SSB Exciter

[Wendymott]

Hi Peeps. I was looking around to build an SSB exciter for HF use, and I came across schematics from "SM0VPO" . I liked it as it was an Audio phase modulator rather than using crystal filters etc. However, I have tried contacting the guy for various help items but nothing as yet.
I wonder if anyone else has delved into this project. The Audio side and phasing filters works very well, however the RF Modulator is giving me a problem. It uses 4 FETs as the switching components with two 1Ko trimmers to balance the carrier to zero. The FETs are closely coupled in pairs with copper sinks, and the torroid is fixed firmly to the pcb, however, I get carrier zero drift, which in effect means that if set at switch on to zero carrier, after about 30 mins the carrier has drifted upwards to about 25 db above noise floor. Am I expecting too much from this arrangement ? or will carrier leak be not so much a problem ? Its operating at 9 Mhz. with RF input from my Marconi 2022. See attached photo and schematic.
Your comments will be gratefully received.

[Skywave]

Hi Wendy,
I think your project is as per this link: http://213.114.140.60/tx/p-ssb.htm

The drift could be due to the VFO signal itself - although since that source is a Marconi 2022, that seems quite unlikely. The most likely cause will be temperature related, so T2 is the primary suspect with the transistors coming a close second. There is also the possibility of a dodgy capacitor, but that seems remote compared to the suspects just mentioned.

Note that T2 has a very low Q: this is important. Try wrapping T2 in a thick wad of cotton wool. If the drift then stops - bingo!

Al.

[Wendymott]

Hi Al. Thanks for your comments. Yes it is the article I am using. T2 is a toroid tightly wound and then hotmelted to keep the wires in situ, I will try the cotton wool. All the components are brand new. The next stage was to encase the module in a copper box, but until I can stop the drift its a waste of resources. I already have a pair of 9 Mhz filters +/- 1.55Khz so it may be the best option if I cannot stop the drift. I don't want to put the device "on air" if its not up to scratch.

[Skywave]

I'm sure you'll find that the cause is temperature related. Classic culprits will be reactive components: L & C. The Vbe of a transistor is temperature dependent too, of course. I would try applying gentle heat to specific components to try to discover which one is the sensitive component - of course, there may be more that one! My money is on an inductor: T2 in this case.

Al.

[Wendymott]

Hi Al. Been messing about with it, too long now. I have slowed down the drift but the problem is I don't know how much RF leakage is allowed, if any. I am looking at using a "H" mixer as suggested by another forum member. I just need to source the IC. As its a long term project, there is no rush.

[G6Tanuki]

Idea: lots of modern mobile-phone cameras can be coerced to giving an infra-red/thermal-sensitivity image. If you've got [or have a friend with] a suitable IR-resolving phone, point it at your circuit and see what's getting significantly warm/drifting.

Must admit, the only time I experimented with your design of SSB-generation it used MOSFETs in the 4-way-mixer and 5400-series TTL [the mil-spec version of 7400] to generate the 4-quadrature LO drive. It wasn't a production-worthy success, the one thing I learned from this six months of fun was that you needed the 4-quad LO signals to be *truly* square-waves and the MOSFET part needed to be driven as a hard-on-hard-off switching mixer. At least MOSFETs are voltage-driven devices so they don't need *much* power to switch them - but even so you need to make sure thet snap-oon and snap-off promptly and cleanly.

[Wendymott]

Hi Tanuki. There is certainly no thermal as in detectable temperature changes. I reduced the drive from the 2022 to hopefully not "overdrive" the fets as I thought that may be a source of the problem, but as yet no one has answered the question Is carrier leakage acceptable or not and to what level. I suspect not.
Thus I may go the "H" modulator route as suggested by David and 9 Mhz filter/s.

[Andrew2]

Hi Wendy.
How far below peak 'wanted' output is the carrier when the balance is OK? If it's way down (say 60-odd dB), then drifting up by 25 dB may not be disastrous.
I'm puzzled by the causes of the drift though. I know ferrites can drift with temperature, but it seems you don't have much of a temperature shift anyway. Does the circuit balance with both balance pots roughly half-way? I ask because it could be a case of a floating gate on one of the fets which is slowly aquiring a charge. Of course the circuit would not balance properly with an o/c gate, but offsetting one or both of the balance pots might force a balanced condition and cause you to think it was operating correctly.
Gawd, I've confused myself now. Hope you get my drift. Ahem.
Good luck with it. I tend to reach for a cute little NE612 these days. Out of the box they will provide about -40dBc suppression and you can get more if you gently apply a few uA to pin 1 or 2.

[Wendymott]

Hi Andy. To be honest I have no idea, as yet, where the balance pots are in rotation. They are both 10 turn, but I did notice that the ceramic trimmers affect the balance. Also, these are supposed to suppress the unwanted side band. I had the thought that one or both these could be the problem. I have contacted SM0VPO and I told him of the problem, but as yet, waiting for a reply.. He is away on business. But thanks for your input. I have bought some FST3125 devices to try the "H" Modulator too.
I will come back to you on the pots later.

