Phase measurement of Synchros

[Ben · Accepted solution]

Some of those varieties were added after my day with them.

In the early days of Gun Fire Control implemented by the US Navy, digital computers had not yet been invented.

See both parts of this series on YouTube to get an idea of how they did it.

The turrets on those monsterous guns you may have seen on battleships were slaved to a Fire Control spotting system where a human kept a target in the cross-hairs and the giant guns would follow the motion as the target moved. But just pointing at the target was not enough. If you shot where a target is now, by the time the projectile gets there, the target will be somewhere else. The math involved is called :The Fire Control Problem". It is a lot of trigonometry with some Calculus and algebra thrown in.

All of the trig functions involved in solving the problem were mimicked using cams, gears, wheels, and shafts. Things like which way was the ship rocking and the target direction were sensed using syncros. While a single synchro can drive an indicator affixed to the shaft of a slave synchro, driving the gun turret required amplifier large enough to turn the guns. That is where the "torque synchro" was required.

I was initially trained to work on the old mechanical computers since most of the fleet still used them. But I did not end up actually fixing them because I was part of the first wave of Fire Control Technicians that needed to maintain digital fire control computers. it had a whopping 24K of core memory, a ASR-33 Teletype, and a cassette tape reader for loading the program ( Operating systems had not yet been invented).

Last bit of trivia

It was the Navy that first assembled a digital computer for the purposes of solving the fire control problem. The first one worked out so well they rushed through a second (I believe it was the Univac 1a) nut there were not enough vacuum tubes available to build he second. SO they replaced what tubes they could with relays. Search on "Grace Hopper bug" to read about how the term "bug" entered into the world of computers.

So your investigation into synchros goes back to the early days of modern naval warfare and a need to kill them before they killed us.

Thank you for the excuse to wonder back to my early days.

Ben

[Dan_Alias]

Ben, I'm very interested in this discussion also... I'll post more toward the end with my specific questions about my application on this topic. However, I saw your statement, "back in the day of Traditional DAQ it was possible to set the inner-channel delay. By default DAQmx will try to spread-out the sampling evenly in order to avoid inter-channel cross talk when measuring from high impedance sources. if that setting is still available, it may help." I too am familiar with the age-old ways of Traditional DAQ... and I think that setting up this style of multiplexed AI is still available in the DAQmx Timing Property Nodes but it is hidden.  You can use a combination of features (Sample Clock rate and Convert Clock rate) to mimic the Traditional DAQ functionality (which can be used to mimic simultaneous sampling even when multiplexed, but is admittedly less accurate).  In the DAQmx Timing Property Node under "More" (I did say hidden) > AI Convert > Maximum Rate (aka AIConv.MaxRate) use that to read your board's maximum Convert Clock rate.  You can then add a second DAQmx Timing Property Node and in the element, select "More" > AI Convert > Rate (aka AIConv.Rate) set the element to write, and wire the read to the write... whammo... that should set your DAQ task timing to the maximum AI Convert clock rate... Assuming that your Sample Clock rate is way slower than the maximum AI Convert clock rate, that bunches all the channel converts together as much as possible, which is the best way to mimic simultaneous sampling on a multiplexed AI DAQ device. I think they deviated from the Traditional DAQ behavior because for multiplexed AI, dividing up the sample period between the number of channels evenly allowed for longer settling time between samples (e.g. AI0 = -10V, AI1 = +10V ... large 20V slew needs more time to settle accurately on the correct voltage).  So the default behavior of DAQmx allows for greater accuracy, but is not good for phase measurements. 

I found the following DAQmx and LabVIEW help topics useful:

• LabVIEW Help "DAQmx Timing Properties" shows a huge table of all the various properties.
• NI-DAQmx Help "clocks" gives a list of all the various kinds of clocks related to AI... some not so helpful figures, but in the related topics, it has a link to...
• NI-DAQmx Help "Multiplexed Versus Simultaneous Sampling" gives a nice figure showing the Sample Clock compared to the Convert Clock.

BTW, in the end we've settled on attempting to implement our synchro application with a NI 9269 4CH AO simultaneous module... so I stopped looking into the above.  But that was what I learned before deciding to just go simultaneous... and hopefully it will help out someone who is wanting to do some synchro AI sampling with a multiplexed DAQ device and cares about phase relationship of the various channels.

[Dan_Alias]

Ah... you've hit the crux of my question.  We are simulating a Torque synchro transmitter (CG) with our DAQmx / hardware... because we are driving a Receiver synchro (CR).  This is extremely common requirement for my industry (flight simulators).  In real aircraft they have 400Hz excitation, and they have both the transmitter and receiver synchros paired... aka self synchronizing (selsyn).  But in flight simulators, we don't have the transmitter side... but we still have to drive the receiver side.  The receiver side is typically what is used to drive the needles on gauges in the cockpit to a correct angular position.  So unlike 99% of the NI folks who perceive synchro's as angular measurement devices only... there are tons of flavors and uses of synchro's, and angular measurement is just one.  Driving the rotor to a particular-specific angle is a really common and important other use case which is generally not even considered on ni.com. Again... I'll hold my question statements to the end of this post.

