Construction Progress

[G1LWX]

Hi Tony
Looking forward to receiving digital from you
Regarding L5 it is it the rfc providing bias to the mimmic.
You would get away with just linking it out I never used one on the board that I made.
I know that it is not the digilite board but the circuit is more or less the same.
If you look at the data sheet it's an optional component anyway
73's Mike

[M0ATV]

Cheers for the info Mike, I may end up actually doing that, also looking forward to working some datv TX your way as well hopefully, if the mar-6 devices ( 9) turn up tomorrow, I'll be at the testing stage ready to go, my LO has a 1249 output, so with a bit of luck should be ok for GB3UD, also I got he WinTV PVR-150 set up & running & recognised in the PC using GBPVR software, will give it a good running through soon & hope all works well.

[G1LWX]

Great
Regarding UD I have put the settings on the Repeater section as they are not standard.
73's
Mike

[w0fms]

As for L5, my experience with using MAR MMICs at UHF and low Microwave is that just about any inductance will have enough reactance at those frequencies will adequately work as a RFC. The board trace almost looks good enough on its own but if you have a small bit of wire and a ferrite bead (or a chip FB), I bet that could be made into a really good substitute for L5 also.

Hello, I'm one of the crazy Americans who will attempt to adapt this to NTSC. It looks EXACTLY like what I wanted to design several years ago but I didn't figure out a cheap high speed interface out to a modulator. I literally have three FT4232H Modules sitting on my desk (for a cancelled project) and I never saw that use for them. (Of course no spare FT2232H's I'll have to buy that!) I'm also somewhat amazed that the DSPic33 is powerful enough to act as the serializer. It had to be a lot easier to write in "C", PIC ASM or a hybrid than trying to do it in Verilog or VHDL like I thought I'd have to do to get there. Nicely done, gentlemen.

Anyway.. I'm going to Americanize the parts list this weekend as there is one more crazy American who I have heard is attempting this.. I'll post after I get it done. I don't think parts availability will be a problem on this side of the pond. The only part that looked iffy is the balun and we can get that from the BATC.

Cheers,

Fred W0FMS

[g7lwt]

Hi Fred,

adapting DigiLite for NTSC may not be as daunting as you think, as the analogue TV input is handled by a WinPVR card. If memory serves me correctly, this card can already handle NTSC and PAL video.

Update: Just found a post from G8AJN that confirms that the only requirement for NTSC is that your WinPVR card supports the format.

Good luck, please let us know how you get on.

Vy 73 de Darren
G7LWT/AK4DB

[M0ATV]

Completed my board except for L5 which I trashed yesterday!. Used plenty of flux and a pointed bit. That mod chips tiny! Think I will have a go at a 70cm varient now.

Regads

Col

I ended up fitting L5 at a 90 deg angle to the pads i.e. lying on its side & soldered it that way, worked first time, hope you managed to get yours sorted in the end Col, Tony

[w0fms]

Hi Fred,

adapting DigiLite for NTSC may not be as daunting as you think, as the analogue TV input is handled by a WinPVR card. If memory serves me correctly, this card can already handle NTSC and PAL video.

Update: Just found a post from G8AJN that confirms that the only requirement for NTSC is that your WinPVR card supports the format.

I never had any doubts that NTSC would work, because at the data level (unless somehow the serializer checked for PAL and there is no reason to do something like this) even the set top boxes don't care. We have access to the IA-9 (former PAS-9) satellite in the Atlantic on C-band, and all of my Set-top-boxes can decode PAL and convert to NTSC fine, and have been able to do so since my first DVB-S box in 2001. The only catch for us in America is that it's really hard to find any PAL equipment like cameras, etc. at all.

One of the really neat aspects of this is that STB's, PCI and USB DVB-S cards are ubiquitous even here in the States, so in many cases if this project becomes popular over here it might be LESS expensive than current AM ATV equipment is. In most of the US there is *NO* ATV activity and in the areas that there is, it's 70cm 8 MHz wide AM NTSC.. yuck. I am not sure due to FCC rules if DATV at <902 MHz is legal in the US. I would almost guess no because of our antiquated digital rules on 50, 144 and 432 MHz here from the 80's that no one wants to update!

Since the PCI and USB cards are available, there is a secondary use for all of this and that is also as a high-speed data modem (or packet if you will). Use a DVB-S dongle in DVB-IP and use the digilite to TX and you could also have high speed links. I could imagine for emergency comms that would have to be a boon.

Hopefully the specs on the dsPIC33 serializer will be open enough to allow us to adapt other PC software (i.e. H.264, data, etc) to feed the DVB-S serializer. I understand everyone's effort on this and desire to keep the software closed source or otherwise controlled.. but I'd love to see the specs for the FTDI/DSPic33 Interface at some point. Hopefully I can convince the developers on releasing a public Interface Control Document for the serializer to PC link?!?

I'd also like to try and write (and/or see someone else) write a DVB-S2 version of the serializer as the Eb/N0 performance is about 3 dB better for that error correction technique and the receivers are still inexpensive.

But the simplicity of how this is all done is rather impressive and I congratulate the BATC team on coming up with this. I wanted to do something like this myself but I never quite was clever enough to figure out a cheap, easy, high-enough speed USB interface and also was convinced that a FPGA would be needed for the FEC/Serializer generation.. and I'm entirely too busy to do that in my limited free time.

I received the board, pic and balun in the mail today. Excellent quality and very quick international shipping by the BATC-- again, commendable!

