Received: from kestrel by gos.Ukc.AC.UK Over Ring with SMTP id aa29275; 19 Jan 89 5:45 GMT Received: from mcvax by kestrel.Ukc.AC.UK via EUnet with authorised UUCP id aa25254; 19 Jan 89 5:45 GMT Received: by mcvax.cwi.nl via EUnet; Thu, 19 Jan 89 06:30:41 +0100 (MET) Received: from abvax.UUCP by uunet.UU.NET (5.59/1.14) with UUCP id AA04027; Wed, 18 Jan 89 17:35:08 EST Received: by abvax.icd.abnet.com (5.59/1.16); id AA14308; Wed, 18 Jan 89 14:18:53 EST From: "Greg F. Shay" Message-Id: <8901181918.AA14308@abvax.icd.abnet.com> To: dss@sun.uucp, "\\"@mcvax.uucp, dmocsny@uceng.uc.edu, csz@well.uucp, houghton@silver.uucp, dmt@mtunb.uucp, craig@scol.uucp, droid@mcrware.uucp, sullivan@asuvax.uucp, mg@ukc.ac.uk, ralph@turtlevax.uucp, jdl@cadnetix.uucp, dcj@sun.uucp, miklyd@ihlpb.uucp, peter@celia.uucp, evans@mhuxt.uucp, jjp@ektools.uucp, cab@hropus.uucp, rt76+@andrew.cmu.edu, lars@ddeswe.uucp, jdm@gssc.uucp, soleil.rob@megatest.uucp, videovax!bill@tektronix.uucp, pmy@jeeves.uucp, dbell@cup.portal.com, steve@media.uucp, knudsen@ihlpl.uucp, jtn@potomac.uucp, mcripps@mtuxo.uucp, tjpadula@phoenix.uucp, waycott@ellymae.uucp, b-davis@cs.utah.edu, gmo@tellab5.chi.il.us, defron@monet.berkeley.edu, brianp@ernie.berkeley.edu, hedley@imagen.uucp, braun@thumper.uucp, brian@orawest.uucp, videovax!johnr@tektronix.uucp, barrett@ektools.uucp, dukbr@dcatla.uucp, nelan@asuvax.uucp, rmd@tellab5.uucp, msr@bunker.uucp, scott@ole.uucp, burris@ihuxz.uucp, gillies@noao.uucp, maher@uiucuxc.uucp, stout!cook@ncar.uucp, booth@vax135.uucp, dpw!roger@esquire.uucp, dat@orville.nas.nasa.gov, edhall@randvax.uucp, bobk@mntgfx.uucp, nsw@cord.uucp, murakami@uiucdcs.uucp, bonzo@xochitl.uucp, bwilson@island.uucp, roger@island.uucp, rgh@inmet.uucp, t-jamesb@microsof.uucp, jimh@hplabs.uucp, lansd@godzilla.eecg.toronto.edu, prbernard@watbun.waterloo.edu, elmer!jbudet@oakhill.uucp, dair@scu.bitnet, whorfin@pixar.uucp, englandr@phoenix.princeton.edu, gson@hupu.uucp, talent@dover.uucp, hpcllcc!chip@hplabs.uucp, luc@kulcs.uucp, hoven@dupond.cd.chalmers.se, bryant@oakhill.uucp Subject: Hot DSPNET info Date: Wed, 18 Jan 89 14:18:51 EST Greetings, This marks the next big step in the DSPNET project. Basically, I have two items to share. 1) Here is the first cut of the hardware design specification. Please comment and give recommendations, or be prepared to live with it as is. I've had good ideas come in already, so don't be bashful. I will consider all input, we can discuss good ideas, and include those that make sense. I have only one hard rule, and that is I intend to meet the $250 limit. A good way to allow fancier things is to make them an option or to design in a connector so that in the future they can be added. 2) The best news is that Motorola likes what we are doing and has made a couple of VERY important things available to our group. I have permission from Motorola to make unsupported copies of the IBM PC/MSDOS Cross assembler and simulator for the 56000 for our group. There are a couple of catches, however. Unsupported means you cannot call up Motorola expecting a lot of questions to be answered or bugs (if any) to be fixed on your behalf (they will listen for any bug reports, but cannot promise quick turnaround on fixes. I was told there should not be many bugs, as this is newest revision software.) Secondly, we are NOT given permission to distribute this software totally freely, it is a special concession for our users group, so let's not abuse it and lose it. This means, no posting to bulletin boards, the net, etc. You can make copy for others you may find interested in the 56000 DSPNET project. In addition, I am responsible for collecting names, addresses and phone numbers of everyone who gets a copy, to give to Motorola. Copying the manual is going to be expensive (about $50 a shot). I will try to find alternatives like copying just the summary of the instructions, etc. I beats the $800 list price, however. If you want just the floppies with a few essential pages of the manual, it will be less $$. I will provide the exact costs of the copies and floppies forthwith. Please contact your local Motorola rep for a 56000 Users manual. Note that the grey 56000 Users manual is marked preliminary and is a little incomplete on hardware in some places, but good for the instruction set and software. A new 56000 manual is coming from Motorola in the next few months. We may have to each buy these, as the big thick manuals are not free to print. DSPNET Hardware specification 1.1 --------------------------------- As soon as we agree on some of the details, I'll make schematics, parts lists and (hopefully) bare printed circuit boards available. You could wire wrap the design if you wanted to, or get the PC board. More on that later. OUTLINE: MAIN COMPONANTS Basic operational system composed of two small boards. Board 1) CPU BOARD 56001 CPU + 20.5 Mhz crystal Boot EPROM Socket, accepts anything from 2716 to 27256, any speed. 