Mike Powell

The prototype DTS2 circuit boards are here. Most of the parts are here as well. What I'm most interested in doing first is simply generating two, 400 Hz output to show that the key concept of this approach is viable. I'll initially leave the power amplifiers off the board so I can debug without fastening the board to a heat sink. It's been awhile since I developed PIC code, and of course the IDE has changed so an early step for me will be reading the Microchip MPLAB X IDE getting started guide.

Not bad, period!

I've also had good luck with TurboCAD. I started with it 12 years ago or so because it was an inexpensive tool for producing EPS files used in dead tree book publishing illustrations. It's since proven to be a useful 2D/3D modeling tool. Most recently I use it to produce DXF files feeding CAMBAM to make G code for CNC operations.

The choice of the PIC was largely driven by familiarity and the fact that I have a number of them on hand. I've been using them for 12 years or so, but there is certainly no reason for someone not to use another type. The availability of a zero (?) cost programmer for ATMega certainly is attractive. A PIC programmer can be had for about US$50 (the PICKit3), but free does have its appeal. (I've built several programmers, but the low cost PICKit with its support for new devices as they were released won me over pretty quickly.)

Hi Peter I use the PIC to tightly control the timing of the output waveform. The PIC will be busy in a loop loading data into the DACs at precise intervals to generate a 400Hz signal with suppressed lower order harmonics. I could probably get reasonable performance from an Arduino if only one or two of the DACs were used, but too many would result in waveform jitter that would eventually turn into random movements of the instrument indicator. If there were a lot of instruments this might mean using a lot of Arduinos which in turn means more USB connections to the PC. With the PIC handling the low level busy work servicing the DACs, a single Arduino can interface the bulk, if not all, of the instruments. Is this the best tradeoff? In truth, I don't know. There are too many variables for me to wrap my head around. (Might be a good experiment for someone.) It is certainly a conservative tradeoff, and at ~$2 not a terrible one if I got it wrong. NOT a dumb question. Thanks for your interest.

I envision the synchro interface card as a dumb peripherial for an Arduino. It might make sense to dedicate an Arduino to managing fairly large clusters of instruments depending upon just where you want to place the processing load of ancillary tasks. For example, the flight sim engine on the main PC may not be able to supply instrument updates fast enough to produce smooth movement of the instruments. (I know this has been a problem with a Falcon derivative.) Jerky motion of a ADI or HSI is seriously distracting. Arduino code may be able to resolve this.

Several years ago I built a synchro interface that took an RS485 serial control input and generated synchro stator signals as outputs. The goal of this project was to allow real aircraft, synchro-based instruments to be used in sims. You can read about the project toward the bottom of this page: http://mikesflightdeck.com/oldnews/oldnews_2006.html Over the past few years technology has changed and there seems to be an increased interest in synchro interfaces so I decided to take another look at my old project and perhaps update it. I saw that the price of a key component of my original design had ballooned. This seemed particularly important to me because there is never a need for just one synchro interface. If you're committed to using real A/C instruments, you'll have a lot of synchro inputs to drive, so low cost is worth pursuing. I also saw that Arduinos had become a common bit of interface glue in even modest simpits. Given their ubiquity, low cost, and capability, utilizing them seemed a no brainer. A key functionality of any synchro interface is the generation of a pair of AC voltages which can be set to various desired amplitudes. For A/C instruments these voltages are nominal 400Hz signals and they are both electrically in phase with the 400Hz power supplied to the synchro instrument. The more accurately the amplitude can be controlled, the smoother the motion of the instrument indicator. For something like a 2 inch engine gauge 7 or 8 bits of amplitude control gives good results. For something like a 5 inch ADI, you'd really prefer something north of 10 bits. A common approach to designing a synchro interface is to use a multiplying DAC, a digital potentiometer, or a voltage controlled amplifier to scale a sample of the 400Hz power. This produces the variable amplitude, and guarantees that the outputs are in phase with system power. Works quite well, but I was looking for a less expensive approach. I decided to generate a pair of amplitude controlled, non sine waves under control of a cheap micro controller, a PIC16F648A (~$2) using a dual voltage digital to analog converter (~$3). Synchros will operate from a wide range of nasty, noisy electrical signals, but I thought if it didn't cost too much I'd add a filter to suppress all but the 400Hz fundamental of the synchro signals the circuit generates. Two individual, four pole filters can be built with a single quad opamp (~$0.40) and a smattering of discrete components (~$0), so adding the filter seemed worth it. Depending on the instrument being driven, several watts of synchro signal might be required so a bit of power amplification is needed. A pair of LM1875 20/30 watt amps (~$3 each) handle that. The PIC micro controller receives its command across a serial com connection. This can be an RS485 physical link if long cables are used, or a TTL input if the commands come from an Arduino a few inches away. The PIC can synchronize with the 400Hz power, or it can send a 400Hz inverter drive signal to another amplifier board capable of generating power for the A/C instrument being interfaced. Here's a block diagram: You'll note that one of the synchro signals is grounded. This is almost guaranteed to generate some angst ridden comments, so here's the explanation: Synchro signals are defined as the difference between pairs of inputs. In an electrically noisy environment with long cable runs and poorly managed electrical grounds, keeping the synchro signals isolated from ground reduces the chances you'll introduce noise into your system. In a sim you're unlikely to face problems quite this bad. In fact, the most likely issue is the possibility that the instrument itself has an internal ground. In that case you'd need to assure the ground in the instrument was connected to the ground on the interface. And in the worst case, you could simply add a pair of isolation transformers to the outputs of the interface. The circuit board will look something like this: (Though not exactly...There's a mistake which I caught just before sending the files out.) Currently parts are dribbling in, and the prototype boards should arrive toward the end of the week. The next big task, aside from soldering a pile of SMD parts on the board, is dusting off some archaic PIC assembly code and adapting it to its new home.

