Analyser's Drawer Of Pirate History

[Dai Pole]

Was looking at some stuff I have saved off Radionecks, and found this:

The circuit was from Germany (where LB still lives). It was used in London as a driver board in the 90s.

The earliest PLL exciters I recall in London were back in the late 70s! Myers (RFL) built one that was meant to be on 92.1MHz, but used to lock somewhere around 104MHz at switch-on. If you gave the rig a thump, it locked in the right place. He looked it over several times, but never found the intermittent connection! It was "lost" to the Post Office after a few weeks.

I used to build PLLs in the late 70s and early 80s - using ¼f generation and multiplying to the output frequency. This meant that the oscillator came inside the speed of simple TTL chips. I used to run the ICs at 6V instead of 5V - they ran hot, but could do the 22 - 25 MHz I needed. The simpler TTL chips (bistables - 7474) were the fastest, so I used a 7474 from the VCO to drive my subsequent divider. The reference was (usually) a 4MHz or 5MHz crystal, and the phase comparator was done using a 7474 and a NAND gate. The rest of the NAND gates were used in the lock detector.

The earliest CMOS PLLs used reference crystals that were related to the required output frequency. The input to the circuit used a prescaler (usually the Plessey SP8629) which divided the output frequency by 100. This went into a 4024 and was divided by a further 8. The reference was a crystal connected to a 4060 which gave division by 16. These two signals were then compared in a 4046. The crystal had to be 1/50th of the required output frequency. Luckily, there were some standard crystals that gave useful frequencies in Band II.

I then introduced a preset divider, using a diode-programmed 4040 and a diode programmed 4060. This gave complete flexibility and allowed PLLs to be built for any frequency. My standard PLL for a long time was a 74HC4024, 74HC4040, 4046, and a 74HC4060, a whole lot of 1N4148 switching diodes, and with a TL072 dual op-amp in the loop filter. These used a 4MHz (probably the most common frequency available) and the whole synthesiser cost about £2 to build!

There was a large run of 1 Watt boards with this PLL, a ½f oscillator, buffer, doubler, filter / amplifier and 2N4427 final on double-sided PCBs about 4" X 2". There were three coils to adjust (VCO and two doubler coils - all S18s) and three trimmers (drive and two in the output match - a green and two yellows). Division ratios were selected by cutting tracks to the diodes. I was given two of these boards recently to have their frequency changed - they were almost 30 years old, but still worked perfectly.

The "Dawson" board was partially a rip-off from this old design of mine. He used a 74HC4059 for his pre-settable divider because he found calculating which diodes to cut too difficult and because he could scrub the numbers off the top of the chips and claim that it was "microprocessor controlled". His version cost about 5 times as much to build as the originals!

Haha, it's funny you say that because Mr. Dawson once told me many years ago that the design was based on another one which he'd worked on to greatly improve their performance.

Shall I have a dig around in my drawer of pirate history and see if I can find some of these?

Here you go then...



This is a driver from the late 80s/ early 90s. Pretty much as Albert describes with SP8629/ 4046/ 4060/4024.



Then we have this, from 1994, using a similar PLL but slightly different RF approach.



Then we have something from around 1998 ish. It uses an interesting approach for the PLL dividers and uses two PIC chips, one with a fixed divide ratio and the other with a variable divide ratio to set the frequency. It gets around the problem of the first two designs where you had to get a crystal cut for each frequency but introduces some of it's own. In the early days of PIC they all ran at 4MHz so to get a 50KHz channel spacing means you end up with a phase comparator frequency of about 400Hz which makes PLL loop filter design more tricky.

That last one was built by "N" as you are well aware.

....I remember when he/you first built those, was a simple VFO along side a seperate PLL board before, maybe one of those Frequency synth things made by Venture Electronics?

Dai, yes you're right, all the girls want to see in my drawers but when they get there they're a bit disappointed because it's all circuit boards!

So Teck, who built these bad boys then???

[sinus trouble]

Lol thanks for posting!! Love all this old stuff!!!
All the new SMT boards around today with two digital buttons to tune up n down takes all the fun out of it!!

[pjeva]

This old designs are good to learn from. Different rf parts, and pll combinations. Nowdays, it is hard to find parts (or too expensive), so many beginner projects start with pic and all-in-one chips. It only takes minutes on internet, and you can build 1w driver board with pll. I don't think it is good approach to learning rf electronics, and after almost 20 years I still use 4059 divider design for pll sometimes. Very important thing to say is that ready made pll chips and pic microcontrollers can be very unstable in high rf field. That is why lot of serious rf equipment still use CMOS circuits. Digital parts and microcontrollers in Elenos and similar are for low (audio) frequency processing and electronic substitute for DIP switches and diode dividers.
Please do post more of old designs if you have some...

