SINUS 5 WATT PLL

[Shedbuilt]

An active loop filter (using an op-amp or two) is easier to configure than a passive one. It does necessitate the use of a few more components, but the results will be worth it!

I think they may doing an element of that - at least for the long time constant, in the Pira design. R2 and C4 look like a feedback network around the SAA1057's PCA ?

[sinus trouble]

As exquisite an op amp would be? I fear it would increase size of the board?
I believe R8 would have given a small voltage to the varicap in case of oscillator failing to start up correctly? I found that R8 nor the original NRG 78L05 was no longer required? Surprising to me also, the unit fires up and locks everytime lol
It is increasingly looking like an added varicap is by far my best option! however I did have a glance at PIRAs VFO that was used with that PLL?

Pira vfo.gif

It is very similar to the configuration i have but with significant value differences?
R4 R5 are 27k whereas i have 6k8s
C2 is 0.22uF and C3 C4 are a pair 100pFs

PS does anyone know is there a way of uploading a short vid to show the board working?

[Shedbuilt]

I believe R8 would have given a small voltage to the varicap in case of oscillator failing to start up correctly?

Yes. Sorry Sinus, I missed the bit which said "optional" - as well as the write up, and misinterpreted its function.

[sinus trouble]

No worries Mr shed
As you suggested in earlier post, i think its worth tweaking the RC values to maybe reach an acceptable compromise! rather than modifiying the tank circuit and creating allsorts of other problems!
For such an inexpensive and simple circuit that already performs beyond my expectations? im probly pushing my luck! lol

[Analyser]

No worries Mr shed
As you suggested in earlier post, i think its worth tweaking the RC values to maybe reach an acceptable compromise! rather than modifiying the tank circuit and creating allsorts of other problems!
For such an inexpensive and simple circuit that already performs beyond my expectations? im probly pushing my luck! lol

Sinus, can you post a link to the Pira PLL section you're using? The only one I could find after a quick check yesterday was the one with a 16F84, some dipswitches and a few other surrounding components. I couldn't see R8 which the Shedmeister was referring to.
In the circuit you posted a post or two above I can see a couple of things, C2 is 0.22uF which looks like a sensible value for RDS, but also the audio passes through that as well.
Also, the varicaps don't seem to have any DC biasing; perhaps they're supposed to be biased by the PLL but since you want a wide lock range the mod level will vary hugely over the band.

I take your point about the circuit being nice and simple, and sure it does work but personally I think there are a few improvements which should really be done to achieve reasonable performance.
The thing about PLL loop stability is that although it looks stable with your 0.47uF audio cap and no mod, how do you know it's stable with mod?

[sinus trouble]

Here is the link Mr Analyser

http://pira.cz/enpll.htm

Here also is a rough guide to what i did

5w stereo schematic.jpg

The 78l05 gives between 3-4 volt on varicap if i remember correctly? and gives the best sound quality? when i tune up the board i always trim VC1 finely to obtain around 4 volt output from the pll feeding the varicap

Im not entirely sure show stable it is with modulation? i dont have access to test equipment
On the other hand? i did once run the 1watt version for two weeks solid with modulation and it seemed very stable? dunno if thats any help? lol

[thewisepranker]

Just had a quick glance. You could reduce power consumption by a fair amount by removing D1, as a 5 W PLL is ideal as-is for the battery powered lot out there. I don't know how efficient it is, but there is probably around 700 mW, perhaps even a Watt, being dissipated in D1.

[Albert H]

Just had a quick glance. You could reduce power consumption by a fair amount by removing D1, as a 5 W PLL is ideal as-is for the battery powered lot out there. I don't know how efficient it is, but there is probably around 700 mW, perhaps even a Watt, being dissipated in D1.

Stephen liked to include an "idiot diode" to prevent reversed connection of the supply. The other, no-loss way of protecting the board is to feed the supply through a suitably rated fast-blow fuse, with a (normally) reverse-biased power diode at the board side of the fuse. A reversed supply connection should cause the diode to conduct heavily, blowing the fuse before damage can happen to the rest of the circuitry - hopefully. Unfortunately, I've seen too many fuses replaced with aluminium foil, nails, bits of wire or anything else metallic!

My polarity protection circuit used to use a car horn or headlight relay (cheap from Halfords, and can switch a lot of current), with the relay coil fed from a diode. The relay contact would remain open, leaving the board without supply - if the supply was reversed - and a red "error" LED would light. If the supply was the right way 'round, the relay would fire, the green "OK" LED would light and the board would be powered through the relay switched contacts.

