[hpsdr] New protocol , new apache hardware?

[Joe Martin]

Alf, 

There are several important areas in which the Atlas-based systems fall seriously short in terms of performance when compared to the later single-board transceiver hardware platforms.  

Examples of these areas include the following:

1) The Atlas backplane bus has a relatively low maximum data possible between boards on the bus.  This fact limits the maximum sampling rate to 384 ksps or less depending upon how many receiver boards are present.  The fact that the Tx (Penelope) and Rx (Mercury) boards must communicate with the ethernet (Metis) board via the backplane bus places a relatively low maximum data rate limit between the Atlas-based radio and the computer, which in turn limits the maximum bandwidth possible for each receiver.  Atlas-based sytems will not be able to run at the 1000T (gigabit) ethernet data rates but instead will always be limited to 100T performance levels, thereby preventing these systems from supporting many of the high performance features that are available with the single-board transceiver hardware designs.  

2) The Atlas-based system utilizes boards that are centered around lower performance FPGAs than later single-transceiver platforms.  In particular, the FPGA used in the Tx board (Penelope) is an Altera Cyclone II series FPGA.  The Cyclone II series FPGAs are no longer supported by current versions of Altera’s Quartus II FPGA development program.  This makes it necessary for developers who wish to support both Atlas-based systems and later single-board transceiver hardware to maintain two versions of Quartus II:  at least one old version of Quartus II that supports the Cyclone II series and a current Quartus II version for working on Cyclone III, Cyclone IV and Cyclone V FPGA based systems.  

3) Although Phil gave the effort a valiant try, it has been shown that supporting PureSignal on Atlas-based systems is not feasible, due primarily to the relatively low data transfer rates among the Tx, Rx, and ethernet boards via the Altas backplane bus.   

4) The number of slots on the Atlas backplane limit the maximum number of independent-ADC receivers to three in a system that has Tx and external 10MHz reference boards present.  

These areas are what come to my mind at the moment, there may be more, but these are sufficient to justify my opinion that the Atlas-based systems are lower performance platforms than are the later single-board transceiver platforms, I think.  

This does not mean that Atlas-based system are worthless; far from it.  To me these facts mean only that perhaps it may be near the time to “freeze” the Atlas-based system developement with respect to adding features that require very high performance because the fundamental hardware design itself is incapable of supporting them.  My opinions may not be shared by everyone, of course, and that’s perfectly okay.   

73, Joe K5SO

> On Nov 14, 2016, at 7:50 PM, Alfred Green <nu8i at cox.net> wrote:
> 
> ***** High Performance Software Defined Radio Discussion List *****
> 
> On 11/14/2016 10:32 AM, Joe Martin wrote:
> 
>> [snip]
> 
>> sometime in the not too distant future, I for one will lose interest in supporting lower performance platforms.
> 
> > [snip]
> 
> Can you elaborate on what you see as deficient, performance wise, with the Atlas-based systems?
> 
> I should point out that I have both an Atlas trx, albeit with a Metis rather than Ozy to give me LAN access, and also a Hermes system (plus a bunch of other SDR platforms) and I'm struggling to find a real reason to dislike my Atlas.
> 
> 73  Alf  NU8I
> Phoenix  AZ DM33xo
> 
> ps. Tnx for all your efforts, Joe, particularly with the RA facility which I use a lot.
>