The data for the Switech X-25 series is similar. A 200 Hz driving frequency is the maximum without the use of acceleration and deceleration sequences. If you use accel/decel you can build up to a 600 Hz driving frequency. However, this is only true when the motor is spinning without the mass of a pointer. Once you add the pointer the rotational inertia increases and the dynamic limits shift downward.

Wicks Aircraft carries MS-35266 slotted phillister head 10-32 screws. Not quite what you're looking for, but being a milspec part, it might have the head diameter you need. http://aircraftproducts.wicksaircraft.com/item/slotted-machine-screws/ms35266-fillister-head-steel-screw/ms35266-65?# Also take a look at SkyGeek http://www.skygeek.com/military-standard-screws.html Random stuff, somewhat more reasonable quantities...

Nothing quite so frustrating as needing an easily available, inexpensive item that comes with a minimum order quantity of 5,000. You have my sympathies. If you find a supplier of smaller quantities, please let us know.

Sweet! Looks like you've got the key elements for pulling you into the sim.

Looks like you're well along to creating a very immersive flight sim. Keep the progress pix coming.

@John I was responding to this post about resistors. Caps present another possibility with their own caveats. Sorry, should have included the quote in my original post.

Placing a resistor (or two) from the wiper contact to one (or both) end of the pot will indeed reduce noise from the pot IF the noise is due to poor or intermittent contact of the wiper. However, there are any number of other ways electrical noise can sneak into a circuit. Adding resistor(s) this way unfortunately also has the side effect of changing the linearity function of the pot's output. Might be best to buy a high quality pot to start with.

Hi Chad, These links may help: http://www.scotthendry.com/hsi_project.htm http://forums.eagle.ru/showthread.php?t=142373 http://mikesflightdeck.com/instruments/servo_instruments.html I seem to recall someone over at Viperpits http://www.viperpits.org/smf/index.php using a small LCD to create small complex instruments like HSIs and ADIs, but I can't remember who. Very few people attempt such projects. As you build your, please post lots of pix. Cheers

