One Last FM Exciter.

[Zozo]

Yes indeed, I have a idea, it's possibly because I'm of Dutch descendant on my fathers side. We maybe think a-like. This goes here, put that there etc

[Zozo]

I had an unexpected delivery today from China. I ordered 50x SAA1057's, which is the PLL Synthesizer IC I used. However the estimated delivery was sometime in late June, but they took less thank 2 weeks to get to me, and not only that I got 4 extra for free

Now comes the fun part in testing them all. First glance they all seem genuine, I'll do the acetone and cotton bud test first.

20210512_105223.jpg

[Krakatoa]

I think you have been lucky, they seem legit by the way the marking looks in color, fonts, code, the package and the legs form is absolutely looking like other Philips IC's from the late 80's early 90s.

[Zozo]

Yes your correct Krakatoa, they all passed the acetone test, and so far I've tested half of them in a test circuit. All that I have tested are working absolutely fine. I'm sure the rest will be too, but I'll still test them all to be safe.

[Zozo]

Here's a quick look at the first MK3 board. I decided to stop using the BFLW-100 wire for the transformer stage, although it's a specially design wire for that purpose, I find it very difficult at times to split it apart. So as you can see in the photo, I've used just standard "twisted pair" ECW.

I don't know if anyone is the same as me when it comes to winding toroidals, but after making a few I also seem to have RSI in the palms of my hands.

The new exposed ground plane around the outer edge should make things a little easier for negative / braid connections.

The last change I made was to rotate the DC Choke inductor to the final stage, and increased the through-hole pitch.

20210513_110028.jpg

[jvok]

Been playing with one of these for a bit now and its a nice board, but I found a small problem with the reset circuit. Its usually fine when you cold start it, but if you power it off then back on within a few seconds the PIC doesn't get reset. This means it won't reprogram the PLL and so either will lock on a random frequency or won't lock at all. Not good if you get a power glitch and your rig starts back up on the wrong channel.

Fortunately its easy to fix:

[*]Change the 2n2 reset cap to 100n
[*]Add a reverse connected diode across the 10k reset resistor - anode to PIC pin 16, cathode to +5v. Can be any standard diode, 1N4148, 1N916 etc

The bigger cap stretches the reset pulse out a bit, and the the diode quickly discharges the cap back into the 5v rail when the power goes down. This makes sure the reset circuit is ready to go again very quickly. The rest of the time the diode is reverse biased so does nothing.

With these changes I found it sometimes takes a couple of secs to start up, but when it does it will lock reliably and quickly every time.

[sinus trouble]

Nice work Jvok!

I have the same problem with the Sinus PLL if there is a sudden glitch in the power supply? However if it is the Pira software Zozo is using? The code resets every 30 mins!

Its far from perfect? But it does work!

[Albert H]

Ever tried a 2SC730 instead of the 2N4427 in this circuit? I found that I could get almost 2 Watts out, which was ideal for some of the FET PAs!

The reset mod that Jvok shows above is exactly what I did to the "Pira" PLL some years ago. I uses a 3n3 cap, but other than that it's the same.

[sinus trouble]

Ever tried a 2SC730 instead of the 2N4427 in this circuit? I found that I could get almost 2 Watts out, which was ideal for some of the FET PAs!

The reset mod that Jvok shows above is exactly what I did to the "Pira" PLL some years ago. I uses a 3n3 cap, but other than that it's the same.

I guess the theory may also work on the SINUS PLL? Its something i will look into!

The 2SC730 is quite an elusive component these days, Even more than the 2N4427! I would be more than happy to give it a go? If i can just get hold of the lil buggers!

[teckniqs]

Ever tried a 2SC730 instead of the 2N4427 in this circuit? I found that I could get almost 2 Watts out, which was ideal for some of the FET PAs!

The reset mod that Jvok shows above is exactly what I did to the "Pira" PLL some years ago. I uses a 3n3 cap, but other than that it's the same.

I guess the theory may also work on the SINUS PLL? Its something i will look into!

The 2SC730 is quite an elusive component these days, Even more than the 2N4427! I would be more than happy to give it a go? If i can just get hold of the lil buggers!

Yeah I think I've recommended them to you a few years ago. I've had almost 3w out of one of those instead of the 1.2w from the 2n4427 in before it.

[XXL]

What frequency do you reckon this would go down to ? I wonder if it would lock to band 1 frequency’s and still put out 1w. Filter probably needs changing though.

[jvok]

What frequency do you reckon this would go down to ? I wonder if it would lock to band 1 frequency’s and still put out 1w. Filter probably needs changing though.

Funny enough I'm actually working on that right now. The saa1057 is only specd down to 70mhz but I've got a feeling it will go lower. I got the oscillator to cover 47-76mhz by changing the toko for a 4.5t yellow one. Will need a firmware change to program the pll for that range which I've done but for some reason isn't locking. Might need to change the pll filter because the loop gain will be different, both because of different oscillator tuning gain and also different divider value. But haven't got that far yet.

Don't see any reason why the pa won't work on any frequency even down to short wave with the right filter values and a bigger load choke.

[Albert H]

The 1057 gets a bit iffy below about 75MHz. There are a couple of approaches to overcome this - the obvious one is to use a lower frequency crystal. I used a 2 MHz rock, and got it to lock reliably down to about 38 MHz! As far as the PLL is concerned, it's "seeing" something within its normal range.....