[Dan_Alias]

Okay... so it seems that hitting reply puts my comments at the end anyway, instead of just below the statement that I hit reply on.  sorry for the confusing 3 posts at the end.  But here's my question....

We need to simulate a Torque synchro transmitter, aka CG.  I've pulled figures from various manuals like Wikipedia, the moog synchro application handbook and edited it with my pretty colors to avoid any infringing...

As I stated previously, in the flight simulation world, we mainly need to drive needles on gauges... where the needle on the gauge is mounted on the rotor of a Receiver synchro (CR).  So we need to drive that CR to a known angle (q).  The math I've found is straight forward and easy to implement:

VS1-3 = Vex sin (θ)

VS3-2 = Vex sin (θ + 120°)

VS2-1 = Vex sin (θ + 240°)

where:

 q is the angle we are going to drive the needle to (which we figure out elsewhere in our simulation code)

Vex is the excitation voltage which is usually 400Hz for aircraft, and in our case is 90VAC, but we've got transfomers to step it down.

Our plan is to use an AI (a cDAQ NI 9221 8CH AI... not simultaneous but see previous post about DAQmx Timing Properties to mimic simultaneous sampling) and 3 AO's from a cDAQ NI 9269 4CH AO. 

The trick is that the best way to read in the Vex AI to get the excitation is probably either a continuous sample (?).  But since I know it is 400Hz, I might be able to do a buffered analog triggered acquisition on some threshold voltage... (?). OR it might be best to do a single sample, and do the math point-by-point...

Regardless of how we get the excitation voltage AI input... I then need to generate the 3 output signals.  Again, I've thought that this might be best done point-by-point... but with various transport delays (cDAQ AI to 9188 chassis to Ethernet to my PC NIC to my LabVIEW application's math, back to PC NIC, Ethernet, 9188 chassis to cDAQ AO... that may take too long.  Not sure what the turnaround time is yet as I haven't been able to purchase hardware and test).  Alternatively, I could have some buffers of sine waves built up, and then do a buffered AO (?)... or... maybe continuously output various sine waves (?).  I'm looking for advice based on any practical experience.  Does building up buffers and then changing the AO waveform cause a glitch between one period of the wave and the next?  There are a multitude of VIs available on the Signal Processing > Waveform Generation and Signal Processing > Point by Point on the functions palette in LabVIEW.  I think I can figure out how to use each tool individually, but I'm looking for some advice on a good overall approach.  So far in other posts, I've only seen people attempting to solve this in cRIO or with North Atlantic Synchro cards.  In the past we've used UEI's Synchro cards and VME Synchro cards, and this the first time we've given NI equipment a shot (at this company) and we've hit this Synchro design issue and are looking for best practices and advice before purchasing.

Thanks in advance!

[ed-64]

Although very late to this party (& don't know if you already got your answer and solved the problem, I had to comment.

First - There is NO phase shift in synchros or generated by synchros.

A Synchro is simply a transformer (single phase) that has a primary winding that you can rotate and three fixed secondary windings physically spaced 120 degrees relative to each other. When you excite the primary winding (Rx) you will see output voltages in the three secondary windings (Sx). The three outputs voltages will be at different AMPLITUDES but will always be IN PHASE with the primary. Transformers (synchros included) do NOT convert single phase AC to 3-phase AC. There is no magic going on inside a synchro. A transformer is a transformer.

If I understood what you were trying to do (and I could be wrong since I only ran across this today and did a quick read of the thread) you want to drive a (receiving) synchro electronically without using a transmitting synchro. To do that, all you have to do is to generate three IN-PHASE voltages that have the appropriate amplitudes that simulate the 120 degree spaced secondary windings of the simulated synchro.

(end of that discussion)

New topic ---- probably NOT useful for what you were trying to do.

Now - let's say that you really wanted to generate a waveform that DID phase shift relative to the primary (Rx), it can be done with extra circuitry. All you have to do is to use op amp circuits to phase shift two of the Sx secondaries by 120 & 240, then sum all three in a summing op amp. The resultant summed output will be a single AC waveform that is phase-shifted relative to the primary (Sx) that coincides with the angle of the transmitting synchro.

That probably wasn't clear so I'll describe it another way ----

S1 - no phase shift

S2 - shift output 120 degrees (op amp circuit)

S3 - shift output 240 degrees (op amp)

Now SUM the above three outputs with a summing circuit & BINGO, you have an AC output that really is phase-shifted relative to Rx. The phase angle is the angle of the transmitting rotor.

We actually did this decades ago.

