Roman Quadrireme Galley by Ian_Grant - 1/32 Scale - RADIO

[bobandlucy]

Very interesting project! And like you, although I became proficient in AutoCad at work as a designer, I always missed the days of drafting on paper- so much more satisfying somehow. . .

[Ian_Grant]

Worries, worries, worries. One step forward and two back. I noticed the oar beam assembly was getting pretty heavy, indeed someone warned me about it in an earlier post. I tossed and turned one night, worried about whether the u/w hull (whose volume I do not know) would displace enough water to support the galley's total weight without sinking too far thus reducing my already paltry freeboard. Ideally it would float above design waterline then I could add ballast. The waterline should bisect the ram.

 

I realized that now that I have the bellcranks moving each end of the beam up and down, a simple dado'ed wood block at each end to allow movement, like in my prototype jig, might suffice.

 

Also, the entire thing was a bit wobbly side to side. The two 17" steel drawer slides have to have some "breathing" allowance in order to absorb tolerances if using them in a drawer scenario. I noticed that they, although 5 inches longer than the 12", only have one extra inch of actual ball-bearing cradle. I decided to replace the 17" slide under one side of the channel with two 12" models. This helped considerably. I didn't replace the other side since there are mounting bolts for the bellcranks etc in the way and I didn't want to change everything.

 

Here is a pic of the new structure, with the old beam and attachments in the background. The weight of the removed metal parts is significant. The most significant factor in the assembly's weight is now the steel drawer slides which are about at waterline. I want to keep the aluminum channel at the base because it will never warp or twist.

 

The up/down movement works well even without my patented packing-tape "lubrication" on the beam. The beam can be thinner/lighter since I am no longer screwing into its ends to attach metal plates. It's flimsier laterally, but the hardwood strips for oar attachment will keep it straight. On the right end is a sweep servo, still just a regular type not the "giant" which will be in the boat.

Now that it has occurred to me I am still worried about displacement. I plan to weigh my battlecruiser, which as I said is a comparable size to the proposed galley, both empty and ballasted to waterline (have to borrow a neighbour's pool since the falling tree destroyed ours). Assume the empty galley weighs about the same as the empty battlecruiser: if this additional ballast amount is less than the weight of two oar drives then I am in trouble. Each 1/4" added to the galley's draft would give on the order of 60 cubic inches additional displacement, or about 2.2 lb. I may have to carve a replica of the lower hull to measure its displacement. Or I just add 1/2" to the draft. Bedford commented earlier that maybe I could "square off" the bilges to create more buoyancy; that may be an idea. But the only galley lines I have are for the "Olympias" and I don't have the 3D tools knowledge to make serious changes. But there's always prototyping hull lines with card bulkheads 🤨.

 

In the meantime, on to oar-making! I've given up on the 3D print idea, far cheaper to just use wood.

Edited November 7, 2022 by Ian_Grant

[Ian_Grant]

Beautiful weather, warm days and cool nights. Perfect for cycling along the canal. Despite this I have managed to make a start on the oars with the aid of some sophisticated jigs 😆.

 

Problem one was to cut a slot centred in the end of the 1/4" shaft, in which to insert a blade. I got a thin-kerf blade for this. Below is the jig; first three holes were drilled during hole/kerf alignment. Labelled three holes are to hold the shafts, in whichever suits since dowel varies a little. Just press the dowel down until it hits the plywood layer at the bottom (hidden, and which is not drilled through).

Next problem is to drill a centred 1/8" hole at loom end of shaft. Jig consists of a piece of wood clamped to the drill press table. Drill a 17/64" hole through it, then change bit to 1/8". Insert shaft into hole from below and drill to depth.

Next chore - cut consistent chamfers at handle end of blades. Another sophisticated jig.

Here are shaft and blade blanks. There are about 100 blades (more than enough) but I'll need to make more shafts.

Finally, a pressed-together oar, so far. The screw eye is soldered into a bit of 1/8" brass tube which will be epoxied into the shaft hole. This extension will prevent the edge of the oar mounting strip on the beam from interfering with the shoulder of the wood shaft at rotation limits. The brass ferrule is pressed on to prevent cracking of the shaft. You can see I added some length to the blades compared to the drawing since they were just too stubby-looking, while maintaining the blade-to-loom pivot length. This blade is the same size as I tested for sweep pull force before; I had reduced it in the drawing to keep overall oar length at what Pitassi states. But now I'm OK with lengthening it slightly.

Now wondering what to do about the squared shaft ends in middle of blade. I tried sanding bevels on one test piece but couldn't get even those two edges consistent, never mind over 88 oars. Might just leave or might do another jig.

