--- In digitalradio@yahoogroups.com, "Rud Merriam" <k5...@...> wrote: > > First, I would not dismiss sound card modes. I think there is much more > that can be done with them. One of the main issues IMO is that they > don't (1) adapt to changing band conditions, and (2) don't utilize FEC > as much as is possible. (We also might need much better sound cards. >
I would say that adapting to changing band conditions and utilizing FEC as much as possible are not inherent limitations of sound card modems, but are simply artifacts of the higher-level protocols. There's the modulation scheme layer, and the encoding layer where the FEC is applied, and then an adaptive layer that comes in where the sender gets feedback from the receiver that things are not going very well and something else should be tried. Now this layer may call for a change down in the modulation scheme layer, to use something more robust and slower when things are not going well, and to hop up to something faster and less robust when conditions permit. I think you're right too about the quality of sound cards. What we get with onboard sound or with the low-priced add-in boards is the lowest grade the consumer will accept. You can't blame the manufacturer for providing that when the consumer will accept such junk - I mean if all you are going to use it for is to listen to loud, intentionally-distorted rock music then you aren't going to care much about signal to noise ratio or linearity. You can find some sound card evaluations on the web; and some have been discussed on this group. There are really good ones, but then the prices are up there with the dedicated DSP engines like the SCS modems. The challenge is to find one that is good enough - and considering all the noise and crud of the HF channel maybe it doesn't have to be very good to be good enough. The big win for sound card modems is that everybody has one already, and that lots of people are writing software that can use them, so lots of experimentation is going on. We've already got dozens of digital soundcard modes that nobody uses because they have not attracted enough attention or shown enough superiority over others to make people want to use them. > Second, further gains could be made using external DSP boards. > These are not that expensive today. What we are talking about here are products that have their own DSP processor and memory and A/D/A conversion, and you can download the firmware that runs the DSP from a PC. The advantage seems to be that the DSP processor is not getting interrupted by all the other activity that is going on in the PC; and the receiver can maintain synchronism with the transmitter over a long period of time. This is presumably important for some modulation and coding schemes - they can spend as much time as they want getting synchronized initially, and then they don't need to spend much effort to stay synchronized afterward. The problem we all realize is that there is no single DSP engine out there that we have all agreed to use and therefore people have an incentive to write software for; so it remains a niche activity. I recall this is where PSK was before the sound card software was developed - it ran on a dedicated DSP engine and so few people had them that there was little incentive for others to get them. If the SCS products were (are?) open to third-party development, so that others might write PSK or MFSK or Olivia modems to run on them, perhaps we would all buy them and quit using the sound cards. I don't think it's realistic to expect SCS or any other vendor to spend their own money to add popular modes to their products, esp. when the sound card versions are out there for free. There once was the HAL PCI-4K DSP modem developed for Clover but later adapted to run RTTY and Pactor-I. Clover was once used for message-passing systems as well as for conversational operation. I don't know how relatively good or bad it was as a modulation scheme compared to some of the others. It seemed to me that a main drawback could have been remedied - it seemed to get stuck trying to send a long block when conditions were too poor and for some reason it could not fall back to sending short blocks which might get through. There also once was the K6STI RITTY software that was designed to run under DOS and for a while could run Pactor-I as well as RTTY. This leads me to wonder if a suitable DSP engine could be made from a PC and a sound card but not running a general purpose operating system. Maybe FreeDOS is a candidate for the operating system, or maybe we just need a single-minded loader that can load and run a single PC program and communicate with another PC over Ethernet or USB or even RS-232. The other PC running a conventional OS would handle the higher-levels of protocol and the user interface. Jim W6JVE
