Saturn: A New Apache Labs SDR Transceiver (ANAN-G2)

A place for Apache HARDWARE discussions, from bare boards to the latest hardware. SOFTWARE discussions belong in the SOFTWARE sub-forums.
User avatar
w9ac
Posts: 290
Joined: Sun Apr 09, 2017 4:01 pm

Saturn: A New Apache Labs SDR Transceiver (ANAN-G2)

Postby w9ac » Mon Mar 27, 2023 1:13 pm

A new SDR transceiver (or Orion board alternative) is in the works from Apache Labs. Among other new hardware features, take note of the PCIe interface and "thin client" remote possibilities. Price TBD.

https://apache-labs.com/al-news/1033/Introducing-the-Next-Generation-Saturn-SDR-Platform.html



Paul, W9AC
Last edited by w9ac on Mon Mar 27, 2023 1:46 pm, edited 1 time in total.
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Saturn: A New Apache Labs SDR Transceiver

Postby w-u-2-o » Mon Mar 27, 2023 1:45 pm

Paul--thanks for posting this.

It's a subject worthy of a new sub-forum, so I have made one! :)

No doubt there are going to be a billion questions. I've already sent a long list of them to Abhi, Laurence and others. I will report what I get back.
User avatar
w9ac
Posts: 290
Joined: Sun Apr 09, 2017 4:01 pm

Re: Saturn: A New Apache Labs SDR Transceiver

Postby w9ac » Mon Mar 27, 2023 1:48 pm

w-u-2-o wrote:Paul--thanks for posting this.

It's a subject worthy of a new sub-forum, so I have made one! :)

No doubt there are going to be a billion questions. I've already sent a long list of them to Abhi, Laurence and others. I will report what I get back.


Thanks, Scott. I edited the first post as I believe Saturn is the name of the new SDR transceiver board, not the model number of a complete transceiver.

Paul, W9AC
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Saturn: A New Apache Labs SDR Transceiver

Postby w-u-2-o » Mon Mar 27, 2023 2:05 pm

I think Apache is going to use the term "Saturn" to refer to both the top level assembly (box) and the new SDR board. We'll see as the project matures.

Note that we have to use the term "new" advisedly. It's the same old, same old but realized in a new way. It's still the same radio, it will still use Thetis/piHPSDR/linHPSDR.

I think it's going to be easier to use and easier to develop on in many ways. Pushing more of the functionality into software is the way to go. Using a lot more commercial intellectual property ("IP", firmware slang for pre-built FPGA firmware code) in the FPGA is the way to go (as long as it doesn't somehow screw over derivative works). The reliability and stability of commercial IP, and the Linux Ethernet networking stack, will solve a lot of the reliability and stability problems that have plagued us over the years.

However, there will be new challenges for the user community. Now there will be the care and feeding of a Raspberry Pi operating system to deal with. Users will have to learn new skills like how to SSH into Linux and use the Linux command line for certain maintenance activities like updating firmware.

It's hard not to want to compare this to the ill-fated Minerva project. Minerva, sometimes known as the "DFC" project, was the brainchild of Phil, VK6PH. It almost accomplished the complete elimination of firmware. You can google it, there are some Youtube videos kicking around. The FPGA on that board was quite vestigial and merely pushed all raw ADC and DAC data over PCI where it was to be processed in a GPU. If you already had an Nvidia CUDA-capable GPU in your PC (and many of us do) you would have been good to go. Minerva was was nearly 100% C code. They had boards and everything. Unfortunately the project fell apart right near the end due to, let's just say, "staffing issues". It was Saturn on steroids in terms of reducing firmware content.
W4WMT
Posts: 325
Joined: Sun Apr 09, 2017 10:12 pm

Re: Saturn: A New Apache Labs SDR Transceiver

Postby W4WMT » Mon Mar 27, 2023 6:47 pm

Doesn't this just turn it into a Flex?
Albeit a Flex that can, also, run Thetis.
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Saturn: A New Apache Labs SDR Transceiver

Postby w-u-2-o » Mon Mar 27, 2023 7:58 pm

W4WMT wrote:Doesn't this just turn it into a Flex?
Albeit a Flex that can, also, run Thetis.

If you mean thin client vs. thick client architecture, the answer is "not yet". However having Thetis split into two parts such that the server could run on the CM4 module and the Thetis UI on Windows would be really great, and would finally allow an openHPSDR radio to be remotely operated almost as easily as a Flex.

