OpenHPSDR protocol

[Art]

Hello Group

In this topic, I want to touch on the features of the OpenHPSDR protocol. I use PCBs of my own design, so I need to change the gateway based on OpenHPSDR.

The first question at the moment is this: 

I can not start the "Disable PA" function in transverter mode. The OpenHPSDR gateway has a FPGA_PTT command, which switches the transceiver to transmit mode. But when you check the box in the desired window, the internal PA is still turned on. The PowerSDR is still transmitting this command to the transceiver. How can you force to turn off the internal PA in transverter mode? Or what command in the gateway can I use to do this?

https://github.com/TAPR/OpenHPSDR-Firmware

["Christoph v. Wüllen"]

In the old protocol, "disable PA" corresponds
to bit7 of C3 in the C&C packet with C0=0001001xb
(bit set: disable PA)

The FPGA should put this on an output (or shift
register) that enables the bias for the PA.

In the new protocol, the "pa disable" bit is
in the general packet, word 58 bit 0
(bit set: enable PA).

For the HL2, it is (unfortunately) different,
here disable PA is in ADDR=0x09 bit 19
(bit set: PA enabled) and this corresponds
to bit3 of C3 in the same C&C packet,
which in the old protocol normally
means "Enable/disable Apollo tuner".

So either your SDR software is HL2-aware, or
you have to reverse engineer and enable/disable
the Apollo tuner in your SDR software to achieve
PA switching on HL2.

> --
> You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/bd0066ae-4f53-4ffa-ac29-44483463b888n%40googlegroups.com.
> <xvtr_disable_pa.png>

[Art]

Thanks for the help. This is very important information for me.

Question 2:

I would like to study the PHY chip and use it instead of the KSZ9021. Now the RTL8211C chip with the RGMII interface has appeared. I found the Verilog code for the GMII version chip. In the QFN version - 64 / 88 is a GMII chip. Searching the group for the word RTL8211 gives several results. And obviously Steve (kf7o) is already familiar with this chip. Perhaps someone has a tweaked code to work with this chip? Since I use boards of my own design, it will not be difficult for me to redo the board routing for a new chip. I can attach modules and individual pieces of code  and make them work. But I can't write code from scratch yet on Verilog. Perhaps it is possible to fix the existing code of the RGMII chip? 

This chip cannot be installed on the HL2 boards, since the functions of the pins do not coincide with the KSZ9021 / KSZ9031 chips.

https://github.com/maxs-well/Ethernet-design-verilog

понедельник, 15 ноября 2021 г. в 13:13:31 UTC+2, "Christoph v. Wüllen":

[Art]

Since the RGMII interface is standard for everyone, we only need to program other values of the internal registers of the new chip. If anyone has a proven solution, please let me know. I think it is possible to do only with other register values, without cardinal alteration of the code.

пятница, 19 ноября 2021 г. в 10:39:40 UTC+2, Art:

[Steve Haynal]

Hi Art,

If you search this list way back more than 5 years ago, you will find that the RTL8211C was considered. I decided not to use it because it was a departure from what was known to work with the openhpsdr code, and would require and additional voltage regulator as the specified core voltage is not 1.2V. Although it does have an internal regulator, I am concerned that the internal regulator may produce too much EMI for a SDR. Any RGMII device should work and only require some small changes to the MDIO accesses. If you are looking for a cheap and available gigabit ethernet PHY, see the B50612D"

https://github.com/q3k/chubby75/issues/16

It is used in this very inexpensive FPGA board:

https://github.com/q3k/chubby75

Which is one of the boards supported by LiteX (a great set of tools and IPs for those wanting to start playing with inexpensive FPGAs):

https://github.com/litex-hub/litex-boards

Note in the picture how this inexpensive FPGA board separates the required magnetics from the ethernet jack whereas the HL2 uses a ethernet jack which incorporates the magnetics into the jack. It is less expensive in terms of parts count to separate these, but more expensive in terms of board area.

I know from private e-mail that you are interested in lowering the price, especially for people like you who live in places like the Ukraine where it can be difficult to obtain parts. The HL2 is already designed to be low cost, with parts available from other sources than DigiKey or Mouser, often from AliExpress. Can you please provide details on how you can achieve a lower price? For example, by replacing part X with Y, I think I can save Z percent. Please relate that to the total price. Saving 50% on one or two parts will not save 50% on the entire design. Do you really think you can shave off more than 10%-20% of the cost? Or are you more of a hobbyist and just interested in building your own boards?

73,

Steve

kf7o

[Art]

Hi Steve.

Thanks for the links. I have already seen this interesting board with FPGA from the Lattice company.

Yes, where I live, there are problems with the supply of Mouser components. There are no problems with supplies from Digikey, but the final cost of the parts turns out to be very high, since you have to pay a large percentage to local intermediaries. Therefore, the only solution is to order components in China. If you replace the parts with those below in the list, then it is possible to assemble the DDC transceiver from Chinese components:

-replacement KSZ9021 / 9031 with RTL8211C /E or B50612D. Now KSZ9021 / 9031 cannot be bought in China at the normal price.

-replacement of slow ADC MAX11613 on MCP3204

-replacing the LAN connector with a Chinese analogue, for example HR911130A

-replacing DC-DC converters with another, I found good converters with a frequency of 130 kHz in China, but more tests are needed. If everything works out, I can give you more information later.

- replacement of MCP4662 with another one, more affordable in China. We must search, there must be something.

-replacement of the codec, AK4951 on TLV320AIC23BPW

I do not think that this will reduce the cost of the transceiver by a large amount, therefore such replacements are only relevant for those places where it is not possible to buy in Mouser, etc.

And since the price of FPGA EC4CE22 has now grown 2-3 times in China, I decided to switch to the BGA version of Cyclones 4. I do not think that it will be available to everyone, since not everyone can solder BGA chips. In the end, this is purely my amateur interest and has more to do with the love of designing new devices than real economic improvements.

I think there is no point in changing something for the HL2 components. 

All the details that I have listed above can be used in a new design. For example, the transition to a new FPGA with a new ADC. I think HL2 has gone through the entire cycle of upgrades and should remain as it is at the moment.

суббота, 20 ноября 2021 г. в 06:37:00 UTC+2, softerh...@gmail.com:

[Art]

I will be able to test RTL8211C / E with a real SDR for interference from the built-in power source. If anyone has an MDIO.v file with settings for this chip, please let me know. In order not to do routine work.

Thanks!

суббота, 20 ноября 2021 г. в 09:31:00 UTC+2, Art:

[Steve Haynal]

Hi Art and Group,

Just quickly scanning your list I see some difficulties. For example, the MCP3204 is a SPI device, not an i2c device. This will require updates to the gateware as well as extra FPGA pins.

The assembly houses I have dealt with in China will not use any Ali Express supplier as a source, no matter how low the price. I have bought several AD9866 devices from Ali Express and my local assembly house found that half of them were unusable. These are parts that can have any number of problems, such as small cracks where moisture seeps in, improper handling of the part, etc. You need to factor in the price overhead of all the bad parts you will end up with if purchasing from Ali Express. Makerfabs is located in China, and knows about Chinese sources. They find the lowest prices from reputable sources.

If you are truly interested in a small batch (<10) of a radio with technology similar to the Hermes-Lite, I would suggest you build a radio cape for the QMtech bajie board:

https://www.aliexpress.com/item/4000323573953.html

https://www.aliexpress.com/item/4000928169941.html

For $50 you get a FPGA with 28K LUTs, slightly more resources than the HL2, with ARM processor and gigabit ethernet. The routes to the connector are length matches so you could even use LVDS that is required by some ADCs and DACs. For $90 you have the option of 85K LUTs, more than 3X what is found in the HL2. The idea is to port Pavel's great work in creating an amateur radio SDR from the red pitaya:

http://pavel-demin.github.io/red-pitaya-notes/

I don't think the port would be too difficult and would mainly involve updating the RTL interface to a new ADC/DAC. It is something I might be interested in helping with provided all parties fully embrace open hardware, keep all files including PCB source files on github, etc.

You could add an AD9866 cape for well under $50. Maybe include an amp and filters for under $100.

So that the group is fully aware, Art's prior work can be found here:

http://www.radioamator.top/hermes-2000/

From private correspondence more than 6 months ago with Art, he has said that this is not a commercial endeavor and has sold only a few units only to friends. Art also shared schematics and gerber files with me as required by the Hermes-Lite 2.0 licensing for derivative work when I asked for them. Art, can you please clarify again if this is still not a commercial endeavor or has it become one? Roughly how many Hermes-2000 units do you sell in a year? Is your Hermes-2000 open source and hardware? If so, how should people obtain schematics and PCB gerber files?

[Art]

Hi Steve

Question on Hermes 2000:

I don't think he deserves much attention. Since the scheme has changed very often, even now it is in beta version. I sold no more than 5 units per year to my colleagues from Ukraine and finished the assembly. If anyone is interested in Hermes-2000, I can send all the files to those who wish. I don’t think it’s available for review in beta without asking questions. The Hermes-2000 project will remain more of an option for acquaintance, although even in the form as it is, it is a completely finished device. The project was created more in order to gain experience and find optimal solutions for DDC architecture.

 Steve, if you wish, you can openly share, upgrade, etc. all files according to Hermes-2000.

XC7Z010 / XC7Z020 is a very interesting option for further development. What can we expect from such a platform? For example, using the XC7Z010 at 28k LE? Will the use of a dual-core processor in the FPGA itself give any advantages over conventional FPGA?

Perhaps a single-board design would be an interesting solution? Yes, not all BGA solders, but now this is not a problem. Of course, if it is interesting and inexpensive enough, I am ready to connect and make a finished version with open materials for review.

суббота, 20 ноября 2021 г. в 20:02:33 UTC+2, softerh...@gmail.com:

[Steve Haynal]

Hi Art,

The XC7Z010 should be very comparable in FPGA capacity to the Cyclone IV. It is newer and faster technology. The XC7Z020 has about 3X the FPGA capabiities. The dual core A9s have a fast connection to the FPGA, so you can move most of the openhpsdr protocol and ethernet packet RTL into C code. This accounts for ~25% of the current design. I wouldn't push the cores to run a full Linux GUI and SDR software, but you should be able to run some low-level FT8 decoders, or TCP server-type applications. 

One of the main attractions of the qmtech board to me is that you don't have to build a complete new PCB but only a daughter card SDR. 

