LMX2541 pll/vco combo - how to modulate?
However I'm uncertain about how such a combo chip is fm-modulated, I would prefer direct (true) fm. Normally the vtune signal between the vco and the pll would be modulated but can I modulate the signal between pin 15 and 16 here as well?
Regards
Martin
Probably it will work, but loading the initial loop filter and CP output definitely some performances as phase noise and spurious emissions will degrade.
Actually the first combo chip PLL/VCO appears about 35 years ago (CD4046), which is still in production.
The 4046 chip is well-known but unfortunately, at least to my knowledge, it cannot generate higher frequencies (vhf, uhf) without external multiplier/divider circuits which will make the final circuit bulky. I think this is still fine for low frequencies though.
The great thing about the LMX2541 is that one design is able to generate about 45-460 MHz which means design it once and use it for several applications...
Can you give me a hint what other possibility there is modulation-wise?
I am afraid that you cannot FM modulate the internal VCO. Seems that is no access to the resonant circuit.
You can try using an external VCO, the chip allowing for this option.
That's possible, you're right, but it's not an option in this case - I need a circuit covering from around 50 to 450 MHz and unfortunately none of the commercially available vcos I've ever heard of covers such a wide band...
You can use a VCO at 2GHz and use DDS to produce any freq less than 800MHz.
I am a little confused about the responses. Unless I am missing something, the Vtune voltage is directly accesable. You can modulate that pin that says "Vtune". Like any of these PLLs, you need to AC couple your modulation onto that pin, and you need so set the Capacitors and resistor of the loop filter between the charge pump out pin and the Vtune in pin so that the PLL bandwidth is much less than your modulation rate. For instance, if you are modulating at 20 KHz rate, you need a PLL bandwidth that is probably less than 2 KHz. If not, you will have the PLL control loop fighting your modulation. You also, in general, need to set the divisor ratio to a pretty high number so that you are only varying the instantaneous phase at the phase detector <<0.2*Pi radians.
There will be some distortion, since the Vtune vs Frequency curve is not linear.
That't exactly what I thought but I'm not the specialist here but hope you're right. My idea is to do it as in the pic (I've seen the modulation coupled in between the series cap and resistor of the loop filter elsewhere).
However I have a problem making the loop bandwidth small; the modulation is between 50 and 3000 Hz but making the bandwidth 500 Hz wide already results in cap values of 27 uF for C2 and 10 uF for C1. What do you think should the bandwidth be?
Regards
Martin
Unfortunately it will not work in this situation to modulate Vtune without distortions and as I mentioned degrading phase noise and spurious emissions.
This because Vtune point is part of the Loop Filter circuit and CP output. Anyway, always worth a try....
Long time ago I've tried to do almost the same thing, doing frequency modulation in that point of the PLL, and the sound was like the voice of Robert Plant from Led Zeppelin in "Dazed and Confused".
No, your audio bandwidth is a little too low for this method. The way described was used for things like TV transmissions from a field crew to the station.
I was able to remember where I saw this method last, below a part of the schematic. This is in fact an analog voice transmitter made for voice repeaters so the same application I'm looking for (well almost - not for a repeater). This tx is or was available commercially.
The fat black line coming from the bottom is the modulation signal from the mic amp. Is this so much different from my schematic or am I missing something?
Regards
Martin
you can try it for sure. I would start off with a PLL corner frequency of 20 Hz. You might have to make that higher. Of course, the higher the corner frequency, the worse the bass performance will be. IF you can get it to lock up with a 20 Hz bandwidth (you would have to be lucky), then you are all set.
The problem is that the free running phase noise of the VCO is going to be not so good due to the high frequency. So running a narrowband PLL loop filter will be like trying to lasso a bucking bronco with some kite string. It simply may not lock up.
In theory you can modulate the loop at the loop filter provided the AC coupled modulation can sink/source current equally well and you modulate well outside the loop bandwidth.
But I'm not sure I would want to design with this type of modulation with a VCO running at 2GHz with narrow loop bandwidth for reasons already given above.
However, with this device it might be worth considering the following alternative:
FM modulate the crystal reference and design the loop to have a high loop bandwidth. eg >25kHz BW.
(has to be narrow FM bandwidth at the reference and with 2500Hz limited audio frequency range)
This is good for close in phase noise and microphony too. It will slightly degrade the far out phase noise though.
It's maybe bad for achieving small tuning step sizes as you have to phase detect at maybe 250kHz. But the device offers fractional N operation which helps here.
But you would only get 'reasonable' FM audio quality with this method due to the non linear tuning of the VCXO reference.
... but you probably only have to tune it 250Hz due to the multiply effect.
Neither method is ideal but the method above might prove the best especially for mobile/portable ham band equipment.
If you do opt for the conventional method I'd experiment with a loop bandwidth in the order of 100Hz and see how far it can be pushed higher without compromising audio quality too much. I think a 20Hz bandwidth is much too low because of microphony and stability issues on a 2GHz VCO.
