Control Bus,
I like CAN bus for hardware real time control.
http://en.wikipedia.org/wiki/Controller_Area_Network
Or possibly its latest incarnation:
http://en.wikipedia.org/wiki/MilCAN
or
http://en.wikipedia.org/wiki/SafetyBUS_p
=====
For high speed data transfer Ethernet.
Every thing in the lab should be coax or better yet Shielded Twisted Pair (STP).
Data and Control busses
I would think fiberoptic would be great for a radiation environment. Plus it's fast. I don't think the transport protocols matter that much, once you've got a good physical network you can pretty much make it work. Understanding system requirements may suggest multiple transport protocols too - some subsystems may work better with a particular kind of connectivity that is poor for other subsystems. Putting glue in to some central control device then makes the whole system optimal.
My problem is that I like inventing protocols. The reality is standards are a good thing. It's just boring!! But using off the shelf stuff just makes life easier.
My problem is that I like inventing protocols. The reality is standards are a good thing. It's just boring!! But using off the shelf stuff just makes life easier.
I have actually invented a few different serial protocols myself.drmike wrote:I would think fiberoptic would be great for a radiation environment. Plus it's fast. I don't think the transport protocols matter that much, once you've got a good physical network you can pretty much make it work. Understanding system requirements may suggest multiple transport protocols too - some subsystems may work better with a particular kind of connectivity that is poor for other subsystems. Putting glue in to some central control device then makes the whole system optimal.
My problem is that I like inventing protocols. The reality is standards are a good thing. It's just boring!! But using off the shelf stuff just makes life easier.
Transport protocols are actually very important. It is not just a matter of how things work when they work. It is also about how they fail.
In a party line system fiber is a pain. It does not do party line stuff well.
If I had my druthers I'd use MIL STD 1553. However, it has two strikes against it. The eqpt. is expensive and the protocol is very complicated.
CAN is a simpler protocol and the hardware is cheaper.
In addition 1553 can't be slowed down to extend the length of the bus. CAN can.
The MIL CAN std is a time division multiplexed bus. Which would make it similar to 1553 in operation. The CAN bus is fully optically isolated. 1553 is transformer isolated.
I did one of the first CAN testers for aerospace in '97. Lots of fun.
The most important thing for control is determinism. Ethernet doesn't have that. It is a crash bus. At high loads packet delivery is very non-deterministic. Even at low loads collisions can occur. Death for real time. Not bad for data collection if the nodes have a nice buffer.
BTW wireless is absolutely out. It is nondeterministic and subject to shut down from enough noise.
I like it, though im not a fan of fiber for this applicationdrmike wrote:I would think fiberoptic would be great for a radiation environment. Plus it's fast.
If i was building the machine i wouldnt use anything less than mill spec connectors, wire and crimps, straight from an avionics shop. Bucket loads of bulkhead connectors designed for pressurized environments, high EM shielding and can be purchased second hand at a fraction of the price and re pinned. Wire is expensive but i can get it at cost and would'nt trust anything less. If WB6'S coils had TEFZEL insulation it would most likely still be here with us today.
MSimon you have been doing some great work on the controll side of things
keep it up. Its good to know your so fluent in this complex mission critical area of the machine.Ive been very busy doing calculations on the main polywell torus coils and their power supplies trying to figure out how to bring them to life and drive them in the real world. Look foward to posting some meat soon.
Purity is Power
Keegan,
Thanks. I have been fortunate to have a career that has spanned a number of industries including nuclear power, commercial computers (clone wars), and mil/aerospace.
BTW just ran into this in RTC magazine http://www.rtcgroup.com
A CAN to fiber converter http://www.ixxat.com
http://www.ixxat.com/index.php?seite=can_cr210_en
CAN-CR210/FO
So there is always that if STP does not serve.
===========
Your point about good wiring and connectors is excellent.
In a test set up it is always wise to go top of the line even if you have to pay full price. So many unexpected things come up. It is wise not to have to fight equipment deficiencies.
I'm not familiar with the kinds of wire preferred these days in nuke installations. Time for some research.
When in doubt we can go full Mil and use Teflon. I like the feel of it. However, it is best to use the heat strippers. Mechanical stripping tends to nick the wires or leave bits behind.
However, consider this is being done for the Navy. They might insist on Mil quality for the full up reactor. With appropriate waivers. i/e. Mil wiring with Mil eqpt. substituted for commercial as the Mil black boxes become available.