See this for more zynq details:

https://www.xilinx.com/support/documentation/selection-guides/zynq-7000-product-selection-guide.pdf

73,

Steve

kf7o

[Vince Vielhaber]

Steve, are you considering doing anything with one of these xilinx
boards? You've got my curiosity up.

Vince - K8ZW.

> *Thanks for the help. This is very important
> information for me.*
> *
> *
> *Question 2:*

>
> I would like to study the PHY chip and use it
> instead of the KSZ9021. Now the RTL8211C chip
> with the RGMII interface has appeared. I found
> the Verilog code for the GMII version chip. In
> the QFN version - 64 / 88 is a GMII chip.
> Searching the group for the word RTL8211 gives
> several results. And obviously Steve (kf7o) is
> already familiar with this chip. Perhaps someone
> has a tweaked code to work with this chip? Since
> I use boards of my own design, it will not be
> difficult for me to redo the board routing for a
> new chip. I can attach modules and individual
> pieces of code and make them work. But I can't
> write code from scratch yet on Verilog. Perhaps
> it is possible to fix the existing code of the
> RGMII chip?

> *This chip cannot be installed on the HL2

> boards, since the functions of the pins do not

> coincide with the KSZ9021 / KSZ9031 chips.*

> <mailto:hermes-lite...@googlegroups.com>.

> To view this discussion on the web visit

> https://groups.google.com/d/msgid/hermes-lite/e7cc1d8b-ff2d-42a6-b84d-7d9a47399db1n%40googlegroups.com
> <https://groups.google.com/d/msgid/hermes-lite/e7cc1d8b-ff2d-42a6-b84d-7d9a47399db1n%40googlegroups.com?utm_medium=email&utm_source=footer>.

--
Michigan VHF Corporation -- nobucks dot net
K8ZW - http://www.hamradio.fun

[Steve Haynal]

Hi Vince and Group,

I don't have any firm plans, but am dreaming of some possible projects. ICs for complete assemblies are hard to find right now, but there still seems to be a lot of cheap boards available. So now might be an interesting time for an inexpensive hack or more cobbled-together experimental SDR. Sort of like the early days with the Hermes-Lite 1 (see this photo gallery) or the radioberry. It is probably time to start sharing thoughts for group discussion and feedback. I've changed the subject of this thread and will make some "blog like" posts over the next few days.

The qmtech bajie boards are an interesting option. They are new, have length-matched routes to the IO, well documented, include gigabit ethernet, and fairly inexpensive. I have added a cheap 100Mbs ethernet port to one of qmtech's other boards and was impressed with their boards. The only complaints about qmtech boards are perhaps inadequate decoupling capacitors and no clock inputs on Altera boards. The prices are often about the same as buying the FPGA alone in small quantities. But $50 or $90 for a zynq board is more than I'd want to pay for part of an inexpensive cobbled-together SDR.

Thanks to crypto mines being closed in China, there is now a glut of inexpensive EBAZ4205 zynq boards, often for $5 in China. There is an interesting blog series covering this board: part 1, 2, 3 and 4. Search for EBAZ4205 on github and you'll find lots of good information including schematics. This board has 3 IO connectors with 14 IOs each. These connectors could be removed and a small SDR daughter board placed instead. This board only supports 100 Mb/s ethernet, but that is what the HL1 had and is very usable. For example, you can still have 4-5 384kHz, 8-10 192kHz, or 10+ 96kHz receivers.  

The AD9866 as used in the HL2 could be used on the daughter card as there seems to be ample supply of these, but I am interested in trying the AD9865. It is a switch from 12-bit to 10-bit but a reduction in price from $22.75 to $11.85 per piece when bought in quantities of 100 from the previous analog.com links. I don't think the extra 2 bits make much of a difference. First, there is a lot of hype when marketing ADCs about the number of bits and dynamic range, but it is important to look at other metrics such as ENOB, or effective number of bits. Compare figure 12 for the AD9866 with figure 12 for the AD9865. Here we see that the ENOB in the LNA range we typically use (+10 to +20) bounces around in the 9.5-10 bits for the AD9866 and 9.0-9.5 bits for the AD9865. So for all practical purposes it looks like there is only about 0.5 bit advantage for the AD9866! This is 3dB or less difference in dynamic range. If you can only set your HL2 LNA to +24 dB before clipping is bad, then you will see the same at +21 dB with the AD9865. There is a lot of debate over how much LNA gain is beneficial, but most of the experts say just enough so that you have signal. That usually is in the +10dB to +15dB range. For these reasons I suspect the AD9865 will perform just as well for most people and cost half as much. The footprint and all control is identical with the AD9866, so a PCB can be built with either AD9866 or AD9865.

With the AD9865+EBAZ4205, I think we could have the core of a DDC/DUC HF SDR for well under $100, closer to $50. A user would still have to make or add PA and filters for a complete HF rig as was required with the Hermes-Lite 1. A daughter board would have the same i2c as on the HL2 to facilitate this.

73,

Steve

kf7o

On Sunday, November 21, 2021 at 9:27:38 AM UTC-8 

[Vince Vielhaber]

The only EBAZ4205s I'm seeing is on AliExpress, prices are between $20
and $40 right now. Looks like an interesting board at first glance.

Vince - K8ZW.

On 11/21/2021 06:46 PM, Steve Haynal wrote:
> Hi Vince and Group,
>
> I don't have any firm plans, but am dreaming of some possible projects.
> ICs for complete assemblies are hard to find right now, but there still
> seems to be a lot of cheap boards available. So now might be an
> interesting time for an inexpensive hack or more cobbled-together
> experimental SDR. Sort of like the early days with the Hermes-Lite 1
> (see this photo gallery

> <https://github.com/softerhardware/Hermes-Lite/wiki/Gallery-of-Hermes-Lite-Builds>)
> or the radioberry <https://github.com/pa3gsb/Radioberry-2.x>. It is

> probably time to start sharing thoughts for group discussion and
> feedback. I've changed the subject of this thread and will make some
> "blog like" posts over the next few days.
>
> The qmtech bajie boards

> <https://qmtechchina.aliexpress.com/store/4486047> are an interesting

> option. They are new, have length-matched routes to the IO, well
> documented, include gigabit ethernet, and fairly inexpensive. I
> have added a cheap 100Mbs ethernet port

> <https://github.com/softerhardware/hazelnut> to one of qmtech's other

> boards and was impressed with their boards. The only complaints about
> qmtech boards are perhaps inadequate decoupling capacitors and no clock
> inputs on Altera boards. The prices are often about the same as buying
> the FPGA alone in small quantities. But $50 or $90 for a zynq board is
> more than I'd want to pay for part of an inexpensive cobbled-together SDR.
>
> Thanks to crypto mines being closed in China, there is now a glut of
> inexpensive EBAZ4205 zynq boards

> <https://www.aliexpress.com/wholesale?catId=0&initiative_id=SB_20211121134942&origin=y&SearchText=ebaz4205>,

> often for $5 in China. There is an interesting blog series covering this
> board: part 1

> <https://embed-me.com/ebaz4205-recycle-cheap-crypto-miner-part-1/>, 2
> <https://embed-me.com/ebaz4205-recycle-cheap-crypto-miner-part-2/>, 3
> <https://embed-me.com/ebaz4205-recycle-cheap-crypto-miner-part-3/> and 4
> <https://embed-me.com/ebaz4205-recycle-cheap-crypto-miner-part-4/>.

> Search for EBAZ4205 on github and you'll find lots of good information
> including schematics. This board has 3 IO connectors with 14 IOs each.
> These connectors could be removed and a small SDR daughter board placed
> instead. This board only supports 100 Mb/s ethernet, but that is what
> the HL1 had and is very usable. For example, you can still have 4-5
> 384kHz, 8-10 192kHz, or 10+ 96kHz receivers.
>

> The AD9866 <https://www.analog.com/en/products/ad9866.html> as used in

> the HL2 could be used on the daughter card as there seems to be ample
> supply of these, but I am interested in trying the AD9865

> <https://www.analog.com/en/products/ad9865.html>. It is a switch from

> 12-bit to 10-bit but a reduction in price from $22.75 to $11.85 per
> piece when bought in quantities of 100 from the previous analog.com
> links. I don't think the extra 2 bits make much of a difference. First,
> there is a lot of hype when marketing ADCs about the number of bits and
> dynamic range, but it is important to look at other metrics such as
> ENOB, or effective number of bits. Compare figure 12

> <https://www.analog.com/media/en/technical-documentation/data-sheets/AD9866.pdf#page=13> for

> the AD9866 with figure 12

> <https://www.analog.com/media/en/technical-documentation/data-sheets/AD9865.pdf#page=13> for

> --
> You received this message because you are subscribed to the Google
> Groups "Hermes-Lite" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to hermes-lite...@googlegroups.com
> <mailto:hermes-lite...@googlegroups.com>.
> To view this discussion on the web visit

> https://groups.google.com/d/msgid/hermes-lite/16e8053d-06ce-483e-a503-7ac76189bceen%40googlegroups.com
> <https://groups.google.com/d/msgid/hermes-lite/16e8053d-06ce-483e-a503-7ac76189bceen%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Steve Haynal]

Hi Vince,

The $5 price is from a reference I read about what vendors pay for these board inside of China, probably at some quantity. The $20-$40 you see on Ali Express is after their markup and for quantity 1. The $5 was from more than 10 months ago so could higher now inside of China. The idea is that Makerfabs could make the daughter board and sell the EBAZ4205 as one shipment. If the price is $5 to $10 inside China, then Makerfabs could sell the EBAZ4205 in the $20-$30 range as a package deal too.

73,

Steve

kf7o

[Alan Hopper]

Hi All,

I did spend a little time last christmas pondering capes for this sort of fpga board using low cost adc's to undersample the vhf bands. Either one per band or selecting the clock speed to put a number of bandpass filtered bands across the adc bandwidth.  I didn't get far with this and did not come up with a plan for tx but it might be food for thought for someone.  You need adcs with high analog bandwidth but can use relatively slow sample rate especially if only doing one band.  I'll try and dig out my notes.

73 Alan M0NNB

["Christoph v. Wüllen"]

My feeling these ARM CPU/FPGA combinations
sooner or later replace the "classical" FPGAs.

