SSB Exciter... HELP !!

[Wendymott]

Hi Peeps... Im having a bit of a problem, understanding what is going on. In no way do I want to unleash something onto 40m that is "iffy". THe problem is the modulation, and levels etc.
I discovered that to ensure correct modulation levels was to use a two tone audio oscillator, which was duly constructed.. I also found waveforms on the web to show the correct level settings. But it was not clear if I should be using both tones at once or singly. I decided that single tones gave me the correct modulation levels as per the photo's I had found, but.... the 1380 Hz pattern shows vastly under modulated RF. I removed the AF amp and filter and went into the MC1496 modulator directly and found a similar level of modulation. I have made some photo's of what I have up to now. and could do with some one with much more knowledgeable than me to point me in the right direction.
The RF monitor point was the output of the 9 Mhz + /- filter..
Using the 750 Hz modulation, shows two nice peaks on the spectrum analyser but rather odd when using the 1380 Hz mod tone.
I will need to do two postings to get the photo's in.

[Wendymott]

Here are the other photo's

[Radio Wrangler]

OK, if that's just at the IF stage of your driver, then you look to have some problems.

You've got the spectrum analyser switched to linear scale, so that makes the differences between signal levels look a lot greater than a log scale does.

For just generally probing around and debugging a single tone will do, but for properly testing an SSB exciter or amplifier, you do need two tones - both on at once, and equal in amplitude.

The first scope picture looks like the output of a balanced modulator. What isn't visible because of frequency/time scaling is that the phase of the RF is reversing between each of the 'bursts' BUT the overall double-sinewave envelope shape which should be there is distorted. The peaks look somewhat blunted maybe the positive peaks a bit more than the negative ones. Maybe you're running it at too high a level? The aim at this stage is to get good, clean, linear modulation. The temptation is to run things hard so less gain needs to be applied later. But gain is relatively cheap. Modulators linear to higher levels are much more difficult to arrange.

I'm not sure what's going wrong with the 1380 Hz scope picture. There seems to be a lot of carrier leakage and it looks like 30% AM rather than SSB/DSB.

The dual tone plot on the spectrum analyser shows the third-order intermod products at about 8% of the wanted tone voltage. This is about 22dB down on the wanted tones.

A good SSB machiine achieves 30dB or better for the whole thing, with the power amplifier dominating this distortion. 20 to 25dB overall is tolerable, getting on to being a splatter-creator.

So the exciter needs to be somewhat better than you want in the overall machine. The first scope picture fits this.

Two-tone testing sort of represents the peak-to-mean ratio of real use, and the spacing allows spectrum analysis of third (and all odd) order intermod products. It also givey you something more diagnostic to see on a scope. Yes it does make the transmitter look to be undermodulated, but that's it doing its job. SSB transmitters in use, transmittiong real audio are very undermodulated because of the characteristics of speech, and the lack of a continuous carrier level. The game is to get the peak levels right and not worry about mean levels. Speech processors work by manipulating the audio to increase the mean level of the signal WITHOUT INCREASING THE PEAKS. The RD sections work harder and get hotter because they spend more of their time at high power, but the maximum power at the peaks hasn't increased. The speech is punchier, there is more Ooomph, but the peaks haven't moved into levels of greater distortion.

Everyone studies the frequency plan of receivers and transmitters, but the level plan is harder work and just as important.

There's a good book "Single Sideband Circuits and Systems" usually saide to be by Sabine, Schoenike etc, but every chapter has a different author, and they're all in the government radio division of Collins radio. There's that nice big sandstone library on what used to be Ramsden Street. It'll probably be an interlibrary loan, though, or maybe a university library job.

David

[Wendymott]

Well............. that certainly covers a lot.. Thanks David. I removed at preset "gain"resistor in the MC1496.... and ran it at full "smoke"....
The basic circuit I am using is the Julian Rosu VA3IUL 20m RX/TX .....
... back to the drawing board...
The reason I went "after" the Audio compressor/filter was that the 1380 hz sine level was maybe 10 db down on the 750 Hz,,, even though the filter is reasonably flat to 3Khz...
Lot to learn.
Ok about the "Big book shop".. I think my library card was written on "parchment"..
Again thanks

[Wendymott]

Ok.. here is the more intelligent waveforms I Hope. Photo's 9 & 10 were taken on the output winding of the 9 Mhz modulator with no filters attached.. see photo 8. The other photo's were taken After the 9 Mhz filters, which are selected by minature Teledyne relay. but just prior to the IF conversion modulator (Pin 1 of MC1496). The gain resistor was looked at but no significant effect to the overall levels until about 1k0 from 10K.
Hopefully this looks better.
AF still directly into modulator, but both tones identical in level but reduced overall.