3 Static memory sites. Able to use either 3 32Kx8 RAMs or 3 8Kx8 RAMs 55ns memory allows 0 wait state. Not available 100ns memory allows 1 wait state. 150ns memory allows 2 wait states. The 56001 has software programmable wait states, so you will be able to configure your system for the memory speed you have with the initial download file. The above memory will be shared by all three 56001 address spaces, P, X and Y. This means it is up to the programmer to assign address space partitions based on address. Midi In/Midi Out circuit and connectors. ?? Midi Through optional (on pc board, but can leave off connector) Power supply connector for external power supply. ?? 25pin (9 pin??) D-shell for RS-232 data lead only connection to host. This would be for those without midi interfaces who are doing stand alone DSP experimentation. Midi people can skip the circuit and connector. Memory expansion bus connector. About 64 pins. (Should there be two?) Could be more static memory or DRAM. No particular plans are in mind, but there are a lot of options. Also could be parallel access periperals. Two I/O interface connectors. About 36 pins each. These carry the 56001 Serial interface lines and the host interface signals. One is intended for the A/D D/A board below. The other can be a custom host adapter or a (future, maybe, SCSI adapter, don't hold me to this yet!) Board 2) ANALOG I/O BOARD Many different options can be built into this. This is why I made it on a separate board, so you can choose the type, cost, and performance you want. The only board I am going to commit to now is the bare bones, cheapest board. I can design better resolution, better filters, etc, but it all costs more. This is going to have to be driven by you. What features do you want? What are you willing to pay for? The bare bones option is the one that gets the whole system in under $250. BARE BONES: 14 bit A/D. 0 to 80Khz sample rate. Includes 4 pole analog input filter and sample/hold circuit. Standard phono jack. 16 bit D/A. 4 pole analog aliasing filter. Standard phono jack. You can adjust the cutoff frequency of the filters by choosing different values of the filter componants. I suggest 15Khz for 32Khz sample rates and 20Khz for 44Khz (CD) sample rates. If anyone is willing to go down to 12Khz for a 25Khz sample rate, you get more processing time vs. reduced bandwidth (this may be acceptable for some synthesis applications.) I'll have the capacitor and resistor values figured out for these three frequencies at least. ----------------------- For additional cost (Which numbers I can get in detail as soon as the first cost roll-up is ready): Stereo 16 bit D/A Roughly $20-30 more over mono D/A. 16 bit A/D About $110 more than the 14 bit design. Limited to 50Khz. A slower 32Khz 16 bit A/D can be used for $70 over the cost of the 14 bit. 7 pole analog filters for about $10 apiece instead of 4 pole. OPERATION: The audio sample rate is programmable internal to the 56001. For the anti-aliasing filters, you will have to choose a filter value and stick with it. I've experimented for quite a while with no antialiasing filters at all and sure, there is a little noise, but for some of what I was doing, I didn't care, so don't get bent out of shape over the filter frequencies. When the 56001 powers up, it will load a bootstrap program from the EPROM into its internal program memory and run it. The bootstrap program will listen on the midi port (opt. RS-232??) for a sys-ex sequence with the second level bootstrap program. This second program can be either the whole application or an extended function monitor that you will make further program downloads to. This second level boot from the host will contain among other things, important parameters like the number of wait states used for the memory, the midi device address, and others. ---------------------------------End of spec--------------------------- Now I have something to ask of all of you, 1) Discuss your comments and preferences and good ideas with me, and 2) just how many of you are REALLY planning on spending your $250 and buying one of these when the time comes? As a disclaimer, I am NOT making a profit anywhere in this, but I do need to cover costs. If I get stuck ordering 20 sets of parts and only 5 are bought, guess who is left holding the bag. --) No need to send money yet, just a head count of those honestly in this project for real. Ok, go to it. Jus' Midi'in along..... Greg   Received: from kestrel by gos.Ukc.AC.UK Over Ring with SMTP id aa20168; 27 Jan 89 9:44 GMT Received: from mcvax by kestrel.Ukc.AC.UK via EUnet with authorised UUCP id aa23775; 27 Jan 89 9:43 GMT Received: by mcvax.cwi.nl via EUnet; Fri, 27 Jan 89 10:39:29 +0100 (MET) Date: Fri, 27 Jan 89 10:39:29 +0100 Received: from abvax.UUCP by uunet.UU.NET (5.61/1.14) with UUCP id AA00455; Thu, 26 Jan 89 02:17:12 -0500 Received: by abvax.icd.abnet.com (5.59/1.17); id AA21794; Wed, 25 Jan 89 18:00:48 EST From: "Greg F. Shay" Message-Id: <8901252300.AA21794@abvax.icd.abnet.com> To: dss@sun.uucp, "\\"@mcvax.uucp, dmocsny@uceng.uc.edu, csz@well.uucp, houghton@silver.uucp, dmt@mtunb.uucp, craig@scol.uucp, droid@mcrware.uucp, sullivan@asuvax.uucp, mg@ukc.ac.uk, ralph@turtlevax.uucp, jdl@cadnetix.uucp, dcj@sun.uucp, miklyd@ihlpb.uucp, peter@celia.uucp, evans@mhuxt.uucp, jjp@ektools.uucp, cab@hropus.uucp, rt76+@andrew.cmu.edu, lars@ddeswe.uucp, jdm@gssc.uucp, soleil.rob@megatest.uucp, videovax!bill@tektronix.uucp, pmy@jeeves.uucp, dbell@cup.portal.com, steve@media.uucp, knudsen@ihlpl.uucp, jtn@potomac.uucp, mcripps@mtuxo.uucp, tjpadula@phoenix.uucp, waycott@ellymae.uucp, b-davis@cs.utah.edu, gmo@tellab5.uucp, defron@monet.berkeley.edu, brianp@ernie.berkeley.edu, hedley@imagen.uucp, braun@thumper.uucp, brian@orawest.uucp, videovax!johnr@tektronix.uucp, barrett@ektools.uucp, dukbr@dcatla.uucp, nelan@asuvax.uucp, rmd@tellab5.uucp, msr@bunker.uucp, scott@ole.uucp, burris@ihuxz.uucp, gillies@noao.uucp, maher@uiucuxc.uucp, stout!cook@ncar.uucp, booth@vax135.uucp, dpw!roger@esquire.uucp, edhall@randvax.uucp, bobk@mntgfx.uucp, nsw@cord.uucp, murakami@uiucdcs.uucp, bonzo@xochitl.uucp, bwilson@island.uucp, roger@island.uucp, rgh@inmet.uucp, t-jamesb@microsof.uucp, hpsadlu!jimh@hplabs.uucp, lansd@godzilla.eecg.toronto.edu, prbernard@watbun.waterloo.edu, elmer!jbudet@oakhill.uucp, dair@scu.bitnet, whorfin@pixar.uucp, englandr@phoenix.princeton.edu, gson@hupu.uucp, talent@dover.uucp, hpcllcc!chip@hplabs.uucp, luc@kulcs.uucp, bryant@oakhill.uucp, greg@numusic.uucp, czei@bagheera.eng.ohio-state.edu, thomash@ed.ecn.purdue.edu, dat@orville.nas.nasa.gov, cogsci.berkele@uunet.uucp y.edu!well!rnm, uunet!unido!laura!hmm Subject: Review of spec 1.1 comments Date: Wed, 25 Jan 89 18:00:47 EST Greetings fellow earthlings, The responses from spec 1.1 have started to taper off, Jan 25. My apologies if I didn't wait long enough for yours to reach me. The response has been overall very favorable. Just about everyone who replied here said they were seriously committed to buying a board, some even two. I am presently working with a printed circuit vendor who is not too expensive, and is willing to work with us. As a result of all of your input, I can firm up the details of the design. My thanks to everyone who took the time to comment. Everyone benefits from a lot of heads thinking of many different angles. Edited summary of responses to spec 1.1: -------------------------------------------- Brian Peterson: Have you seen the Electronic Musician article (Jan. '89) "Experimenting with Brickwall Filters"? It describes a -78 dB/octave switched-capacitor filter. The filter uses two LTC1062 switched-capacitor filter chips. (....) The design given allows cutoff frequencies of about 17kHz to 20.5kHz. If it could be controlled by the DSP chip, then we'd have a nifty software adjustable filter.(.....) The price for a pair of LTC1062's is given as $15 (Maybe less) Would a filter like this be worth the cost??? | I have looked at these filters in the past and they are very good. The only | real problem is they are somewhat noisy (typically S/N at -75 to -80 db or so) | 16 bit audio has a 96 bit S/N ratio. 12 bit audio is -72db. It's a shame to | inject the noise of a 12 bit system into a 16 bit system! | I've got the LTC1062 databook in front of me and it lists 62uVrms of noise | for a 2.6Vrms signal, translating to 20*log(62E-6/2.6) = -92db S/N. That looks| ok. However, the filter response curve for operation in the 10Khz to 20Khz | region shows that the maximum attenuation of high frequencies is between | -30db and -40db (it's leaky). Analog filters can be low noise and have | maximum attenuations on the order of well over -100db. | I'll check the pricing out. I think the LTC1062 may be an alternative for | the 4pole filters. The ability to tune them would be nice, I agree. | A circuit to allow software tuning would cost $15 extra, adding circuitry | to the main board. Bear in mind what I said about the antialiasing filters, | Unless you are really into an audiophile frame of mind, a little noise may | not be noticable (compare the typical noise in a synth, the D-10, MT-32 | or K-1 (and others) and you will see). We don't need to build the best | audio section in the world for the bare bones setup. P.S. Is this mailing group a moderated letter, or should we just toss our ideas out to the whole group? | For the time being, if there is a specific question/comment, you can mail to | me and I will summarize and retransmit to the group. Use the newsgroup if | you want. I am trying to avoid too much more growth in the group for the time | being, so I for one am not going to actively solicit more members on the net. | By all means, if you know someone or have a users group you know of, get them | involved and you be the coordination point for them. ------------------------ John Miller: I would pay the extra for the 16-bit stereo versions with the 7-pole filters and generally highest-quality