Page 2 of 2

Re: New NRG prototype

Posted: Sun Nov 04, 2018 9:29 am
Here’s my next PA, using a modern LDMOS MRFE6VP5300NR1, this one takes 1watt input for 350w out. The output LPF I designed for a rolloff of 115MHz and uses thicker 16swg tinned copper wire. 2nd and 3rd harmonics are about -60dBc, the coils also get quite hot!


Re: New NRG prototype

Posted: Sun Nov 04, 2018 1:30 pm
by Albert H

The CMOS PLL is really simple (a bit like me!). First of all, it has to be realised that the ordinary CMOS is only good to about 8 MHz, so you have to use 74HC or 74HCT versions of the chips in the faster parts of the circuit. Let's assume that we're using a VCO at ½f, feeding a buffer and a doubler. We'll take our RF sample from the buffer stage (I used to use a buffer with a BF494 / 199 with 10k in the collector and 1k2 in the emitter, and take the sample for the PLL from the emitter).

The ideal "prescaler" chip is the 74HC4024. This is good to at least 70MHz, but we only need it to work to 54 MHz. The logic all works off a 5V regulator (78L05). The input pin of the '4024 is biased with a couple of 10k resistors - one up to 5V and the other down to ground - keeping the input pin (pin 1) at 2V5. This increases the sensitivity of the IC, and obviates the need for a pre-amplifier - there's enough signal at the emitter of the buffer transistor to make the IC switch reliably - the emitter feeds pin 1 through a 1n capacitor.

There are several outputs from the 74HC4024 - division by 2, 4, 8, 16 and so on. I generally used divide by 4, so that the frequency range for 87.5 to 108 MHz becomes 10.9375 to 13.5 MHz in 12.5 kHz steps.

We need a 12.5kHz reference, and the cheapest crystal that's widely available is 4MHz. We need a crystal oscillator on pins 10 and 11: simply connect a 1MΩ resistor pin 10 to 11, add a couple of 33pF capacitors - from each pin to ground - then put the 4 MHz crystal from pin 10 to 11. If you power the IC up with 5V at this point, you'll see that you get 4MHz at pin 9 - a useful test point. If the frequency is fractionally low, reduce one of the 33pF capacitors to bring it up a bit, and I'm sure you can work out what to do if the 4MHz is a bit high....

We have to divide 4MHz by 320 for the reference. Luckily, 320 is easily achieved using the outputs from the 74HC4060, and a bit of diode programming. It's done with 2 diodes and a 4k7 resistor: we use the divide by 256 (pin 14) and divide by 64 (pin 4). Connect the cathodes of two 1N4148 diodes to those pins, connect the anodes together to pin 12 (reset) and connect the reset pit to pin 16 through the 4k7 resistor. You take the output from pin 14, and you'll find that it's a squarewave of exactly 12.5 kHz and virtually 1:1 M/S ratio.

We now have to divide our pre-scaled signal down to 12.5kHz for comparison with the reference. Let's assume that we want our rig to be on 98.4MHz. The output from the prescaler will be 12.3000MHz, so we have to divide the pre-scaled signal by 984 to get 12.5 kHz. The cheapest way to do this is to use a diode-programmed 74HC4040. If you wanted to get clever, you could use BCD-coded rotary switches and 74HC4059 - the number shown on the switches would be the output frequency - but this would add to the expense and make the rig easily re-tunable by thieves!

The 74HC4040 is programmed in the same way as the '4060 - using the 512, 256, 128, 64, 16 and 8 outputs, six 1N4148 diodes and a 4k7 resistor. The output is taken from the 512 (pin 12), and again will be a squarewave with a very close to 1 : 1 M/S ratio. When the buffer of the exciter is connected through the prescaler and the pre-set divider, we'll get a (nominal) 12.5 kHz.

The reference and the output of the preset divider are fed to pins 3 and 14 of a 4046 PLL IC. We can "get away" with using an ordinary CMOS 4046, since we're well inside the frequency range that it can handle, but for the sake of convenience and because of economies of scale (I use the 74HC4046 in lots of other stuff), I stuck with the HC version of the chip. Pin 5 of the IC needs to go to +5V to disable the internal oscillator, and we take our output from pin 13.

