Use of DDS for carrier oscillator / BFO?

[cncjerry]

One of the hold-ups for my home brew transmitter is the carrier oscillator (like the BFO in a receiver).  The oscillator I am using is a CXO with a trimmer which enables me to set the CXO pretty precisely for the filter I am using, one of the KVG9 types.  The problem is though that it isn't very stable especially when I switch sidebands back and forth which uses another crystal.   Sometimes the oscillator doesn't start and you wouldn't find it out until you try to transmit. The switching is at DC level which is common practice.  I could buffer the CXO but I don't think that is the issue.  It's almost like the crystals I am using are being pulled (with the trimmer) to far out of their natural resonant frequency to set the correct frequency for the following filter.  The alternative would be to use a DDS like and AD8951 or AD8954 even.  This would raise the phase noise which I might be ale to tweak out a little by frequency selection.

So would the DDS phase noise be a problem (vs the CXO) on a transmitter?

Thanks

[Theboel]

Hi,

May I suggest 2 alternative :
1. Use SI5351 from silabs, with 3 output You can use as triple conversion.
2. Use Epson SGR-8002DC-PHB-ND buth this tiny thing have a problem, you can not reprogrammed and You can not pull or push the Freq.

[vk3yedotcom]

You should be right for amateur work at ordinary output power levels. 

Especially if the DDS works OK in a receiver (which due to the microvolt level signals amplified and mixed requires a clean local oscillator). 

Yesterday I was trying a Si5351A based VFO ( http://www.ozqrp.com/CDVindex.html ) as a local oscillator in a 7 MHz direct conversion receiver and it sounded as clean as the previous non-DDS oscillator.

In a transmitter provided that your spurious products are 60dB or more below your signal then it is unlikely that any DDS-induced distortion products will be audible except very locally.

[Z80]

Why not fix the VXO?  They are very simple and stable and perfect for a superhet BFO.  The problem you are having is possibly being caused by using 2 crystals, a few pF of stray capacitance can excite the unused crystal and cause it to pull the active one as its frequency is very close.  You only need to use one for the few KHz of shift needed, just diode switch the LC trimmer networks and it should be rock steady.  You don't say what mixer you are using, but the oscillator in the SA602 / 612 can be a bit feeble when pulling a crystal.  A single transistor circuit is fine for driving a 602, an extra buffer transistor is recommended if driving a diode ring.

[janoc]

One of the hold-ups for my home brew transmitter is the carrier oscillator (like the BFO in a receiver).  The oscillator I am using is a CXO with a trimmer which enables me to set the CXO pretty precisely for the filter I am using, one of the KVG9 types.  The problem is though that it isn't very stable especially when I switch sidebands back and forth which uses another crystal.   Sometimes the oscillator doesn't start and you wouldn't find it out until you try to transmit. The switching is at DC level which is common practice.  I could buffer the CXO but I don't think that is the issue.  It's almost like the crystals I am using are being pulled (with the trimmer) to far out of their natural resonant frequency to set the correct frequency for the following filter.  The alternative would be to use a DDS like and AD8951 or AD8954 even.  This would raise the phase noise which I might be ale to tweak out a little by frequency selection.

So would the DDS phase noise be a problem (vs the CXO) on a transmitter?

Thanks

Devices like Si5351 or AD9850/9851 are just fine for a ham transciever. There is phase noise but it is way below other issues you may have. Here is a comparison of Si5351 vs an LC VFO and a VXO:
http://soldersmoke.blogspot.fr/2015/08/si5351-phase-noise-tale-of-3-oscillators.html

Check also the comments, there are some good pointers too.

I have built receivers using both AD9850 and Si5351 and never had any noise issues with them. So I really wouldn't worry about that too much unless you are building a high power transmitter with it - but then the phase noise of something like a DDS or Si5351 is likely going to be swamped by your microphone noise anyway!

[CJay]

Why not fix the VXO?  They are very simple and stable and perfect for a superhet BFO.  The problem you are having is possibly being caused by using 2 crystals, a few pF of stray capacitance can excite the unused crystal and cause it to pull the active one as its frequency is very close.  You only need to use one for the few KHz of shift needed, just diode switch the LC trimmer networks and it should be rock steady.  You don't say what mixer you are using, but the oscillator in the SA602 / 612 can be a bit feeble when pulling a crystal.  A single transistor circuit is fine for driving a 602, an extra buffer transistor is recommended if driving a diode ring.

