5kc switch ...resulting on an easy to use radio and less things changed from a std radio ie no switches changed ....i still have some wires to add as no offsets connected but its in and working.still have a strange bug where i program the avr then connect it to the pc software and make adjustments goes fine .but when i disconnect i am unable to reconnect i have to reflash the device....its no biggie and i can work around this and maybe i need just to reboot the pc ...this needs further testing on my part and dont think its a problem with the boards or software....i still have mabe 5 cb and a few hf sets to repair before i shut down all repairs completely to give my self time to work on my own projects like this 1 of yours ,
so i will get time soon to do and test a version for most radios with your coding .
pll02a
mb8719
upd858
mc145106
dont have a upd2824 based radio but i can buy or beg for 1 lol
thanks again for putting the time and effort into this project to help out the radio community.
i for 1 will use the information to help out others as i never charge for radio repairs or mods and if i do install these for any 1 they will be at cost.
when i buy in cheap radios broke and then add mods like this and resell them is how i make pocket money to keep me in spare parts

[lbcomms]

i did notice under testing the lsb icon dont light on the debugging app so i may have a wiring issue

Check the voltage change at pin 11 (usually labelled "A1") of the nano module. This in turn goes to pin 24 of the microcontroller IC. It's configured as a digital input (port C can be set to either analog or digital), so anything under 2.5V is logical "0" and 2.5V to 5V is logical "1".

You could have a wiring error, bad resistor, short to ground at the Nano pin, open pin / trace on the Nano, or a blown microcontroller IC. Should be easy enough to trace with a multimeter while flicking the radio between USB and LSB modes. If going to LSB gives you more than 5V at pin 24, the micro is dead and will need replacing. This can be caused by using too low a value for the current limiter, such as 10 ohms instead of 10K.

the option for the coarse tune would be perfect if a -/+ 5kc was added

Had multiple requests for this one, it'll be in the next version, before the 87xx and 2824 variants are done. The coarse 64Hz step (+/- 7.5 KHz) option will become 40Hz steps (+/- 5.0KHz) and the coarse 128Hz step (+/- 15 KHz) option will become 64Hz steps (+/- 7.5KHz).
The +/- 15KHz option was pointless anyway as you could just go a up or down 10K with the channel selector, and was only there as a demo that it's possible to have 30KHz of slide in a drift free radio.

strange bug where i program the avr then connect it to the pc software and make adjustments goes fine .but when i disconnect i am unable to reconnect

Sorry, haven't seen that one. Could be a flaky programmer, USB port, or Nano module. Is it repeatable / consistent or intermittent?

pll02a

That one will be a much more complex installation if you want to avoid a reduction in the RF performance of the radio. The Unidens and most others just use the downmixer to feed a PLL divider, so the signal doesn't have to be exceptionally clean.

The Cybernets are different. The difference signal feeds the PLL in the usual way, but the sum also becomes the local oscillator. Any junk on the downmix oscillator signal is then added to the LO, degrading the receiver performance by causing out of band noise / signals to be within the passband of the crystal filter and therefore audible to the user, and increasing spurs on TX.

There are a few ways to overcome the problem, but for now we'll be getting the other chassis types finished and get back to the Cybernet afterwards.

mb8719

Will be very similar to the 858. Software change only. Numbers are changed slightly due to the different divide by N codes in the PLL chip.

mc145106

That chip is used in lots of different chassis. If the radio has a mixing scheme similar to the 858/87xx/2824 it'll work, but I'll need to know some details such as the radios 1st IF (usually 7.8 or 10.695 MHz), whether its LO is high or low side injected, and the PLL "N" code values (such as 91 to 135 for FCC channels 1 to 40, giving an Fin frequency of 0.91 to 1.35 MHz).

dont have a upd2824 based radio but i can buy or beg

They are fairly common down here, we've got a waiting list of about ten of these to be done, so this will be well covered for us. They are somewhat narrow banded without a tuned mixer mod, most of the locals only want to get up to around 27.6 to 27.8 so most will be straightforward. With a tuned mixer mod though, they can cover 26.0 - 29.0 with good sensitivity / transmit power...

[dt307]

mod.jpg

not finished yet got all the options left to wire.

traced bad lsb to blown a1 pin . new micro and working now.

just been talking on it for last 2hrs as is (yes i know will be off frequency until all wiring and config done) was just testing.

in 2hrs no noticeable ssb drift.

no digital noise.

audio report was stunning and no difference between usb /lsb

so far its perfect!

for me removing low low and adding 3xhi for 10m better or remove the ref frequency limits i could do it that way to shift up by 450 :-)

[cb4ever104]

Looks good Chris . I already had a Maurer pcb for my Adams , but I may go down this road with my AR144s/146 GTL . I'll wait for Sue to post her 2824 results .

[lbcomms]

for me removing low low and adding 3xhi for 10m better or remove the ref frequency limits i could do it that way to shift up by 450

The ref limit can't be changed by that amount, as too many other frequencies are dependent on that one being accurate. If that's out, the UK/5K/10K450K shifts will be out too, as well as the centering of the two clarifier inputs.

But - there's no need to. The digital inputs are independent of each other, and will add together if you put two or more high. So "HI" and "-5KHz" on channel one will give you 27.410 MHz (5K down from 41), "LO" and "HI" on channel 20 will give you 27.205 MHz (the HI and LO cancel, leaving you on mid!).

So for HI HI HI, just put "HI" and "HI HI" (use a pair of diodes or a second pole on the band switch) positive at the same time

In a real install though, you'll need some serious broadband mods if you want the FCC channels and 3xhi in the same radio...

[dt307]

all understood about bands thats awesome.

small bug report

the +10kc is actually -10kc

all wired and working well

thanks

[lbcomms]

Thanks for that, I'll see if we can replicate it here today and change if necessary when we change the last two selections for the coarse clarifier ranges...