[hpsdr] Ozy-Metis (RPi2) gateway announcement

[Bob Wisdom]

Hi Alf,

The current discovery implementation is limited. Not having a Metis or
multiple SDR's to observe, I couldn't work out 100% what was really
supposed to happen in practice,  so thank you for your insight!

The libusb does have to be there. Again, from memory there are a couple of
libusb's and if you follow the info found in ghpsdr3-alex you should be
fine. After finishing I recompiled the code on the U9600 x86 laptop running
Ubuntu with no bother. The good thing about the laptop was I could use the
USB capture capability of Wireshark to have a look at the stuff coming from
Ozy (see 2 below).

The gateway will probably work on the Pi B1 with recompilation for the
lower ARM core. The Pi B1 cpu is a bit underpowered but I seem to remember
surprisingly low utilisation on the Pi2 when running 'top'. I believe the
Pi's use a single USB port from the CPU, multiplexed to both Ethernet and
USB ports, so ultimately it cannot be as good as dedicated USB and Ethernet
ports on gateway hardware (maybe the BB Black?). However, on my Pi2 and a
single HPSDR I tested PowerSDR at 384000, and also KISS, with both wide and
normal displays; they were fine. I have not tested any other SDR apps to
date.

As for outstanding issues: 1) I don't think the gateway will currently work
with multiple Mercury boards in the same rig -  I did not write it to
account for more than one. 2) There is an occasional sync issue which
appears, particularly after switching clients. I haven't been able to track
it down and it seems that the USB is getting extra data in front of the
0x7f0x7f0x7f sync bytes from the Ozy. 3) The I2C control channel over USB
seems to take up too much time - I have mitigated it in the implementation
by reducing the frequency of I2C transactions, but its still a bit of a
nag. The main source file 'listener.c' notes these things too.

The great thing about emulating the Metis is that it's much easier to
fiddle around with the code than on a FPGA. It's a small source, so I'm
confident it can adapted or enhanced to suit most needs without a major
headache. I'm on another project now too, but  ready to help if I can be of
any assistance.
GL
Bob VK4YA

On 25 May 2015 at 01:51, Alfred Green <nu8i at cox.net> wrote:

>  On 5/23/2015 7:40 PM, Bob Wisdom wrote:
>
> Hi Alf,
>
> I hope it's of use... I think BB Black should just be a recompile with the
> correct ARM CPU options in the Makefile.  It might even be okay as is,
> since RPi2 uses a newer core than RPi.
>
> The code for the Ozy-Metis gateway discovery responder might be a bit too
> friendly for multiple HPSDR radios on the same LAN in its unchanged form -
> I  have only one radio.
>
> Currently, upon receipt of a discovery packet it will stop transmission
> (if started), reply to the new initiator, setting that new initiator as the
> client - and wait for a start command. It should be pretty easy to change
> the code, and I guess it *should* be changed,  to become more compliant
> with the Metis discovery spec.
>
>
>  Hello Bob,
>
> Tnx for the response. It turns out my RPi is one of the older B+ models
> rather than a -2. I have had a quick look through your code and don't see
> any particular reason why it shouldn't work. Before I spend much effort on
> this please let me know if there is a major roadblock. Obviously it needs
> to be recompiled against the target, but is there anything serious missing?
> I tried running it on a couple of my BBB units (latest models) and while
> it appears to Make ok it has trouble with my libusb installation. I will
> dig into that further.
>
> As for the discovery issue, that definitely needs to change. I can
> envision a setup where we have multiple hpsdr-like units on a LAN,  and
> they should respond correctly. Currently, PowerSDR does not handle the
> situation properly, but KISS and my homebrew derivatives do a better job.
> As I see it, the client units, such as the RPi, should respond either with
> 'I'm here' or 'I'm here, but busy'. Both replies are covered by the Metis
> spec.
>
> I've also found out that my spare Atlas board is no longer working for
> some reason, so I don't currently have a source to drive the USB side of
> the translator. I could retrofit my Metis/Mercury/Penny/Excalibur system
> with the Ozy card if necessary, but I need to see what firmware versions I
> need to put on the beasts.
>
> Anyway, thanks for putting this out there. As if I need another project...
>
> 73  Alf  NU8I
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openhpsdr.org/pipermail/hpsdr-openhpsdr.org/attachments/20150525/ada41aa2/attachment-0001.htm>