Making the transition would imply that we loose much of
what has been done in the past. However, there is a
distinct advantage (at least with my RedPitaya) which is
of this class of hardware:

- all the TCP interfacing is done with an easy to maintain
C code
- most of the "automatic band filter selection" logic etc. etc.
is done in C, and very easily so
- one can easily implement things such that one can connect
both via UDP and TCP (This has been done for RedPitay, and
piHPSDR supports both connections)
- for any protocol, the TCP connection is more versatile and
one can use 100 MB/sec AND 1 GBit/sec.

So there is a big downside but also some pros for moving to
the "classical CPU/FPGA" combinations.

Christoph DL1YCF.

> To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/43cdf760-8df9-1872-961b-f6a1f0920d07%40michvhf.com.
>

[Roger Critchlow]

I was thinking while I looked at the test build for the keyer how the keyer project has neatly dodged the chip shortage.  

It seems that any surplus china can create, china will also figure out how to use up.  Isn't it optimistic to count on any juicy surplus lasting long enough for us to get through a design, test, revise, release cycle?  

Maybe there's a simpler hermes lite to be built with the parts that are available?  I can't read FPGA part numbers, but if I sort digikey's listing of in stock FPGA's by the number of parts available the first hundred results range from stock of 110,000 to down to 1187, lots of obsolete parts sitting in trays or on reels gathering dust.  Could you break the HL2 RTL onto multiple smaller FPGAs, assuming that the more available parts are the smaller, less capable ones.  The modem chip service and ethernet interface are just logic and memory, only the DUC/DDC chains need multipliers,  Any number of DDC RX could daisy chain the incoming sample stream, mounting as daughter cards on RAM sockets, and each could operate as an independent HL2 RX running at its own sample rate.  You'd have to multiplex/demultiplex UDP endpoints over one ethernet interface, but that's just more logic and memory.  But I can't read FPGA part numbers ot their specs and I don't know which dev tools they need, so I'm just talking through my hat here.  But we're hams, we're supposed to build our radios out of junk, right?

-- 73 -- rec -- ad5dz --

To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/8f84e621-62c2-4784-8010-2f5b59029ad9n%40googlegroups.com.

[Esteban Benito]

Hi Steve and all,

Please take a look at those inexpensive boards too:

https://github.com/q3k/chubby75

Regards.

--

You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/16e8053d-06ce-483e-a503-7ac76189bceen%40googlegroups.com.

[Scott AK5SD]

Steve,

If you are looking to replace the AFE, then you might want to check out TI's AFE7222. It is $19.82 in 1k quantity from TI, active, and in stock. The 100 piece price would be in the mid to upper 20's. For the money you get two 12-Bit 65MSPS ADCs (think diversity reception), two 130MSPS DACs, a 12-bit auxiliary ADC with two multiplexed inputs and two 12-bit auxiliary DACs. The ADCs have an ENOB of 9.8 bits and the SFDR is 85dB.

73,

Scott

AK5SD

[Steve Haynal]

Hi All,

Thanks for all the comments and discussion.

Alan, I think it would be good to make VHF/UHF possibilities easier in any future project, whether by undersampling or mixer. I would want to modularize more so that the ADC/DAC was separate from most of the frontend. This would allow for easier addition of undersampling filters or mixers. Also, separating the FPGA-engine from the ADC/DAC would allow experimentation with a wider range of ADC and DACs. Art recently posted a list of interesting ADCs, some with better undersampling potential than the AD9866. A more modular approach was discussed in the tangerine sdr project but I don't think ever came to fruition.

Christoph, I agree with you about the advantages of moving to a newer FPGA/CPU combination. It has just been in recent years that the zynq has come down to an interesting price point in my opinion. I use the zynq some at my day job. In fact, I am currently evaluating Xilinx's new versal technology. This continues to move in the direction of integrating more specialized hardware inside a FPGA. There are dual A72 cores, hardened memory and PCIe blocks, a hardened NOC, etc. This particular unit has 1968 DSP blocks! Too bad the card currently costs more than $10000. Maybe in 10 years this will be a common inexpensive hobby part...

Roger, I am also surprised that parts are hard to find but dev boards are often easier to find. I just purchased a Raspberry Pi pico, but there is no rp2040 chip stock listed on www.octopart.com. The same is true for the Teensy. You can find boards but not the processor. The i.MX RT series is a nice powerful inexpensive family of MCUs with USB 2.0, and I'm interested in using one of the smaller family members like the RT1010 for USB 2.0 interfacing instead of a FT2232H part. The same goes for the FT2232H, you can find boards but individual parts are rare and very expensive.

Esteban, I have a ColorLight 5A-75B and have run a few experiments on it. The gigabit ethernet design on it is an interesting example of a very inexpensive design. I like that the Lattice ECP5 flow is well supported by open source tools. The problems I have with this board are first that the part only has 28 18x18 multipliers. Compare this to 66 in the Cyclone IV we currently use and 80 in the Zynq 7010. Also, all the IO have a 74HC245 on them. The datasheet shows switching times in the tens of nanoseconds for this part. The AD9866 interface operates at 153.6MHz or 6.5ns. I worry that variation through the 74HC245 will be larger than this and timing closure will be very difficult.

Scott, the AFE7222 and AFE7225 have been on our radar for many years. They are definitely interesting candidates. One attraction of the AD9866 for me is the built-in LNA with -12dB to +48dB range. The AFE722* parts don't have a builtin LNA. One would have to add a gain block, maybe two LT6402-20 devices. But these are fixed gain, so one might want to do what is done on the original hermes and add an i2c (or other control means) attenuator. Once you add these additional parts, you have higher parts cost, more board area, more pins to solder, etc. (Maybe there is an easy and inexpensive way to do this that I am missing, which I'd love to hear about.) If we consider quantities of 100, the AFE7222 is in the $27 to $29 range. Add another $10 to $15 for two LNAs and step attenuators to bring you into the $37 to $44 range. Compare this to $45 for two AD9866s or $25 for two AD9865s with only two chips (less area) on the board, plus the ability to reduce costs by having only one ADC/DAC. We know two of these can be linked for diversity. Given this reasoning, the AFE7222 is interesting but has never been compelling to me.

One family of parts which has always intrigued me is the AFE580* series of ultrasound AFEs. In particular, take a look at the AFE5803. This contains 8 14-bit 65MSPS ADCs with LNA for $65 in lots of 100 from TI. Imagine what might be possible with 8 coherent HF receivers! In reality it would be hard to have 8 decent HF antennas and instead one might reserve 3 or 4 inputs for HF and use the others with mixers for VHF and UHF. This would be different from existing SDRs as you could concurrently receive over a wide range of frequencies, instead of the typical slice of 10-60MHz at a single point in the spectrum. You could design and optimize some ADC frontends to perform well for specific bands. For instance maybe one ADC has a frontend and antenna designed for 630M and 2200M. Another for 160M-40M. Two others for diversity on 30M-10M. One for 4M and 6M. One for 2M. And the last two for 70cm and 33cm. Monitoring of all these bands could happen concurrently. 

This project would require a Xilinx FPGA board with length matched LVDS. At 14 bits and 61.44MHz, a LVDS pair would be operating at 860Mbs. Most other inexpensive FPGAs don't go above 800Mbs for standard LVDS. I think the $90 Zynq 7020 bajie board would work. You would want the 85K LUTs and 220 DSP slices in the larger device given the 8 ADCs. The AFE5803 has a builtin LPF with highest cutoff of 30MHz, so (I think) one could only use a mixer for higher frequencies and not under sample. There is a very wide selection of mixers out there such as the lt5560 or the one used in the ubiquitous RTL dongle. There are also many interesting variable LOs such as the max2870. But VHF/UHF frontends are not my expertise and I would have to rely on others in this group for a good design. I would probably make the AFE5803 board have generic inputs to each of the ADCs paired with an i2c or spi bus. Then one could add a small custom mixer board. Once you add some mixer boards, a final price would be $200 - $300 for this receiver.

The AFE5803 has some interesting active termination options for the ADC inputs. One could possibly use a very minimal frontend. There are some noise trade offs with the various termination options. Again, this is an area I could benefit from others expertise. 

There is also the less expensive, 12-bit, and more DIY solderable AFE5801. But this has a LPF with highest cutoff of 15MHz. The LPF doesn't appear to be bypassable although there is an undefined setting in the datasheet. 

Please remember that this is only brain storming. I'm not committing to anything yet. There are still a couple more ideas I want to share over the next few days.

73,

Steve

kf7o

[si...@sdr-radio.com]

Yes to VHF/UHF!

 

Just FWIW my ELAD DUO transceiver & Q5 Signal transverter handle a local with 112dB SNR. Signals can be very, very strong on VHF so ENOB will be most important. SHF and above you’re usually happy to just see a signal.

 

In tomorrow’s episode – egg sucking for beginners 😊 .

 

Simon Brown, G4ELI

https://www.sdr-radio.com

[Steve Haynal]

Hi Simon and Group,

The Q5 signal transverters look very nice, but are expensive relative to the HL2. I don't think anything I am envisioning would be that nice or complete. I am not a VHF/UHF expert and would have to rely on others for help here. I am thinking more about modularizing the ADC/DAC from the frontend so that alternate frontends can be swapped in, and then provide i2c/spi to the modular frontends to facilitate mixer use if needed. And this is all at very low power, no thoughts of an amp.

There are also quite a few SDR radios in the VHF/UHF space. LimeSDR, HackRF, Adalm-Pluto just to name a few. If you search for SDR on aliexpress, ebay, or amazon, you get quite a few hits. For example, how good is this SDR, and does it make sense to buy 8 of these instead of making an 8 receive-only SDR with the AFE5803?

This is just a DIY hobby group with no intent to compete with commercial gear. I think there will only be interest if a niche is filled. Historically the Hermes-Lite has been a low cost decent DDC/DUC HF QRP transceiver. Maybe that is what should be built upon.

Another interesting new AFE I ran across recently is the AFE5401-Q 12-bit 4-input 25Msps ADC for automotive radar. I'm not sure if there is a good application. Perhaps two HF/6M receivers using a Tayloe mixer? Or a polyphase mixer?

https://www.pa3fwm.nl/technotes/tn17c.html

73,

Steve

kf7o

[Vince Vielhaber]

We'll find out soon enough (prolly January). It looked interesting and
cheap so I ordered one. They also have one that goes to 2GHz with only
one antenna port.

Vince.