[Clint1000]

I would like very much to keep this going.  As its been months since I looked at it ( waiting for hardware to be built) I have gained a little more ( very little ) knowledge of synchros.  Some of it is still confusing.  The test set is 1980 vintage and I have been tasked to upgrade it.  The test set tests a wing flap.  The flap goes up (+30 deg) to down (-30 deg).  Via hydraulics the flap is commanded to move "x" degrees.  Inside the flap is moog synchro.  The dwg for the syncho is labeled "CT" so I'm assuming its a control transmitter.if that matters.   The device is a Moog Synchro..see attached JPEG.  The old way , right or wrong, and the way I have to keep that they did it was to move the flap a "known" amount of degrees and measure the movement w the synchro.    They excite S3 relative to S1 and S2 ( they are tied together) w 5.25vac@ 400 hz.  Measurement is taken off R1/R2.( old analog circuitry)

To verify the flap is reporting the correct position an external device is placed over the flaps to measure its actual position..this too is a moog synchro.  The difference being is R1/R2 is excited w the 5.25 vac @ 400 hz and the measurement is take off S1, S2 and S3...again an analog circuit.

My question is on the external device that verifies the actual position of the flaps can I use a PXI-6229 and measure single ended voltage , say S1 rel to S2 or do I have to take 3 measurements differential and compare phase shifts relative to the input voltage on R1/R2??  If I use single ended measure what does the voltage tell me??  I know moog or North Atlantic has tables but wasn't sure I was using them correctly.  (voltage vs position )

Thxs

[ed-64]

Not sure if this will help but here you go.

Since you only have to measure 60 degress of synchro rotation, (maybe) you can get away with just measuring the voltage of one leg of the synchro (S1-S2, S2-S3, or S3 to S1). But you will have to look at each output to pick the best one.

What you might try is to first look at the voltages on a scope & choose the one who's  AC voltage AMPLITUDE stays between zero and peak AC as you cycle the measurement device through its 60 degree throw. To quickly see if you have the resolution necessary for your testing, put an AC digital voltmeter across what you are looking at on the scope an watch the numbers. If you see that the numbers are sensitive enough to measure the flap angle to your required accuracy, then take the single measurement (S1-S2 or S2-S3 or S3-S1) with the NI stuff.

Another way of looking at it is that the three Sx AC outputs of the synchro are the equivalent to having three potentiometers reporting position. All three provide an *amplitude* that changes with position. You just have to choose the one that stays between 0 & peak/max.

You again mentioned "phase". Although the term "phase" is used regularly in synchro documentation, again, there is no phase shift in/through a synchro. Only the single phase amplitudes change with shaft rotation.  If you measure/display S1-S2, S2-S3 & S3-S1 simultaneously on a scope, you will not see any phase shift. You will only see the three different ac amplitudes. Zero crossover of all three ac waveforms will occur at the same point (NO phase shift).

[Clint1000]

Thank-you!  The phase I think my documentation was eluding to was the Sx relative to the input r1>r2...no?

If the center of my flap is zero degrees and "up" is +30 deg and "down" -30 deg, using your scheme how do you determine up or down position w the S1>s2, S2>S3 etc

[Minions]

We work with three phase supplies and monitor each in a similar fashion.  I created a vi that allows you to envision the resultant rectified voltage.  That is where it veers from the LVDT conversation.  See if this helps any.  The frequency input was for future usage.  The rectified waveform below is the [(Vmax-Vmin) - Vdiodes].  Choose one of the other forms for viewing then select the input, and Amplitude.

[ed-64]

Again, there is no phase shift of anything in/through a synchro. A synchro is the equivalent of a transformer that has one primary winding & three secondary windings. The only "feature" of a synchro (transformer) is that you can rotate the primary winding because it's on the shaft. Rotating the shaft only changes the angle of the primary winding to the three secondary windings (that happen to be physically spaced at 120 degrees (NOTHING to do with electrical "phase"). So as you rotate the shaft the relative angle of the primary winding and the secondary windings change causing the IN-phase AC AMPLITUDES to change as the angles change. It's the AMPLITUDES that tell you the angle of the shaft.

As you move the flap from +30 to -30, all that the synchro does is rotate through 60 degrees. It does not know plus from minus. That's up to you to do the math to convert synchro (Sx) amplitude to the equivalent angle. In it's simplest form it could be done with a lookup table but since this is linear, you can do the math on the single voltage that you measure. Put a scope on it & take a look.

Using the table idea to think it through, it would be something like this --- (using arbitrary voltages, not yours)

(Measuring S1-S2 or S2-S3 or S3-S1 AC Amplitude

Again - you have to choose the correct one by scoping that stays bet 0 and peak/max)

.5 v = -30 deg

.6 v = -29 deg

.

.9 v = -26

1.0 v = -25

.

1.5 v = -20

2.0 v = -15

.

3.5 v = 0

3.1 v = +1 deg

3.2 v = +2

.

etc

Am I understanding correctly what you are trying to do??