I am awaiting samples on the modulator IC's and need to order the rest of the parts. I'm considering making a replacement of the MAR-6 MMIC to a more readily available but similar spec Avago MSA-0885 part (in the states, Newark, Digi-Key, Mouser).. this would require also changing the bias resistor and the 7808 to a 7810. An ERA-50 from Down East Microwave in the states also looks like a superior improvement. DEMI still sells the MAR-6 also in small quantities if one doesn't need $50 worth of Mini-Circuits direct parts. MAR-6's used to be one of my favorite "input" MMICs but it's getting a little long-in-the-tooth compared to more modern devices such as the ERA-50.

I'll write up my "US Distributor" parts spreadsheet and make it available somewhere.. maybe one of my parts blogs.. maybe by Monday..it's the Iowa QSO party this Sat so that will likely take precidence, as will the family.

Cheers again, Fred W0FMS

[G4EWJ]

The PC / serialiser interface is very simple. There's no protocol or packetisation on the data, you just open the DATA com port and send bytes to it. The data can be anything.

Apart from optionally doing the FEC, the serialiser is a fairly dumb device. It just splits whatever it receives into I and Q streams. All the hard work is done by DigiLite Transmit.

The only oddity is that when the serialiser is not doing the FEC, it expects bit zero to be the most significant. I started writing the dsPIC software before I knew which way round it went. It's more efficient to shift data out of bit zero, so I left it that way.

The transmission parameters are sent to the serialiser on the CONTROL com port at 57600,N,8,1,noflow. A:SRF4000,3<CR> sets the symbol rate to 4000 and the FEC to 3/4. There are about 25 symbol rates built into the serialiser. Not all of them are on DigiLite Config, but there is a way of setting them.

In time I'll write up the full spec and put it on here.

Brian

[g8ajn]

The changes to the values are 'improvements' that have been made during the course of the first batch of boards being assembled. There is no reason to worry about the differences as they are minor changes, the 1n2 decouplers were chosen instead of the 1nF as they were already in use on the filters, the value is non critical. Slightly better matching into the MMIC was noticed with C24 being increased in value. So dont panic if the values are sightly different, I will get round to re-doing the overlay asap, if in doubt use the values in the circuit diagram.
I would suggest the ICs are fitted before any other components, whilst this is probably counter-intuitive it does make the fine pins easier to work with while there are no other parts fitted.
Dave G8AJN

[w0fms]

Hey.. in gathering parts up I noticed at least two inconsistencies: The 74AC574MTC specified in at least one of the parts lists is a TSSOP-20 part according to the Fairchild datasheet. It's also at End of Life. The version 5.6 board is layed out for a SOIC-20W footprint. I ordered the MTC part.. and it'll probably be wrong. Luckily I dug deep into old IC's I got surplus and found an ON Semi 74AC574DW part which fits nicely.

Also if you follow the Farnell part for the DSPic3333FJGP802 part it specifies a SOIC-28 part (I/SO) this should obviously be a (I/SP) version for the narrow 28 PDIP part. This isn't an issue if you bought from the BATC shop, which you should because it's only pennies (in both US and GB currency) more to do so. I decided to buy enough parts to populate a second board, because it's cheap to do and I do have a PICKIT3 and a ZIF-socket adapter for DIPs so I can burn these turkeys. So I tacked a blank one on to my Mouser order.

Also I hope that the Farnell part number for the TLE2426 part is correct because there seems to be a few variants of that part that operately slightly differently.. the one that it cross referenced to is the "virtual ground" TLE2426ILP part. Can a designer verify that is correct?.. hopefully it is as that's what I ordered 4 of!

Last thing..the "add on board" for programming the Israeli oscillators.. from the parts list, a 16F628 PIC is specified but on the Digilite web site, a 16F628A part is specified. I bought a 628A part as I actually have one of those synths (and may order another since they are available again) set to 1152 MHz.

Of course, I'd recommend maybe in a future version building the synth onto the board.. but I'll duck out of this suggestion before I annoy the designers.
Doing that could mean that the DSPic control channel could also set the frequency...

I'm working off of three different parts lists now to and that could be part of my issues. I still intend on making a clean "US source" one for those of us European challenged!

Still it's turning out to be a fun project to build.. thanks again guys.

I'd also have gone with leaded electrolytics instead of the chip tants for all but the critical 10uF part on the DSPics. Oh.. I work with PIC32's all the time. A chip tant is about all that will work with the internal 2.5V core regulator (1.8V in a PIC32) -- low ESR is essential with those devices.. for those building out of junk box parts you can't substitute that part with a leaded or chip electrolytic, for example. The reason I would have gone with leaded elecrolytics for the rest of the regulator bypass caps is that a well stocked junk box could save most builders $10 (or 10 pounds) in chip tants.. boy, those things are pricey. And they tend to explode when they fail.

For those of you in the states: Two sources of parts will be necessary regardless. I recommend Mouser for all but the Analog Devices parts who they don't carry. Arrow appears to be the cheapest for those but DigiKey or Newark are other options.
Meaning the AD828's (which I had surplus parts of) and the AD8346's which I'm attempting to sample at this time.

I'm going to try to substitute the Avago MSA-0886 device available from Mouser for the MAR-6 in my unit. This requires a 10V bias and different resistors.. it doesn't consume any more current, but the bias point is at a different point. The output 1 dB point is slightly higher but really close to the MAR-6 and gain is similiar.. it should perform about the same. That means that a 7810 needs to be in place of the 7808 and the resistors need to be two 120 ohms in parallel. This isn't tested yet.. I'll let the group know when I do it.

73,
Fred W0FMS