everything. | For REAL high quality output, some of the oversampling D/A filters out | of CD players can be used (available from Signetics). Costs more, obviously. | (I'll guess about $50-$100 extra). I already have a DSP56000 User's Manual (the grey, preliminary one) but i'm sure I can get the new one through the local sales office. Is this the $50 manual you are referring to? | The $50 for reprint cost manual is not either of the user's manuals, but I | was referring to the manual for the Assembler and simulator software. A | whole section describes the instruction set, which MAY be duplicated in the | users manual in the same detail. If so, we could lose 30% of the cost | assuming you get a users manual for the instruction set reference. ------------------------ Don Jackson: 1) Daniel Steinberg & I are fairly committed to at least 1 board. When you firm up the spec, we can firm up our commitment. |That's all I ask for. 2) The 56000 serial ports should come out as 'NeXT' compatible if you do it. People will be building lots of nifty A/D D/A circuits for the NeXT box, so if you have compatible serial ports you will have access to lots of fun toys. | I'm bringing out the 56000 serial port pins raw to the I/O connector. If NeXT | adds some hardware to that, we can design an adapter (assuming they tell us | what their design is. 3) Are you going to put this on a 'IBM-PC' card? If not, you should put space on the board for lots of RAM. You don't have to stuff it to keep the cost down. | The boards are meant to be stand alone. Blank board space is not free. PCB | costs depend on the area of the board and the number of holes. This is why | I suggested the memory expansion connector. Those who want more memory will | have to pay for two mating connector halves and the additional board, while | those that do not carry no additional cost. | The 56001 can support a maximum of 64Kx24 in each of 3 memories, P,X and Y. | The DRAM interface will operate by mapping a window into the DRAM memory | with a register holding the window address. This basically allows unlimited | address space. The DRAM circuit has additional overhead, so it definitely | will belong on the extension card. Perhaps though, I will allow for a | fully populated static ram implementation on the main board ( 3 x 64K x 24) | or at least some large fraction of that, if the cost of the larger PCB and | holes does not put us over budget. Bear in mind that 3 x 64K x 24 is | 18 32Kx8 memory chips at $20 each (100ns 1W.S.) which is $360. This is | not really enough memory to do a very powerful sampler (only 4 seconds or | so at 32Khz sample rate). If you are thinking about a sampler, you'll | want a meg or two of DRAM and a fast host interface to load it. 1 Meg x 16 | DRAM,(32 seconds sample @ 32khz) will run you $500 today (at $32 per 1MBx1 | chip). IF you could buy them (they are scarce and hard to find) four 256Kx4 | DRAMS would get you 256kx16 sample memory for $128 or so, (8 seconds of sample | at 32Khz). | I will worry about the DRAM interface later, if enough people want it. -------------------------- Michael S. Czeiszperger: (...) The file format is obvious (For storing downloadable sys-ex files): The Standard MIDI File format is now part of the MIDI spec. Now, about the hardware. Do you think the 56000 will be able to do reverb in real time? | Yes. Different reverb models use a varying number of delay/summation nodes. | The number of delay nodes is limited by memory, summation nodes are limited | by summation speed. With the 56000, you will definitely run out of memory | before summation speed. I would guess 4 to 6 nodes would make a decent | reverb, 8 to 12 for a real pro job. Design of reverb models is a little | known art that I know just a little about. Lexicon knows. Look at their | parameter configuration panel. And what about the audio input? I'd like to see it go stereo at 16 bit so the system is perceived as "professional" quality. | It looks like the audio section is tending to two designs. The original | quick and dirty mono (possible stereo out), and a full 'almost pro' stereo | in/stereo out for more money. Is there a 56000 compiler that runs on the mac? | Yes, the development system does have a Mac version. I do not have this from | Motorola, however, and I do not know if they will be as free with it. | We need a strong showing of Mac people who are seriously interested so I | have some backup for asking Motorola to extend their deal to the Mac version | too. If not, we will have to write a simple cross assembler ourselves or | find a table driven cross-assembler that can be modified. ---------------------------------- Ed Hall: My only comment is to ask whether you might go with the 7-pole filters to start with, since they don't seem to cost that much more. | Point taken. Final decision will be made after final cost. ---------------------------------- Forrest Cook: Build the board on an IBM-PC type card so that power supplies are