news@ncsuvx.ncsu.edu (USENET News System) (11/28/89)
availability
Expires:
Sender:
Reply-To: aras@ecerl2.UUCP ()
Followup-To:
Distribution: world
Organization: North Carolina State University
Keywords: VME, 680X0
From: aras@ecerl2.ncsu.edu ()
Path: ecerl2!aras
I am trying to set up a VME based system for a real-time robot control
application. I had previously posted a request for VME info. Now I have
found several large companies out there and would like to hear about any
experiences you net folk may have with them. The companies are below:
1. MATRIX: Produces VME and STD bus boards They don't have 68030's yet
2. MIZAR:
3. FORCE
4. THEMIS:
5. MOTOROLA
6. HEURIKON:
We have several special requirements such as I/O capability directly off
the processor using daughter boards with parallel and serial I/O which
reduces the list of available candidates.
Has anyone been using boards from these companies, how are their tech
support, the boards etc. Also, for people integrating imageprocessing
with VME, are there any image processing boards which down load data not
just through the VME but also through an external channel so that the
image data does not bog down the single bus?
Thanks.
Caglan M. Aras | cma@csl.ncsu.edu | Experts know more &
N. C State Univ. | aras@eceris.ncsu.edu| more about less &
ECE Dept. Robotics Lab | | less till they know
| | everything about nothing!luis@octopus.tds.kth.se (Luis Barriga) (11/29/89)
>>From: aras@ecerl2.ncsu.edu () >>I am trying to set up a VME based system for a real-time robot control >>application. I had previously posted a request for VME info. Now I have >>found several large companies out there and would like to hear about any >>experiences you net folk may have with them. The companies are below: For some time ago I used a board (I think it was from Motorola) called VME134/68020. I found some bugs in it: the dissassembler failed on some complex instructions although the board could still execute them. There was something worse: an instruction like "MOVE.L A6,16(SP)" was not executed. It was a board error since the a SUN could execute it. I've heard the they do not produce it anymore (hopefully). -- ________________________________________________________________________| Luis Barriga The Royal Institute of Technology | Dep. Computer systems (TDS) | e-address: luis@tds.kth.se S-100 44 Stockholm | SWEDEN | ________________________________________________________________________|
stevea@mcdclv.UUCP (Steve Alexander) (12/01/89)
In article <LUIS.89Nov28175600@molly.octopus.tds.kth.se> luis@octopus.tds.kth.se (Luis Barriga) writes: > >>>From: aras@ecerl2.ncsu.edu () >>>I am trying to set up a VME based system for a real-time robot control >>>application. I had previously posted a request for VME info. Now I have >>>found several large companies out there and would like to hear about any >>>experiences you net folk may have with them. The companies are below: > >For some time ago I used a board (I think it was from Motorola) called >VME134/68020. I found some bugs in it: the dissassembler failed on some >complex instructions although the board could still execute them. >There was something worse: an instruction like "MOVE.L A6,16(SP)" was >not executed. It was a board error since the a SUN could execute it. >I've heard the they do not produce it anymore (hopefully). >-- It is pretty obvious that you are confused. Motorola (my employer) makes a CPU card called the MVME134, DY-4 makes a DVME134 cpu card, Ferranti makes a DVME134 and I seem to recall one vendor making a VME1xx cpu board series, but I can't recall right now who it was. I have been using the Motorola MVME134 for a couple of years now running System V Unix and Psos. I can assure you that the 68020 on the MVME134 has no trouble executing any valid 68020 instruction. It is possible that the `problem` you refer to is in the on board firmware called the MVME134-Bug. In addition to providing primitive disk and serial line I/O facilities, there are memory dump/modify commands which include a single line assembler and disassembler. These were not intended to be used as a developers assembler & disassembler tool, but merely as a quick & dirty way to decode & modify the contents of memory. This limited ROM-based assembler does not accept SP as a mnemonic for the stack pointer. This is clearly documented on page 4-5 & 4-6 of the MVME134BUG/D1 User's manual (did you RTFM ?). I have just tried the instruction: MOVE.L A6,16(A7) on my MVME134 and it executes as expected. Unfortunately the illegal syntax: MOVE.L A6,16(SP) does not generate a syntax error and interprets the destination as location hex 16. I guess you can only fit so much software on 32KB of ROM. I should mention that the Sun assembly sytax differs in a number of ways from the Motorola assembly syntax for the MC680x0 and anyone attempting a port should consult the appropriate manuals. The MVME134 is still happily in production on Motorola's JIT Fab line in Tempe Arizona. -- Steve Alexander | Evolution is a very messy business. Motorola Cleveland |
piner@pur-phy (Richard Piner) (12/02/89)
In article <4667@ncsuvx.ncsu.edu> news@ncsuvx.ncsu.edu (USENET News System) writes: >I am trying to set up a VME based system for a real-time robot control >application. I had previously posted a request for VME info. Now I have >found several large companies out there and would like to hear about any >experiences you net folk may have with them. The companies are below: >1. MATRIX: Produces VME and STD bus boards They don't have 68030's yet >2. MIZAR: >3. FORCE >4. THEMIS: >5. MOTOROLA >6. HEURIKON: >We have several special requirements such as I/O capability directly off >the processor using daughter boards with parallel and serial I/O which >reduces the list of available candidates. > >Has anyone been using boards from these companies, how are their tech >support, the boards etc. Also, for people integrating imageprocessing >with VME, are there any image processing boards which down load data not >just through the VME but also through an external channel so that the >image data does not bog down the single bus? I put together a VME system to run an STM and process the images. Some of the details are in the October issue of Review of Scientific Inst. I used a combination of Mizar, Datacube, and Heurikon stuff. The Mizar boards were just simple I/O boards. We used their boards because they had some simple cheap boards and that was all we needed or could afford. The CPU and system came from Heurikon. Their stuff is top rate. We just got a