[G0HZU_JMR]

Looking back at your initial scope plots in post #2...

If you feed 750Hz AF into the modulator (without a crystal filter at the output)it will produce a two tone output envelope with tones at RF+750Hz and RF-750Hz which will look like your first scope plot. What you see there is the USB and LSB tones.

The job of the SSB filter is to shave one of these RF tones away and just leave USB (or LSB). So it 'should' look like a steady single tone coming from the SSB filter. But you still have both tones!

Looking at your first two scope plots it looks like you have used the incorrect RF frequency to drive the modulator so both of these tones are squeezing through the SSB filter at the same amplitude. It is probably wrong by a kHz or so?

It really looks like both of these LSB and USB tones are passing through your SSB filter so you must have the wrong RF frequency. You need to set it such that one of these tones gets rejected in the SSB filter and the other one gets passed as the wanted sideband. Try experimenting with the RF frequency into the modulator.

In your second scope plot you can see that one of the tones is partially suppressed and this makes it look like an AM signal. What you should be seeing here is a single continuous (unmodulated) tone at Fc + 1380Hz if you wanted the USB output. But because you have the RF frequency incorrect you are allowing the unwanted sideband to also leak through to the output at a significant level and this makes it look like an AM signal on the scope.

Once you get the RF frequency correct you can use both AF tones together and then you will see the classic SSB envelope leaving your SSB filter.

[Skywave]

What has been said about getting the RF freq. correct is valid, of course. But don't overlook the 9 MHz filter: it may be defective or its source and load impedances may not be correct.

Al.

[G0HZU_JMR]

I also agree with Al that that it's worth checking the health of the SSB filter. Is it OK and terminated correctly

Looking at the latest spectrum analyser plots without the SSB filter it does look to me that you have a lot more AF (harmonic) distortion in your system than I would expect. It's probably 25 years since I last played with the classic old MC1496 chip but I think it is capable of much better performance than this.
You look to only have 20-30dB suppression of harmonic terms and I'd expect this to be more like 40-50dB on a typical design.

You can see that the SSB filter gets rid of the worst of this but it won't get rid of all of it. Some of the harmonics will fall inside the normal comms bandwidth and the quality of your audio won't be ideal. It might sound slightly muddy or 'processed'. But it might still go unnoticed by anyone listening on a typical SSB receiver with its own linearity issues wrt AGC performance etc

[G0HZU_JMR]

Maybe the xtal filter has a different passband response than you expect so this is causing the issue with the incorrect filtering response?

Maybe this diagram below will help? This is what you would expect to see if the USB filter had a passband of 9.0003MHz to 9.0022MHz.
You would have to use a carrier frequency of 9.000MHz in order to pass USB correctly and see the 750Hz and 1380Hz tones as in the image below.

[G0HZU_JMR]

Your 9.0015MHz spectrum analyser plot (with SSB filter) looks to me to be misaligned as in the image below. i.e. the carrier frequency is wrong for the passband of the SSB filter so the unwanted sideband is able to leak through and only part of the wanted sideband AF range of 300Hz to 2200Hz is able to get through.

[G0HZU_JMR]

Looking closer at your info I think you may have a pair of SSB filters that are designed to be used with a fixed carrier of 9.000MHz.
i.e. you have an LSB filter and a USB filter. So you can generate both USB and LSB signals by using both of these filters in a switched arrangement but with a FIXED carrier of 9.000MHz in both cases. It's an expensive way to do it because you need two filters to get the choice of USB or LSB in your radio.

Or, to save on the filter cost, you could just use the USB filter on its own and switch the carrier to approx 9.0025MHz to create an LSB signal using the USB filter. That's how a commercial rig would do it for an SSB filter that was about 2.2kHz wide. You would have to correct for this frequency shift by sidestepping the main first LO by 2.5kHz to keep the radio on frequency when swapping from USB to LSB.
It's a bit niggly to have to do this but this is still cheaper than using two crystal filters. This single filter approach is used in plenty of ham or CB radios that can switch between USB and LSB.

[Radio Wrangler]

I agree.

The spectrum of the modulator output shows a symmetrical picture with two mirror-image sideband structures surrounding a null in the centre where the carrier would have been. So the modulator balance looks good.

In the spectrum taken after the filter, the passband shape of the filter can be seen as a template BUT the central null from the modulator can be seen well within the filter passband. It should be off to one side and far enough out to almost be invisible.