Agree, if it were my radio and I had that fault I'd fix the VXO or at least try to work out why it was misbehaving, but, the spirit of experimentation and possibly getting more for less, I'd also be curious about using a DDS or Si5351...

[cncjerry]

The CXO is driving a 1496G.  If you have a copy of the "Solid State Design for the radio amateur" I am using the CXO on page 203 upper left corner.  It's not the first I've tried.  The crystals should be set to 8.9985Mhz and 9.0015Mhz using 60pf trimmers.  There is also a 90pf on the combined output.  I just can't get the three trimmers set at a point where both the oscillators start reliably. It is a little better if I bring the voltage up a little which is the same as dropping the value of the 1k resistor at the top of the circuit.  I've messed with this quite a bit and thought, what the heck, throw the DDS in there with an ouptut filter set to 9Mhz and forget it.  But then again if I wanted a digital transmitter I wouldn't be building this one which is a combination of circuits with most coming from the 1976 handbook.   I have an output chain to take it to 25w PEP, not like I am going to use it to drive my 3-500Zs or nothing.  By the way, I did use them to boost my 10W K2 prior to adding the 100W output stage and it worked fine.

[Z80]

I had a quick look at the datasheet for the chip you are using and it gives an example circuit for a single rail balanced modulator (attached, snipped from the OnSemi datasheet).  It looks like the input is high impedance and they have terminated it with a 51 ohm resistor.  Your circuit has a 510 ohm termination which may be designed or may be a typo.  Changing that resistor will have a big effect on the reflected impedance at the oscillator collector and if it is a typo then it may be putting too light a load on the collector?  The datasheet seems to suggest a carrier input level of 160mVrms for optimum carrier rejection @ 10MHz.  Some rough calculations would suggest a turns ratio for L1 L2 of about 14:1 for a 12V supply (With the MC1496 terminated @ 50 ohms).  What is the suggested ratio in your book?
I have also attached a VXO circuit I built recently if you have no luck with the one you have.  It uses diode switching, just apply voltage to the input of either L3 or L4 through a suitable resistor to give a couple of mA of bias to the diode you want to turn on (Values of L3 / L4 are not critical they are just chokes).  You could increase the value of R4 to lower the operating current for 12V.

[cncjerry]

Z80, I am impressed. I don't know why I didn't think of the loading before.  Even when I tested it outside the circuit I used a 1K resistor which probably exacerbated the problem.

The output transformer turns ratio is 15:1 (doubly impressed).

In the circuit you posted, would I be able to switch the crystal in/out?  It looks like you are switching just the trimmers.


Thanks you, very helpful.

[CJay]

Disconnect the top of C2 from the circuit, add a second crystal, X2,  to top of C2 and to the junction of X1/R2/R3 then applying switching voltage to D1 through L4 will switch X2 in.

If you need both trimmers for each crystal then just duplicate L1,2,3,4, C1,2 and D1,2 for a second crystal connected to the same X1/R2/R3 point.

Of course, you can only have voltage on one set of switch inputs or 'odd' things will happen.

[Z80]

Ok, if the book says 15:1 then I'm pretty sure the value of the load resistor is wrong.  Try replacing the 510 ohm between pins 8 and 10 with 51 ohm as per the datasheet example and see if your oscillator starts behaving.
Regarding my circuit, yes you are correct I am switching the trimmers (one for USB and one for LSB), there is only one crystal needed (in fact you really don't want to use 2 as they tend to interfere with each other as they are the same frequency).  The circuit is designed as a BFO and has enough pull for sideband selection with a standard 2.5KHz bandwidth and I used it in a recent 'junkbox' SSB transceiver.  If you want to try it, then build it as is.  Apply 12V through a resistor (2K2 should be ok) to either L3 or L4 and adjust the relevant trimmer for either USB or LSB as desired.  The transistor isn't critical any jellybean will do at this frequency.