First Test - Poor Results

Discussion about this major DATV Project. See https://wiki.batc.org.uk/The_Portsdown_Transmitter
Nick
Posts: 121
Joined: Sat Oct 03, 2015 10:09 am

First Test - Poor Results

Post by Nick » Sat Jan 27, 2018 2:30 pm

Hi,

I have just completed the Portsdown build and have some issues with the transmitted signal.

The transmitter output was fed to a spectrum analyser and the MiniTiouner is used for receive. The signal took quite a while to lock and produce an image, and the constellations are poor - image attached.

Maybe there may be a problem with my MiniTioune software as when typing in the frequency or SR value there is a significant lag from entering the numbers on the keyboard to them appearing in the text box - is this normal ?

Looking at the Porstdown output on the analyser there are several harmonics of a significant level, but the output from the LO Filter is very clean - no harmonics visible, so it seems these are being generated in the Filter/Modulator - image attached.

I have been going through other threads looking for any clues to this and am now wondering if it could be how I have installed the board interconnections. To keep the layout tidy all my looms are twisted single cores, and many of these looms (Breakout to ADF4351, Breakout to IQ on Filter Mod, Breakout to SR Select on Filter Mod, Breakout to Shutdown Button/LED and 4 Band Decoder to Panel LED's) are for a great part all tied together.

In the forums I found mention that the IQ wiring should be short and not near other wiring, and also found a mention that the wiring to the ADF4351 should not be twisted as mine is.

Before I pull apart the wiring and replace it can anyone confirm that this may be the issue, or could there be any other possible cause's.

Regards,

Nick - G4NKV
Attachments
70Cms Receive.jpg
70Cms Receive.jpg (443.7 KiB) Viewed 3779 times
70cms Analyser.jpg
70cms Analyser.jpg (135.25 KiB) Viewed 3782 times

g0mjw
Posts: 2330
Joined: Sat Sep 20, 2014 9:15 am

Re: First Test - Poor Results

Post by g0mjw » Sat Jan 27, 2018 3:00 pm

Hi Nick - don't twist I & Q together, assume that isn't what you meant. Certainly don't twist the I2C lines.

How are you receiving this? Try a direct cable connection through attenuation. Your problem might be multipath.

Interesting background.

Mike

G8PEF
Posts: 176
Joined: Mon Oct 30, 2017 6:42 pm

Re: First Test - Poor Results

Post by G8PEF » Sun Jan 28, 2018 10:01 am

Nick I'm by no means the expert on this, but I've been staring at the pics, and the spectrum and constellation do look a bit grubby, and if you say the LO is clean, then it's odd.
It looks like the sort of grubby I'd expect if it were the Pi operating in ugly mode - are you sure it's actually using the Filter-Modulator to TX?
Interested to see what Dave and Noel make of this.

John

G8GKQ
Site Admin
Posts: 2798
Joined: Sun Mar 22, 2009 1:21 pm

Re: First Test - Poor Results

Post by G8GKQ » Sun Jan 28, 2018 3:01 pm

Hi Nick

I've refrained from jumping in, because my first 2 thoughts have already been covered by Mike and John. - Thanks guys!

To answer your other questions in order:

The Moving Balls test card always takes quite a while to lock. The Camera in H264 or any of the MPEG-2 modes lock a lot faster.

MiniTioune can have a significant delay in screen update on less capable PCs, but still reproduce clean constellations - so I don't think that's your problem.

The harmonics from the F-M board are normal and are not the cause of your problem. You always need a good low pass filter after your PA as stated here https://wiki.batc.org.uk/Filters.

Wiring is unlikely to produce the constellations that you have. I would suspect something to do with the coupling between the F-M board and the MiniTiouner. Also, the level on the analyser looks low - I would expect around +9dBm at 437 MHz https://wiki.batc.org.uk/Portsdown_Output_Levels. Or perhaps you had an attenuator in-circuit?

To enable further fault-finding, please post a photo of the Information screen (on Menu 3), and another of the inside of your transmitter. Oh, and check that you do not have any RP-SMA leads in the signal path!

Good luck!

Dave

Nick
Posts: 121
Joined: Sat Oct 03, 2015 10:09 am

Re: First Test - Poor Results

Post by Nick » Sun Jan 28, 2018 10:55 pm

Mike/John/Dave,

Thanks for input.

As John realised I did have an attenuator (20dB) in circuit. When I removed this it showed that output from the LO Filter was not as clean as I first thought. As a result I connected the output of the ADF4351 to a 70cms filter obtained from e-bay (the one in the Wiki). This gave a cleaner output, but when fed to the Filter/Mod the output from the Filter/Mod showed no real difference to using the LO Filter.

My main concern was the harmonics from the Filter/Mod, so I am relieved that this is normal.

As there was no real difference on the Tx output using the LO Filter or the ebay one I would assume that the LO Filter is probably OK.

I have attached 2 images showing the LO Filter output on 2mtr and 70cm. Are these as should be expected ?

I have also attached an image of the output from the Tx using the LO Filter without the 20dB attenuation inline.

Also attached photo of information screen and one of inside of Tx.

For receive all I was using for the MiniTioune input is a short length of wire via a 12" length of coax from the receiver input. This was placed near the coax from transmitter to the analyser.

As for feeding the output of Tx to input of receiver I have not tried that yet. I can provide 60dB of inline attenuation - would that be sufficient to prevent damage to receiver ?

Hopefully this may cure the constellations when I try it.