Similarly, I used a red LED to indicate "out of lock" and a green one for "locked". There was also a green LED for "PA on" - fed through another car relay and only enabled a few seconds after the PLL had locked. The PA also had a red LED for "error" for over-SWR, and the power to the PA would be switched off after a couple of seconds.

I also had a simple LED indicator for the modulation level - red to "too high", green for OK, and another couple of LEDs monitoring the supply voltage - useful with battery-powered rigs. When the bi-colour red / green LEDs became available at reasonable prices, I used those.

The whole rig had a row of LEDs - a row of green, and everything was OK. Any red LED would show that the rig was off, and what had caused the shut-down. Adding the indicators cost a matter of a few pence and meant drilling a few extra holes in the panel, but could really save time when someone with little electronic knowledge was trying to set up a rig! A red LED would warn that the supply was wrong, or he'd forgotten to connect the aerial, or set to mod too high (or low!).....

One thing to remember - nothing can be certified "foolproof" until you've used a real fool to test it!

[Analyser]

Here is the link Mr Analyser

http://pira.cz/enpll.htm

The 78l05 gives between 3-4 volt on varicap if i remember correctly? and gives the best sound quality? when i tune up the board i always trim VC1 finely to obtain around 4 volt output from the pll feeding the varicap

Im not entirely sure show stable it is with modulation? i dont have access to test equipment
On the other hand? i did once run the 1watt version for two weeks solid with modulation and it seemed very stable? dunno if thats any help? lol

That's the PLL circuit I found.
So, you're saying that R33 and R34 are 6k8, and that C25 is 220nF and C9 is 470nF on your circuit (which is the NRG diagram like above)?
Looking at Pira's PLL filter. the values look sensible but I suspect there is an interaction between R3, C5 (on the Pira PLL) and the components you have following it causing instability.

I would try this; Increase value of your audio coupling cap (C9?) to make the PLL unstable. Reduce the value of C5 (Pira PLL). See if that affects the instability/ gets rid of it.
Increase R33 (NRG diagram) quite a bit. Note that the mod level will decrease.

Overall the best solution would be to add an extra varicap so you could get the PLL loop perfect and then adjust the mod circuit to suit. If you wanted to do that you could add another varicap in parallel with the existing one and reduce the value of C27 accordingly. The mod varicap would need to be DC biased again at your 4v or whatever.

[Shedbuilt]

Here is the link Mr Analyser

http://pira.cz/enpll.htm

The 78l05 gives between 3-4 volt on varicap if i remember correctly? and gives the best sound quality? when i tune up the board i always trim VC1 finely to obtain around 4 volt output from the pll feeding the varicap

Im not entirely sure show stable it is with modulation? i dont have access to test equipment
On the other hand? i did once run the 1watt version for two weeks solid with modulation and it seemed very stable? dunno if thats any help? lol

That's the PLL circuit I found.
So, you're saying that R33 and R34 are 6k8, and that C25 is 220nF and C9 is 470nF on your circuit (which is the NRG diagram like above)?
Looking at Pira's PLL filter. the values look sensible but I suspect there is an interaction between R3, C5 (on the Pira PLL) and the components you have following it causing instability.

I would try this; Increase value of your audio coupling cap (C9?) to make the PLL unstable. Reduce the value of C5 (Pira PLL). See if that affects the instability/ gets rid of it.
Increase R33 (NRG diagram) quite a bit. Note that the mod level will decrease.

Overall the best solution would be to add an extra varicap so you could get the PLL loop perfect and then adjust the mod circuit to suit. If you wanted to do that you could add another varicap in parallel with the existing one and reduce the value of C27 accordingly. The mod varicap would need to be DC biased again at your 4v or whatever.

Indeed. Not sure if I'm on the right track, but I think the output loading may be making the PCA of the SAA1057 go unstable - in which case I think the output loading may also be interacting with Pira's R2 / C4. Reducing the output loading is the easiest thing to try, and I think either of the above is a good starting point. Increasing R33 may affect the audio response as well as level, but will help to reduce the influence of the audio coupling cap on the PLL.

[Analyser]

That's right Shedmeister. The Pira R2 & C4 form the main PLL loop filter and R3/ C5 form another filter to further reduce phase-comparator leakage on to the carrier. The bandwidth for the second pole of the filter is commonly a multiple of the cut-off of the first filter (x5 - x10) and indeed the Pira's circuit has a cut-off of ten times the main filter. This is to avoid significant phase changes at the main loop frequency and therefore cause instability of the loop.
The problem with Sinus' circuit is that there are other components effectively in parallel with R3/ C5 which form another pole on the audio side. A quick fix might be to increase R33 and try to isolate the audio from the PLL filter, or play with R3/ C5 but the correct solution, I think, is to fit a second varicap.

