Are my constellations any good?

[g0nzo]

Chaps, I've been looking through the wiki,and it has answered pretty much all my questions. And got me this far.
But.... as this is my first time playing with the minitioune and davt, I have no idea if the constellations I am getting are good or bad.

Attatched is one example.
This is 437MHz, 333ksym. The portsdown filter/mod is connected directy to the Serif tuner, via as many attenuators as I could find. So there should not be any qrm getting in there.

What's the verdict chaps?

Jules

[g0nzo]

A bit more info......

The expected MER, from the wiki, should be 10dB better than I am getting. So suggests that this is not as clean as it should be.

This is just a standard build, by the book. Using the pre=populated/set-up filter/modulator board, from the BATC shop.
The output at 437 (from the modulator) is around 12mW, so that suggests that the drive levels and LO filter board are ok.

The only thing I have different to the wiki instructions, my LO board shows a 3v3 feed on the header, rather than the 5V that the wiki wiring suggeses. So I've fed the LO with 3v3. But the LO sounds clean on an SSB RX, so that looks ok.

Another issue I am seeing is, the minitioune seems to often refuse to get lock. The tuning trace is fine, but the carrier lock and timing lock are hammering between 0% and 100% a few times a second. All the tune/PLL settings are as installed.

[g0mjw]

Hi Jules,

Awful - absolutely terrible! Not really. Its not a very good MER for a direct connection, should be over 20 as its still a fairly strong signal but when you get to using it over the air, that't probably not going to be the limitation. What it should look like is more like this https://flic.kr/p/Hv3Mms

Off air though, here is a picture from Arthur https://www.flickr.com/gp/ad6xy/q93831 you will see it looks quite similar.

Where that would be a problem is when looking at higher order modulation, which the Portsdown can't send https://www.flickr.com/gp/ad6xy/bCy548

I would see what is causing that spreading. Is it the TX or is it the RX? Does it change with frequency? Did you follow the instructions?

Mike

[G8GKQ]

Hi Jules

I'm not sure that your constellations are so bad. The Serit Tuner seems to be more critical of the very sharp IQ filtering on the Portsdown than the old Sharp tuner was.

Some things to try:

1. First, the not so obvious: Have you selected IQ output mode, rather than QPSKRF that you might have used for early testing?

2. Reduce your symbol rate to 300KS on both the Portsdown and MiniTioune. If you get 2dB or so of improvement in MER, it suggests that the IQ filter clock might be at the edge of tolerance.

3. Increase your symbol rate on both to 1000KS. This uses an LC filter and may tell us whether the problem is as described in (2).

Please post your results here and we'll work out what to do. Failing that, are you going to the FRARS rally? We'll be there with some test gear and knowledge.

Dave

[g0nzo]

Dave,

Ok, here is the raw info for you to be going on with. I've not started digging through the schematics myself yet.

At 330ksym, I'vere-run, for comparison. And get the same MER of 17dB. As I'd popped a bit more atten in, in case I was overdriving the Serit.
The 300ksym gives an MER of 20dB. Pic attached.
At 1000ksym I get 20dB as well.

I will be ar FRARS Hamfest, though I won't be there till later in the day. But if you don't need to run off, we could have a try.


Mike: I only have the Portsdown and the Minitioune, which have only ever been tested with each other. So it is possible that it is the tuner. I have beed in and added some extra C around the regs, in case there was any hooting that could be getting in. But seems stable. On the scope, the supplies look OK.

Cheers for the help both.

Jules

[G8GKQ]

OK Jules

Sounds like you have some minor problems there, but nothing to be worried about at the moment. Probably the filter clock AND some minor MiniTiouner snag.

I would suggest that you continue with building the rest of your system and we'll try and identify the exact issue at FRARS. We can try different transmitters and MiniTiouners to locate the last few dB.

Dave

[g8cpf]

Hi Jules,
Also worth checking is the "cleanliness" of the supply to the Synth.... It can make a big difference, especially at low symbol rates..... Try running it from a battery as a test.....
de Mike G8CPF

[g0mjw]

Hi Jules,

Firstly - 17dB MER is not that bad - but we strive for perfection.

It could be the receiver - its best in the 1-2 GHz range so you could test that. Possibly you didn't use the same regulator and have some noise. The only line that really matters on the receiver is the analogue 3.3V. You don't want a lot of noise on that. Its very important to use low ESR caps. Checking it against another source would be useful. Go and see Paul perhaps?

Things you could try :

Bypass the narrow IQ filter by pretending a higher symbol rate on the control pins. Its quite easy to get these the wrong way round.
Check the supply to the LO chip. It takes 5V input. Boards differ but that's where you might get phase noise introduced.It looks like that may be happening here.
Make sure the 7805 is producing 5V - it needs 2V of headroom.
Don't run the IQ lines close to others. They should be twisted pairs to reduce pickup.
On the receiver, try running the board from a linear power supply, e.g. a bench one, not more than 5V, via the alt input.

Mike

[g0nzo]

Sorry if this is a bit of a running comentry, on the build. But was not sure how good a signal/constellation/MER to expect in practical terms. And aware that this project may still be work in progress.

I found a hooting reg on the LO filter, with a few 100mVof noise on the 5v. Fixed, but that looked not to be the problem. Results same.
The header on the synth was marked 3v3, so I took 3v3 from the breakout board. I suspect that not to be the cleanest and there was some chirping on the synth, listening with an SSB RX.
Now have the synth and LO filter running off the filte/modulator 5v reg. So both getting a nice clean supply.
My error here may have being trying to feed the LO board through it's 3v3, which may have been an output.

Result, better pattern. MER of 19-20dB on 437MHz 333ksym.
I still see the 300ksym at a higher 23dB MER. So poss still that filter issue Dave mentioned. But we are getting there.
24dB at 2000ksym. So l;ooking better on higher rates.

I have twisted pair for the IQ feeds, though as those are opto's, I didn't see it as critical. But using LAN wire for the signals, it was twisted anyway, so I used it in that config.
Have doubled checked the band and rate switch wiring. Don't think I've made any sillies there.

Also, checked the tuner psu's. All look clean.

Mike, I'm actually ahead of Paul on this project. A suprise as he is usually the one throwing everything at them.


Jules

[g0nzo]

Also the constant 0%-100% cycling on the carrier lock looks to have settled down.
So that may have been the LO stability?