I just dug out my 80s Book on satellite receiver design and compared it to your schematic. The diseq chip is not in there (at that time there was a fixed +15V power supply for the outdoor unit and a separate microwave polarizer) but the rest pretty much matches. Except 2 things that are missing. There’s a circular mixer (the by-now ancient HPF511) consisting of 4 schottky-diodes for tuning/mixing in a variable LO and there’s a bandpass filter after this mixer. The bandpass is for suppressing harmonics, in this case going into the second IF and the FM Baseband demod. From the Quorvo datasheet i assume this HF mixer and the LO synthesizer all all in this chip, and need to be recreated to replace the chip.
yes, saw that too - it’s no replacement for bullseye thats for sure. Just throwing different ideas in the ring.
If you already have one, I can send you a proto for testing.
I have a couple of these installed here and have 2 spare units in storage. I can hook them up for testing. As @Tysonpower mentioned they will most probably not work over extended lengths of time. I was thinking of somehow bypassing the DC3 tuner circuit and doing the reception test on there, since the rest of the HF Circuit is pretty much the same as on DC4.
That should be possible. You’ll just hotwire into the RFIO/pin-22 of the SX1281.
Since Dreamcatcher 4 was first announced, there have been at least 4 different iterations. It all works, but there is still one problem which is preventing a finalization of the design and move to mass production.
The device functions without issue in under in the US and Canada. That’s because the frequency from the satellite is 12089 MHz and the output from the LNB, after the 9750 MHz LO, is 2339 MHz. That is smack in the middle of the sweetspot of the SX1280. RSSI and SNR are as good or very slightly better than Dreamcatcher 3.05.
The problem is with the satellite frequency in Europe. 12,623 MHz has an IF of either 2873 MHz or 1873 MHz. Both values are outside of reception band of the SX1280, so we use the 1873 MHz, as it comes out of the LNB at a much stronger signal.
This primary IF from the LNB must now be mixed with an LO on the Dreamcatcher. In the previous version, 3.05, we used a very expensive device from Qorvo, which integrated a synthesizer and mixer into a single IC. The last six month have been spent in trying alternative ways to achieve the same results, without spending $10 on a dedicated IC.
We came close by using a single Schottky diode as a mixer, but the SNR is worse than the 3.05 by 3. As you all know, losing that much signal is a big deal and I don’t want to ship a product that does not have equal performance in the North America and Europe.
A few filter changes and chip swaps were performed in the following iterations. Each time was actually even worse performance. Right now, three different boards are being designed to test alternatives in parallel.
- Using the existing mixer, removing the LNA in front of it, and putting a SAW filter right after the mixer.
- Replacing the diode mixer with the LTC5560 from Analog Devices.
- Replacing the diode mixer with the GRF7001 from GuerrillaRF.
I’m allergic to the second option, although it will likely provide the best performance. The LTC5560 is the most expensive IC and also requires more costly passives to function.
Once these boards are built, I’ll be sending samples to @caveman99 and @Tysonpower for testing.
If I can’t solve for the lost 3 SNR after these iterations, then I think we’ll just further reduce the bitrate across both beams so that reception with a bare LNB is not an issue in Europe.
1 Like
thx for the info, looking forward to testing the new Boards on the EU Beam 
The new set of prototypes have been ordered. I should have them in about two weeks.
1 Like
New prototypes are on their way to me by DHL. Should have them in hand on Thursday.
1 Like
Sounds great!
Boards in hand. I’m testing them tomorrow.
2 Likes
Let’s hope for good results…
I’m using the most recent firmware posted on Github (the binary blob). For some reason, using the 9750 LO from the LNB is not allowing reception, but when the 22 kHz tone is activated and selects the 10600 LO, reception is possible.
The performance is similar to an old known-good prototype which has no mixer on the Dreamcatcher. The RSSI is a bit lower than I was expecting (-87 dBm RSSI). The SNR is -14, which is likely the cause of the CRC errors. I’ll repoint and to get a better SNR, which should resolve the CRC errors. For the current radio settings, I believe the required SNR is -11.
Tagging @caveman99 and @Tysonpower in case they have any confirmation on their end.
The CRC errors were definitely related to the low SNR. I minor nudge of the antenna brought SNR down to -7. Now I’m getting files.
1 Like
Dreamcatcher 4 with the integrated LO/mixer might be a bust. Fortunately, we just tested a dCSS LNB that worked really well. Assuming that the dCSS LNB passes another couple rounds of testing, Dreamcatcher 4 will basically be just the SX1280 and ESP32-S2.
Sorry that it’s all taking so long. But at least there is progress!
3 Likes
That sounds good in terms of the dCSS LNB, looking foward to giving it a try
LoRa is a great idea (meybe mesh like pylife, pineDio etc)
Why not using bigest processor like Raspberry Pi 2040 or normal STM ARM?
is any reason?
Is any software/hardware on board for positioning device? Meybe normal diode : too left, to up etc.
1 Like
@fajne Components-selection is primarily based on the cost of parts and cost of assembly. The Pi 2040 does not include wifi, so another IC or module would be necessary. I’m not aware of any STM microcontrollers that have similar processing capacity as the ESP32-S2 and also include dual-mode wifi (client and access point) and USB.
There is an LED on the board and recently software support for SNR-related beeps was added.
2 Likes
I don’t understand, one diode? How to use it?
I’m talking about turning it to the left, then you need to turn it to the left, one diode is not enough. @fajne is right
It is important that the system could do something else besides answering. For example, it could extract data or send it by local mail. Hence my suggestions that it should be a better processor. Or simply a raspberry pi zero