new system from them a couple of months ago. The CPU card does it all. SCSI, serial i/o, parallel, 4 Meg of memory, etc. The system only cost $8K (+ $700 for OS9/68K and C compiler). They also have UNIX and VRTX, but they cost more. For what it's worth, I heard that Heurikon has bought OS9000 and is porting to some of their systems too. I think Heurikon can meet your needs. We bought a five slot system, and four of the slots were open. To process our images, we are using Datacubes MAX-Graph. That card plus software plus a monitor cost under 3K$. It works well, but there is still some sort of bug in the driver's interrupt handler. Datacube makes all kinds of frame grabbers, displays, convolutions, etc. The boards are connected through cables on the front panels and commands can be queued in each board. If you want to get some impressive image processing done and don't mind writting some complicated code, you can make it go real fast. Datacube's number is 508-535-6644. They support both UNIX and OS9. Hope that helps. R. Piner
sbrandt@herds.SRC.Honeywell.COM (Scott Brandt) (12/06/89)
In article <4667@ncsuvx.ncsu.edu> news@ncsuvx.ncsu.edu (USENET News System) writes: >I am trying to set up a VME based system for a real-time robot control >application. I had previously posted a request for VME info. Now I have >found several large companies out there and would like to hear about any >experiences you net folk may have with them. The companies are below: >5. MOTOROLA I have been using Motorola MVME-133xt and MVME-147 boards running vxWorks for the past two years, with a very high degree of satisfaction. The boards work fine and the support (when needed) has been very good. We accidentally burned out 8 boards the day before a demo and got replacements for them within 3 days. >6. HEURIKON: My only experience with these guys has been to discover that their benchmarks cannot be trusted. >We have several special requirements such as I/O capability directly off >the processor using daughter boards with parallel and serial I/O which >reduces the list of available candidates. > >Has anyone been using boards from these companies, how are their tech >support, the boards etc. Also, for people integrating imageprocessing >with VME, are there any image processing boards which down load data not >just through the VME but also through an external channel so that the >image data does not bog down the single bus? I work in the Signal and Image Processing department, and have been working on prototype real-time image processing systems for the past few years, and we have been using boards from Datacube with extreme success. These boards are VME based image processing modules which use a proprietary image bus for real-time (rs-170) image processing. The modules perform various functions and can be pipelined together to implement algorithms of any degree of complexity. I helped design and build a system consisting of about 80 of these cards. I would very strongly recommend that you look into Datacube products. Their software stinks, but their hardware is superb. If you would like some more information about them, I would be glad to answer any questions you may have. Scott Brandt
michaeli@stout.ucar.edu (Matt Michaelis) (12/07/89)
In article <4667@ncsuvx.ncsu.edu> news@ncsuvx.ncsu.edu (USENET News System) writes: >I am trying to set up a VME based system for a real-time robot control >application. I had previously posted a request for VME info. Now I have >found several large companies out there and would like to hear about any >experiences you net folk may have with them. The companies are below: > >1. MATRIX: Produces VME and STD bus boards They don't have 68030's yet . . . > >We have several special requirements such as I/O capability directly off >the processor using daughter boards with parallel and serial I/O which >reduces the list of available candidates. > >Has anyone been using boards from these companies, how are their tech >support, the boards etc. We are using the new Matrix MDCPU-320 for our real-time data acquistion system. The board has plug in daughterboard capability as well as a daughterboard which has 2 serial ports and 32 I/O lines. The board itself has two more serial ports and 8 I/O lines. Because of our needs for a reasonably low powered ethernet board we went this route. There is an ethernet daughterboard available that uses significantly less power than the standard CMC Ethernet board which is ubiquitous but is also a true power hog. The bad news is that we have had to be a Beta test site for Matrix. This is because we wanted the board ASAP and it has just been released this fall. We have had a few problems as you might expect from using a Beta board but that brings us to the good news. Matrix has been extremely helpful and extremely cooperative in tracing down bugs. They have shipped us new hardware Fed Ex to rectify problems. They were also friendly and helpful before we even bought the boards. We talked to their tech people before we bought anything from them and were able to get some useful information as to whether their products were what we needed. The bugs in the board have just about been taken care of. I do not know of any hardware problems that we still have with the boards. Based on our experiences with Matrix, I would recommend them to someone looking for a VME vendor. One company that I would be rather cautious of is Dynatem. They are about to come out with a 68030 board (if it is not out yet). We have had some dealings with these folks too. We have their DICB-16 serial board. This is an 8 or 16 channel smart serial card with 68HC000 processor. The board can be run either as a master or slave, but the memory is all dual ported. Thus if another board has to read the serial data out of RAM, then the 68000 just has to wait until it has the bus back. We have had lousy luck with their tech support. The guy that was trying to program the board had to abandon it due to strange hardware problems that Dynatem could not find. In fairness to Dynatem, the people who designed this board left the company and took all technical knowledge of the board with them. However, the documentation for the DICB-16 is sparse and marginally useful. Their DCPU-1 board (a 68HC000 processor board) also has lousy documentation. On the other hand, before we bought the DICB-16 we talked to someone who allready had some and thought that they were great. It is the only 16 port smart serial board that I know about. In addition Dynatem is the only company that makes low power VME boards.