Incidentally, the SAA1057 loop filter is a bit odd (to me at least). Generally you see one of two types, one with two resistors and a capacitor and the other a charge pump type with two capacitors and a resistor. In the SAA the filter would be the first type with the components in the feedback loop of an active filter, but we only get to change R2/ C4 and don't get to see the value of the other resistor. Maybe I'm missing something or not read the datasheet thoroughly enough...

[Shedbuilt]

That's right Shedmeister. The Pira R2 & C4 form the main PLL loop filter and R3/ C5 form another filter to further reduce phase-comparator leakage on to the carrier. The bandwidth for the second pole of the filter is commonly a multiple of the cut-off of the first filter (x5 - x10) and indeed the Pira's circuit has a cut-off of ten times the main filter. This is to avoid significant phase changes at the main loop frequency and therefore cause instability of the loop.
The problem with Sinus' circuit is that there are other components effectively in parallel with R3/ C5 which form another pole on the audio side. A quick fix might be to increase R33 and try to isolate the audio from the PLL filter, or play with R3/ C5 but the correct solution, I think, is to fit a second varicap.

Incidentally, the SAA1057 loop filter is a bit odd (to me at least). Generally you see one of two types, one with two resistors and a capacitor and the other a charge pump type with two capacitors and a resistor. In the SAA the filter would be the first type with the components in the feedback loop of an active filter, but we only get to change R2/ C4 and don't get to see the value of the other resistor. Maybe I'm missing something or not read the datasheet thoroughly enough...

I think it differs from an Op Amp in a couple of ways. The PCA of the SAA1057 is a current amplifier, where the gain is software configurable (4 bits CP0 to CP3).

[Analyser]

I think it differs from an Op Amp in a couple of ways. The PCA of the SAA1057 is a current amplifier, where the gain is software configurable (4 bits CP0 to CP3).

I wonder if there's any literature on this type of loop filter to be found on the interwebs. If I get some time tonight I'll have a search.

[thewisepranker]

Stephen liked to include an "idiot diode" to prevent reversed connection of the supply. The other, no-loss way of protecting the board is to feed the supply through a suitably rated fast-blow fuse, with a (normally) reverse-biased power diode at the board side of the fuse. A reversed supply connection should cause the diode to conduct heavily, blowing the fuse before damage can happen to the rest of the circuitry - hopefully. Unfortunately, I've seen too many fuses replaced with aluminium foil, nails, bits of wire or anything else metallic!

I'm not convinced that the fuse and reverse-biased shunt diode would catch it fast enough.

I do condone the idiot diode, however I think there's a better way of doing it:



In normal operation (forward bias), with Vin anywhere between 5 V and 60 V, the body diode starts conducting like a normal diode. Unless there is a direct short between the Vout (the source) and return (drawn as ground, which is where the gate is referenced), the gate will always be negative with respect to the source, turning the MOSFET on and putting Rds(on) in parallel with the relatively large equivalent series resistance of the body diode. I mention that Vin should be above 5 V, as much below this won't be enough to drive the FET into or close to saturation. The 15 V Zener protects the gate and can be any value between around 10 V and 18 V.

When reverse biased, or in other words when the input is connected backwards, the body diode is also reverse biased in the same way a normal idiot diode would be and Vgs is always positive, which keeps the FET off.

Typically, P-channel devices are twice the price of N-channel devices and have on resistances around five to ten times that of their N-channel counterparts. I stumbled across the IPP80N06S4L-07 a while ago when looking for a device for just this application and found them on eBay in TO-220 packages for £1 each (well, £5 for 5 + 99p delivery) from the UK. Rds(on) is only 6.4 mΩ when Vgs is only -10 V. They take this measurement at a drain current of 80 A, however Rds(on) reduces as Idrain reduces. This means that it is ideal for 12 V (or 13.8 V) circuits.

[Analyser]

Stephen liked to include an "idiot diode" to prevent reversed connection of the supply. The other, no-loss way of protecting the board is to feed the supply through a suitably rated fast-blow fuse, with a (normally) reverse-biased power diode at the board side of the fuse. A reversed supply connection should cause the diode to conduct heavily, blowing the fuse before damage can happen to the rest of the circuitry - hopefully. Unfortunately, I've seen too many fuses replaced with aluminium foil, nails, bits of wire or anything else metallic!

I'm not convinced that the fuse and reverse-biased shunt diode would catch it fast enough.