On 11/24/2021 12:43 AM, Steve Haynal wrote:
> For example, how good is this SDR

> <https://www.aliexpress.com/item/1005002925249933.html>, and does it

> make sense to buy 8 of these instead of making an 8 receive-only SDR
> with the AFE5803?
>

[Robert Nickels]

On 11/24/2021 12:05 PM, Vince Vielhaber wrote:
>
> We'll find out soon enough (prolly January).  It looked interesting
> and cheap so I ordered one.   They also have one that goes to 2GHz
> with only one antenna port.
>

Most are probably aware that the makers of SDRPlay receivers are the
same folks who designed and produced the Mirics RF chipset used in all
the RSPs.   This guy BG7YZF found some in the China distribution channel
and designed a cost-reduced SDR receiver that since it uses the same
chipset naturally runs with their software.     Had he refrained from
printing their name and logo on the PC board what he did would have been
legal, but he's obviously not concerned since he puts his callsign on
everything.

The board works ok, he eliminated the RF switches so you have to attach
the antenna to the proper input connector.   Like anything built from
surplus parts, future availability of ICs is unknown.

73, Bob W9RAN

[si...@sdr-radio.com]

Also,

A big downside of the Mirics chips is they have massive gain, much of the support involves stopping them being overloaded.

ADC / DAC is the way to go.

Simon Brown, G4ELI
https://www.sdr-radio.com

-----Original Message-----
From: herme...@googlegroups.com <herme...@googlegroups.com> On Behalf Of Robert Nickels
Sent: 24 November 2021 18:18
To: Vince Vielhaber <v...@michvhf.com>; herme...@googlegroups.com
Subject: Re: Thoughts on Future SDR Projects (was Re: OpenHPSDR protocol)

--
You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/ab81176e-6df0-24de-6422-74421f19524d%40gmail.com.

[Steve Haynal]

Hi Group,

I've moved most of the future SDR ideas to a wiki page:

https://github.com/softerhardware/Hermes-Lite2/wiki/Future-Projects

So far we've covered ideas which I lump under the "Hazelnut SDR." There are still some ideas to post about a Hermes-Lite 2.1 and a Hermes-Nano.

Here are a couple more interesting SDR projects:

https://www.crowdsupply.com/search?q=SDR

https://www.crowdsupply.com/krakenrf/krakensdr

https://www.crowdsupply.com/cariboulabs/cariboulite-rpi-hat

73,

Steve

kf7o

[si...@sdr-radio.com]

Hi Steve,

 

I probably have more SDR than anyone else, trust me when I say that Lime / Hack / Pluto don’t come remotely close to HL2 in either RX or TX.

 

Yes, Q5 is expensive but very good and there aren’t a great number of HF SDRs which can drive it – ELAD DUO, ANAN 10 come to mind. A lower cost SDR like HL2 is ideal here.

 

I agree totally about a modular front end with ADC / DAC, some interesting chips are emerging from China, slow boats permitting.

 

Simon Brown, G4ELI

https://www.sdr-radio.com

 

[Alan Hopper]

Hi Steve and all,

I've been going around in circles in my plans/dreams for a vhf radio(s). Sometimes I wonder if just building a frontend and amp for the lime etc. is simplest but I have never been convinced  with the rx performance and very much prefer network connected radios.  

I also go round in circles between very simple individual radio per band vs more complex multiband design.

I quite like the idea of a simple radio per band but this rather depends on whether you have an antenna per band, I guess it is quite common to share a single white stick for the vhf bands. I want to be able to receive on all bands at once and  assuming an adc fast enough to directly sample all at once is unaffordable some effort is needed to share an antenna between multiple adc inputs whether it is a single radio with multiple adcs or multiple radios.

I've been pondering a multiband pa, one thought was to build an amp per band but share the heat sink and psu as only one band will be used at a time, the idea was to avoid the complexity of band switching, again this makes most sense with an antenna per band.

The AFE7222 looks interesting and could do two bands,  I had been looking at some slower adcs with higher bit count for undersampling, I can't find my notes of part numbers at the moment.  The balance of processing gain vs raw bit depth is interesting.  adc linearization is possibly easier with faster low bit depth(smaller lut).

73 Alan M0NNB

[Art]

Hi Steve.

These are very interesting ideas for further development. To use AD9865 instead of AD9866 is it enough to change the gateway? I look at the datasheet for both chips, and they differ only in 11 and 12 pins. Is it enough to change the data bus in the gateway ?

input [4:0] ad9866_rx,

output [4:0] ad9866_tx,

четверг, 25 ноября 2021 г. в 09:41:16 UTC+2, ahop...@googlemail.com:

[Steve Haynal]

Hi Art,

It is a little more complicated than that. The AD9865 uses all 6 bits by default but in a different pattern than the AD9866. You have to enable the 5/5 nibble bits for TX and RX as described on page 25 of the datasheet:

https://www.analog.com/media/en/technical-documentation/data-sheets/AD9865.pdf#page=25

You'll also have to update the Verilog for packing/unpacking the 5 bits into a 10 bit word as it currently assumes 6 bits into a 12 bit word. You can then pad the 10 bit word to make it a 12 bit word and use the same DSP path, but would probably want to optimize and save resources in the future by making it a true 10 bit DSP path.

73,

Steve

kf7o

[Steve Haynal]

Hi Group,

I have added the "Hermes-Lite 2.1" ideas to the wiki page:

https://github.com/softerhardware/Hermes-Lite2/wiki/Future-Projects#hermes-lite-21

Please read over these and feel free to comment on this thread.

The Hermes-Nano ideas should be posted in a day or two. After that, I'd like to run some sort of survey, perhaps a monkey survey, to gauge interest (not a vote for what happens) regarding these various ideas.

73,

Steve

kf7o

[Art]

Hi Steve.

I will say a few words on the topic of HL2.1:

-DC-DC ST1S10 are quite available on Aliexpress. And maybe you need to leave them. But I found a HX1314G synchronous converter chips that operates at 130 kHz and does not give any interference, even if the BPF/LPF are on the same board. I don’t know if the DC-DC regulators need to be changed. If they are tested enough, then they can be use. I use HX1314G, since they do not need to adjust the frequency of the PWM signal, they work with standard inductances of 7x7mm and produce a current of up to 1.5 A.

-would like some alternative for the slow ADC MAX11613 and digital resistor MCP4662. You cannot buy them in China. As it turned out, these are the most problematic parts for ordering. It is possible to simply support other chips in the gateway. I have not yet considered this issue in detail. I know that there are available SPI ADCs, I need to look with the I2C bus.

пятница, 26 ноября 2021 г. в 22:13:10 UTC+2, softerh...@gmail.com:

[Steve Haynal]

Hi Art,

To build 100 units or more at a time, I can't rely on AliExpress for individual parts. There has been numerous discussion on this list about the quality of individual parts from AliExpress. They are often seconds, pulls, mishandled or have water damage. Makerfabs is saying the ST1S10 is difficult to find from reputable suppliers so that is why I am considering other options.

Also, when we build 100 units or more at a time, they are built in China, so you can buy the ADC MAX11613 and MCP4662 in China. Makerfabs has not said either of these are difficult to source. It does not make sense to me to find alternatives for these if it is only for one or two DIY builders building one or two units.

73,

Steve

kf7o

[Art]

Okay, Steve. I understood. I think that who needs to - he put the right one. Moreover, it is not so difficult.

I studied datasheets for RTL8211 chips, and it is better to use RTL8211E, it has a 1.05 V power bus current almost half as much (about 200 mA instead of 400 mA). Also, just in case, I found an affordable linear regulator if the built-in DC-DC regulator is noisy. This is a series of APL5912, 5913, 5914, 5915, etc. RTL8211E chip has been tested and has all the documentation for it, including the wiring diagram from the manufacturer. Of course, this is not the most important thing. If you are confident in the B50612D chip from Broadcom, then you can install it. But for some reason there is little information about him. And even on the manufacturer's website, they don’t know him. Perhaps this is a Chinese replica.

пятница, 26 ноября 2021 г. в 22:51:51 UTC+2, softerh...@gmail.com:

[Steve Haynal]

Hi Art,

For information on the B50612D see the B50610 datasheet as described here:

https://github.com/q3k/chubby75/issues/16

If you really want to replace the MAX11613 and MCP4662, I would consider an inexpensive MCU. Most microcontrollers have at least 4 slow ADC inputs. The resistor network is just setting a low current Vbias level. This could probably be done with a microcontroller PWM or DAC output plus a couple of transistors. Once a MCU is on the board, you have options for UART and other IO like Bill is pursuing with the RP2040. You can always substitute a manual POT for the bias resistors and set the bias by hand.

Regarding the RTL8211E, I don't like the idea of additional voltage regulators. Frankly, I think there are too many in the current HL2 design. You have to also consider the price of the RTL8211E + Voltage regulator + extra PCB space + extra soldering fees per pin + your time for this change.

[Art]

Hi Steve.

Thanks for the information. I saw a datasheet for the B50610D chip. I think it would be nice to use it. Moreover, this chip is now available and inexpensive.

Also noteworthy is the option with MCU. In addition to the above functions, some other functions can be assigned to the MCU, for example, additional Alex/ Apollo signals, etc.

суббота, 27 ноября 2021 г. в 00:09:35 UTC+2, softerh...@gmail.com:

[Alan Hopper]

Hi all,

just found some of my undersampling notes where I optimised the sample rate to fit 6m,4m and 2m to give the biggest gaps between the bands:-

clock 56.832Mhz min gap 7.104

50 52 -> 4.448 6.448

70 70.5 -> 13.552 14.052

144 148 -> 21.344 25.344

so a AFE7222 could do 3 bands simultaneously with one adc leaving the other one free for hf or diversity.  It it realistic to make bandpass filters to do this?

73 Alan M0NNB

[Steve Haynal]

Hi Alan,

Could you clarify 3 bands simultaneously? I thought with under sampling you had to have a bandpass filter for each Nyquist zone of interest, but maybe I am missing a trick. I would think you need a switchable bandpass filter allowing for one Nyquist zone at a time on one ADC, and the other could be used for HF. Are you considering transmit too?

73,

Steve

kf7o

[Alan Hopper]

Hi Steve,

I may be missing something, but my thought was as long as you have a bandpass filter per band and they don't alias to overlapping frequencies you can cover more than one nyquist zone at once.  I've not really thought much about tx.

