Would you mind showing your SNR and RSSI if you place the LNA between Morfeus and DC4?
will give it a try if it works 
Got a RSSI of -86 and a SNR of -6 to -8 today with a 60cm Dish, just the Dish on my DC3 gives me +9 SNR.
What is your RSSI with a DC3?
on the DC3 i have with the Dish RSSI -77, SNR +9, without the Dish -88 RSSI
Did some tests with LNA configs, strangely the LNA even when off improves the Signal compared to no LNA, what is quite strange.
60cm Dish standing at open Window.
LNB->Morfeus->DC4 -6 -99 (small Packet Loss)
LNB->Morfeus->LNA(off)-<DC4 -6 -96 (why is this better? LNA has no Power)
LNB->Morfeus->LNA(on)->DC4 -4 -88 (Current go to config gor me)
LNB->LNA(off)->MOrfeus->DC4 -11 -104 (Packet Loss)
LNB->LNA(ON)->Morfeus->DC4 -7 -86
I don’t understand how this configuration is better than with the LNA before the frequency conversion. Is this configuration, the LNA is amplifying all of the junk (spurs, intermods) from the Morfeus. I would have expected LNB–>LNA(on)–>Morfeus to provide the best results.
Can you receive with a bare LNB in either of the configurations with an LNA(on)?
I wasn’t able to get the bare LNB working with the DC4 in any Configuration, on the Old beam (upcovnerting) it worked, to me it seems that the morfeus likes upconverting more then down.
Even more strange to me is that the Signal is better with a LNA that is off, i can see the Signal on my SDR better then when i connect my SDR directly to the Morfeus as well, so i isn’t some strange bug of the DC4.
The final DC4 will have a built-in Frequency Shifter?
There are greater losses at the IF from the LNB, as the frequency increases. That is one factor, though probably not the only factor.
There is an LNA in the next version of Dreamcatcher, which acts as a sort of filter. Our RF engineer said that the simulation results with an LNA were better (produced less conversion loss), compared to the use of an LC filter. The LNA reflected back some of the signals, but I can’t describe the exact detail of what is happening. Also, the simulation assumed that the LNA was powered. I’m guessing that the LNA is still reflecting some signals, even though it is not powered.
Yes, the goal is for performance with the new frequency converter to be similar to that of DC3. But so far, it has not been working. If this next version does not show decent results, I’ll need to go back to using the Qorvo part.
The next board and firmware does enable the 22 kHz tone for the LNB, so the IF will have a stronger signal. The IF will require up conversion.
12,600 MHz - 10,600 MHz LO = 2,000 MHz IF from LNB
Here might be something to consider concerning the LNA (when off) causing an improved signal strength. Installing the LNA, on or off, improves the impedance matching in the cable run between the LNB and the receiver input. Maybe the Morfeus unit is causing an impedance mismatch that’s improved with the LNA installed. Something to chew on. Good luck.
LNB IF matching is a mastery in itself, especially with low-end LNB in the market that are not really well designed. So specifically not the bullseye… BUT. i have seen harmonics from oscillators within the ZF band up to a point that ‘mirror’ transponders are so strong a DVB receiver could lock them. The Broadcast-LNBs are fairly good designed to suppress that but there are a few models out there that use existing mass market circuitry shoehorned to lower or higher frequencies that shift these harmonics into the reception band.
TL/DR: to find out how the combination of LNA and morpheus handles in this respect you’d actually have to look at the spectrum. Every (wanted and unwanted) intermodulation product means energy/noise which will be amplified by the LNA and influence reception in some way.
Another point to consider with these LNA circuits is the quality of the power supply / bias-T - if that is wonky, you’re amplifying the harmonics of switching power supply as well. That could be the reason why it’s better with power off.
1 Like
We are getting our asses kicked with this new design for the secondary frequency conversion. The latest schematic is right here.
When tuning into SES-2 at 12,089 MHz, the reception is fine. RSSI is strong and SNR is comparable to DC3. In this case, the LO on the LNB is 9750 MHz and the IF from the LNB, which goes into the Dreamcatcher, is 2339 MHz. The SX1280 can easily demodulate at that frequency.
The latest 2104-version of Dreamcatcher 4 has actually received some packets in Europe, but the SNR value is considerably worse. After the 22 kHz tone is activated, the LO on the LNB jumps to 10600 MHz. The 12,623 MHz (subtract 10,600 MHz) gets turned into an IF of 2023 MHz. We then mix in another LO, which is 400 MHz. In this particular instance, DC3 was at an SNR of -10 and DC4 was at -16. The RSSI of DC3 was -84 dBm, whereas the DC4 was at -58. This is with a bare LNB.
The 2104-version attempts to replace the very expensive Qorvo RFFC5072 on the Dreamcatcher 3 with a frequency converter made of separate components. You’ll see an LNA immediately after the F-connector, followed by a Schottky diode and filter network. The circuit is designed for an LO source from 300 - 1100 MHz, which is also in the circuit. During bench testing with a 12 GHz transmitter, the performance of DC4 is fine, but
I would like to start selling these pretty soon, so sales may be limited to US customers until the frequency conversion circuit is figured out. I’m happy to share a few prototypes with anyone who has bench testing equipment to analyze the problem and with an antenna pointed at Eutelsat 8WB.
1 Like
Yeah the Qorvo is really Expensive, 13€ (100pcs) at mouser for such a little thing is A LOT!