[Dai Pole]

Well it's Analyser we have to thank for all those juicy pictures, great to see, especially with Albert H's explanations.

I agree that all the old stuff is great to learn from, and with the availability of new tiny logic devices perhaps there will be a resurgence in this way of making a PLL, especially now that PLL chips are getting harder to find and more expensive too! After all these years of Albert drumming it into our heads that CMOS logic is still a viable solution, I've come to appreciate the elegance in these designs. They're easier to understand how they work too.

I don't know how much stuff I saved from Radionecks, and it's scattered all around my hard drive, but if I find anymore stuff I'll stick it up.

[thestig]

I wonder if Albert will come to join this new improved RN site? As himself I mean.

[Dai Pole]

I thought the whole point in this incarnation of Radionecks was that it intended to be as much like the old one as possible? Otherwise we'd all be over at Radio Anoraks where I've heard Albert holds a mean candlelight supper.

[shuffy]

I wonder if Albert will come to join this new improved RN site? As himself I mean.

If Albert came as himself, you wouldn't know who it was. I looked at this "anoraks" site a while back, didn't seem to be much going on?

[shuffy]

with the availability of new tiny logic devices perhaps there will be a resurgence in this way of making a PLL, especially now that PLL chips are getting harder to find and more expensive too!

I quite agree Dai! Didn't we have a chat on the old site about how the NRG exciter (which is great apart from the size) could be made smaller? That's quite viable for the PLL and PA (I've been using MRF4427's for a while now) but not sure about the oscillator.

[sde-1104]

I wonder if Albert will come to join this new improved RN site? As himself I mean.

If Albert came as himself, you wouldn't know who it was. I looked at this "anoraks" site a while back, didn't seem to be much going on?

He is more than welcome back on here if he wants to sign up

[Dai Pole]

I quite agree Dai! Didn't we have a chat on the old site about how the NRG exciter (which is great apart from the size) could be made smaller? That's quite viable for the PLL and PA (I've been using MRF4427's for a while now) but not sure about the oscillator.

Yes Shuffy, we did indeed. I'd noticed that it may be possible to build the non 4059 version of the NRG PLL in a very tiny space due to the availability of similar or work alike chips being available from NXP in DHVQFN14 & DHVQFN16 format. I've also made two layouts of the full NRG oscillator section - both SMD & TH - that take up a much smaller footprint than the original, with the TH version having space for three different package types of varicap. The only reason I haven't continued with the project is because I've run out of money, hence why I rarely ask any questions at the moment, so I am concentrating on getting even deeper into C programming for now - which is free!

How much power do you get out of the MRF4427? I remember looking for more information on that particular chip a while back and couldn't find much.

[shuffy]

I'd noticed that it may be possible to build the non 4059 version of the NRG PLL in a very tiny space due to the availability of similar or work alike chips being available from NXP in DHVQFN14 & DHVQFN16 format.

Bit of a tricky one though, this. If you're making a small number of exciters (i.e. they're going in TX boxes which are going to get nicked from time to time) then you want them cheap and easy to build but I'd say smaller than the NRG PRO boards. So whilst SMD is definitely a winner for size, I'd rule it out for that kind of application.

Sinus probably has the right idea with his NRG VFO/Pira combination. I'm actually tempted to try it as I've never used Steve Moss' oscillator (the design I prefer is similar to the one shown evolving in the analyser history thread) and I have a few SAA1057s knocking around. I still think it's larger than it could be, hence my comment about the MRF4427. If I was building rigs at a rate of knots then it probably would be worthwhile getting a contract manufacturer to build SMD PLLs using a design like the NRG. But whilst I'd like a nice SMD PLL board, it would probably just be a curiosity as I'd put so much effort into building it that I wouldn't want it getting stolen.

How much power do you get out of the MRF4427? I remember looking for more information on that particular chip a while back and couldn't find much.

I started using these as they're cheap and the metal ones were getting scarce. In terms of power they're comparable with the metal ones, so in the NRG circuit I'd expect somewhere in the high hundreds of mW without too much trouble but careful how hot they get in this particular application. I only use them at around 300mW because the next stage in my current design is an RD06HVF1 which can easily amplify this up to around 8 or 9W to drive a bigger device. The MRF4427s run quite happily at that power without any special heatsinking arrangements.

[Dai Pole]

Bit of a tricky one though, this. If you're making a small number of exciters (i.e. they're going in TX boxes which are going to get nicked from time to time) then you want them cheap and easy to build but I'd say smaller than the NRG PRO boards. So whilst SMD is definitely a winner for size, I'd rule it out for that kind of application.