73 Alan M0NNB

[Steve Haynal]

Hi Alan,

I get it now. Very clever. So the main challenge is building a filter (or set of filters) with three passbands (50-52MHz, 70-70.5MHz and 144-144MHz) that reduce down to feed a single ADC channel all at the same time. 

73,

Steve

kf7o

[Alan Hopper]

Hi Steve,

yep that the idea.  I've now found the sw I used to pick the frequency, here are other options, the later ones might allow hf to 10m.

clock 44.544Mhz min gap 2.300

50 52 5.840 7.840 flipped False

70 70.5 17.820 18.320 flipped True

144 148 11.520 15.520 flipped False

clock 44.928Mhz min gap 2.912

50 52 5.456 7.456 flipped False

70 70.5 18.588 19.088 flipped True

144 148 10.368 14.368 flipped False

clock 45.312Mhz min gap 2.144

50 52 5.072 7.072 flipped False

70 70.5 19.356 19.856 flipped True

144 148 9.216 13.216 flipped False

clock 45.696Mhz min gap 1.376

50 52 4.688 6.688 flipped False

70 70.5 20.124 20.624 flipped True

144 148 8.064 12.064 flipped False

clock 52.608Mhz min gap 5.104

50 52 0.224 2.224 flipped True

70 70.5 17.776 18.276 flipped False

144 148 8.672 12.672 flipped True

clock 52.992Mhz min gap 3.568

50 52 0.608 2.608 flipped True

70 70.5 17.392 17.892 flipped False

144 148 9.824 13.824 flipped True

clock 53.376Mhz min gap 2.032

50 52 0.992 2.992 flipped True

70 70.5 17.008 17.508 flipped False

144 148 10.976 14.976 flipped True

clock 55.296Mhz min gap 1.148

50 52 2.912 4.912 flipped True

70 70.5 15.088 15.588 flipped False

144 148 16.736 20.736 flipped True

clock 55.680Mhz min gap 2.684

50 52 3.296 5.296 flipped True

70 70.5 14.704 15.204 flipped False

144 148 17.888 21.888 flipped True

clock 56.064Mhz min gap 4.220

50 52 3.680 5.680 flipped True

70 70.5 14.320 14.820 flipped False

144 148 19.040 23.040 flipped True

clock 56.448Mhz min gap 5.756

50 52 4.064 6.064 flipped True

70 70.5 13.936 14.436 flipped False

144 148 20.192 24.192 flipped True

clock 56.832Mhz min gap 7.104

50 52 4.448 6.448 flipped True

70 70.5 13.552 14.052 flipped False

144 148 21.344 25.344 flipped True

clock 57.216Mhz min gap 6.336

50 52 4.832 6.832 flipped True

70 70.5 13.168 13.668 flipped False

144 148 22.496 26.496 flipped True

clock 57.600Mhz min gap 5.568

50 52 5.216 7.216 flipped True

70 70.5 12.784 13.284 flipped False

144 148 23.648 27.648 flipped True

clock 62.208Mhz min gap 1.148

50 52 9.824 11.824 flipped True

70 70.5 8.176 8.676 flipped False

144 148 20.352 24.352 flipped False

clock 62.592Mhz min gap 1.916

50 52 10.208 12.208 flipped True

70 70.5 7.792 8.292 flipped False

144 148 19.584 23.584 flipped False

clock 62.976Mhz min gap 2.684

50 52 10.592 12.592 flipped True

70 70.5 7.408 7.908 flipped False

144 148 18.816 22.816 flipped False

clock 63.360Mhz min gap 3.452

50 52 10.976 12.976 flipped True

70 70.5 7.024 7.524 flipped False

144 148 18.048 22.048 flipped False

clock 63.744Mhz min gap 3.920

50 52 11.360 13.360 flipped True

70 70.5 6.640 7.140 flipped False

144 148 17.280 21.280 flipped False

clock 64.128Mhz min gap 2.768

50 52 11.744 13.744 flipped True

70 70.5 6.256 6.756 flipped False

144 148 16.512 20.512 flipped False

clock 64.512Mhz min gap 1.616

50 52 12.128 14.128 flipped True

70 70.5 5.872 6.372 flipped False

144 148 15.744 19.744 flipped False

73 Alan M0NNB

[Michael Durkin]

Bandpass filters for vhf/uhf can be had from hobbypc... 20 us each and im sure it costs less to build them .

--

You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/f8acd139-9e0d-430d-9f30-8fd08c92899en%40googlegroups.com.

[Steve Haynal]

Hi Group,

I have added the "Hermes-Nano" ideas to the wiki page:

https://github.com/softerhardware/Hermes-Lite2/wiki/Future-Projects#hermes-nano

Although there hasn't been much comment or discussion so far, it has been helpful for me to write up notes, even if they are very rough. I will organize an online survey for feedback, partly because I want to figure out how to organize a survey. Again, the survey is just to gauge interest in the various ideas and is not a commitment do implement any idea. Please add your comments or ideas to this thread.

I have been enjoying this time with lighter Hermes-Lite 2.0 support. There is support that happens on this list, in private e-mail, and with Makerfabs as they build units. This chip shortage may be a great opportunity for me to get away from the Hermes-Lite 2.0 and start working on some fun projects which I really enjoy, without any pressure to produce them or fix user problems. A very possible outcome of all this thinking and discussion may be for me just to go off on my own for a while and work on some ideas. Maybe you won't hear anything from me for months, except for the CW Keyer project!

73,

Steve

kf7o

[Art]

Hello everyone.

I found an example of including the B50612 chip. I would like to discuss a couple of questions:

- In HL2 and some other circuits, the digital core is powered from a separate source, in our case from 2.5 V. In the circuits of Hermes, Angelia in the OpenHPSDR project, this power is not shared. And 3.3V is used to power the analog core and digital interface. Are there real benefits to sharing power? Or, in practice, you can use one power supply (eg 3.3V)?

-In addition to programming the required registers in the mdio.v file, are there any other changes to the gateway?

воскресенье, 28 ноября 2021 г. в 05:20:57 UTC+2, softerh...@gmail.com:

[Steve Haynal]

Hi Art,

Thanks for the schematic link.

In the HL2 the KSZ9031 digital core is powered at 1.2V. See the DVDDL and AVDDL power pins on the HL2 schematic. The digital IO banks (power source for external IO signaling level) is set to 2.5V in the HL2. There are two benefits to this. First, power consumption and heat is reduced. See the KSZ9031RN datasheet where it lists required current when operating with 3.3V IO and with 2.5V IO. Second, it is easier to deal with high speed signal integrity issues with lower IO voltages. The RGMII interfaces operates at 250MHz which is pretty fast for the type of PCB used for the HL2. There are probably no signal integrity issues with 3.3V IO, but you are in a better place if you can use lower voltage IO. The Cyclone IV FPGA used in the HL2 requires 2.5V for powering the PLL. I did not add an extra 2.5V just for ethernet. I took advantage of the existing 2.5V power supply for the FPGA and used it to gain advantages for the ethernet.

The MDIO interface on ethernet PHYs have a standard subset and then often some manufacturer extensions. The KSZ9031 does a better job of adhering to the standard subset than the KSZ9021, but if I remember correctly, we are using some manufacturer-specific settings for both. These manufacturer-specific settings will have to be disabled or updated for the B50612. One would have to closely compare the MDIO register map of the B50610 with the KSZ9031 to know for certain. Unlike the ethernet MAC IP we use, I think most generic ethernet MAC IPs try to use only the standard subset so are compatible with a wider range of ethernet PHYs.  

73,

Steve

kf7o

[Steve Haynal]

Hi Art,

Even though the B50612D is common on AliExpress, I'm not sure it can be found with reputable vendors in China. It appears to be a part only used in large volume products, and not available for general use. The AliExpress sellers are probably selling excess stock. One way to check for general part availability in China is to see if it is on www.lcsc.com.

73,

Steve

kf7o

[Alan Hopper]

Hi,

I think we would need sharper filters than the hobbypc ones (but without the power handing), they need to cut out of band signals so they do not alias with other bands or with themselves.  The best case for the 6,4,2m setup gave 7Mhz between bands so there has to be significant attenuation beyond 7MHz, steeper filters allow more flexibility in selecting clock speed and might allow more bands. Even with just one undersampled band you need to bandpass to the samplerate/2 ~32Mhz.  Any filter experts out there, what is reasonably achievable?

73 Alan M0NNB

[Steve Haynal]

Hi Group,

I have posted the survey to gauge interest in these project ideas:

https://forms.gle/HsxKB4aWdpTuRFxP7

Please read over the project ideas:

https://github.com/softerhardware/Hermes-Lite2/wiki/Future-Projects

and take the survey. You can share the survey and wiki links with others who might be interested.

I will keep the survey open for 1 week.

73,

Steve

kf7o

[James Ahlstrom]

Hello Alan and Group,

Consider the clock frequency 56.832 MHz and the 2 meter band 144 to 148 MHz. Samples are periodic at the clock frequency, so the 144 to 148 MHz band appears at 26.496 to 22.496 MHz. Unfortunately the frequencies from the adjacent Nyquist zone from 136.2 to 140.1 map into this same range. So we need a filter that passes 144 and up but rejects 140 and down. That is not possible. The Nyquist zones must be separated effectively.

Jim

N2ADR

[Alan Hopper]

Hi Jim,

thanks for the feedback on this, my question is what filtering is practically possible, we can then find out how many bands and what sample rate and speed of adc can do anything useful.

73 Alan M0NNB

[James Ahlstrom]

Hello Alan and Group,

Filtering for VHF and up transverters is a problem. For the 2 meter band and a superhet architecture the first filter is usually a weak one or two pole coupled resonator. It can't be too selective because losses add to the noise figure. So maybe you get 30 dB at half the band center. Next is a bullet-proof amp and a sharper filter. Then the mixer. The filters must reject the image. For a 116 MHz mixer the image is at 86 MHz. So we must accept 146 MHz and reject 86 MHz. That is OK for an analog filter.

To get a feel for LC filters just get filter software and play. Sharper filters mean higher losses. Losses are unacceptable at the antenna, but lack of filters there means you need robust amplifiers and mixers to survive the large unwanted signals.  Coupled LC filters can have bandwidths of 10%, but the increase in attenuation away from the center is slow and just focusing on bandwidth is misleading. Also, filters must be tuned, and that is hard without a spectrum analyzer. There are alternatives to LC filters. SAW filters are available, but not for amateur frequencies AFAIK. Crystal filters are too narrow. Helical resonators are nice, but too large below at least 400 MHz.

