50MHz/6m as a future option?

[Max]

My first post in this group. The project looks very appealing and I am sorely tempted save for the omission of 6m capability (although even so, I am still very tempted!). I do understand that 6m has been omitted due to technical restrictions of the re-purposed lower cost FPGA, but just wondered how likely it would be for 6m to appear in any future incarnations of this project, or if the whole Hermes Lite project is wedded totally to this particular FPGA and hence unlikely?

I guess what I am asking as a novice in this area: Are there any other FPGAs now out there or on the horizon that might change this situation in, say, Hermes Lite 3, or is a "Hermes Lite 3" an unlikely project anyway?

Thank you for reading.

73

Max

[Marc OLANIE]

Hi Max

 

The “around 40 MHz” limitation of the HL2 is definitely not tied to any kind of fpga problem. It  depends of  the max sampling rate of the ADC. In this case, the AD9866

https://www.analog.com/en/products/ad9866.html

is a 12 bits 80 MSPS. If you do the math,  80:2=40… and no more (Shannon theorem)

 

If you wish to extend your BW to 54 MHz, you’ll need at least a 110 or 120 MSPS ADC…. And these are definitely more expensive.

 

The main advantage of the 9866 is that it integrates the ADC for RX, the DAC for TX, a low noise variable gain amplifier and the ability to run in full duplex mode. All of this for approximately 20 USD apiece.

 

A 12 or 14 bit 125MSPS ADC alone could cost around 100 USD, without driver (LNA) and without DAC… this is definitely not the same budget.

 

VY 73’

Marc f6itu

 

PS : you could have a look at the Odyssey II or the Red Pitaya boards. These are able to cover from 100 kHz to 54 MHz without troubles, and are compatible with the OpenHPSDR environment, thanks to the people who work on the gateware. But as I already mentioned, theses kind of board using bigger ADC are far more expensive than the HL2

 

 

 

De : herme...@googlegroups.com <herme...@googlegroups.com> De la part de Max
Envoyé : dimanche 1 décembre 2019 11:14
À : Hermes-Lite <herme...@googlegroups.com>
Objet : 50MHz/6m as a future option?

--
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/7a709d16-7fb2-4d92-818f-524f3447cbed%40googlegroups.com.

[Ulf Tjerneld]

Hi Max,

I had the same concerns before ordering my first HL2 but I concluded that if I wanted to have fun today at a low cost, then I have to live without 6M. Or I could spend more on a Red Pitaya. My recommendation is to go with the HL2, because it is a very simple setup (unless you scratch build it) and the cost is so low that you can later jump onto a Red Pitaya or some other project that will appear. Prices are going down all them time. I would easily say: go with the HL2, have tons of fun, and then go from there 😊

/Ulf

SM0NOR

 

Sent from Mail for Windows 10

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/DM5PR0102MB3365B98A031B47065454AD90E3400%40DM5PR0102MB3365.prod.exchangelabs.com.

 

[Don Solberg]

After a lot of Internet research, the HL2 was by far the best performing direct sampling QRP SDR that I could find.  I purchased a build 8 unit and have done a lot with it.  I have used it as a QRP HF transceiver, driving my home brew solid state amp.  I use PowerSDR and have found that this combination can make a very high quality desk top station.  My primary rig is an Elecraft K3 with an LP-PAN SDR and NAP3 software for the panadapter, so I have a pretty good station to compare the HL2 to.

I have used the HL2 as a second receiver, interfaced to my K3 for CAT control, and using an MFJ1708B SDR switch.  The HL2 gave me everything that I wanted in a new receiver: a really good panadapter display, and 3 receivers, with one of those being on another band.

I am now experimenting with Pure Signal and am seeing great results with my solid state amplifier.

I liked the HL2 so much that I ordered a second one.  This way I can dedicate one for the K3 panadapter, and the other as the exciter for my new LDMOS amplifier.

With all that said, are there additional features that I would like to see in the future, sure:

• 15 watt output, for additional drive to my amplifiers
• 6 Meters
• 14 or 16 bit ADC
• Dual ADC's
• Preamp for high frequencies

I really wish there was a single board SDR XCVR that filled the gap between the HL2 and an Anan 7000, in the $1K price range, but I haven't found anything better than the HL2.

We all owe a great deal of thanks to Steve and the others who have brought the HL2 to this point.