However that's for a future that may or may not occur. For now the team merely took the functionality out of the existing firmware and split it across a different FPGA and a general purpose processor, in this case a Raspberry Pi Compute Module 4 (CM4). Altogether the hardware looks like a regular 7000 to Thetis. In the case of piHPSDR it also looks like a 7000, but the excess capacity of the CM4 can be used to host piHPSDR internally. Either way it's still thick client.
W4WMT
Posts: 325
Joined: Sun Apr 09, 2017 10:12 pm

Re: Saturn: A New Apache Labs SDR Transceiver

Postby W4WMT » Mon Mar 27, 2023 10:14 pm

Well, if it gets all the Ethernet stuff out of the FPGA for good, it was probably worth it for that alone :-)
Still though, a Pi seems like gross overkill. Monday Morning Quarterback said that :-)
K1CWD
Posts: 10
Joined: Wed Jan 18, 2023 3:51 pm

Re: Saturn: A New Apache Labs SDR Transceiver

Postby K1CWD » Tue Mar 28, 2023 4:22 am

IMO cutting over to Xilinx is an excellent choice. Using PCIe is a no-brainer, and the Raspberry PI is interesting, but there are a lot of choices for that portion of the system. Could have maybe used a Zynq and PetaLinux, but understand the design choice of the PI.

One area that I would be concerned about will be putting a large attack surface on the radio from a cyber perspective, but this isn't something the amateur radio community typically cares about. It will require endless OS updates for the radio, but everyone is used to living with that by now.
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Saturn: A New Apache Labs SDR Transceiver

Postby w-u-2-o » Tue Mar 28, 2023 1:27 pm

The Pi was a good choice. The main problem with the Pi is that it's a bit underpowered. It would have been nice to see something a bit more beefy. The other problem with the Pi is its popularity. Apache will have supply chain problems sourcing them.

Other reasons the Pi was a good choice is because it is very well supported, and because now there is some wiggle room for experimenters. There are really interesting pin-compatible alternatives from Pine64, Banana and, most particularly, Radxa. Some of these include neural processor chips.

Zynqs are great for super-embedded stuff, but they are weak sauce where horsepower is concerned and drive both development complexity and cost up considerably. It was a very good decision by Laurence et al to divorce the GPP from the FPGA.

I am curious which flavor of CM4 Apache is going to spec. because it comes in a variety of RAM and memory configurations. I'd like to have one spec'd with the onboard WiFi so that the GigE interface can be dedicated to radio-only and the WiFi interface used for maintenance and other functions.
W4WMT
Posts: 325
Joined: Sun Apr 09, 2017 10:12 pm

Re: Saturn: A New Apache Labs SDR Transceiver

Postby W4WMT » Tue Mar 28, 2023 2:19 pm

Laurence's YouTube video is a bit unfocused, so it's a little hard to see the block diagram. But it's my impression, from looking at the bullet points, that the DDC is no longer using CIC decimation filters? So it's doing all the decimation in FIR? If so, that's pretty cool!
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Saturn: A New Apache Labs SDR Transceiver

Postby w-u-2-o » Tue Mar 28, 2023 4:00 pm

Capture from the video. The Pi appears to have WiFi and not the eMMC, but who knows what they'll actually ship. You can also see the SD card on a board edge that might not be easily accessible when (not if) it eventually fails. If I ever have one of these and it does not come with a CM4 with eMMC the CM4 is definitely getting replaced with one that does.

Capture3.JPG
Capture3.JPG (260.16 KiB) Viewed 7680 times
laurencebarker
Posts: 219
Joined: Mon Nov 11, 2019 7:39 pm

Re: Saturn: A New Apache Labs SDR Transceiver

Postby laurencebarker » Tue Mar 28, 2023 4:30 pm

Saturn does use CIC filters. It does have unused multiplier resources - even after having 10 DDCs - but the FIR filters consume block RAM too. I'd be surprised if it was viable to use purely FIR downconversion unless we reduced DDC count; and we did have people asking for that many.

I don't know what the plans are for the pi4 CM. My choice would be with wifi, 4GB RAM and external SDCard (as I have now). We've used both flash memory variants in testing. Recovering one where the flash is on the pi requires a switch to be changed on Saturn, then plug a USB cable into the side connector, then execute special software on your PC. Then remember to change the switch back. It can be done, but harder to explain to users.