Hi John, My knowledge of what stepping motor models are readily available is unfortunately dated. I saw your post and thought I could easily recommend a few options, but after visiting a few of my favored surplus providers and finding no stock, all I can do is offer general thoughts. My preference is to not drive a motor directly from a microcontroller. The thought of switching an inductive load that way makes me cringe. I know that instrument movements like the VID series can work okay, but moving up to a motor with higher torque gets increasingly dicey. Using a driver protects the microcontroller as well as opening up the possibility of using motors with a variety of voltages. A quick look at Ebay turned up a NEMA 14 hybrid stepping motor with a 400 step per rev rating. This can be easily half stepped to deliver 800 steps which will give you 0.45 degree steps if you mechanically direct drive the roll motion. You also get the advantage of using the motor bearing to support the gauge mechanism. [Hybrid stepping motors are those robust motors used in small CNC machines. You can see the motor laminations on the sides. Hybrid motors come in a variety of sizes. A NEMA 14 is 1.4" inches square.] If you don't use a hybrid motor, you're most likely left with selecting some type of "can stack" motor. These have two sections that look like steel cans spot welded together. They are generally smaller and cheaper than hybrid motors, and they offer much larger step sizes. A 15 degree step is common, some offer a 7.5 degree step. To get smooth motion of your instrument you'll need to use gearing. Half stepping a 7.5 degree motor and using a 5:1 gear reduction would delivery fair results in a small instrument. I got fair results using Delrin gears. I was also lucky enough to find motors that already had a small pinion gear on the shaft. I notice that there are a few offerings of stepping motors on Ebay with reduction gearing. These may work, but a potential problem is that the backlash is not specified. I believe I read somewhere that some of these geared motors were originally designed to operate valves. A few degrees of slop in turning a valve may not be an issue but would probably be objectionable on an A/C instrument. Personally, I would lean toward a 400 S/R NEMA 14 hybrid motor. If you decide to go with a can stack motor and gearing, I think I have a couple of small ones with pinion gear in my junk box which I could send your direction. And, BTW, that's a really nice looking project.

The 2N7000 and VN2222 are excellent discrete switching FETs for driving LEDs. No gate resistor or heat sink needed. A 2N7000 cost .07~.08 in small quantities from Jameco.

As far as I know, magnetically held switches are hermetically sealed, individual toggle switches like these Honeywell units: http://www.digikey.com/catalog/en/partgroup/et-series/26208

I strongly recommend the use of a robust signaling approach like RS-485/422. It may be overkill in many situations, but its use covers a multitude of sins. As the total system grows it becomes susceptible to all kinds of random problems. (Works great until the daughter uses her hair dryer/neighbor's garage door opener operates/etc.) Any time you have a system with lots of wiring, uncertain grounding and shielding practices, multiple power supplies, residential AC power, etc. you will eventually be chasing intermittent problems that detract from the fun stuff. Gadroc's ideas have merit.

Outstanding

Using the analog input with a potentiometer is a clever solution. Going forward with this idea... Rather than adding detents to the potentiometer, you might use a mechanical rotary switch with fixed resistors forming a voltage divider which acts electrically like the potentiometer.

This is all too true. The Momus design allows room for builder choices. If the machine is built with 200 steps per revolution motors and 8X micro stepping is used then the tool step interval is 0.00125". But this alone doesn't tell you how well the machine will perform. There are other factors such as backlash and spindle run out. Overall this design seems to be well thought out. There is a balance between the rigidity of the structure and the positioning accuracy of the tool. There are a number of DIY CNC designs that appear to offer 0.0001" accuracy or better, but sit on a framework that you just know will flex a quarter inch once a significant load is placed on the spindle. The Momus design avoids this. Momus builders have posted videos on the CNCZone support forums of projects they have done using this machine. It apparently works very well with wood. It's also been used to mill aluminum with what appears to be very good results. So, how well will this design meet the needs of hobby sim building? In truth, I don't yet know. It appears to be one of the better, if not the best, DIY CNC machines in its class. (And, yes, I know I used a bit of weasel-wording there.) I chose it as an entry into CNC based on the quality of the plans and on an expectation that the result would quite likely be useful in sim building. When I finish some other project work I will complete the machine, run some tests, and let you know if I've been disappointed. Progress pictures on my Momus CNC are here: http://mikesflightdeck.com/oldnews/oldnews_2012.html and here: http://mikesflightdeck.com/oldnews/oldnews_2013.html

I think I have about $1200US in it now. I bought the motors and electronics from CNC Router Parts for about $540. I think you could find less expensive alternatives on Ebay. I've bought no software as yet. I will at least initially use LinuxCNC which is free.

If you do decide to build a CNC machine, I suggest you look into http://momuscnc.com/ I'm building one now, and I've been impressed by the thought that went into its design. There is also a well populated builder's forum at http://www.cnczone.com/forums/momus-design-cnc-plans/

Using bar stock results in a very realistic looking structure. It's time consuming, (I cut everything by hand with a hacksaw) but not particularly difficult. I did it mostly to see if it could be done that way. I particularly wanted to recreate the multiple levels of the instrument placement. This approach probably wouldn't be a good choice if the MIP sits in front of a monitor. I got most of the metal from OnLineMetals, so not "too expensive".