I could easily see a 3 year crash program to get this installed on a ship if Dr. B.'s experiments work out in the coming months.
I was looking up the weights of some shipboard nuclear plants and they run from about 1,500 to 2,500 tons. Think of what a boost it would be to reduce that to 100 to 200 tons. Or even 500 tons. Some of the left over space could be converted to crew morale features and he rest for more stores. Plus the turbine room could be fitted with more water tight bulkheads for better battle damage survivability.
Thanks. I have been fortunate to have a career that has spanned a number of industries including nuclear power, commercial computers (clone wars), and mil/aerospace.
BTW just ran into this in RTC magazine http://www.rtcgroup.com
A CAN to fiber converter http://www.ixxat.com
http://www.ixxat.com/index.php?seite=can_cr210_en
CAN-CR210/FO
So there is always that if STP does not serve.
===========
Your point about good wiring and connectors is excellent.
In a test set up it is always wise to go top of the line even if you have to pay full price. So many unexpected things come up. It is wise not to have to fight equipment deficiencies.
I'm not familiar with the kinds of wire preferred these days in nuke installations. Time for some research.
When in doubt we can go full Mil and use Teflon. I like the feel of it. However, it is best to use the heat strippers. Mechanical stripping tends to nick the wires or leave bits behind.
However, consider this is being done for the Navy. They might insist on Mil quality for the full up reactor. With appropriate waivers. i/e. Mil wiring with Mil eqpt. substituted for commercial as the Mil black boxes become available.
I could easily see a 3 year crash program to get this installed on a ship if Dr. B.'s experiments work out in the coming months.
I was looking up the weights of some shipboard nuclear plants and they run from about 1,500 to 2,500 tons. Think of what a boost it would be to reduce that to 100 to 200 tons. Or even 500 tons. Some of the left over space could be converted to crew morale features and he rest for more stores. Plus the turbine room could be fitted with more water tight bulkheads for better battle damage survivability.
The Navy is starting the transition to all electric drive for surface ships. Superconducting motors.
Think of being able to feed power to the motors such that the props are always just short of cavitation. Tremendous acceleration and maneuverability gains. Plus you eliminate the shaft alleys which are a below the waterline unsegmented compartment liability.
Polywell, unlike fission/steam turbine plants. should be able to control the output power on millisecond or less time scales. Over at least a 100 to 1 range.
I could see the Naval power plant guys salivating to get one of these. To start. Then it is outfit the entire fleet ASAP. Naval architects have one central element that all ships must be designed around: "the tyranny of the drivetrain."
The guys with the hardest job re:retrofit will be the reactor shielding guys. How to keep the shielding up and the weight down.
Here is a bit on electric ships and superconducting motors.
http://www.memagazine.org/backissues/me ... /hunt.html
Think of being able to feed power to the motors such that the props are always just short of cavitation. Tremendous acceleration and maneuverability gains. Plus you eliminate the shaft alleys which are a below the waterline unsegmented compartment liability.
Polywell, unlike fission/steam turbine plants. should be able to control the output power on millisecond or less time scales. Over at least a 100 to 1 range.
I could see the Naval power plant guys salivating to get one of these. To start. Then it is outfit the entire fleet ASAP. Naval architects have one central element that all ships must be designed around: "the tyranny of the drivetrain."
The guys with the hardest job re:retrofit will be the reactor shielding guys. How to keep the shielding up and the weight down.
Here is a bit on electric ships and superconducting motors.
http://www.memagazine.org/backissues/me ... /hunt.html
Why would you want to change the power output of the generator? Isn't there a way to vary the power consumption of the motor and keep the powerplant constant?Polywell, unlike fission/steam turbine plants. should be able to control the output power on millisecond or less time scales. Over at least a 100 to 1 range.
No.JohnP wrote:Why would you want to change the power output of the generator? Isn't there a way to vary the power consumption of the motor and keep the powerplant constant?Polywell, unlike fission/steam turbine plants. should be able to control the output power on millisecond or less time scales. Over at least a 100 to 1 range.
Electrical power generated has to be consumed unless you have a place to store it.
OK, I don't know how big boats are wired up and I'm assuming it works like a commercial power grid - the power co monitors consumption and makes gradual changes but usually does not turn down the power to force everyone's lights to dim.
So, on this electric boat the polywell generator's tied directly to the electric motor and other things on the boat (lights, general wall power, instruments, etc) are run off another generator or batteries or ???