73,

Don

K9AQ

To unsubscribe from this group and stop receiving emails from it, send an email to herme...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/7a709d16-7fb2-4d92-818f-524f3447cbed%40googlegroups.com.

--
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 herme...@googlegroups.com.

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

Don,

 

If you have $1k available pick up a used ANAN 10 (16-bit). This is a cracking SDR, especially running Protocol II and Thetis.

 

Not exactly single board, but not too big.

 

Simon Brown, G4ELI

https://www.sdr-radio.com

[Steve Haynal]

Hi Max and Group,

Thanks for the discussion and support of the HL2. As Marc said, the primary reason for no 6M is the 80MHz max sampling rate of the AD9866 which limits the first Nyquist zone to 40MHz. However, it is possible to do at least 6M receive with undersampling. If I recall correctly, Alan and Taka experimented with this using the Hermes-Lite 1 and had success. The AD9866 was not meant for undersampling applications so does not have enough bandwidth to say undersample all the way to 2M, but 6M was possible. In the standard HL2 configuration there is a physical LPF that prevents undersampling. But that filter can be removed and other filters substituted via connectors RF4 and RF5. You would need a bandpass filter for the Nyquist zone you are interested in, plus a means to switch that filter out for a LPF when you want just HF frequencies. Analog to digital converters with higher sampling rates do not need these two filters as the first Nyquist zone covers both HF and 6M.

I'd like to say a few words about the HL2 design philosophy. Originally I started this project because I wanted a DDC/DUC architecture replacement for my SoftRock ensemble transceiver with similar price tag. Once you add in the cost of a good soundcard, plus multiple SoftRocks to cover all the bands, I think we've come close with the HL2. The cost is kept down because we use a commodity part, the 12-bit AD9866, which is meant for the cable modem market. This ~$30 part combines DAC, ADC and adjustable LNA. This is significant savings compared to discrete components for all these functions. Furthermore, the low cost and commodity performance of the AD9866 is balanced with a low cost 4-layer PCB, $35 FPGA, less stringent clock due to the lower sampling frequency and fewer bits, as well as minimal extra features. This all adds up to a well balanced and decent performing complete QRP HF rig in my mind, with no extra expenses or over specification in any area.

I don't think the right direction is to push the HL2 up. As already mentioned, there are other good SDR efforts that fill this space. There are all the Apache labs and openhpsdr radios. There is the opensource Angelina clone, the Odyssey II. There is a lot of interesting activity with the Red Pitaya and in particular the new Stem Lab 125. I have very little interest in a higher end HL3, and instead have more interest in pushing costs down give similar performance.  If I were designing a HL3, I'd take a serious look at the AFE7222 and AFE7225 as well as all the recent entirely open source flows and effort around the Lattice ECP5. See OrangeCrab, TinyFPGA EX, ULX3S, Supercon Badge, Litex IP and flow, NextPNR, Project Trellis, and Yosys.

There are also other active design efforts underway. If you want to get involved with the creation of a new SDR, checkout the Tangerine SDR.

73,

Steve

kf7o

 

[Max]

Thanks to Steve and all others for your replies. Yes, all pretty much as I had guessed. Steve, I can quite understand the reasons for the way the project has been directed and all for very noble ideals. In any case even without 6m it is still very appealing. It's just that this would have been "icing on the cake", but really one cannot complain at all given the alternative more expensive options.

Had not heard of the Odyssey but it seems (hope I have my facts straight) to have ground to a halt in terms of making a built PCB available and also of course would be more costly. A different beast I also fully understand.

No problems. Just now trying to agonise over pulling the trigger before the current Build V9s all get sold!

Thanks again for all the input. All comments fully digested and most appreciated.

73

Max

[Christopher KB3CS]

someone mention 50/28 MHZ TRANSVERTER BOARD at http://transverters-store.com/50mhz.htm ?

there does not seem to be any boards to be had from the ody-sdr.com site, though the board design and BOM is right there in the Odessey-2 github.

  - 1z (base 38) -

On Sunday, December 1, 2019 at 5:30:23 PM UTC-5, Max wrote:

Thanks to Steve and all others for your replies. Yes, all pretty much as I had guessed. Steve, I can quite understand the reasons for the way the project has been directed and all for very noble ideals. In any case even without 6m it is still very appealing. It's just that this would have been "icing on the cake", but really one cannot complain at all given the alternative more expensive options.

