The D-STAR buzz continues to build in Colorado with a few repeaters on the air and more to come. For an overview of D-STAR, take a look at the article I wrote for CQ VHF magazine. I decided to dig into the digital modulation format that is used in D-STAR so I could understand it better. (Moving Forward!) At first, I figured that this newfangled digital modulation had nothing to do with FM but later realized that this is not completely true. (No, FM and D-STAR do not interoperate.)
It turns out that D-STAR uses Gaussian Minimum Shift Keying (GMSK), the same modulation format that GSM mobile phones use. What is that, you say? Let’s start with Frequency Shift Keying (FSK)…when the digital signal is a logical one, a particular frequency is generated. When the digital signal changes to a logical zero, a different frequency is generated. On the receive end, we just keep an eye on the frequency and decode the digital signal accordingly. In concept, we could generate this FSK signal by hosing the digital signal into an FM modulator. Minimum Shift Keying is a special case of FSK where the frequency shift (and the phase changes) are carefully controlled (modulation index of 0.5) to keep the phase discontinuities low and the bandwidth minimal. Add a Gaussian filter on the front end of this and you have GMSK. (The gaussian filter smooths out the digital transitions and gives an even narrower bandwidth.) I found this GMSK tutorial to be very helpful.
On the air, a GMSK signal has a constant amplitude, just like FM. It will switch back and forth between two frequencies as the digital signal goes high and low. The digital signal can be recovered using an FM detector but the output of the detector is the digital format.
Of course, GMSK is only part of the D-STAR modulation system. The other key component is the vocoder, which is the AMBE-2020™ Vocoder from Digital Voice Systems, Inc (DVSI). This chip performs the magic of smooshing the digitized voice signal into a reasonable number of bits per second so that when they are turned into GMSK they fit within the rather narrow bandwidth (6 kHz nominal). Since this vocoder chip is proprietary, it has caused some hams to grumble about the lack of an “open” vocoder algorithm. I think they have a point but it does get overblown. I have heard statements that this proprietary algorithm adds $100s to $1000s of dollars to the price of a D-STAR radio. My sources tell me the chip sells for about $20, a very reasonable price for this functionality.
I’m in the process of turning this investigation into my FM column for the Winter issue of CQ VHF, so there will be more detail there. I hope this short note gets your brain thinking.
73, Bob K0NR
Hi Bob,
I do think the cost issues have been overblown by some, but there are some truly unavoidable issues with proprietary hardware such as the AMBE vocoder. The vocoder chip is provided by one company, so they completely control its availability and pricing. If they decide to end-of-life the hardware and/or algorithm and replace it with a new incompatible protocol, there’s nothing anyone can do about it. The algorithm is proprietary/closed source, so no amateur can learn from the code or improve upon it. And even worse, with all the sunk costs in AMBE based gear and little to no chance of a competing compatible technology, we’re pretty much locked into AMBE until the protocol or chip is end-of-lifed. I’m unsure if JARL or Icom made the decision to go with AMBE, but I’m surprised that in this day and age of open source and community supported software that a proprietary algorithm and hardware was chosen. I’m not saying D-STAR is bad because of the issues with AMBE, but we need to recognize the issues with proprietary hardware in amateur radio and prevent its use in the future. While it may be inexpensive today, it may not be tomorrow.
73
Goody
K3NG
Goody,
Thanks for the reply. No question, all other things being equal, it would be better to have the vocoder algorithm in the public domain. Bruce Perens K6BP makes a strong argument that an Open Source vocoder needs to be developed: http://codec2.org/
From what I have read, DVSI has the lead in vocoders and the AMBE algorithm is arguably the best or one of the best in the industry. (Or I maybe I just believe their marketing 🙂
73, Bob K0NR
I’m with Bruce (and have been on Linux/Open Source Software) for many years. When I got into Amateur Radio the idea was that we could learn some stuff and share some stuff. Open Source Software just seemed like a logical continuation of the same thing.
Let’s act like real hams and build something we can share.
73, Phil, YN8PH/WA6SWR
Hi Bob,
Happy holidays! For giggles I took a look at D-STAR pricing in my AES catalog. The Icom UT-118 module which upgrades most of their D-STAR capable rigs to operate D-STAR runs about $220. Ironically, their lowest end monoband 2m mobile rig is only $170. I’m guessing that the UT-118 card has only the $20 DVSI AMBE chip and not much more, though I’ve never had one in my hand. It seems strange that an entire rig costs less than the option card, though perhaps this is the Bic razor/ink jet printer cartridge type business model at work.
So while the DVSI AMBE chip indeed may not be adding a significant cost to the system, it would seem Icom is adding more cost, perhaps more than most would expect. Considering their D-STAR repeaters and networking accessories aren’t too cheap and it’s my understanding that it’s next to impossible to homebrew a D-STAR repeater, it would seem that Icom is making a marketing mistake. You would think they would want to price the equipment with minimal margins so they could get the market for D-STAR going faster, before Yaesu or Kenwood decide to enter and take a piece of the pie. Considering how difficult it is to get 2m frequencies (at least here in the east) and the significant costs of a duplexer, feedline, and antenna and the logistics of getting an additional antenna on what is often a commercial or public safety tower site, turning up a new D-STAR repeater is not a trivial task. The $1400 cost for a D-STAR repeater probably isn’t helping D-STAR coverage and market penetration.
I’m sure DVSI was probably the best off-the-shelf technical solution. I won’t belabor the point on open source as that has been rehashed many times by others, but I do hope Bruce Perens is successful in developing an alternate vocoder. Whether the D-STAR community will be willing to replace $220 and $330 rig daughtercards or the logistics of how such a swap out is accomplished is another discussion for another day. 🙂
Anyway, sorry to divert the discussion away from the technical focus you initially had; thank you for looking at the technical aspects of the D-STAR system.
73
Goody
K3NG
Goody K3NG,
I’ve also noted the pricing on the basic ICOM 2M FM rig (IC-2200H) and the D-STAR option UT-118. What I see here is strong price competition on the basic 2M FM rig, with the other players offering a similar price. I am still amazed that you can buy a full-featured, synthesized 2M FM rig for less than $175.
D-STAR is a different competitive picture with ICOM being the only major manufacturer involved. In general, when manufacturers have a differentiated offering, they try to extract more profit dollars. Should they lower the price to try and drive adoption? Maybe so. That leads to questions like “who are the early adopters and how price sensitive are they really?”
Another point of view I have heard is that the D-STAR option provides capability that is roughly equivalent to an AX.25 packet TNC. A Kantronics KPC-3 Plus costs about $190 these days, so that is another comparison you could make.
Or compare it to APCO 25 gear and D-STAR starts to look very affordable.
All of these are imperfect comparisons…in the end, the buyer decides how much a product is worth to his or her pocketbook.
73, Bob
Pingback: Irlp System Guidelines At Ke9v.Net