[Wendymott]

Hi peeps. Just built a "H" Mixer, using the FST3125 and driven by a 74HC86 with a balance pot. The problem is, I cannot null the carrier (9 Mhz). The balance pot has no effect on the null, but adjusts the mark space ratio of the square wave drive. The Ferrites used X3 are from old TV ant input balun transformers. My suspicions are the Ferrites, but has anyone else any ideas. Incidentally the ferrites are wound trifilar, 4 turns, on little "pallets", to aid easy replacement.

[Radio Wrangler]

Diagram and photos?

David

[Wendymott]

Hi David. I thought you may ask for them. The 10T preset is temporary, if it works ok I will get the correct one.

[Wendymott]

Incidentally David and others. I'm having a teensy bit of difficulty sourcing Binocular ferrite beads. I did see some in Farnell, but today, when looking again they do not seem to be there. Anyone any ideas please, any UK source would be appreciated.
Found some on Ebay.

[Terry_VK5TM]

http://www.jabdog.com/cores.htm

Second to bottom of the page box/listing.

Terry VK5TM

[Radio Wrangler]

Ha! beaten to it Terry!

The jabdog stall should be at Rishworth in 2 weeks time. Plenty of ferrite bits to choose from on multiple stands.

And if I remember, I might chuck a couple of 1.4meg xtal filters in the car.

David

[Wendymott]

Hi guys. Many thanks for the info. Rishworth am I missing something ??

[Radio Wrangler]

Evidently!

http://www.gqrp.com/rishworth.htm

A little radio rally so good you get people from Sweden, Germany, Nederlands, USA.

It used to be held in Rochdale in George Dobbs' church hall, until George retired.

Kits, components, surplus stuff, old stuff, test gear, honest-to-goodness junk. No computer stuff, no new japanese big-3 transceivers. An old-fashioned radio rally.

Within wellie chucking distance of Huddersfield, and now it's the Yorkshire side of the border, you no longer need an exit visa with a Lancastrian endorsement to go!

David

[Wendymott]

Oh THAT Rishworth...... Brilliant....will definitely be there....

[Wendymott]

Well gentlefolk.... My first try at the "H" Modulator, was a complete disaster ... well not so much disaster as not working to my understanding of balanced mixers. and the fact that I had not soldered one transformer point.. "red face"....
However after discovering my error, I tried to balance the carrier to zero again, but again the carrier will not zero. My understanding is that the carrier should zero on the "IF" port, until another modulating input is applied, then the resultant rf will be shown on the IF port. However.. I have attached 3 photo's that show what is going on...The first shows the carrier RF input "10 Mhz" squared, the balance pot does indeed balance the squarewaves.. note one waveform is inverted by the scope to show the near identical waveforms.
Photo 2 shows the RF peaks at 10 Mhz, and harmonics at 20 Mhz and 30 Mhz.
Photo 3 shows the 20 Mhz harmonic at a much reduced level due to the balance control..... The question is.... is my understanding wrong or correct.. If wrong is the modulator working ok ? I suspect it is not working as it should.
I have bought different ferrite cores but it is not worth fitting them if it is working correctly... over to you....

[Radio Wrangler]

It's a double-balanced mixer, which means there are two dimensions of balance in addition to the mark-space ratio of the LO drive.

Adjusting the mark-space ratio should null the even harmonic terms of the LO. And this is what you've seen.

Beyond this is the 2-way balance of having equal impedance in the four arms of the bridge. It is a transformer coupled bridge (rather than the ring type bridge used for diodes)

To adjust the balance here, you would need to add a low value resistor in series with one of the switches, and add a trimmer each to the other three. Similarly some trimmer capacitors will reduce the quadrature term of LO leakage.

BUT FIRST, get an idea of how much LO leakage there is. How many dB is that 10MHz LO component smaller than the amount of LO drive?

Say 5v p-p squarewave, the fundamental is going to be around +18 dBm worth of swing going into each gate.

The variant of this mixer I've used uses the SD5400 quad D-MOS transistor chip and a 74ACT flip flop with Q and Q bar driving the gates. I made my own transformers. One straight-forward balanced to unbalanced (though it was really balanced-to balanced with the spare winding being a transmission line capacitive counterpoise.) The other transformer used a counterpoise on the unbal winding, but the rest were balanced pairs- all wound together as one hexafilar winding. The SD5400 drains were hard grounded, the gates driven from flip-flops and no power supplies anywhere near the signal circuits. Purely passive and not floated. The snag to the grounded SD5400 is the need for a few volts of negative bias for the substrate. With care in the transformer, it was balanced enough for use as a mixer without any twiddles.

As an SSB modulator, you may want more LO rejection. Id dispense with one transformer and use balanced audio from an opamp circuit. I'd feed balanced audio into the RF transformer centre taps, via a pair of 3rd order lowpass filtrs in order to wind up with a capacitive low Z (at RF) feed into those centre-taps. A balance control affecting the relative gains of the balanced audio paths will give you one mixer balance control,but you'll still need to adjust the impedances in each switch a little for ultimate balance.

David