Had not heard of the Odyssey but it seems (hope I have my facts straight) to have ground to a halt in terms of making a built PCB available and also of course would be more costly. A different beast I also fully understand.

[...]

[rentwist]

Same guy but seems to be a greater variety (including built, tested and packaged) on his eBay store: https://www.ebay.com/sch/m.html?_ssn=transverters-store&_from=R40&_sacat=0&_nkw=50+to+28&_sop=15

73,

Robert, WA2T

[Sebastien F4GRX]

That would be a great move!

Sebastien

[Christopher KB3CS]

interesting. AFE7222 $22. AFE7225 $31. both dual channel. you'd gang them together to get twice the Nyquist BW? or create a coherent two channel RX?

existing Altera EP4CE22 FPGA is roughly $50.

the Lattice ECP5 line is not as granular (fewer part variations) but it rapidly surpasses the EP4CE22 in LUTs and Multipliers, etc.

the mid-range device price looks roughly $25 and less.

hmm!

  - 1h (base 28) -

[Steve Haynal]

Hi Chris,

Where did you find the AFE7222 for $22? These parts are designed to operate as IQ quadrature input/output. One thing I'm not sure about is if you can bypass all/some internal block. For example, they have an internal fine mixer with NCO. Also, these parts do not have and ADC driver with gain plus attenuator. That will run an extra ~$10 per channel for decent external components to do this.

73,

Steve

kf7o

[Christopher KB3CS]

looking more carefully this time, not just to get an idea of what price range it's in, i have:

$26.68644 @ 250 units

$22.16214 @ 2000 units

$38.44 single piece

.. at Digikey.  i gather the application is primarily automotive, so the price breaks make sense to me.

in any case, the prices look quite roughly similar to the AD9866.

  - 1h (base 28) -

On Tuesday, December 3, 2019 at 10:51:25 AM UTC-5, Steve Haynal wrote:

Hi Chris,

Where did you find the AFE7222 for $22? These parts are designed to operate as IQ quadrature input/output. One thing I'm not sure about is if you can bypass all/some internal block. For example, they have an internal fine mixer with NCO. Also, these parts do not have and ADC driver with gain plus attenuator. That will run an extra ~$10 per channel for decent external components to do this.

[...]

[Steve Haynal]

Hi Chris,

Okay. The additional external LNA and attenuator required by the AFE7222/7225 is a significant increase in expense as that will cost $20 for two channels, unless someone has a clever idea. Also, this part is not easily available in China. If you search for AD9866 and AFE7222 on aliexpress you will see hits only for the AD9866.

Since we are just brainstorming, I wouldn't consider the Cyclone IV technology for a new design. If you want to stay close to the Cyclone IV, use the Cyclone 10. They are newer and less expensive.

Another interesting options is the $81 Z-turn lite board which is now sold by Mouser. It should be possible to design a small companion board for that FPGA card which connects an AFE7222 or even AD9866. If done right, it should be possible to leverage much of the work already done for the Red Pitaya but at half the hardware cost.

73,

Steve

kf7o

[Duncan Clark]

In message <0039c9c6-1983-48ab...@googlegroups.com>,
Steve Haynal writes

>However, it is possible to do at least 6M receive with undersampling.
>If I recall correctly, Alan and Taka experimented with this using the
>Hermes-Lite 1 and had success. The AD9866 was not meant for
>undersampling applications so does not have enough bandwidth to say
>undersample all the way to 2M, but 6M was possible. In the standard HL2
>configuration there is a physical LPF that prevents undersampling. But
>that filter can be removed and other filters substituted via connectors
>RF4 and RF5. You would need a bandpass filter for the Nyquist zone you
>are interested in, plus a means to switch that filter out for a LPF
>when you want just HF frequencies. Analog to digital converters with
>higher sampling rates do not need these two filters as the first
>Nyquist zone covers both HF and 6M.

Another place to look is what G4JNT has done for his ELAD FM-Duo,
undersampling for both 4m and 2m. Published in the RSGB's RadCOM 3months
or so ago?

http://www.g4jnt.com/FDM-DU0_Notes.htm

In particular:

http://www.g4jnt.com/FDM-DUO_at_VHF.pdf

PCB layout is available - http://www.g4jnt.com/Elad-VHF-PCB.zip

Running his methodology using the 78.4MHz clock I came up with the
attached.