Regards,

Nick - G4NKV
Attachments
LO Filter Output on 2mtrs.jpg
LO Filter Output on 2mtrs.jpg (134.13 KiB) Viewed 3587 times
LO Filter outut on 70cms.jpg
LO Filter outut on 70cms.jpg (134.83 KiB) Viewed 3588 times
70cms Output using LO Filter.jpg
70cms Output using LO Filter.jpg (138.26 KiB) Viewed 3589 times
Info Screen.jpg
Info Screen.jpg (772.29 KiB) Viewed 3590 times
Portsdown Internal.JPG
Portsdown Internal.JPG (800.31 KiB) Viewed 3590 times

G8GKQ
Site Admin
Posts: 2798
Joined: Sun Mar 22, 2009 1:21 pm

Re: First Test - Poor Results

Post by G8GKQ » Sun Jan 28, 2018 11:07 pm

Hi Nick

60 dB of attenuation should be just right. It’s what I usually use. Interested to see your results with that.

LO filters look to be working as expected. Nice build!

Dave

Nick
Posts: 121
Joined: Sat Oct 03, 2015 10:09 am

Re: First Test - Poor Results

Post by Nick » Sun Jan 28, 2018 11:40 pm

Dave,

That was a quick reply - thanks....

Attached are images of the results with a direct connection. I get different results between MiniTioune v0.6 and v0.7.

Does one of these look OK ?

On the face of it v0.6 looks better, but maybe I have some settings wrong in v0.7.


Regards,

Nick.
Attachments
MT v0.6.jpg
MT v0.6.jpg (290.86 KiB) Viewed 3573 times
MT v0.7.jpg
MT v0.7.jpg (297.11 KiB) Viewed 3573 times

G8GKQ
Site Admin
Posts: 2798
Joined: Sun Mar 22, 2009 1:21 pm

Re: First Test - Poor Results

Post by G8GKQ » Mon Jan 29, 2018 12:18 am

Hi Nick

The results are not far off what I would expect. Your F-M board is from batch 2, so would have been tested with a Sharp tuner and MiniTioune V0.6d, which I suspect gave an MER of around 24 dB (you could check the test sheeet).

Since that batch were tested and sold, we have found that the Serit tuner is more critical on the waveform being absolutely correct and the batch 3 tuners were tested on a Serit and MiniTioune 0.6d, with a pass criteria of 18 dB MER at 333KS. We are now working on a simple modification for Batch 4 which improves this figure by slightly increasing the I and Q filter bandwidths, and users will be able to modify boards from previous batches. I suspect your board might be at the end of the tolerance range. To complicate matters further, the MER calibration changed in MiniTioune V0.7b.

Try 1000KS symbol rate, you should get around 23 dB MER with MiniTioune V0.6d, and a better constellation if all is well. If that works, I do not think that you have a problem, but it will be worth you implementing the batch 4 modification when it is finalised.

Dave

g0mjw
Posts: 2330
Joined: Sat Sep 20, 2014 9:15 am

Re: First Test - Poor Results

Post by g0mjw » Mon Jan 29, 2018 6:45 pm

It looks nicely built.

What are you powering the LO from? Is it a clean supply or could there be switching noise on it? I can see what looks like a lot of filtering but not sure of your layout or what switching supply you are using.

I assume the Minitioune produces a better result with other signals and the fault, if there is one, really is in the Portsdown.

Mike

Nick
Posts: 121
Joined: Sat Oct 03, 2015 10:09 am

Re: First Test - Poor Results

Post by Nick » Tue Jan 30, 2018 3:34 pm

Hi,

Thanks for feed back - much appreciated.

Dave - I tested using 1000KS and got better results - see attached image. My board was tested with a Serit tuner and MiniTioune V0.6d. Given MER's for 23cm are 25.

Mike - the LO supply is the same 5.1V supply used by the Pi, and the regulators are LM2596 based boards from e-Bay.

Further tests show there may be a problem with my RF Output Switch. Up until last night I had only used 437Mhz, so I decided to see what receive was like on 2Mtr and 23cm. On 23cm the constellation looked something like a spiral galaxy, so I checked the signal on the analyser. The output level was very low, around -23dBm. The test sheet showed it should be 3dBm. I then took the output directly from the Filter/Mod card and the output was in the order of 4dBm.

Now as the outputs on 4Mtr, 2Mtr and 70cm via the output switch are in the correct ball park level wise I thought perhaps the ADG904 output switch RF4 was faulty, so to test I wired the MSB and LSB inputs to ground so that the 4Mtr output (RF2) was permanently selected no matter what band was selected for transmit. Monitoring this output (RF2) showed that the outputs on 4Mtr, 2Mtr and 70cm were near enough the same level wise as their normal switched outputs. I now expected to get full output on 23cm, but that was not the case - it was still low at -23dBm.

Receiving 23cms directly from the Filter/Mod gave a much better constellation than when receiving from the RF Switch - images attached.

Looking at the data sheet I see that there should be an insertion loss of only around 1.5dB at this frequency so it looks like there is a problem here - any ideas what it could be ?

I have not checked what the 70cms signal looks like on the MiniTioune direct from the MiniTioune yet, but expect it will be much better.

Regards,

Nick.
Attachments
70cm using v0.6 from RF Switch.jpg
70cm using v0.6 from RF Switch.jpg (298.97 KiB) Viewed 3285 times
1249 using v6 at RF Output Switch.jpg
1249 using v6 at RF Output Switch.jpg (293.23 KiB) Viewed 3286 times
1255mhz using v0.6 direct from Filter-Mod.jpg
1255mhz using v0.6 direct from Filter-Mod.jpg (297.16 KiB) Viewed 3286 times

Post Reply

Return to “The Portsdown Digital ATV System”