not an extra hassle. This also solves the problem of putting the card into a box. It would be nice to have it talk directly to the PC buss, but that might exclude non PC people. | That is the whole problem. For now, I think enough people with a wide | variety of systems are interested. As a possible follow up project, I | can design a PC card based system, but 1) it definitely will be second in | line and 2) will wait a while until the activity of the midi version | dies down. Also, the fewer number of people buying one means the overhead | costs of haveing a PC board made go up per unit. If you can find 10 to 20 | interested buyers minimum, I'd be willing. With the extra space filled with holes for prototyping additional hardware onto. | Use the connectors. Stereo output seems kind of appealing. I would be interested in simulating a moving doppler shift sound effect that would require 2 channels. Stereo input would be nice, but I could live without it. I can live with a 14 bit input A/D and the 16 bit D/A chips. | All falls into the 'bare bones' board provided I put stereo D/A on it. It would be extremely nice to be able to use the C language. I Don't know what's out there as far as C compilers, but maybe we could get a group discount. | Motorola has the C compiler for the 56000 now. Thats asking a lot from them | though. Well, let's set our sights high, if we have a strong project showing | maybe we can get it. I think Motorola will be watching how far we get with | what we have now. ------------------------------------------------------ Peter Farson: 1. Are there any 56000 cross assemblers for the Macintosh? Since I have a Macintosh and don't have a PC or clone, this is a big concern for me. | See above comment. Will one DSPNET unit be able to have more than one processor? Or maybe two or more units could be coupled together via an expansion connector ? | I hadn't planned on multiple processors, although Motorola shows that | multiple 56000's can be connected together using the SSI port. I am | planning to use the SSI port for the A/D and D/A. A couple of possibilities | are to operate the SSI port in the TDM (Time Domain Multiplex mode) and | reserve a few slots for interprocessor data. Another possibility is to | build the A/D and D/A to plug into the (parallel interface) memory expansion | connector, leaving the SSI free. Yet another would be to somehow hook the | multiple 56000's together using the 8 bit parallel host interface port that | I an not planning on presently using, ( a future SCSI interface could go here) | The basic nature of this stage of the DSPNET project is not really including | multiple processors. How many others would be serious about multiple | processors? I really would like to have two analog inputs. One of the things I want to experiment with is digital vocoding and other types of digital operations on two sounds. | Sounds like a great application. You'll need the better quality audio | board for dual A/D's. I think that the analog board should be as isolated as possible from the cpu board. I vote for using opto-isolators for all signals going between the two boards and using independent power supplies. | I'll consider this. You always could use the same supply for both if | you didn't want to bother with two supplies. The total cost for the basic system is targeted at $250. But what is the breakdown? Just as you are concerned about being left holding the bag when people who said they were interested all of a sudden disappear, I, and probably at least some of the rest of the people interested in this project, feel a little funny about sending $250 in the mail to a private individual that I have never met except through correspondence over computer mail. |I'll provide a complete breakdown, and you can buy just the pieces you want. |I'll even provide complete design, schematics and parts lists for just an |SASE for those who really want to do it themselves. |I really imagine I'll only be supplying the PC boards, but some parts have |quantity discounts, so if we combine orders, we can get better prices. So anyway, how will this work? Do we buy bare boards, and maybe an eprom, from you and populate it ourselves? How much will the bare boards cost? What about cases and power supplies? Supplying only a portion of a complete system reduces the risk to both parties, but may result in a lot of duplication of effort among end users. Maybe the design could have a certain "recommended" enclosure and power supply in mind. |I'll be as flexible as possible. |I'm seriously considering handling the money for this whole project as a |club fund. That is, all money sent in, minus costs of what is send out |may leave a balance in the user's group fund. We can adopt a mutual agreement |among paying members that at any time the majority vote to do so, the group |will be dissolved and any balance distributed evenly among members. |I'll be keeping detailed accounting records for tax reasons (I dont want money |I may handle to look like income to the IRS), so I'll be accountable for the |groups funds in a legal sense (the treasurer). |If we agree to form a semi-official club, it would help out a lot for up |front overhead costs of the printed circuit board manufacture if we would |each give $10 or so club membeship dues, and in return you would get official |membership and the right to vote, examine financial records, etc. | |Another advantage to the offical club organization is that it avoids problems |about carrying on commercial business on USENET and avoids problems with |my employer using USENET where money is involved. |Please comment on the club organization idea. -------------------------------------- Bob Kelley: I can't stand standard phono jacks. BNC connectors? |Do you mean screw-on BNC connectors like those used in video equipment? |I was not aware BNC's were used commonly for audio. I would have guessed |balanced XLR connectors were the audio connector of choice. XLR's are |heavy, and without a supporting box, I see a problem of the weight pulling |apart the board connectors or knocking them over. I could leave holes for |circuit board mount XLR or BNC if a couple of people want it. (name which |one, not both.) You meant 44.1KHz, not 44 KHz. Right? |Right. CD rate is 44100hz. I would definitely prefer the best A/D and D/A even if the cost went up significantly. Which parts were you considering? |The higher end parts I presently know about and am evaluating are: |D/A: Burr Brown PCM54 series 16 bit. $15 | Burr Brown PCM5? series 18 bit. $??? guessing $30 for now. | Signetics TDA1541 stereo 16 bit, serial interface. $25. | Signetics TDA7220 stereo 4x oversampling digital filter, for use with the | above serial stereo D/A. This would be a kick-a** D/A setup comparable | to the best CD players available. $25. |A/D: Burr Brown PCM75 series 16 bit. $120 ea. | Burr Brown PCM78P 16bit, supposedly low cost. My data book doesn't list | this yet. $50?? have to see. | Crystal CSZ5116JD16 Self-calibrating 16 bit A/D. About $128 ea. | Crystal CSZ5116JD32 Slower (32Kh max), less expensive than above. $90 ea. | |If the prices seem reasonably low to you, don't forget by the time the whole |board is done you have a lot of other little costs that all add up too. Each |of the above has a little different interface requirements, voltage reference |requirements, clocks, power supplies, etc. So don't draw too many conclusions |about what is obviously best yet, it's not that easy. | |If anyone has any suggestions of other high-end parts to consider, and if the |prices are reasonable, be sure to let me know and I'll check it out. For |the bare bones A/D and D/A, I've got a low $ design that I don't think can |be beat for price/performance. ------------------------------------------------ Paul R. Bernard: The only problems I can foresee are hassles from Customs while trying to get it across the border. | Yes I agree, why don't you find out for me what is involved. At the end of your last message you said that you're not making a profit anywhere. I do think that you deserve something for all the work you're doing. |See my above comment on the problems of running commercial businesses on |USENET from my place of employment. I hope we all share code and have a |good time swapping sound effects and synthesis programs. That will be a |reward. 1) In my own particular case I can see a situation where I might want to connect the box to a machine that doesn't have a MIDI port. It would be nice to be able to download code from either the MIDI port or the RS232 port with the minimum of effort. I can't decide from the specs whether MIDI and RS232 would be on the same serial port or not. If so then I suppose a switch would be best. Otherwise the boot PROM could watch both both ports for data, taking the one that sends the correct SYSEX data first as the "console". | I planned on using the same serial port for both. I hadn't considered |someone wanting to use both at the same time. We can make the boot PROM |program auto-baud (detect the baud rate and set it automatically) on |reset so a switch may not be necessary, just don't try both at the same |time. Alternatively if you have a midi keyboard but no midi interface on |your computer, you could 1) connect the RS-232 (or flip a switch). | 2) Download a program, 3) connect MIDI (or flip the switch back), and | 4) 'play' the new program from the MIDI keyboard. This would mean that the | serial port software should consider checking the other baud rate if it | starts getting overrun errors. In this way, it could automatically switch | baud rates. 