You don't have the right carrier frequency for this filter, (or you don't have the right filter for the carrier crystal)

David

[Wendymott]

Hi peeps.... All comments noted.... The Crystal filters are 8.9982 Mhz and 9.00180 Mhz. Now I see the error... I am using two Xtals.. at the same frequency as the filters...and obviously should be using... either one crystal at 9.0 Mhz or One 9 M filter and two crystals offset... I told you I was a novice at this...but learning....
I should have studied the circuit I was using as a reference, as it only has ONE Filter and two crystals for U and L sidebands.
One filter would lessen the PCB area considerably and the need for one less Teledyn relay
I will try both methods and report back.

[Wendymott]

Hi Peeps... Thanks to all your inputs I think the problem is nearly "cracked". I replaced the filters with a 9.000 Mhz filter and fed the mixer with RF from my Marconi 2022. It was a revelation.. The actual carrier frequencies to get best Single sideband are 8.9960 and 9.0040 Mhz at about 50mV.
The second problem was my two tone generator see photo.. Obviously the OPamp is not happy with the two sinewaves being combined at the input..... I replaced the generator with a recently purchased Two output generator and combined the outputs..
As can be seen the result is far superior........... but I will return to the home built item to get it right.
So.. I will still try a 9.00 Mhz RF oscillator and switch filters, as they were bought for such a job.
But again Thanks for all who commented..

[Jon_G4MDC]

To be honest Wendy when you posted the circuit of the 2 tone gen my first thought was it might work but I suspect audio harmonics will plague you.

Nothing to limit oscillation amplitude so it builds up until clipping limits it. Those oscillators won't have sinewave outputs.

[Radio Wrangler]

I remember the circuit as a pair of 'phase shift' oscillators. As Jon says, these have no amplitude control mechanism other than banging into their endstops and the distortion levels will be terrible.

You really need a pair of proper sine generators like Wien bridge oscillators with thermistor or FET amplitude control.

Another way is to make your tones on a computer and play them via a sound card.

David

[Jon_G4MDC]

Thanks for the reminder David.
I last used Tcube by I2PHD for this - seemed to work OK on a WinXP laptop. Shortcut was hidden among the other junk
but it's still there and it still works.

http://digilander.libero.it/i2phd/tcube/tcube.html

[G0HZU_JMR]

Back in my student days I built the two tone generator in the 1981? ARRL handbook. This uses a Wein bridge circuit for each tone made from an opamp, a bulb and some R and C parts.

The outputs are then filtered and summed in another opamp. I used this for many years to help repair/modify various CB and ham radio transceivers.

I recall that there were one or two issues in the ARRL circuit and PCB layout but I eventually got it working OK. Back in the 1990s I even used this generator at my place of work for a research task. I recall that I improved the performance a bit by fitting a better summing opamp. It gave better IMD performance than any of our (combined) AF sig gens at work!

I just looked at it on a PC soundcard and the distortion performance is still pretty good. However, I think some of the caps may have aged because I think it delivered about -80dBc spurious performance back in the 1990s. It seems to be about -65dBc in the image below. I think a cap in one of the LPFs may have aged and lost capacitance. It must be 30-35 years old...

[Wendymott]

Hi Peeps..... Thanks for the continued interest.....Yes I did come to a similar conclusion..These phase shift generators are ok for single tones, but not for mixing...As I have a commercial generator now, that is what I will use.. see photo.. if anyone is interested I can PM the details..

Ref the actual SSB exciter.. Two avenues of direction.. 1. to use 1 X 9 Mhz filter and 2 x L O.... and 2. 2 X ( Mhz +/- 400 hz) filters and 1 LO..
It looks like 1 Filter and 2 x LO's as if I use 2 filters and select with a Teledyne relay, the 8.998 Mhz filter is still not properly isolated and the upper sideband is not as well attenuated, The 9.0015 is fine.......Removing the 9.0015 filter makes the 8.998 filter act correctly.
Using a single filter makes finding the exact crystals for Upper and lower sidebands to match the crystal, filter which I found was +/- 400 hz, may be difficult..
The ones I bought from a member were 9.0015 and 8.9985 Mhz which by my calculation are 1.5 Khz and not 400 hz.
Am I being too exact ?

[Jon_G4MDC]

1.5kHz is usual. In reality it depends on actual width of the filter.

You are trying to position the BFO on the upper and lower shoulders so that given it's amplitude response it produces typically 300Hz-2.5kHz transmitted LSB or USB.

Some filters are tailored to be sharper on one side or the other. If you have a pair USB and LSB I would expect them to have different response shapes - not just different centre frequencies.

If you have the two filters I would think sorting the termination and switching and a 9MHz BFO is the way to go. It also means you don't have to dodge the main LO frequency around to correct final freq calibration.