 

Edited August 15, 2022 by Ian_Grant

[Ian_Grant]

Forty four oars (enough for one side) in production. Here are assembled shafts, blades, ferrules. Blades are from cherry cutoff I had lying around which will be beautiful when varnished, but shafts are plain "hardwood dowel" as sold at Home Depot etc. Lee Valley has nice cherry dowel but only 1/2" and up. Might try to stain shafts.

 

I must have miscalculated how many strips to cut up for blades; I still have 77 left to choose from, for 44 more oars (plus some spares).

Now soldering screw eyes into short brass tubes. Tedious cutting the brass. Once finished these will be epoxied into the holes drilled in the loom ends using a simple jig to keep consistent projection of the eyes. Then all can be mounted on the oar beam for a trial of two full remes rowing, controlled by the arduino. But this exciting test will have to wait until we return from holiday in Hawaii.

[Ian_Grant]

Continuing with test build for two full remes.

 

I used TinkerCAD to sketch a section of the side of the ship, 1/16" thick, pierced by four cylinders representing an oar shaft at the four extreme positions of the oar. Goal is to find minimum size and shape for thole hole which allows rowing without being too much of an entrance for water. Here is a screen image.

Then I change the cylinders from "Solid" to "Hole" and "Group" the lot. This leaves the red plate with the ensuing hole.

I drew two cases, for the lower and upper remes, since the angles are different i.e. the lower oars' vertical range includes horizontal, whereas the uppers are angled down whether in or out of the water. Here is the result:

The shaded ellipses at the bottom by my judgement are more or less equivalent to the cutouts in the plates. You can see that the upper reme requires a taller opening, since these oars pass through at a steeper angle.

 

For the lower reme, a suitable "ellipse" can be formed by using a 7.5mm (0.295") brad point drill bit to drill a pair of holes offset by 5/64" (gives an oval 0.295" x 0.373"). My old eyes had trouble seeing the 64ths marks on my ruler so I just used a slightly heavy 1/16".  Using a jig on the drill press I was able to drill a nice line of 22 of these ovals in a strip of 1/16" plywood, which is what I plan to use at this location on the actual model.

 

This strip was then mounted on the baseplate of the jig, with height over waterline, and spacing from the mechanism, matching the planned values for the model. Once this was in place, the lower oar attachment strip was mounted on the beam such that the fully down beam position yielded horizontal oars, and all 22 oars were attached. Here is a little video. I just used the 2-channel RC set to move in each dimension. We're leaving on vacation tomorrow and I don't have time today to hook up the Arduino, get myself back up to speed on the software, and alter the servos' movement ranges to suit the new mechanical setup.

 

 

The beam only has to move up and down by 3/8" to move the oars from in the water (measuring against the blue surface level) to horizontal. This is exactly as calculated in earlier drawings, thanks to the reduction in loom length to 1.5" from 2".

 

I mention that the oars have a lot of play in the sweep direction. This could be problematic in terms of having another set of oars from the upper reme interlaced with them at only 5/8" spacing. It would be ok during the power stroke because all oars would be pressing on the forward ends of their thole holes, and they'd all be aligned nicely, assuming I drill the thole holes and mounting screw holes accurately. But once lifted into the air on the return stroke they're liable to be flopping all over and perhaps hitting each other. My first reaction was, "Damn, I'm going to have to add thole pins and strap the oar shafts to them like in the real thing". I'm reluctant to get into this, with the prospect of strings breaking and continual upkeep with poor access. All I really need is some way to "gently" continuously nudge the shafts to the forward ends of their holes, where they will naturally place themselves during the power stroke. I thought of lashing all the oars in a reme together just inside the bulwark with a rubber band tensioning the forward end; doesn't sound too practical either. Perhaps a rubber band on each oar? Nah! Then I thought of having a rubber o-ring placed just aft of each oar, with its aftermost quarter siliconed to the inside of the bulwark, or a rib of the model, and its free end just pushing the oar shaft against the forward end of the thole hole. When the shaft has to pivot aft for the return stroke it can just flex the o-ring into a vertical ellipse. There wouldn't be appreciable additional stress on the sweep servo since the o-ring is right at the pivot point. Hmmm, maybe.

 

It has now been a whole year since I started farting around with this 😒, albeit I spent much of last winter on rigging  another model. The Admiral must be sick of hearing about it. I really really hope I can start on the actual hull this winter. Just want to add 2nd reme to this jig and satisfy myself that I can make a bireme work, whether with o-rings or some other idea. (monoreme woudn't be an issue, perhaps that's why videos I see are all monoreme RC models??).  Or just maybe I will add the 2nd reme and there won't be a collision issue 😬. One can hope.

 

Thanks for reading!

Edited June 2, 2023 by Ian_Grant

[rookie]

Looks great!