We did consider Zynq processors; the Ultrascale+ ones are comparable to current pi4 processors. They were expensive, but the recent Kria system on module makes them look more attractive. But I imagine there will be higher performance CM4 before there will be new FPGAs. You can see our rationale in a presentation at Friedrichshafen last year:

https://www.youtube.com/watch?v=JUw4-0-1lzI
Laurence Barker G8NJJ
K1LSB
Posts: 639
Joined: Wed Feb 05, 2020 5:25 pm

Re: Saturn: A New Apache Labs SDR Transceiver

Postby K1LSB » Tue Mar 28, 2023 7:05 pm

Laurence,

What advantages or improvements does the Saturn offer compared to a current 7000DLE running Thetis on a late model computer?

TIA,
Mark
K4IBC
Posts: 64
Joined: Sun Apr 09, 2017 7:31 pm

Re: Saturn: A New Apache Labs SDR Transceiver

Postby K4IBC » Tue Mar 28, 2023 7:37 pm

Since data is PCIe between the FPGA and the CM4, does that mean that the firmware is going to be the same for protocol 1 and 2. Then it would be the matter of running a P1 or P2 app in the CM4. Or will this be strictly a P2 radio.

Yeah I know P1 is outdated and no longer necessary.

Just thinking to myself that by offloading the Ethernet from the FPGA would eliminate some of the firmware issues with timing and sequence errors. It simplifies that portion of the radio.

Where do you see upgrades happening the most the firmware or in the CM4 code?
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Saturn: A New Apache Labs SDR Transceiver

Postby w-u-2-o » Tue Mar 28, 2023 9:13 pm

K1LSB wrote:What advantages or improvements does the Saturn offer compared to a current 7000DLE running Thetis on a late model computer?

Not Laurence, but in terms of RF performance this is a clone of the 7000/8000.

In terms of functionality that is driven by the client software. It will support piHPSDR, linHPSDR and Thetis. Pick your poison. If it's not in the UI you can't do it. If it's in the UI it will work. When this launches in all cases you select a rig type of "ANAN-7000".

This evolution of the hardware does, however, promise a few immediate advantages. We'll see what Laurence might say, but the obvious ones are:

- Full GigE stack in Linux should make networking much more reliable and robust.

- You can host piHPSDR or linHPSDR on the CM4 and have an "all in one" machine. However right now that's also a disadvantage in that neither app is as feature rich as Thetis. Nevertheless some folks are really turned on by the idea of something that more closely approximates something like a more conventional radio.

It's the future advantages that are really compelling. For example, the ability for non-firmware folks to now program many of the things that used to be Verilog but are now in C. The ability to support a thin client/server architecture.

K4IBC wrote:Since data is PCIe between the FPGA and the CM4, does that mean that the firmware is going to be the same for protocol 1 and 2. Then it would be the matter of running a P1 or P2 app in the CM4. Or will this be strictly a P2 radio.

The presentation says this will be a strictly P2 radio. However, because the GigE stuff is now executing in the CM4 on a proven GigE hardware platform and a proven Linux networking stack, all of the trouble folks might have had with GigE reliability should be largely eliminated. The protocol does still rely almost entirely on UDP, though, so if your network is dropping UDP packets you are still going to have to fix that.

Just thinking to myself that by offloading the Ethernet from the FPGA would eliminate some of the firmware issues with timing and sequence errors. It simplifies that portion of the radio.

Yes, exactly. It doesn't simplify the radio, but it does simplify the FPGA.

This assumes, of course that everything is rainbows and unicorns on Day 1. Who knows what challenges are lurking in the PCIE firmware code, for example.

Where do you see upgrades happening the most the firmware or in the CM4 code?

We haven't had an "upgrade" in the firmware in a long time, not since the advent of QSK in P2. There really haven't been many upgrades in the firmware at all. We went from P1 to P2, added QSK in P2 (for certain hardware), and there was an improvement made in annunciating the firmware version to the software. That's it. There's been many bug fixes to P2, however (thanks, Rick!).

From that perspective we won't see many upgrades to the firmware or the CM4 code very often. However...now that the CM4 code is much more accessible to mere mortals, i.e. non-firmware engineers, perhaps that will change. And since the Xilinx is now supporting 10 DDCs, it'll be a while before both the CM4 and thick client code catches up.

This might all change, however, if some group of developers decides to bite off on a real thin client/server re-spin of everything.
User avatar
KC1LKO
Posts: 31
Joined: Thu Jul 23, 2020 11:45 pm

Re: Saturn: A New Apache Labs SDR Transceiver

Postby KC1LKO » Wed Mar 29, 2023 1:42 am

It's the future advantages that are really compelling. For example, the ability for non-firmware folks to now program many of the things that used to be Verilog but are now in C. The ability to support a thin client/server architecture.