I haven't thought too much about under sampling with the HL2. Under sampling increases the noise level because noise from a wide bandwidth folds into the desired Nyquist zone. And the clock jitter becomes more important. Also the noise figure of a typical ADC is 30 dB. This requires on the order of 30 dB of gain to get a low noise figure. Fortunately the HL2 has a low noise preamp. But the gain reduces dynamic range.

I wish I had a better answer. I have thought about transverters, but not about under sampling, and have no magic bullet. A good start would be to find SAW filters for the amateur bands. I haven't found any yet.

Jim

N2ADR

[Alan Hopper]

Hi Jim and group,

firstly there was a bug in my sw so the frequencies in previous posts are all 384Khz too high.  I believe the funcube dongle pro+ has a 2m saw filter and I did find this  Low Insertion Loss Filter 144.5 MHz (microsaw.fi) so they do appear to exist.

Just food for thought:-

if we are less greedy and go for just 2m

clock 64.896Mhz min gap 28.416

144 148 14.208 18.208 flipped False

which gives us 28.416MHz before any aliasing

if we go for the slightly more expensive  AFE7225 with max 130MHz adc and 6m and 2m

clock 129.792Mhz min gap 25.792

50 52 50.000 52.000 flipped False

144 148 14.208 18.208 flipped False

It is a shame that 2m ends up in the middle of the hf band 

There is a useful guide in the datasheet as to the required clock jitter for given input frequency.  I'm really just trying to get a feeling for the relative merits of undersampling vs using a mixer. 

The idea of handling multiple bands with one adc is not restricted to undersampling, maybe multiple mixers could also be combined.

73 Alan M0NNB

[James Ahlstrom]

Hello Alan and Group,

To get an idea of what can be done with LC filters I designed a four pole coupled resonator filter with a center frequency of 146 MHz and an 8 MHz bandwidth. The Q of inductors was 100, capacitors 1000. The loss in-band is 6 dB, and adds directly to the noise figure. The 60 dB attenuation points are 132 and 165 MHz. Tuning this filter will be difficult, and it must be carefully constructed to approach these numbers.

Regarding SAW filters, we see that they can certainly be constructed for amateur bands, but the market may not be big enough to justify marketing them.

Jim

N2ADR

[Alan Hopper]

Hi Jim and Group,

thanks, that helps a lot in understanding the trade offs.

For a brute force vhf solution there is https://www.ti.com/lit/ds/symlink/ads5403.pdf which is still just about affordable, not sure if it would push the fpga into a new price bracket though.

73 Alan M0NNB

[Matthew]

All,

I have been following this discussion with interest.

From a practical point of view, I agree with Jim's comments regarding filters. When I was building/designing my transverter for my HL2 a few months ago I learnt a lot about VHF filter design. The numbers Jim quotes sound realistic. However, if I recall, the Coilcraft variable inductors only had a Q of around 60. Cap values are typically in the region of < 1 pF to 5 pF for filters at 2m. I'm not sure if the ones I bought had a Q as high as 1000. You would probably need to be very specific about the exact cap part number to ensure consistent Q of the filter in production batches. I have seen on a bench in the past that high Q filters can be susceptible to temperature affecting the tuning.

I get a feeling that VHF/UHF users demand high performance from their equipment. With this comes cost and complexity. I imagine trade-off must be made to balance performance and cost. Does this community have VHF/UHF design experts willing to contribute? I like the idea of "offloading" this onto a separate card to allow options for different user needs. This also moves the design risk to a separate board to allow for RF screens etc. I like the idea of a low cost SDR starting a renaissance in 2m (non-FM) activity and generating something to listen to on the band. However, the Pluto SDR doesn't seem to have initiated this.

Each of the projects described look interesting. I agree with comments about the practicality of diversity antennas with 8 ADCs. However, I feel that for the modern ham in an urban environment 2 ADCs to allow noise cancellation should very much be on the agenda.

I am interested in the comment around RF3 for HL2.1 and PureSignal. I use this for a dedicated RX antenna input, it would be a shame if we lost this through the use of RF3 more dedicated to PureSignal. I have often pondered whether the TX/RX switching relay should be on the N2ADR board and if the bare HL2 board should be dedicated RX and TX inputs/outputs. Then the N2ADR board could support a dedicated RX antenna connection. It looked at the practicality of this a while ago and I think it is possible with a small SMT relay. I have never got round to trying to re-route the PCB.

I first saw Ultrasound ICs around 12 years ago and thought they would make a great ham radio project. This sounds really fun to experiment with.

Re: offloading processing to the PC. The HPSDR folks talked about this a few years ago and I think prototyped something? I'm not sure this made it into a product? My concern with RPi offloading is you become tied to specific hardware/methods. I once tried to teach myself CUDA, but didn't really progress far. If we look at when the HL2 was in the early years we had the RPi1. It has been through so much progression since. Designing something to couple closely with specific computing hardware seems like an obsolescence challenge waiting to spawn. A fun project nonetheless.

In amongst all this, the conversation has been very much focused on hardware. In gateware/software terms, it feels like there is a good level of support for the current protocol in various SDR software packages. Is it envisaged the protocol would remain as is? I have never found anything the succinctly details the advantage of protocol 2 for the HPSDR radios. I know in the past SoapySDR has been talked about as an interesting concept to support/interface with.

Steve, I look forward to reading about the outcome of your period of going silent!

73 Matthew M5EVT.

[Art]

Hello Group.

Since this topic is the most popular now, and there are many respected constructors here, I had one question.

I want to use OPA2674 (AD8012) + RF transistors in the SOT-89 package in one of the versions of my boards. Most likely there will be transistors like in HL2 or similar. I do not have a HL2 board and I cannot measure the level of harmonics at the output. At 5 W, their level should not exceed 50 mW. Which is equal to 20 dB of suppression.

Please tell me, did someone measure the level of harmonics without a low-pass filters? Most likely the 2nd harmonic will be suppressed well. Are the harmonics higher? It is difficult for me to buy an inductance smd, for example AISC-1210. Therefore, I would like to consider the possibility of operating a 5W output stage without a low-pass filters. Unless, of course, the level of harmonics will not exceed the permissible level.

понедельник, 29 ноября 2021 г. в 19:33:02 UTC+2, Matthew:

[Alan Hopper]

Hi Group,

just found this Nooelec - Nooelec SAWbird+ 2m Barebones - Premium Dual Ultra-Low Noise Amplifier (LNA) & SAW Filter Module for 2-Meter Amateur Radio Band Applications. 145MHz Center Frequency which might be any easy way to prototype an undersampling receiver.

73 Alan M0NNB

[James Ahlstrom]

I have been following the discussion of future projects with interest, but have been busy with other projects. I will do the survey, but here are my thoughts so far.

I don't like assembling multiple boards, and I like the all-in-one HL2 as it is. It is frustrating that we can buy boards for cheap but can't get chips, but boards come and go. I think I remember in the HL1 days that boards would go out of production when an "improved" version became available. And then we had to write new software to use the new board and figure out how to support multiple hardware versions.

Some of the ADCs mentioned have sampling rates of 65 Msps. That is inadequate for sampling ten meters. There has to be headroom so that anti aliasing filters can be effective. To sample a 30 MHz signal we need a clock of 60 MHz plus at least 10 or 20 MHz. And don't forget to use a clock that is a multiple of 48 kHz.

I really prefer the AD9866 no matter what Figure 12 says. Going to ten bits to save ten bucks doesn't work for me. Even if it is just window dressing, and I am not sure it is.

So that brings us to HL2.1 and I find that exciting. First I think I know that there are FPGAs with another core for an ARM CPU. And I think I know that we can program a CPU into an FPGA if there are enough resources. It would certainly be pleasant to program some or even most gateway logic in C instead of Verilog.

But we need to point to real improvements for the user:

On the hardware side, we currently have RF3 in parallel with RF2. This is awkward, and I would like to see a switch to select the RF input so that RF3 can be connected to another device. The switch would need to support Pure Signal, transverters and other applications. Always using RF2 for Rx input risks sending 5 watts to your transverter. I am ambivalent about adding another I2C interface or more IO for slow peripherals. That runs unrelated IO through the HL2. An alternative is to use a separate box that is connected to Ethernet and leave the HL2 out of it. In practice that is a 10 or 100 Meg Ethernet peripheral next to the HL2 and a small Ethernet switch to connect everything. Peripheral du jour anyone?

On the software side it would be nice if HL2.1 worked over WiFi. That requires a bigger Tx buffer. A different FPGA may have more memory for a bigger buffer. Alternatively we could start to support Tx sample rates of 8 or 12 ksps so that a smaller buffer becomes 6 or 4 times bigger. More logic would be required to interpolate the rate to 48 ksps and onward. For remote CW over WiFi, we could support a Tx sample rate of 1 ksps. One means key down and zero means key up. Envelope shaping is done in the HL2. The jitter is one millisecond. An alternative is time stamps, easily written in C. And users sometimes have network problems. I think there are features in IPv6 that can fix this. I worked on this once but dropped it. We don't need all of IPv6, only some of it.

Jim

N2ADR

[Mike Lewis]

Regarding the I2C through the HL2.  I do not see a need to add ‘more’ as we can add devices on the existing bus and use the existing mechanism.  I don’t think any proposed uses require great speed?   A useful addition could be an i2c line driver chip to make it easy to add things that may be outside the box and help isolate the internal bus for better reliability .

 

I would avoid using a separate ethernet connected box for simple things that the i2c can handle via the HL2.  The HL2 can, will be, and is, used remote and a separate control box on an ethernet requires an ethernet switch or another long cable.  These add more RFI exposure, more equipment, space and cost to house, and a small switch likely requires 5 or 9VDC so another odd power supply beyond the control circuit needs.   My use case it the HL2, transverters, amps and antenna switches located in a box in the back, attic, or roof of a house, truck, or RV.  Today all one needs is a single ethernet to the control point because the i2c and/or serial from the HL2 extend your shack.

 

Some jumpers to remove RF2 and relay would be nice.  For my split IF transverter use I tap RF1 an RF3 on the IO strip but still listen to the relays, I did not want to modify the HL2 board.

 

 

Mike

K7MDL  EL87sm & CN88sf

--

You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/39749b0c-89af-4551-81ea-d53b25c85595n%40googlegroups.com.