6m should be possible, if I have done it right, at 26.8MHz and maybe 4m
at 7.6MHz.

I 'simply' need to etch the pcb, build it and see what happens. As I
have Minicircuits UK 10mins drive away I have the MMIC's already and
will need to just re-jig the filter values etc.

On the list of things to do.

I am an 6m DXer so 6m is a very high priority for me. I don't like those
Russian transverters - I don't think the 6m one is high enough
performance when I have several EME capable stations around me. 400W and
2 x 6ele 2 or 3 miles away is a lot to handle on any 6m rig. Add in my 2
x 6ele and one needs something bomb-proof. A G4DDK Anglian like
transverter would be a much better bet.

Duncan
--
Duncan Clark
G4ELJ

[Christopher KB3CS]

Cyclone 10CL040 runs $39 single piece and 325 I/O pins. (assuming you meant the Cyclone 10 LP series, not the GX).

the $81 Zynq boards looks like it uses the Zynq-7007S SoC and i don't think you'd find that at all adequate since the Artix-7 FPGA within is a series that does not appear to be equivalent or better than the Cyclone EP4CE22 except at the extreme.

the Zynq-7020 SoC board is more like $140. a Kintex-7 within the SoC is available. the low end of the Kintex-7 series appears to exceed the Cyclone EP4CE22.

  -  2'ell' (base 26) -

On Tuesday, December 3, 2019 at 12:34:08 PM UTC-5, Steve Haynal wrote:

Hi Chris,

Okay. The additional external LNA and attenuator required by the AFE7222/7225 is a significant increase in expense as that will cost $20 for two channels, unless someone has a clever idea. Also, this part is not easily available in China. If you search for AD9866 and AFE7222 on aliexpress you will see hits only for the AD9866.

Since we are just brainstorming, I wouldn't consider the Cyclone IV technology for a new design. If you want to stay close to the Cyclone IV, use the Cyclone 10. They are newer and less expensive.

Another interesting options is the $81 Z-turn lite board which is now sold by Mouser. It should be possible to design a small companion board for that FPGA card which connects an AFE7222 or even AD9866. If done right, it should be possible to leverage much of the work already done for the Red Pitaya but at half the hardware cost.

[...]

[Christopher KB3CS]

skepticism is good. i think the referenced 20m/6m transverter resembles the "T-Kit" Ten-Tec transverter of legend.

the EME stations are pointing at the Moon, not your antenna array, i should think. and their narrow beamwidths mean, from your observation point, an attenuation similar to an RF sampler, or do you expect otherwise?

  - *2h* (base 28) - [correction of previous] 

On Tuesday, December 3, 2019 at 4:24:46 PM UTC-5, Duncan Clark wrote:

[...] 

I am an 6m DXer so 6m is a very high priority for me. I don't like those
Russian transverters - I don't think the 6m one is high enough
performance when I have several EME capable stations around me. 400W and
2 x 6ele 2 or 3 miles away is a lot to handle on any 6m rig.

[...] 

[Reid Campbell]

Hi Duncan,

Have you seen https://github.com/ji1udd/Hermes-Lite/tree/6M

There is a potential problem with 6M and the second harmonic of 26.8Mhz.

Cheers

Reid Gi8TME/Mi0BOT

[Steve Haynal]

Hi Chris,

Why do you think the Zynq-7007S is not equivalent to the EP4CE22? Here is a comparison with the Zynq-7007S numbers first:

Logic Elements: 23K vs 22K

Memory bits: 1.8Mb vs 0.6Mb

DSP: 66 full slices vs 66 18-bit multipliers

Plus the Zynq-7007S doesn't need to spend FPGA logic for ethernet or other control as that would be in software. That saves 20%-25% of logic elements.

I think the Zynq-7007S is a step up from the EP4CE22.

73,

Steve

kf7o

[Christopher KB3CS]

53.6 MHz is nearly the upper band edge.  terrestrial SSB use trails off by 50.3 MHz or so. 

there's still some old-timers with 50.4 MHz AM gear.

  - 133 (base 7) -

On Tuesday, December 3, 2019 at 6:05:45 PM UTC-5, Reid Campbell wrote:

Hi Duncan,

Have you seen https://github.com/ji1udd/Hermes-Lite/tree/6M

There is a potential problem with 6M and the second harmonic of 26.8Mhz.