^^^^ THIS :-)
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Saturn: A New Apache Labs SDR Transceiver (ANAN-G2)

Postby w-u-2-o » Wed Mar 29, 2023 8:09 pm

So it looks like Apache has set both the name and the price for the new unit, the ANAN-G2.

Prices for with and without a front panel shown on their website:

https://apache-labs.com/1001/Ham-Radio-Products.html
laurencebarker
Posts: 219
Joined: Mon Nov 11, 2019 7:39 pm

Re: Saturn: A New Apache Labs SDR Transceiver

Postby laurencebarker » Wed Mar 29, 2023 8:11 pm

I never started Saturn from the perspective of it being better than Orion. Simply that it was different, and would enable different things. My original plan - which I may yet get back to - was a smaller lighter QRP radio. That was going to be Zynq FPGA; and ultrascale+ one could still be a good choice for that.

So what's different?
Well, it should have eliminated the ethernet interface issues; that's now provided by the Raspberry pi motherboard. It has 10 DDC that someone might have a use for. Linhpsdr could use those straightaway.

As Scott says, it could have a "think client" UI. Only today one of my work colleagues was telling me about the delights of writing user interfaces for web clients. Well that happen? I'm not a software developer so I'm likely to need some help!

There are unused FPGA resources. One of the things I'm interested to know is, could I put the panadapter code into the FPGA to offload the raspberry pi? It would make no difference if you were using it with Thetis; with a low power QRP radio it could make a big difference.

Will there be a P1app? Well there's a half written one already, that was used in early testing. Then I wrote the P2app - expecting to find it really hard - and it just worked so I've never been back to P1. It could be done.

The PCI express interface has been very stable. It uses Xilinx IP and "just works". On one side it plugs into PCI and has a linux device driver; on the other I get two bus interfaces - one for register accesses and one for "direct memory access" bulk data transfers. The linux device driver was unsupported on ARM and I had to remove two bugs but it's been fine ever since.

When it ships I hope pihpsdr will have a direct interface to the data, rather than through protocol 2; at the moment CPU cycles are being tied up moving data to an intermediate standard and moving data through unnecessary UDP packets. Rick has been working at that; some of you will know that he is also tied up with other firmware upgrades for the current series of hardware.

We are likely to have to change the codec for obsolescence reasons. There are no direct replacements, so some work will be needed all in CM4 software. Because we have various version numbers programmed, the software will be able to work out what driver it needs. A side benefit is we are likely to get far more control over mic gain that 0dB or 20dB boost.

There is a lot more flexibility built into this approach. Where that exists and clever people get access to it, novel applications will emerge.
Laurence Barker G8NJJ
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Saturn: A New Apache Labs SDR Transceiver (ANAN-G2)

Postby w-u-2-o » Wed Mar 29, 2023 8:21 pm

Thanks for the response, Laurence.

There is a lot more the current CODEC can do than what it is doing, including control mic gain. It's simply not supported in the firmware and exposed to the client software.

It's hard to tell from the photos but the XLR looks like a mini-XLR male with 3 pins, is that correct? That's pretty obscure outside of the world of professional wireless microphones. Why did the team not decide upon a more straightforward full size female XLR? What preamp IC was used?
K4IBC
Posts: 64
Joined: Sun Apr 09, 2017 7:31 pm

Re: Saturn: A New Apache Labs SDR Transceiver (ANAN-G2)

Postby K4IBC » Wed Mar 29, 2023 10:12 pm

Yeah the technical specs say Mini XLR. and I assume the PTT, Line In, and DIgi in are 3.5mm TRS. Shame they didn't add one more and give us line level outs. I am surprised at the price point. It is much lower than I expected.

Since the back is the same for both radios with or without screen and knobs, I am tempted to go knobs. Learn to use with and without PC. A computer disaster wouldn't mean being off the air.
User avatar
KC1LKO
Posts: 31
Joined: Thu Jul 23, 2020 11:45 pm

Re: Saturn: A New Apache Labs SDR Transceiver (ANAN-G2)

Postby KC1LKO » Thu Mar 30, 2023 12:42 am

When it ships I hope pihpsdr will have a direct interface to the data, rather than through protocol 2


This is the sort of stuff that enables magic down the road.

73
KC1LKO
K1CWD
Posts: 10
Joined: Wed Jan 18, 2023 3:51 pm

Re: Saturn: A New Apache Labs SDR Transceiver (ANAN-G2)

Postby K1CWD » Thu Mar 30, 2023 1:23 am

Won't Saturn quickly cannibalize sales of other products like the 7000DLE?
K1LSB
Posts: 639
Joined: Wed Feb 05, 2020 5:25 pm

Re: Saturn: A New Apache Labs SDR Transceiver (ANAN-G2)

Postby K1LSB » Thu Mar 30, 2023 9:18 am

K1CWD wrote:Won't Saturn quickly cannibalize sales of other products like the 7000DLE?