Sinus probably has the right idea with his NRG VFO/Pira combination. I'm actually tempted to try it as I've never used Steve Moss' oscillator (the design I prefer is similar to the one shown evolving in the analyser history thread) and I have a few SAA1057s knocking around. I still think it's larger than it could be, hence my comment about the MRF4427. If I was building rigs at a rate of knots then it probably would be worthwhile getting a contract manufacturer to build SMD PLLs using a design like the NRG. But whilst I'd like a nice SMD PLL board, it would probably just be a curiosity as I'd put so much effort into building it that I wouldn't want it getting stolen.

Yes you're right, DHVQFN are too small to be practicable - it's been a while since I thought of this the first time around, and as such had forgotten what I was originally thinking, but I have been looking at my notes due to this thread and have remembered.

Initially I was looking at the NXP DHVQFN stuff, but after I realised the package size was smaller than a grain of rice I started searching for alternatives in larger packages. That's when I came across the TI Little Logic series. They have a nice selection of gates and flip flops in SOT-23 5/6 format - that's not too small to solder is it?! I can hand solder SOT-23 transistors.

In the NRG PLL, IC2 uses one JK flip flop out of a dual package, wasting the space the unused one occupies; IC3 uses two XOR gates out of a quad package, again wasting the space the other two occupy; and IC4 is a dual D flip flop which could be split into two single D flip flops in order to use the TI chips and economise on size. The 193 & 4060 chips could use the standard SOIC versions, so not much of a space saving there, but with the TI Little Logic chips able to fit around them I think it would be a lot smaller and not too difficult to make. No?

[shuffy]

In the NRG PLL, IC2 uses one JK flip flop out of a dual package, wasting the space the unused one occupies; IC3 uses two XOR gates out of a quad package, again wasting the space the other two occupy; and IC4 is a dual D flip flop which could be split into two single D flip flops in order to use the TI chips and economise on size. The 193 & 4060 chips could use the standard SOIC versions, so not much of a space saving there, but with the TI Little Logic chips able to fit around them I think it would be a lot smaller and not too difficult to make. No?

A good point, might be worth a try. For me, the SOIC IC's aren't so much of a pain, it's the unmarked discretes like capacitors - messing around with tweezers etc. I reckon soldering takes longer, but on the plus side there are other advantages like no drilling. So, swings and roundabouts.

I think board development would take longer though. I wouldn't want to prototype RF in the SMD domain. Not really high frequency or complex stuff, anyway! Iterating and rework would be a right pain - and if you prototype with normal sized components which are easy to iterate with, you're not guaranteed a successful transition to SMD because of layout. For something like the NRG exciter though, it shouldn't be too much of a problem.

[sinus trouble]

Some good points raised!!
The good points of the steve moss design is it rejects harmonics (apart from around 50mhz) and isolates the output from load conditions! the bad points are its bulky!
so the question is? could it be made smaller and easier? i think yes!
The BF199s and tank coils in the oscillator could be replaced with equivalent SM components and easily soldered with a dab of solder paste and some heat!
the only drawback i see is the tuning caps?

[teckniqs]

the only drawback i see is the tuning caps?

[reverend]

Love all these old designs. Similar to mine which used an SP8629 and then the usual 4060/4059/4046 combination like the badly over modified one below.

synth23_pic.jpg

Has anyone tried using the 4040 as a Div by N with diodes. Does this work well?

The Veronica frequency doubler design could be made a lot smaller by replacing the 6 individual 2 turn coils with a single 10 turn coil tapped every 2 turns wound on a suitable toriod (i.e T37-17). You could also take the 50 MHz signal for the PLL from a single turn on the same toroid. Just my two penneth.

[Bton-FM]

Love all these old designs. Similar to mine which used an SP8629 and then the usual 4060/4059/4046 combination like the badly over modified one below.

synth23_pic.jpg

Has anyone tried using the 4040 as a Div by N with diodes. Does this work well?

The Veronica frequency doubler design could be made a lot smaller by replacing the 6 individual 2 turn coils with a single 10 turn coil tapped every 2 turns wound on a suitable toriod (i.e T37-17). You could also take the 50 MHz signal for the PLL from a single turn on the same toroid. Just my two penneth.

Yes Albert has and he raves about it. The schematic for it is on here somewhere.

I made a board which had the NRG doubler oscillator with the 4060/4020/4046/4040 PLL but that fell through and I never finished building it. Now, I'm designing a board with an @ frequency ocscillator and TSA5511 because it's simpler, cheaper and less effort than winding 6 coils exactly the same. You might see a post about it soon ...