If the current Approach without a rf mixer chip isn’t performing as it needs to maybe a cheaper RF Mixer could be an alternative? Something like this maybe: https://eu.mouser.com/ProductDetail/Qorvo/RF2053TR13?qs=tkvX2tz6JTF6HrxTU11iOw%3D%3D
If the 22 kHz tone is somehow injecting noise into the LNB’s signal, what happens if you operate the system without the 22 kHz signal? A 12,263 MHz signal with a 9750 LO should output 2873 MHZ. Mixing that with 400 MHz would bring it into the 2.4 GHz band range. Or is there something in the LNB preventing that? Just some thoughts. Good luck.
The problem is not the 22khz tone, this is actually generated by a mass-market IC that is used for this puropse in millions of satellite receivers. The Problem is that the EU frequency (and possible other regional frequencies in the future) requre a secondary mixer and that’s generating the headache. If the LNB LO could be shifted finer that 9750/10600 as it is now this would not be a problem and we could adjust the primary mixer in the LNB to output in the right range.
I see one design problem in total. The Universal LNB’s and especially the Bullseye use the Satellite BSS and FSS frequencies and convert the input frequency to a IF of 950-2150 MHz. Every Satellite receiver can tune that range. BUT the input for the SX1281 is higher than that. So we’re technically using the Broadcast LNB’s outside of their spec. The US frequency is already in the high band (+22kHz) for a commercial sat receiver.
The Sat frequency blocks are: 10700-11700 MHz low band (LO 9750) -> 950-1950 MHz IF and 11700-12750 MHz High Band (LO 10600) -> 1100-2150 MHz. The Bullseye will actually get lower than that cause it has an extended lower frequency range of 10489 MHz, but only in lowband.
Maybe the solution could be to shift the SX1281 down a fixed number of 500 MHz to bring it within the defined reception range of the LNB. USA would mean switching to highband reception but it should be an easier RF design than a programmable or dynamic freq shifter.
Changing the LO on the LNB requires changing the reference crystal. There are some LNBs which have tremendous frequency flexibility, but those dCSS models are very expensive. We could add multiple reference crystals in a future Bullseye, which would be cheaper than a dCSS LNB, but that doesn’t solve the near-term problem.
This is actually not much of a problem. There are basically only two LNB ICs on the market now (with monstrous market share): RDA and RafaelMicro. They are very similar components and both can output their IF at 2400 MHz without much added loss. For example, the difference in received power between a 1489 MHz IF and 2339 MHz IF is only 2 dBm.
On SES-2, for North American coverage, we use 12089 MHz without 22 kHz, so it’s the 9750 MHz LO. Yes, the LNB is being used outside of spec, but the specs are all pretty conservative and focused on DTH marketing.
I do have some good news to share. The Dreamcatcher 4 is now receiving with 100% valid packets. The 22 kHz is on, so the LO is 10600 MHz, which results in a primary IF 2023 MHz. We then add 500 MHz to produce a secondary IF of 2523 MHz.
The downside is that we are still not at equivalent performance to the Dreamcatcher 3.05. SNR in that device is 10.5, whereas DC4 has an SNR of 13.
I haven’t yet looked into the noise floor or a bunch of intermods/spurs causing interference at 2523 MHz.
I have to correct myself there - i’ve had a bit of a brainfart late at night. Commercial satellite receivers of course can TUNE the IF band (or digitize it in whole in case of an FBC tuner) whereas we have a narrow target frequency band of the SX chip, namely the 2,4 GHz ISM band. So my idea of using a fixed mixer is moot. We of course need an adjustable IF cause that is what is actually “tuning” in our case.
Still, every IF produces harmonics and you have to watch out to not cancel out your signal with a harmonic from another IF. Or use appropriate filtering to suppress the unwanted spikes.
dCSS is pretty cost prohibitive, and i really doubt you could get a clear narrowband signal from that. Running a Unicable II setup myself, it’s not the most reliable thing in the world either.
I do enjoy the “find the electron” type exercises. In the schematic, I see a 1100 MHz lowpass filter on the RF output line from the 500 MHz oscillator. I assume that’s there so the 2023 MHz signal from the LNA is stopped. But I don’t see any filtering to stop the 500 MHz from making its way to the LNA. The 500 MHz is going to jump right through the C15 coupling cap and reach the LNA’s output, which is some sort of impedance matching network according to the block diagram. I couldn’t find the LNA’s schematic to see what that looks like inside. The frequency mixing occurs at the output of L2 where the 2 signals meet but the 500 MHz signal also travels to the LNA, where it could be reflected or maybe even re-injected, depending if the impedance matching has some sort of feedback setup. You could end up with 2 different mixed signals, slightly out of phase heading toward the SX1280’s RF input. The RSSI would remain the same but the SNR would be lower because of induced phase noise from the out of phase signals. Has anyone considered adding filtering to keep the 500 MHz signal out of the LNA and see if it improves the SNR? Just some thoughts. Good luck.
Just an observation. In Europe you can buy (even at a decent price) so-called wideband LNB’s without the need to utilize the 22kHz signal to switch LO. They come with an LO of 10400 MHz and convert the entire Sat-band of 10700-12750MHz to 300-2350 MHz. They also don’t have a 14/18V switch for polarization but instead use one connector for V and one for H. These LNB’s are meant for directly driving dCSS switches with a reduced amount of cabling to the outdoor unit and are used to supply larger houses with satellite TV.
In out Case (Eutelsat W8) the math would mean 12623-10400=2223 which should still be in the ballpark for the SX chip, right?
An example for a wideband LNB would be Wideband 40mm LNB with Horizontal/Vertical ports, LO 10.40 GHz (EU) - Inverto.TV which is sold in the market for 15-20 €. If i hooked one of these up to my DC3, how could i test bypassing the mixer on that board?