What does it matter? Apache Labs gets the sale regardless.

Mark
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Saturn: A New Apache Labs SDR Transceiver (ANAN-G2)

Postby w-u-2-o » Thu Mar 30, 2023 1:48 pm

K1CWD wrote:Won't Saturn quickly cannibalize sales of other products like the 7000DLE?

That's a good question.

They could get stuck with a number of unsold inventory of 7000's. Perhaps we'll see some sort of "fire sale" of 7000's at some point. They also do seem to be "trickling out" the 7000's, i.e. they appear to be made in small lots, so any sale of remaining units would probably be small.

At any rate, Apache has made their decision so we can only assume they have a plan they are happy with.
K1CWD
Posts: 10
Joined: Wed Jan 18, 2023 3:51 pm

Re: Saturn: A New Apache Labs SDR Transceiver (ANAN-G2)

Postby K1CWD » Thu Mar 30, 2023 1:52 pm

K1LSB wrote:
K1CWD wrote:Won't Saturn quickly cannibalize sales of other products like the 7000DLE?

What does it matter? Apache Labs gets the sale regardless.

Mark


Well I would think that any interest in the community supporting a radio with little to no further sales or EOL announcement would diminish rapidly. Trying to fix or add features to the existing Orion board would not make a lot of sense given how difficult it is compared to something like the new Saturn architecture.

Maybe I need to sell my 7000 now before the price on the used market drops precipitously!


73,

Craig K1CWD
K1LSB
Posts: 639
Joined: Wed Feb 05, 2020 5:25 pm

Re: Saturn: A New Apache Labs SDR Transceiver (ANAN-G2)

Postby K1LSB » Thu Mar 30, 2023 1:56 pm

There has been no mention of any "EOL announcement" for the 7000.

Mark
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Saturn: A New Apache Labs SDR Transceiver (ANAN-G2)

Postby w-u-2-o » Thu Mar 30, 2023 2:22 pm

A technical concern comes to mind this morning: how will safe Pi shutdown be achieved on units that are headless? Just slamming the power off tends to lead to SD card corruption.
K1CWD
Posts: 10
Joined: Wed Jan 18, 2023 3:51 pm

Re: Saturn: A New Apache Labs SDR Transceiver (ANAN-G2)

Postby K1CWD » Thu Mar 30, 2023 2:34 pm

w-u-2-o wrote:A technical concern comes to mind this morning: how will safe Pi shutdown be achieved on units that are headless? Just slamming the power off tends to lead to SD card corruption.


I have had pretty good luck with embedded Linux due to the journaling file systems available and taking some measures such as disabling write caching (which of course hurts file system performance to some degree), and disabling a lot of the more noisy system logging. You can get even better reliability with other techniques depending on how far you want to deviate from the baseline distro.

I wonder who will provide the active maintenance of whatever modified distro is provided since the community will have to take ownership if the Apache Labs business model stays the same.
laurencebarker
Posts: 219
Joined: Mon Nov 11, 2019 7:39 pm

Re: Saturn: A New Apache Labs SDR Transceiver (ANAN-G2)

Postby laurencebarker » Thu Mar 30, 2023 4:50 pm

"There is a lot more the current CODEC can do than what it is doing, including control mic gain. It's simply not supported in the firmware and exposed to the client software."

There is indeed more you can do, but not programmatically change the gain. See the functional block diagram in the data sheet, section 1.3.

You can control line in gain, but the mic doesn't go through that path. The only adjustment for mic gain is a 0 or 20dB switch. You can feed it into headphones as a sidetone level and that's adjustable (and enabling that, as an option, is the kind of thing we can more easily do with Saturn).
Laurence Barker G8NJJ
Joe
Posts: 190
Joined: Tue Aug 22, 2017 5:58 pm

Re: Saturn: A New Apache Labs SDR Transceiver (ANAN-G2)

Postby Joe » Thu Mar 30, 2023 5:24 pm

I have been following the postings and very promising for sure. I just had a quick question, in the video there is mention that the Saturn and Orion form factor are the same, is there a possibility to order just the Saturn board ? I guess you know where I am going with this or say you had an Orion board failure and with getting eol parts these days would the Saturn be a replacement obviously with additional I/O work.

Thanks,
Joe
WD5Y

Return to “Apache-specific Hardware Discussions”