[James Ahlstrom]

Hello Mike and Group,

Regarding I2C through the HL2. I do see your point and I don't feel too strongly about this. I am not opposed to I2C via HL2 but I want to point out an alternative. If we have I2C in the HL2, and the interface, microcontroller (if any) and buffers are in the HL2 box, then I see that I2C via HL2 is convenient. The only problem is noise from the MCU.

But if we are talking about I2C bus extenders going outside the HL2 box, then I am not so sure. The bus extender must be going to a separate box with its own power supply and interfaces to other equipment. The noise from the MCU is isolated. We need a small switch for the Ethernet connections, but that is about USD 20.

Option 1:  Ethernet to HL2 to I2C bus extender to external box with MCU, serial port, band-to-voltage converter, etc.

Option 2:  Ethernet to                                                  external box with MCU, serial port, band-to-voltage converter, etc.

Jim

N2ADR

[Roger David Powers]

I think my thoughts are aligned with James and Matthew. 

I am somewhat concerned about too much hardware diversity and ending up without the kind of software app and library support we may be taking a bit fore-granted with HL2.  Perhaps more so with regard to Hazelnut which is by definition a modular SDR.  

Clearly HL2.1 would be backward compatible with HL2, not sure if we'd do that for Hazelnut or Nano.  

To be honest I haven't thought about what improvements would be needed in either an on-the-wire protocol or a programming API for a HL2 follow on.  As above I think HL2's use of HPSDR protocols made it so a lot of app and library support came without the same amount of effort needed if HL2 had rolled its own protocols/APIs.

I happened to be looking at Caribou SDR today which was mentioned in the Nano write up, and it is supporting the Soapy API.  

Regards,

RDP

[Graeme Jury]

The future HL project is complicated by its own versatility and so many people want to use it in so many different ways with the list below touching on some of them.

1. Use as a stand alone QRP radio for WSPR, SOTA etc.

2. A driver for any number of peripherals, linear, transverter etc.

3. A complete QRO radio contained in a single case.

4. A test instrument and experimental radio.

Each of these use cases places a different demand on the supporting PC consoles, firmware and hardware and puts a big demand in time on the people supporting this, as requests for this property or other rolls in. Mostly the developers want to service these requests as generally it means that their product will get better and better but sometimes it only adds value to the requester and has to be gently refused. The choice of using a (very close) subset of the Hermes protocol has proven to have been a huge asset with PC Console software immediately available and yet more was re-written to accomodate the Hermes-Lite/Hermes and it now has a sustainable user base and should be kept. There has been times where we have stolen function bits from Apollo or Alex filters or the like to support popular peripherals but nothing which destroys compatibility.

Where we go in the future will be driven by the final use of the HL and here is a look at what each of the four nominated cases might bring.

1. This fits with the current HL2 and will appeal to the advanced to not so technical user as it is complete and fairly easy to get going – load a Hermes compatible PC console onto your computer, plug the HL2 into the network, power and antenna and you are good to go. For the portable QRP operator the RaspberryPi and a 7” screen will have you hill topping in very little time. The radio of the future should work with current software and not change the current concept of radio and filter in one box with some external lines and data out/in. Different FPGA’s or LNA’s are “black box” and won’t noticeably alter the performance except perhaps more Rx for WSPR but aerials are usually the constraint there.

2. Here we are appealing to the technically competent to exceptionally talented and providing a base that will drive all the external devices that they love to build, ranging from a Hardrock kit to an aesoteric fully protected max limit amplifier with auto band changing and bells and whistles. These people will be looking for a performance HL which will support undersampling with a grunty FPGA and the best LNA – external attenuators if needs be. A wee bit canny with the cost but not an over riding factor as long as it is high performance and good value for the dollar and maintains backwards compatibility. We don’t want to lose the development investment in the HL2 or diverge off so far that there is the old HL2 group and the new ones who don’t quite match.

3. The dedicated home brewer will be looking at the new HL2 as a board to form the heart of their home brew transceiver. They will tend to the case, filters, amplifiers and control system building up a device that will be the envy of Yaesu Icom et al. The board needs to be a performance board and cost is a secondary consideration. The firmware and good PC console software is what counts. The operator will probably sit in front of this radio in a ham shack and will be able to plug in and turn knobs.

4. The owner of this HL version which is a good match to the “Hermes-Nano” will be looking at a cheap board which runs a decently performing radio, not necessarily in a case. It will need to produce +37 dBm on one output and +17 dBm (ish) on another. Two Rx inputs are required with one being able to be jumpered to an aerial changeover on the 5W output or jumpered to an sma for separate receive port and the other input through an isolating pad, dedicated to pure signal or similar use. Many users of this board will be avid experimenters tinkering with devices like Arduino or RP2040 MCU and a good buffered I2C as posted in here would be an advantage. Bringing as many outputs/inputs to connectors would be an advantage

The following is my personal wish-list. Some features that are already in HL2 are not mentioned but I would not like to see them go as I already use them.

1. ALC
   Facilities to accept an ALC signal from an amplifier in the range 0 to 3 volts (or suitable alternatative) and reduce the RF output proportionately. The amplifier control circuitry is responsible for setting the control voltage range and any threshold. Maybe an opto coupler would be appropriate as an input.

2. Feedthru control signals via I2C.
   When operating the HL remotely which many users do, it would be nice to be able to use the HL to send and receive control information to be able to do things like monitor the shack voltage/current, switch antennas, operate an antenna tuner and receive a “tune complete” signal as a sample of possibilities. The HL2 would be powered up all the time and able to respond to control commands. The control console should be separate to the PC console software driving the HL to avoid having several authors having to customize software to cope with the control signals which are really peripheral functions rather than radio functions. Currently I control and monitor with a separate application but the HL2 already has an ether connection to my remote site and this is a duplication.

3. Separate pure signal input
   A dedicated input presenting 50 ohms forming the shunt path with a 470 ohm series leg to the RX in as an L pad.

4. +17 dBm for transverters
   A separate sma connector perhaps switched on by the protocol 1 address 9, 8th bit but there may be better suggestions

5. 5 Watts (+37 dBm)
   Exactly as at present with interaction on the transverters +17 dBm so it is off when the other is on.

6. Onboard I2C signal level buffering
   The buffer here here should be on board to enable a robust extension of HL filter switching to high power amplifiers vi I2C with good RF immunity. If the remote control is introduced the same I2C bus as the filter switching probably could be used and only a single I2C line needed

7. Separate receiver input
   An sma connector for a separate RX input. This should be connected to a jumper so that the RX in can be connected to the changeover relay or directly to the sma to manage the cases where an amplifier etc has separate RX and TX connections and attends to the aerial switching - a very common scenario.

8. PTT leads the RF by 20 mSec
   Give time for antenna switching etc. to occur. There is much to consider here as break-in CW is affected and most of the PC console software has addressed this anyway but it needs consideration.

73, Graeme ZL2APV

[si...@sdr-radio.com]

Graeme,

 

Agreed 99%, especially support for a separate receiver input as this is essential for me and also for transverters.

 

PTT lead time should really be a software option I think, rather than hardware.

 

Simon Brown, G4ELI

https://www.sdr-radio.com

 

From: herme...@googlegroups.com <herme...@googlegroups.com> On Behalf Of Graeme Jury

…

[Graeme Jury]

Hello Simon,

Thanks for your comment with which on further reflection I agree. I was thinking of the case where the HL was keyed from its front panel. As you infer the PC Consoles in current use have had a lot of work and thought put into the keying and definitely won't put the system into a state where it is hot switching. If needs must it would be possible to delay the bias on the linear  so it is not producing RF until switching has occurred.

While I am on pie in the sky wishes, the perfect portable radio would be an HL plugged into a tablet either via wifi or USB3 which seems to be on most tablets nowadays. This might be a project on its own but would be a trail blazer in the field of portable SDR operation. Maybe Santa will come some Christmas soon but he had better get a move on coz I'm 80 next birthday.

73, Graeme ZL2APV

[Mike Lewis]

Why wait?  Just use a USB ethernet adapter on your tablet today.

 

From: herme...@googlegroups.com <herme...@googlegroups.com> On Behalf Of Graeme Jury

Sent: Friday, December 3, 2021 23:31
To: Hermes-Lite <herme...@googlegroups.com>
Subject: Re: Thoughts on Future SDR Projects (was Re: OpenHPSDR protocol)

 

Hello Simon,

--

You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/52b23c74-edff-453f-99f6-c258dfa1d782n%40googlegroups.com.

[si...@sdr-radio.com]

Yes,

 

I’ve often wondered why SDRs (and other stuff) don’t have a USB socket where you connect a Wi-Fi dongle, I’m going cable-free here. For TX Wi-Fi does have drawbacks, you must know what you are doing.

 

Also, ALC is bad idea IMO for solid-state stuff – best option is to turn the gain down on the exciter. If seen some shocking signals using ALC.

 

80 – that’s ancient? I’m only 64 (I think).

 

 

Simon Brown, G4ELI

https://www.sdr-radio.com

 

From: herme...@googlegroups.com <herme...@googlegroups.com> On Behalf Of Graeme Jury
Sent: 04 December 2021 07:31
To: Hermes-Lite <herme...@googlegroups.com>
Subject: Re: Thoughts on Future SDR Projects (was Re: OpenHPSDR protocol)

 

Hello Simon,

--

[Sid Boyce]

Hi Graeme,
May be not quite pie-in-the-sky. Looks like it has already been done.
https://uk.pcmag.com/networking/132701/how-to-connect-a-phone-or-tablet-to-the-internet-using-an-ethernet-cable
I have been considering it and found this
https://www.youtube.com/watch?v=kNkLT2ukUFQ
73 ... Sid.