So, on this electric boat the polywell generator's tied directly to the electric motor and other things on the boat (lights, general wall power, instruments, etc) are run off another generator or batteries or ???
DeviceNet
DeviceNet is a deterministic protocol that can be implemented through an ethernet cable. Much of the vacuum system instrumentation (MFSs, temperature controllers, ion source power supplies) come with devicenet interfaces. You can also get DeviceNet interface modules for most of the major PLC brands (Yokogawa, Omron, A-B, etc.).
DeviceNet is fast enough for all of the vacuum process applications that I am familar with.
DeviceNet is fast enough for all of the vacuum process applications that I am familar with.
DeviceNet and CAN
Yes, DeviceNet uses CAN as the physical layer. DeviceNet is the communication protocol that runs on CAN and is starting to become popularly used in semiconductor and other high-vacuum process tools. The advantages is that much of the control and instrumentation such as power supplies, MFCs, and vacuum gauges are now coming in DeviceNet versions to make implementation easier.
PLCs are best for the automation control. They are robust and deterministic real time. The only other option is to go with a RTOS and control software and implement it on industrial hardware like PC-104 or the like. PLCs are a lot easier to work with and have become much faster in the past 10 years.
You can get 16 bit resolution PID loop controllers from Eurotherm and Yokogawa. The PID modules that go with PLC systems are typically 12 bit resolution, but some are 14 bit.
The polywell device that you guys are trying to make is going to create huge amounts of EMI. Perhaps an optical bus for the physical interconnects is in order here.
PSII (plasma source ion immersion) is a PVD thin film deposition technique that utilizes rapid switching of high voltages (30-50 kV, 1kHz switching rate) to generate a pulsed bias for implantation effect. A search of PSII literature on the net may suggest ways of handling the EMI problem.
It seems to me that the electro-magnetic polyhedral coils that comprise the WB6 and WB7 will need active water cooling. The company I worked for manufactured ion-plating systems that used water cooling of the plasma source, target, and electromagetic coils. And our ion plating systems used a lot less power than these polywell devices.
PLCs are best for the automation control. They are robust and deterministic real time. The only other option is to go with a RTOS and control software and implement it on industrial hardware like PC-104 or the like. PLCs are a lot easier to work with and have become much faster in the past 10 years.
You can get 16 bit resolution PID loop controllers from Eurotherm and Yokogawa. The PID modules that go with PLC systems are typically 12 bit resolution, but some are 14 bit.
The polywell device that you guys are trying to make is going to create huge amounts of EMI. Perhaps an optical bus for the physical interconnects is in order here.
PSII (plasma source ion immersion) is a PVD thin film deposition technique that utilizes rapid switching of high voltages (30-50 kV, 1kHz switching rate) to generate a pulsed bias for implantation effect. A search of PSII literature on the net may suggest ways of handling the EMI problem.
It seems to me that the electro-magnetic polyhedral coils that comprise the WB6 and WB7 will need active water cooling. The company I worked for manufactured ion-plating systems that used water cooling of the plasma source, target, and electromagetic coils. And our ion plating systems used a lot less power than these polywell devices.
Kurt9,
I actually favor LN2 cooling. It reduces supply power and voltage. It keeps everything cool. It doesn't ionize. It doesn't separate into an explosive mixture under reactor flux conditions.
It costs $1 a liter. To run a reactor with a total loss system takes 400 liters a minute. If you evaporated 1/5 of that to cool the other 4/5s you could run the reactor for 5 minutes.
Now this would be for an experimental reactor.
For a full size machine it would be superconductors all the way.
I actually favor LN2 cooling. It reduces supply power and voltage. It keeps everything cool. It doesn't ionize. It doesn't separate into an explosive mixture under reactor flux conditions.
It costs $1 a liter. To run a reactor with a total loss system takes 400 liters a minute. If you evaporated 1/5 of that to cool the other 4/5s you could run the reactor for 5 minutes.
Now this would be for an experimental reactor.
For a full size machine it would be superconductors all the way.
Am I wrong in reading the MgB superconducting threadMSimon wrote: I actually favor LN2 cooling. It reduces supply power and voltage. It keeps everything cool. It doesn't ionize. It doesn't separate into an explosive mixture under reactor flux conditions.
viewtopic.php?t=285
as an endorsement of that technology? It would seem to be, based on the temperature of superconductance (39K for MgB superconducting, vs ~77K at the freezing point of nitrogen). Endorsing liquid nitrogen cooling would seem to be at odds with MgB superconducting magnets.