Maybe you can use some baseband modulation chip IC, to move audio freq 50~3KHz to say about 10K~13Khz, then you can directly inject this signal on Vtune of VCO. I used to use PLL loop bandwidth as 100Hz. You can use 2.2uF tan capacitor to inject the modulation signal.
OR You can use ADF4350, 137M~4400M, then you can use freq divider by 3 to get 46M~1000MHz.
Read the data sheet, and you are able to modulate with 2FSK via the DATA pin:
3.11 DIGITAL FSK MODE
The LMX2541 supports 2-level digital frequency shift keying
(FSK) modulation. The bit rate is limited by the loop bandwidth
of the PLL loop. As a general rule of thumb, it is desirable to
have the loop bandwidth at least twice the bit rate. This is
achieved by changing the N counter rapidly between two
states. The fractional numerator and denominator are restricted
to a length of 12 bits. The 12 LSB?s of the numerator and
denominator set the center frequency, Fcenter, and the 10
MSB?s of the numerator set the frequency deviation, Fdev.
The LMX2541 has the ability to switch between two different
numerator values based on the voltage at the DATA pin.
When DATA is low, the output frequency will be Fcenter ?
Fdev and when the DATA pin is high the output frequency will
be Fcenter + Fdev. A limitation of the FSK mode is the frequency
deviation can not cause the N counter to cross integer
boundaries. When using FSK mode, the FDM bit needs to be
set to zero.
So, what you could do is to sample the audio with a high speed 1-bit ADC, use the 1 bit output to drive the DATA pin, and you are there. At the other end you just lowpass filter the demodulated output. You would want a very high loop bandwdith if doing it that way.
Thanks for your follow-ups. I think it is correct that the traditional method is no good in this case, at least I wasn't even able to design a loop filter with narrow bandwidth (using Webench), ridiculous capacitor values of 22000uF were the result... Maybe a loop filter specialist would do better but it seems not to be easy. Because the frequency range of the modulation should go down to 67 Hz the bandwidth would need to be really narrow...
The method with the modulated crystal could be a possibility. Would the modulation be applied in the same way as in a crystal-controlled tx using a varactor diode? I tend to use a crystal reference and not an oscillator.
The other method with the Delta-Sigma adc sounds interesting, but I knew nothing about these so far. It is a bit higher tech than I intended but still worth considering. First of all I had a look at the websites of the usual suspects (Analog, National, TI) and was looking for a small and simple DS-adc which could do the job, however it's not easy to decide without experience. Would a ADS1202 or ADS1203 do? Else, what do I need to look for?
Furthermore, what would a very high bandwidth be? 1 kHz? More? Less?
Many thanks!
Martin
The pll bandwidth, if you are going to modulate the divisor ratio with the 1 bit DATA pin, would have to be fast enough so that the highest audio frequency you want gets through un-molested. For example, if you want to send 20,000 Hz with this basically digital modulation method, Mr. Nyquist suggests that you would need at least a 40,000 bandwidth. So I would, in that case, start off at 50 KHz and see how it goes.
As a point of refrence, a standard FM broadcast radio station only tries to transmit up to 15 KHz audio bandwidth.
For what I'm looking for at the moment, that is speech-only transmission (two-way radio), the audio frequency passband usually goes up 2.5 or 3 kHz only. So theoretically this would only need a bandwidth of 6 kHz which means that these 2 adcs are more than wide enough, but I'm also unsure about other things to look for.
Surely this wouldn't be compatible with other ham radios even if it did work. Plus it would need huge channel bandwidth? Who would the OP be able to communicate with?
The idea is that you design for a high loop bandwidth so the VCO is fast enough to faithfully track the modulation of the reference. Therefore, you get good FM in theory.
You could do it with a varactor feeding a crystal oscillator. However, the audio quality might not be as good you hope as there will be a degree of non linearity in the tune volts vs frequency.
If you are designing a dual/triple conversion radio on receive I suppose you could use the second LO (receive) as the reference for the Tx PLL and lock this LO (design as a VCO) to your fixed crystal reference. Then modulate the LO to get more linear sounding FM on Tx.
The phase noise and jitter on TX would suffer but it might still meet your requirements and it would save on cost and complexity as you would be reusing part of the receiver for Tx.
Despite all of the above I think you would be better using a different synthesiser strategy. Eg a more conventional design for each Ham band rather than this one that tries to cover all of them.
I don't understand what you mean, my understanding is that the resulting modulation coming from the pll unit would be as analog as any other fm and it wouldn't be visible to the "outside world" that it has been generated this way. Furthermore, why would channel bandwidth need to be huge?
No doubt there are more sophisticated ways than this. I just pulled the manual of the FT817 and had a look at the pll unit - this one is brilliant but hey, they even had a custom pll chip made! It's a question of expense and return - I was looking for a multi-band design that is as simple as possible but still good enough for jazz (I mean fm). I'm not very confident that vcos built with discrete components will work good enough and the ready-made ones do often not cover the frequencies needed and are expensive as well.