Cheers

Reid Gi8TME/Mi0BOT

[...]

[Christopher KB3CS]

yes, those specs cited are true. 

my path to expressing a value judgement was multifold:

1. i encountered detailed specs on the Artix-7 product line faster and in more detail than i could find for the Zynq-7007S.
2. i compared the Artix-7 line to the Kinetix-7 line.

3. i found the Zynq boards readily on offer were the -7010 and -7020 boards and did not encounter a -7007S board on offer.

4. you have expressed reaching the limits of the EP4CE22 in the present project work.

  - 81 (base 9) -

On Tuesday, December 3, 2019 at 6:36:06 PM UTC-5, Steve Haynal wrote:

Hi Chris,

Why do you think the Zynq-7007S is not equivalent to the EP4CE22? Here is a comparison with the Zynq-7007S numbers first:

Logic Elements: 23K vs 22K

Memory bits: 1.8Mb vs 0.6Mb

DSP: 66 full slices vs 66 18-bit multipliers

Plus the Zynq-7007S doesn't need to spend FPGA logic for ethernet or other control as that would be in software. That saves 20%-25% of logic elements.

I think the Zynq-7007S is a step up from the EP4CE22.

73,

Steve

kf7o

[...]

[Duncan Clark]

In message <c1f328f8-8f75-4982...@googlegroups.com>,
Christopher KB3CS writes

>the EME stations are pointing at the Moon, not your antenna array, i
>should think. and their narrow beamwidths mean, from your observation
>point, an attenuation similar to an RF sampler, or do you expect
>otherwise?

They work EME but are also terrestrial DXers. We all chase the similar
DX in the narrow DX window. Not totally unusual to have a few Carribean
stations in just 20Khz or so of the DX window.

[Duncan Clark]

In message <acf8e194-ccce-4202...@googlegroups.com>,
Christopher KB3CS writes

>53.6 MHz is nearly the upper band edge.  terrestrial SSB use trails off
>by 50.3 MHz or so. 

That's what happens over here too. I don't think it matters too much
whether the 2x issue is at 47.whatever MHz or 53.6. Have to use a very
narrow filter which I think can be done. I'll find out more when I play.

[Duncan Clark]

In message <60df90ab-427f-2fd8...@gmail.com>, Reid
Campbell writes

>Hi Duncan,
>
>Have you seen https://github.com/ji1udd/Hermes-Lite/tree/6M

I hadn't seen that. Thanks for the link.Very interesting.

[Christopher KB3CS]

here's a new something to consider (Trion T20 is midrange and only one or two steps away are very capable chips):

   https://www.efinixinc.com/products-trion.html

would basing your multiRX idea on one of their test and eval boards make good sense? (open toolchain exists)

  - 133 (base 7) -

On Wednesday, December 4, 2019 at 2:23:01 PM UTC-5, Christopher KB3CS wrote:

yes, those specs cited are true. 

my path to expressing a value judgement was multifold:

1. i encountered detailed specs on the Artix-7 product line faster and in more detail than i could find for the Zynq-7007S.
2. i compared the Artix-7 line to the Kinetix-7 line.

3. i found the Zynq boards readily on offer were the -7010 and -7020 boards and did not encounter a -7007S board on offer.

4. you have expressed reaching the limits of the EP4CE22 in the present project work.

  - 81 (base 9) -

[...]

[Steve Haynal]

Hi Chris,

Thanks for the link. Did you find any eval boards at $20 or less? All I could find were the small 4K LE parts.

We need at least 15K LE, >50 18x18 multipliers and 500 kbits of RAM to have enough resources. One thing I like about the QMTech boards is that they provide this at under $20. Now they have a $26 version with gigabit ethernet:

https://www.aliexpress.com/item/33007471265.html

Some FPGAs I am watching are:

The new Lattice Nexus family with first parts in the Crosslink-NX series

Lattice ECP5 mentioned before

Microchip PolarFIRE SoC which is similar to the Zynq but with opensource RISC-V

Intel Cyclone 10

Low cost low end Zynq-based boards

Lichee Tang

It is interesting to look at and learn about these different low cost FPGA options, but I don't have any plans to use one in a next version HL2.

73,

Steve

kf7o