The pin 13 output is a series of "phase pulses" which - when filtered - give a DC voltage that's proportional to the error between the two frequencies coming into the '4046. If both frequencies are identical, we end up with 2V5, and this will fall for a higher frequency divided signal, so feeding the control voltage back into the tuning input of the VCO means that the VCO will "lock" to a multiple of the crystal frequency. Remember that we want to apply "deviation" to the VCO signal, and our PLL will "see" this as a tuning error, so will try to correct it, unless we slow down the changes in the tuning control voltage to lower than the minimum modulating frequency. We can use a passive filter to achieve this, and a further benefit is that no 12.5kHz ripple appears on the rig's output - it would sound like a high-pitched whistle.

That's how the CMOS PLL works. The four ICs you need can be bought for <£1 and the crystal is around £0.30. The 1N4148s are a penny each and the passive components might come to a further 10p. It's really cheap, flexible and completely reliable. The 78L05 is available for around £0.25 - remember to bypass its input and output to ground with 10n capacitors to keep it stable with RF around. It's also good practice to bypass the supply pin of each IC with (say) 10n to keep things stable.

Re: New NRG prototype

Posted: Sun Nov 04, 2018 10:26 pm
by Albert H
4th Paragraph should say that the reference oscillator / divider IC is a 74HC4060. As the crystal is only 4 MHz, you could get away with the lower speed (ordinary CMOS) part, which might be slightly cheaper, but I've found that the diode programming doesn't work as well with the ordinary CMOS - the edges are more clearly defined and faster with the 74HC versions.

Re: New NRG prototype

Posted: Sun Nov 04, 2018 10:48 pm
by Albert H
One further thing - we used to fit all the programming diodes around the 74HC4040 and then snip a leg of the ones we didn't want. The other option we used was to lay the PCB out so that the diode programming could be done with simple solder blobs bridging tracks. DIP switches to achieve the same thing would have more than doubled the price of the PLL!

We applied the most parsimonious engineering principles to the design of those rigs - we avoided the use of heatsinks wherever possible, as they were expensive. Regulators were always just bolted to the case (usually with the tab isolated to prevent ground loops). One run of 2N2866, 2N3375, 587BLY PAs were built on to aluminium extrusions that were used for office room dividers or cubicles. We found loads of these in a skip in Harlesden, and quickly realised that a 40cm length of this stuff would have a thermal resistance low enough to work, and it was the perfect shape to duct air. A fan was screwed to the end of a pipe made from two lengths of this extrusion, and a paper filter (cut from a vacuum cleaner dust bag) went on the other end, so muck wouldn't be drawn down the tube. The exciter went inside - in its own tinplate screening box, and the linear PSU - a cheap 24V secondary E-I transformer, a 10A bridge rectifier and a big cylindrical smoothing capacitor also fitted inside the tube. Several stations used these weird-looking tubular rigs. We made about 40 of them them over about 8 months, until we ran out of the aluminium extrusions. Cost of these 90 Watt, mains powered, PLL-controlled rigs to construct? About £40 each! We sold them for quite a lot more......

Re: New NRG prototype

Posted: Sun Nov 18, 2018 1:31 pm
by gurudattapanda
Please provide the new NRG schematics? is it same as early 7 watt FM tx? what modification needs to be done for RD mos?

Re: New NRG prototype

Posted: Sun Nov 18, 2018 11:46 pm
Mostly the input matching is different with the addition of a second bfr91 to get the 50mW or so drive that’s required for the rd06hvf1. The schematic will only be distributed with the product, so you’ll need to purchase one when I get them in production - if I put the schematic up here then it would get copied. The new design will also use a PIC with code protection enabled, anyone trying to copy the schematic will need to write new PIC firmware or use a different PLL design.

Re: New NRG prototype

Posted: Mon Nov 19, 2018 1:26 am
by 3metrejim
I can write PIC firmware, no problem (8 bit PIC, in assembly - used to do it as a full time job years ago). I wouldn't worry too much about your design being copied, as no-one (pirate type, that is) is willing to pay to have code written specially for something like this - would be a couple days work probably.