On 04/12/2021 07:31, Graeme Jury wrote:
> Hello Simon,
>
> Thanks for your comment with which on further reflection I agree. I
> was thinking of the case where the HL was keyed from its front panel.
> As you infer the PC Consoles in current use have had a lot of work and
> thought put into the keying and definitely won't put the system into a
> state where it is hot switching. If needs must it would be possible to
> delay the bias on the linear  so it is not producing RF until
> switching has occurred.
>
> While I am on pie in the sky wishes, the perfect portable radio would
> be an HL plugged into a tablet either via wifi or USB3 which seems to
> be on most tablets nowadays. This might be a project on its own but
> would be a trail blazer in the field of portable SDR operation. Maybe
> Santa will come some Christmas soon but he had better get a move on
> coz I'm 80 next birthday.
>
> 73, Graeme ZL2APV
>
> On Saturday, December 4, 2021 at 7:25:33 PM UTC+13 si...@sdr-radio.com
> wrote:
>
> Graeme,
>
> Agreed 99%, especially support for a separate receiver input as
> this is essential for me and also for transverters.
>
> PTT lead time should really be a software option I think, rather
> than hardware.
>
> Simon Brown, G4ELI
>

> https://www.sdr-radio.com <https://www.sdr-radio.com/>
>
> *From:*herme...@googlegroups.com <herme...@googlegroups.com> *On
> Behalf Of *Graeme Jury
>
> …
>
> 1. ALC

> Facilities to accept an ALC signal from an amplifier in the
> range 0 to 3 volts (or suitable alternatative) and reduce the
> RF output proportionately. The amplifier control circuitry is
> responsible for setting the control voltage range and any
> threshold. Maybe an opto coupler would be appropriate as an input.

> 2. Feedthru control signals via I2C.

> When operating the HL remotely which many users do, it would
> be nice to be able to use the HL to send and receive control
> information to be able to do things like monitor the shack
> voltage/current, switch antennas, operate an antenna tuner and
> receive a “tune complete” signal as a sample of possibilities.
> The HL2 would be powered up all the time and able to respond
> to control commands. The control console should be separate to
> the PC console software driving the HL to avoid having several
> authors having to customize software to cope with the control
> signals which are really peripheral functions rather than
> radio functions. Currently I control and monitor with a
> separate application but the HL2 already has an ether
> connection to my remote site and this is a duplication.

> 3. Separate pure signal input

> A dedicated input presenting 50 ohms forming the shunt path
> with a 470 ohm series leg to the RX in as an L pad.

> 4. +17 dBm for transverters

> A separate sma connector perhaps switched on by the protocol 1
> address 9, 8th bit but there may be better suggestions

> 5. 5 Watts (+37 dBm)

> Exactly as at present with interaction on the transverters +17
> dBm so it is off when the other is on.

> 6. Onboard I2C signal level buffering
> The buffer here here
> <https://github.com/pressla/I2C-Bus-extender> should be on

> board to enable a robust extension of HL filter switching to
> high power amplifiers vi I2C with good RF immunity. If the
> remote control is introduced the same I2C bus as the filter
> switching probably could be used and only a single I2C line needed

> 7. Separate receiver input

> An sma connector for a separate RX input. This should be
> connected to a jumper so that the RX in can be connected to
> the changeover relay or directly to the sma to manage the
> cases where an amplifier etc has separate RX and TX
> connections and attends to the aerial switching - a very
> common scenario.

> 8. PTT leads the RF by 20 mSec

> Give time for antenna switching etc. to occur. There is much
> to consider here as break-in CW is affected and most of the PC
> console software has addressed this anyway but it needs
> consideration.
>
> 73, Graeme ZL2APV
>

> --
> You received this message because you are subscribed to the Google
> Groups "Hermes-Lite" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to hermes-lite...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/hermes-lite/52b23c74-edff-453f-99f6-c258dfa1d782n%40googlegroups.com

> <https://groups.google.com/d/msgid/hermes-lite/52b23c74-edff-453f-99f6-c258dfa1d782n%40googlegroups.com?utm_medium=email&utm_source=footer>.


--
Sid Boyce ... Hamradio License G3VBV, Licensed Private Pilot
Emeritus IBM/Amdahl Mainframes and Sun/Fujitsu Servers Tech Support
Senior Staff Specialist, Cricket Coach
Microsoft Windows Free Zone - Linux used for all Computing Tasks

[James Ahlstrom]

Hello Graeme,

Thanks for your very detailed and thoughtful response.

With regards to "2. Feedthru control signals via I2C." and "Currently I control and monitor with a separate application" could you please provide more detail about what you are doing now? How do you connect your separate application?

Jim

N2ADR

[ron.ni...@gmail.com]

You can use a table or mobile phone already, by connecting via either a USB to ethernet adapter, or for iOS devices, a Lightning to ethernet adapter.  I operate my HL2 for portable ops this way from my iPhone 12 Mini.

The SDR app I modified to operate an HL2 from an iPhone or iPad is in Apple's App store; but if you are on this list and have an HL2, please ask for a free copy.  I also have an (unfinished) FT8 app in beta test.

73,

Ron

n6ywu
http://www.hotpaw.com/rhn/hotpaw/

[Graeme Jury]

Hello Jim,

Here is the detail of what I am currently doing. I have 3 attachments with "Remote Switch.jpg" being the hardware sitting in my garage along with my HiQSDR and HL2. Basically it is an ESP01 connected to my local lan via wifi and passing its traffic via I2C to an arduino on the motherboard. The arduino decodes the commands sent from the slave software in my spare room and does the switching and echos the command back to my PC so it knows that it was successful. The rest of the box contains the hardware - switching relays PSU etc.

The file "Remote Monitor.png" is the circuit of my remote hardware switching box.

The file "remote.png" is a screenshot of the app that I wrote in Qt5 to control the remote hardware. I have shown it with the modal setup window opened where I can change the IP address and port and also the titles on the 5 buttons which control 5 relays in the remote plus the titles on the antenna switch which goes to a relay box where the antennas come into the garage so I can select the relay I wish to use. My antenna is a trap dipole for 80, 40, 30 plus a magnetic loop which covers 30, 20, 17,15. The plan is to map the antenna relays to my bandswitch so that when I change bands the appropriate antenna is selected. All my code is available on my  github although very poorly documented I'm sorry.

73, Graeme ZL2APV

[Graeme Jury]

Hello Ron,

Thanks for the information and your very kind offer. Some years ago I went over to the dark side and became Linux only with an Android phone. As far as I know there is only the ghpsdr3 app by Alex Lee or the HPSDR app by John Melton for Android. Both of these apps have not been maintained for some time and don't perform super reliably although I have had them going.

I was thinking that probably HL could operate directly with the USB3 rather than a USB to ethernet converter although this is a great suggestion to get going right now as also suggested by Mike and Sid and I will pop down and get on tomorrow. An SDR is a great device to track down noise and having it fully portable connected to a small loop with its sharp null is a great tool.

Thanks guys for all the comments and advice.
73, Graeme ZL2APV

--
You received this message because you are subscribed to a topic in the Google Groups "Hermes-Lite" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/hermes-lite/JTdxzmjO_zI/unsubscribe.
To unsubscribe from this group and all its topics, send an email to hermes-lite...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/373b3f48-5269-4b87-9af8-7f052741d5dbn%40googlegroups.com.

[James Ahlstrom]

Hello Graeme and Group,

Thank you again for the detailed response. I am not operating remotely, but my QTH also has relay boxes that I need to switch. For another point on the curve, here is my setup. I have a gateway box wired to my LAN. It accepts UDP traffic and transfers the data to one of three local buses, I2C, RS485 and my implementation of a one wire bus. My peripherals use one of these low-cost buses so I don't need Ethernet in every box. The I2C bus is useless because I didn't use a bus extender, and it fails in an RF environment. The other two work fine.

I use Quisk software with a special hardware file. All my software is in the n2adr subdirectory of Quisk. For example, when I change bands the hardware file is notified with ChangeBand(self, band). I send the new band to the gateway with UDP, the gateway sends the band out on RS485 or 1Wire, and the peripheral responds by switching a relay.

I am exploring why you and others want the HL2 to forward I2C. Your setup currently uses WiFi. It could also use the existing wired Ethernet connection and a small Ethernet switch to do the same thing. If you use the HL2 to forward I2C, then each software you use (Quisk, Spark, etc.) will need to send the I2C commands. In general, these would need to be custom for each QTH. Each SDR software would need to create a way to send arbitrary custom I2C commands. But there is the advantage that the band would be known. A separate program would get around the custom problem and require no software mods if it sends I2C on port 1025 instead of 1024. But why is this better than direct Ethernet?

I don't want to add an external I2C bus to the HL2 unless we have a complete hardware/software design, the design is superior to all other alternatives and the design satisfies all user requirements. IMHO we don't have a complete design yet.

Jim

N2ADR

[Mike Lewis]

I liked i2c because it was from the existing HL2 to reduce external cable count and is already supported in software today at least for the J16 pattern section mechanism.  Without SDR app modification or external programs, that limits the i2c usefulness to the 7 or 8 io ports.   

 

Being a user of several SDR app, piHPSDR on RPi mostly, running external apps would only be attractive if they had the means of keeping in sync with the SDR app and capable of being fully headless for seamless remote ops.  True CAT port data may be enough to do most things with some added messages as it does not today share data LO and IF frequencies.

 

Since this is future talk, and we accept changes will be needed to likely all SDR apps, then any better solution will work. It could be an internal i2c to 1 wire conversion inside the box for example.

 

The existing serial port in the HL2 (and related protocol extensions) could be revised to pass through the actual CAT port data from the SDR app rather than just the locally inferred band data as it does today.  Then you can add any useful data like LO, offsets, etc.  With the concept of addressed serial messages one could do shack in the box functions like rotator control.

 

Mike

 

[ron.ni...@gmail.com]

For home operation, I have colocated with the HL2, a switch, a WiFi router, a Raspberry Pi 4, and a whole bunch of network cables and wall wart power supplies.  The Raspberry Pi has plenty of IO to control other stuff.

But for portable ops, I would like to ditch the wall warts and network gear, and just use one ethernet cable from the HL2 to my iPad, yet still have the option of some kind of IO to control maybe a portable auto-tuner, or etc.  Ether an I2C or a serial port tunnel through the HL2 allows.  I write a lot of my own software, so as long as the protocol is documented (port 1025, etc.), I can use it.  Much preferred to needing even more stuff to have to pack or set up on some park bench.

73,

Ron

n6ywu

[Steve Haynal]

Hi Graeme,

I think this point:

1. PTT leads the RF by 20 mSec
   Give time for antenna switching etc. to occur. There is much to consider here as break-in CW is affected and most of the PC console software has addressed this anyway but it needs consideration.

is already covered. You set TX buffer latency to 20ms and then PTT will trigger 20msec before RF. This is true for both software or direct CW connection to the front panel. Is this not the case for you?

73,

Steve

kf7o