2) One way to get a MIDI Through without the connector is to wire it to the two unused pins on the MIDI out. I've only ever seen it used on one machine (The Atari ST). | How many people want MIDI Through? A vote! 3) I don't know if this can be done easily or cheaply but being able to select 1 of 4 different filter sections (25Khz, 32Khz, 44Khz sample rates and no filters) with jumpers or DIP switches would be more useful than having to decide ahead of time what my own personal sample rate is going to be. | This can be cheap using the little jumper blocks connecting the appropriate | set of capacitors. Buying multiple capacitors is more expensive than one set, | so if you wanted to live with one freq, you could not populate the others. | I'll try to work this in. The analog filter section tends to be a tight | squeeze on the PC board, though. 4) I'm not willing to spend additional money to get 16 bits of A/D at any sample rate, 14 bits of A/D is plenty for me. I could even live with just 12 bits if necessary. I can't decide if stereo D/A would be all that useful or not, but I certainly would not complain if the final design had stereo. For me however, the money would probably be better spent on memory. | A voice on the other side of the fence. Bare bones is for you. ----------------------------------------------- Ralph Campbell My thoughts are that I would like to do two things. 1) build the low or medium end system on PC board as you have planned. 2) Build a high end system that will plug into my multibus bassed workstation. If I have to wirewrap, I'd just as soon wirewrap #2. This way, I can experiment with something simple and see what I might want in a high end system. | You might consider taking the schematics and modifiying the host interface | to use the Multibus. Keep the rest the same. If the PC/MSDOS cross assembler is binary only, I can't use it. I also hope its in C but can convert. | I know we don't have the source for the cross tools from Motorola. My | original plan all along was to come up with a simple, portable, cross | assembler written in C to be adapted to different hosts. There should | be public domain cheap table driven assembler sources out there somewhere. | I'll be tied up with the hardware for a while, perhaps someone out there | will check into this and help make it happen. I've got the opcode summary | in the cross assembler manual when you want it. -------------------------------- Chip Chapin: As a S/W-type engineer with some H/W background, I'm looking to this project to greatly accelerate my education in programming for DSP. And I think it will be great fun. HP California Language Lab | A candidate for the above mentioned cross assembler? Maybe, Chip? -------------------------------- Brad Davis: Could the memory sites (both eprom and static memory) have sockets (or holes) for 1meg devices (128Kx8) or even 4meg devices? As you probably know the JEDEC pinouts for Nx8 devices allow for a jumper (or two) to select the proper device. This possible extra memory on the host board could expand the usefulness of the base system at small incremental cost. We could even make the jumpers and larger sockets optional, just put the address lines and holes on the CPU pc board. | The 56000 only supports 64K address space (16 address lines) times | 3 address spaces (X,Y and P) The full address decoding to allow the | options you are suggesting gets into more logic than I had planned. | Presently, The EPROM can be up to 32Kx8. Perhaps allowing more | sockets for further 32Kx8 will be sufficient. Overall, the memory | expansion board should not add more than $15+$5=$20 for the bare | PCB and connector for memory expansion. I think that is not too bad | of an incremental cost for memory expansion. As I explained way above, | for decent sample memory size (in the MB's) you will need a pageing | address register hardware and will find DRAMs less expensive per bit | for large quantities of memory. 9 pin RS-232 (on the IBM standard) is ok with me. | Sounds good. Smaller is cheaper from the PCB view. 14 bit A/D, mono (maybe stereo) slow (12khz) D/A, 4 pole (maybe 7 pole) filters. I am doing synthesis (sound effects) and so processing time and bigger storage in terms of seconds are important to me. | Agreed. Bare bones + memory expansion board is your recipe. -------------------------------- Andreas JH Gustafsson: Have you looked into using the Burr-Brown PCM78P 16-bit A/D converter? There was an ad for it in Electronic Design, August -88, pages 101 & 155, where it was offered for $39.90 a piece in lots of 100. | Good eye, I'm going to call BB on this. I don't think a full 25-pin RS232 connector is necessary, even for me who is going to use RS232 rather than MIDI if possible. | Ok, 9 pin planned for. Otherwise, the spec looks very good to me. Don't try to improve on it too much, it's more important to get the thing out before the end of the decade. |I've got a bias toward action. It's the hacker in my blood. And keep the bus loading down so that the thing will run with 0-wait-state RAM:s when they get cheap enough. |Without a backplane, I don't forsee a capacative load problem. ------------------------------------- Robert Marsanyi: Stereo would be nice, since a lot of what one does with these things can involve experimenting with imaging and location. 