That idea about the toroid is a shout I'd like to see if it works. It would make it easier to do an SMT version of the oscillator!

[Albert H]

Believe me when I tell you that the push-push oscillator with six two-turn coils has been built in dozens of differing configurations. Stephen and I spent a lot of bench time getting that oscillator optimised. I suggested the change to FETs for the "sniff" and "balance" transistors (as in the Pro III), to reduce any loading on the oscillator itself - we got an improved output with both more p-p volts and a purer waveshape. The "Veronica" version from "up the road" removed the balancing transistor from one side of the oscillator, with entirely predictable results - he has about as much ½f getting to the next stage as he has the required frequency, What a maroon! (to quote Bugs Bunny).

The 74HC4040 programmed with diodes works flawlessly. You have to remember that the output from the divider isn't (quite) a 50:50 M/S ratio, so if the phase comparator you're using demands that, you have to divide the 4040 output by 2 with a bistable. One phase comparator I really like uses a dual bistable (74HC74) and four NAND gates. It does demand 50:50 inputs for purest results, so I feed its inputs from an extra 74HC74, using one side for the 4040 divider, and the other for the crystal reference divider (a 74HC4060 also diode programmed in the same way as the '4040). It adds one IC, but the advantages are immense - the phase comparator gives a very clean DC control voltage (minimising control loop ripple), and it also gives an absolutely solid locked / unlocked output, which you can trust to switch the RF output.

The other thing about the push-push oscillator - that always really annoyed me - was the trimmer to set its range. I came up with some modifications that switched in capacitors as you went down the band, using the programme pins on the '4059 to switch them. It was unwieldy, but it removed the trimmer! Incidentally, the idea of the 4059 (instead of the cheaper three dividers) was mine. It allowed the use of BCD-coded rotary switches to set the frequency - nice and obvious, and much easier than the row of DIP switches (which always caused problems for the hard-of-thinking).

The other problem was the chain of six identical two-turn coils. Stephen had them made in bulk by a spring manufacturer, which alleviated the tedium of making them ourselves, but still didn't get rid of the need for them. I tried etched coils - with reasonably good results - but the boards were always microphonic, and the oscillator phase noise figure worsened because of the reduced "Q" of the inductors. We tried making them with moulded chokes - they worked, but had to be hand-selected, and carefully mounted the right height above the board. I tried a tapped 12-turn coil on a toroid with great results (and a physically tiny layout), but winding and tapping the coil was incredibly fiddly!

Also - we tried every PLL IC we could get hold of. Philips, Motorola, Toshiba, Mitsubishi and many more, and none of them gave results as good as the discrete logic version. I did try changing the phase comparator (I never really liked the '86 gate used that way) - I built a couple of prototypes using the 4046 as the comparator, followed by a dual op-amp for the loop filter. It worked really well.

Another thing I tried to eliminate the oscillator trimmer was to provide a 33V varicap supply rail. I built an oscillator (using the redundant gates in the '86), with a diode pump and a 33V zener (the TV tunerhead type), and scaled the capacitors around the oscillator to suit. I could get the oscillator to tune the whole length of the band (with tuning voltage going from around 1.5V to 31V), but the mod sensitivity was wildly variable up the band, and the oscillator p-p output voltage varied dramatically! I overcame the modulation issue by use of a second varicap, biased from the control voltage scaled down through a resistive divider, and with some amount of fiddling was able to get a reasonably similar mod sensitivity across the band. Unfortunately, the increased sensitivity of the tuning varicap made the board more sensitive to hum and noise, and I wouldn't have wanted to use it in the same box as a big PA!

A further refinement to the push-push oscillator that I used with a lot of success was the addition of a SAW bandpass filter module after the output. This wouldn't allow any ½f (or 1½f, 2f etc) through at all. The ones we used were Japanese, not too expensive, and could handle about a Watt pushed through them. The improvement in spectrum was remarkable!

If anyone wants to give the CMOS PLL a go, I'll be happy to answer (sensible) questions. It's by far the cheapest PLL, doesn't require special crystals (I usually use a 4MHz rock which costs about 30p in quantity), and can be built for pretty much any frequency. The practical upper frequency limit (using the 74F series ICs) is around 120MHz, but using the ordinary 74HC or HCT stuff is around 60MHz. If you want to go higher (up to 1GHz), you can use one of the cheap ÷64 prescaler ICs to get you into the range the PLL can handle. The only real disadvantage with this approach is that you have to know what you're doing. The average solder-jockeys throwing together junk rigs for the pirate market aren't usually big on thinking much.....