To move that exciter down to Band 1, you're going to need to change the VCO coil and the output coils. If you (basically) double all the frequency-determining component values, the rig will operate at half frequency. Halving the crystal will get you there, but the lock-up time will double! You'll find that the '4427 will get close to 1.5 Watts because the gain is much higher at lower frequency - I had to reduce the supply voltage a bit to get the power back down to around a Watt - you're outside the safe operating area (the "SOAR") if you run more than about 1.3 Watts.

My Band 1 exciter uses a pair of BF199s for the oscillator (at ½f) and tune their collector load to the output frequency. The "power" stage uses a pair of 2N2219s in parallel, and it gives about 1.6 Watts at 12V. The PLL is just three 74HC CMOS ICs with a TL072 as the loop filter and modulation buffer. The whole thing fits on a 2" X 4" PCB, and fits nicely into a little Eddystone diecast box! I also made a higher power version with a pair of 2SC1971s for the output stage, with 2N2219s in the oscillator, and that gives 10 Watts easily at 12V, using the same PLL circuit. There were several London-based stations used those Band I boxes back in the 90s.

I always filled the Eddystone boxes of the 10 Watt ones with epoxy resin to stop people fiddling with them or copying them - I'll put the circuit up here if anyone's interested, since it's rather "old tech" these days.....

I found an old "airchain" system the other day when clearing out some old gear. Everything was built in neat little Eddystone boxes. There was a complete Band IV link transmit / receive pair, stereo coder and three-band audio processor, and even an early RDS coder using an EPROM and some logic! There's a station over in Ecuador that needs some gear, so it may all get shipped next week.....

[Albert H]

Looking at your board - it might be worth making space for a screen between the output filter and the rest of the board - it'd clean up the RF a lot. It could be worth laying it out a bit more "tightly", to shorten the RF-carrying tracks. The PLL layout is pretty much optimised (though Jvok is right about the reset circuit), but you could do quite a lot to shorten the RF side of the board. It would also be good to include an active modulation buffer stage. The board wouldn't need to be any bigger, but with a bit of extra hardware, you could turn this into a brilliant exciter. It's close as it is, but you can always improve it!

[sinus trouble]

Yeah I think I've recommended them to you a few years ago. I've had almost 3w out of one of those instead of the 1.2w from the 2n4427 in before it.

Haha! Yes indeed you did! I will have to keep an eye out for any available which are decent quality!

[sinus trouble]

Some excellent points Jvok and Albert!

Ive always been a bit sceptical of using the SAA1057 below its rated frequency, Yet it would be interesting to see what developments become?

[jvok]

I think I might have been a bit hasty with the reset thing, it still sometimes doesn't lock. It's definitly better with the reset circuit, but still not 100% reliable. Need to see if I can beg/borrow/steal a logic analyser to check the 1057 is getting programmed properly each time. Maybe I can hack something together with an arduino to sniff the bus? Otherwise I'm thinking maybe a loop filter issue, but kind of doubt that because of how it sometimes locks to the wrong frequency rather than just not locking at all.

On the band I front, I just couldn't get it to lock so I went back to the band II board to see how low that will go. But I can't for the life of me get it to lock below 82mhz. 82.0mhz is fine, 81.9mhz or lower just won't lock. Control voltage is about 3V at 82mhz with the toko run all the way in, so its not running out of tuning range. Wonder if its related to the unreliable lock issue? Maybe need to fix that first otherwise I could just be banging my head against a brick wall.

If excess gain turns out to be an issue in band I, I think it might be better to lower the drive power rather than the supply voltage? Otherwise you're pushing the amp into saturation which will make harmonics worse. Could maybe be done by taking some turns off the drive transformer. But not much point worrying about that yet when I can't even get the pll to work.

Would love to see those band iv link circuits you've been promising albert...

[jvok]

Well I'm an idiot...

I must have been looking at the pinout upside down or something because I got the /MCLR pin confused with the OSC1 pin. So that "reset" cap I changed is actually the RC timing cap, and all I was doing was making the PIC run slower. Would explain why it took longer to start up.

These are the correct mods this time:

[*] Add a 10n cap from PIC pin 4 to ground (pin 5 is ground so you can use that)
[*] Add a diode e.g. 1n4148, 1n916 etc. across the 10k resistor, with anode towards pin 4 and cathode to 5v

With these changes its more reliable although I've still seen it refuse to lock once or twice. More work needed I think.

[sinus trouble]

Well I'm an idiot...

I must have been looking at the pinout upside down or something because I got the /MCLR pin confused with the OSC1 pin. So that "reset" cap I changed is actually the RC timing cap, and all I was doing was making the PIC run slower. Would explain why it took longer to start up.

Ahh!! I thought you was slowing the clock rate at first!

Makes more sense now! A delay on the MCLR (Reset) in theory would do the trick!

[jvok]

It's weird how slowing the clock down did make a difference though. Reading through the code it looks like it programs the pll with high gain first then goes back and sets it to low gain after. I think the idea is to do the initial lock up quickly then switch to lower gain so the pll isn't fighting the modulation at low frequencies. But now I'm thinking maybe it's not spending enough time in high gain mode to lock reliably. Maybe the original pira vco locked faster than this one so it was fine in the original design but not here. And then me accidentally slowing down the clock made it spend longer in fast lock mode which made locking more reliable.

Also noticed when I had the slower clock it would actually lock much faster. As in there was a delay after power on where nothing seemed to happen because programming the pll took much longer, but then once it was programmed it would snap to frequency almost instantly. But now with the normal clock speed it starts locking straight away but can take 10 seconds to fully lock. Which I think confirms the theory that its not spending enough time in fast lock mode.

Either way I think slowing down the clock is a bodge and makes the boot time to long. Instead I'll try modding the code to wait longer in fast lock mode before changing the gain.