14-bit A/D fine, 16-bit D/A great, flexibility in filtering is appreciated. The protocol you're envisaging for booting the system sounds good - looking at the RS-xxx line as well as the MIDI connections for boot code would be nice. |All very well and good. I built a graphics environment for software synthesis that I stopped further development of until availability of of low-end DSP equipment, so I'm ready!! |Great! I really hope you get this going and make it available to us. If |if works well, I wouldn't be shy to ask a few $$ for your time per copy. |I would hope to set a precedent that for a reasonable price ($15-$25), |software will become available and others will be slightly motivated to |package up the stuff they write and send it to people. If you always |write stuff for free, it gets to drag on you when you have to make |copies for 50 people. |On the flip side, perhaps for our DSPNET group we may want to set a precedent |of sharing code at no charge, even for bigger packages. |Yet another alternative, since no one person can do it all, is to open up |a description of your graphics environment and have us each work on parts |of it as a group project. This forces nice things like device independent |screen output interface formats and file access. I would like to a few |group software projects happen as kind of an experiment in group cooperation |and group management of a loosely coupled/loosely structured organization. Saw your posting on the Well, by the way, courtesy of Carter Scholz. |Thanks for the boost, Carter. ------------------------------------------ End of edited summary spec 1.1. Will send out spec 1.2, as well as more detailed design info. F8 F8 F8 F3 F8 F8 F8 F8 F3 F*!? F8 ...... Greg Shay 7508 McKinley St Mentor, Ohio 44060 ..uunet!abvax!gfs ..cwjcc!abvax!gfs ..decvax!abvax!gfs   Received: from kestrel by gos.Ukc.AC.UK Over Ring with SMTP id aa22157; 1 Apr 89 10:43 BST Received: from mcvax by kestrel.Ukc.AC.UK via EUnet with authorised UUCP id aa20360; 1 Apr 89 10:43 BST Received: by mcvax.cwi.nl via EUnet; Sat, 1 Apr 89 11:39:37 +0200 (MET) Received: from abvax.UUCP by uunet.UU.NET (5.61/1.14) with UUCP id AA08975; Fri, 31 Mar 89 17:30:17 -0500 Received: by abvax.icd.abnet.com (5.59/1.17a); id AA12280; Fri, 31 Mar 89 16:58:24 EST From: "Greg F. Shay" Message-Id: <8903312158.AA12280@abvax.icd.abnet.com> To: dss@sun.uucp, "\\"@mcvax.uucp, dmocsny@uceng.uc.edu, csz@well.uucp, houghton@silver.uucp, dmt@mtunb.uucp, craig@scol.uucp, droid@mcrware.uucp, sullivan@asuvax.uucp, mg@ukc.ac.uk, ralph@turtlevax.uucp, jdl@cadnetix.uucp, dcj@sun.uucp, miklyd@ihlpb.uucp, peter@celia.uucp, evans@mhuxt.uucp, jjp@ektools.uucp, cab@hropus.uucp, rt76+@andrew.cmu.edu, lars@ddeswe.uucp, jdm@gssc.uucp, soleil.rob@megatest.uucp, videovax!bill@tektronix.uucp, dbell@cup.portal.com, steve@media.uucp, knudsen@ihlpl.uucp, jtn@potomac.uucp, mcripps@mtuxo.uucp, waycott@ellymae.uucp, b-davis@cs.utah.edu, gmo@tellab5.uucp, monet!defron@ucbvax.uucp, brianp@ernie.berkeley.edu, hedley@imagen.uucp, braun@thumper.uucp, brian@orawest.uucp, videovax!johnr@tektronix.uucp, barrett@ektools.uucp, dukbr@dcatla.uucp, nelan@asuvax.uucp, rmd@tellab5.uucp Subject: DSPNET status Date: Fri, 31 Mar 89 16:58:22 EST Greetings DSPNET, Rumors of my death have been greatly exaggerated. I must sincerely apologize for my recent lapse silence. As you all well know, time is our most scarce commodity, and I have had precious little in the last month. Which brings up a good point; I am probably holding those of you back who have more time than I at present to progress on the DSPNET project. At present, I have hand drawn schematics for the 56000 CPU board, for the pieces that actually had to be designed (i.e. the decoding, the MIDI interface, etc.) I have not drawn up an all inclusive schematic with a picture of the CPU, memory chips, connector pinouts, that's easy but just takes time. I'm still waiting for Motorola to send me the signed agreement saying we can share the development code. They have promised me they would send it, but maybe they are waiting for some more action on the hardware. A couple of you have offered to put schematics into CAD format, and I intend to send out copies of the hand drawn pages to them. I had intended to follow through with the printed circuit board myself. I can still do so, but again, someone else may have more time to crank it out faster. Looking at my schedule at work and home, I can't say I'll be able to have it done within the next month or two. Since we are a rather loosely knit group, it is harder to pull together and each contribute, but the opportunity to do so is at hand. I had hoped to do it all and share the results, but I can see my pace is not up to what many might like. I'll be out of town the first half of next week, so I'll not get a chance to answer e-mail until the end of next. Thanks, Greg Shay 7508 McKinley St ..uunet!abvax!gfs Mentor, Oh 44060