I do condone the idiot diode, however I think there's a better way of doing it:



In normal operation (forward bias), with Vin anywhere between 5 V and 60 V, the body diode starts conducting like a normal diode. Unless there is a direct short between the Vout (the source) and return (drawn as ground, which is where the gate is referenced), the gate will always be negative with respect to the source, turning the MOSFET on and putting Rds(on) in parallel with the relatively large equivalent series resistance of the body diode. I mention that Vin should be above 5 V, as much below this won't be enough to drive the FET into or close to saturation. The 15 V Zener protects the gate and can be any value between around 10 V and 18 V.

When reverse biased, or in other words when the input is connected backwards, the body diode is also reverse biased in the same way a normal idiot diode would be and Vgs is always positive, which keeps the FET off.

Typically, P-channel devices are twice the price of N-channel devices and have on resistances around five to ten times that of their N-channel counterparts. I stumbled across the IPP80N06S4L-07 a while ago when looking for a device for just this application and found them on eBay in TO-220 packages for £1 each (well, £5 for 5 + 99p delivery) from the UK. Rds(on) is only 6.4 mΩ when Vgs is only -10 V. They take this measurement at a drain current of 80 A, however Rds(on) reduces as Idrain reduces. This means that it is ideal for 12 V (or 13.8 V) circuits.

I like the idea and just did a quick search for the datasheet. It's showing this as an N-channel FET and also the diode is the opposite way round.

[thewisepranker]

Whoops, you're right. Most people wouldn't bother. The body pin arrow also points the wrong way on an N-ch FET.
I was using that particular N-channel device in a low voltage cutoff circuit for something battery powered. The performance-related numbers I quoted are correct for that device, the N-channel one, so put a + in front of the Vgs numbers. The schematic is correct apart from the part number of the FET.
I bought both the N-channel FETs that I misquoted above and the ones I meant to quote from eBay and got the part number from my purchase history. The N-channel ones were at the bottom of the list and the P-channel ones were truncated. I had a 50% chance I suppose.

The ones I meant to quote were IPP80P03P4L-04. Rds(on) = 4.4 mΩ (conditions: Vgs = -10 V, Id = -80 A).

I can't find them on eBay at the moment, but they're available from RS, MOQ of 5 for about £1.12 each, part no. 823-5554.

The IPP80P03P4L-04 is rated at -30 V (Vds), however there are other FETs for higher voltages if required. Infineon seem to be leading the way at the moment with the P-channel FETs. Not much seems to come close regarding nice, low Rds(on) devices.
I've noticed that the Arduino lot tend to use IRF devices for everything, which lends itself to being the go-to device, hence for P-channel you'd choose IRF9530, with an Rds(on) of 300 mΩ. Still, this isn't terrible.
Has the IRF540N become the new 2N3055 in terms of ubiquity?
Interesting too that Infineon bought International Rectifier last year. I digress.

You could use the N-channel FET I originally mentioned if you really wanted, if sourcing a decent P-channel device is problematic, the only real point being that the gate drive is a bit more complicated. There are numerous ways of doing it - a favourite of mine is using a photovoltaic diode array, also known as a diode output optocoupler. The TLP190B is an example and only costs about a quid.
PV arrays can require a conventional transistor output optocoupler in parallel (and driven as "not") with the PV array to short it when you want to command off due to the gate charge, if you need it to go off quickly.

[Albert H]

I stumbled across the IPP80N06S4L-07 a while ago when looking for a device for just this application and found them on eBay in TO-220 packages for £1 each (well, £5 for 5 + 99p delivery) from the UK. Rds(on) is only 6.4 mΩ when Vgs is only -10 V. They take this measurement at a drain current of 80 A, however Rds(on) reduces as Idrain reduces. This means that it is ideal for 12 V (or 13.8 V) circuits.

You must remember, back in the late 90s when we were working on the NRG boards, the very low Rds(on) FETS were either very expensive or made of unobtanium. We tried the reversed diode and fuse combination on the PLL III board, and reversed the supply several times, always without damage (except to the fuse!). The speed of the fuse failure was (just) fast enough to protect the semiconductors on the board. The logic was fed through a 7805 and had sizeable smoothing capacitors across the rail, and these wouldn't charge (backwards!) fast enough to allow the voltage to rise much.

The diode and horn relay approach is even more foolproof, and has the advantage that you can do anti-tamper switching (and other clever things) with it. Those cheap relays were also great for power switching to a PA stage as well, to allow delayed switch-on and out-of-lock power-down.

[teckniqs]

You must remember, back in the late 90s when we were working on the NRG boards

Back when Wise Pranker was the tender age of 7.

[thewisepranker]

I was 9 in the extremely late 90s. Just!

[yellowbeard]

I was 9 in the extremely late 90s. Just!

Whippersnapper!