 

If it does not work out, you may have a future as a hand-model     

[Ian_Grant]

I hooked up the Arduino to see the entire reme rowing. There was bit of a rattle when moving up and down at the ends of the stroke which I attributed to the loom eyes shifting along the flat head screws mounting them to the oar strip (the screws had to be left loose to allow the oars to pivot up and down since the oar strip's upper surface remained horizontal as the beam moved).

 

The oar strips were changed to the hinged design, allowing the loom screws to be tightened as theoretically the looms only need to swivel on the strip, not tilt up and down.

 

Here are some pics of all 44 oars in both remes installed with the new strips. In the second one you can see the strips tilted, with one of the hinges quite visible. I like the third one which hints at how the galley hull will look with the oars emerging; of course the actual hull will have much nicer brackets supporting the outrigger. I'm happy that the angles all seem to match my revised drawing of oar geometry.

What I need to do now is determine the range limits of the servos, in terms of their control pulse widths in microseconds. For this I need a little test program, and for that I need an Arduino book again because I need to use a function I have never tried. Again, I find the on-line Arduino help not very helpful at all; I much prefer to have a physical book I can leaf through. Back to the library........... Once I know the servo ranges I can modify the code and see the whole shebang row 🤪.

 

On another topic I borrowed a cat weigh scale from my wife's vet clinic in an effort to determine what kind of volume the galley's lower hull will need.

 

My old battlecruiser "Lion" mentioned earlier in this log is about the same length and beam as proposed for the galley. It weighs 4.16kg empty (ie no motors, battery, or fixed ballast), 6.85kg ballasted to waterline. My calculated estimate of her underwater displacement, by dividing the hull into rectangles and triangles on paper, was very close to this.

 

Let's assume the empty galley hull being of similar wood construction weighs about the same as the Lion, say 4.2kg. The oar mechanism with oars and servos weighs 1.87kg, two thus weigh 3.74kg. Galley hull plus oar drive weighs about 8kg. Add about 0.34kg for the NiMh battery and we're at 8.34kg, which is 1.5kg heavier (!!) than the Lion.

 

I estimated the displacement of the galley underwater hull by similarly dividing it into rectangles, triangles etc, and came up with about 3.7kg which is 4.64kg too low. Bedford was right, again, back when he suggested I make the bilge fuller to gain volume. I'll need to probably abandon all pretense of a "correct" underwater hull in order to get the volume I need. Must double-check my galley volume calculation first, since it's quite low.

 

My plan now is to come up with a keel depth and beefy midships section shape to get me the required volume, then carve two half hull representations of the transition from this cross section to pleasing bow and stern. Then I can use a contour gauge to get intervening sections to cut bulkheads. Best way I can think of rather than fiddling with drawings and templates. Or spending months mastering CAD.

 

 

Edited November 7, 2022 by Ian_Grant

[Ian_Grant]

While driving to the cottage today I wondered about eliminating the heavy metal drawer slides since I only need 3/4" of movement each side of centre position. What if I retained the aluminum base beam and just had it slide on Lee Valley UHMW "Slippery Tape", intended to lubricate drawers in carcasses? It would just need a couple of short guide blocks to retain it laterally, with a bit of a lip to retain it vertically. Guide blocks could have the slippery tape too. There would be nothing to get rusty.....

 

Will weigh the slides intended for the other side in the morning. Now all I need to do is change the beam to baltic birch ply and all the servocity stuff (which I paid a fortune for) will be "made redundant". But I like the idea of the aluminum base channel sliding on the tape, which only its two flanges would contact as opposed to the full face of a plywood base. Also the ply base's edges would be hard to totally smooth out.

[rookie]

This is the same size as your Lion?

Is that not 5 feet long?  

 

That's a lot of oars

[Ian_Grant]

The oars take up about 30" of the proposed 54" hull; that's the size of the "Lion".

[rookie]

I guess I remember it bigger

sounds cool 

[Ian_Grant]

Well, 4-1/2 feet, 5 feet; near as dammit after 40 years out of your sight!  😀

[Ian_Grant]

**** This post is mainly a note to myself as an aide-memoire in the event that I lose the napkins I did these calculations on ****

 

So the set of three drawer slides under the aluminum channel weighs 0.38 kg.

Eliminating them on two sides saves 0.76kg. Almost exactly half the weight in excess of that of the "Lion".

 

I estimate I get 1kg (2.2lb) additional displacement for each 1/4" increase in the galley hull's depth.  That's without making the bilges fuller. To add the 4.64kg of additional displacement required, keel depth (if keeping the slides) must increase by 1.2". That seems a lot.

 

I can gain an additional 0.7kg by making the bottom flatter and the bilge fuller.

 

Remember that Lion is all 1/32 ply skin and no fiberglass (since all ply panel joints fall on ribs or stringers). If I plank the galley which means F/G exterior it could be heavier.

 

God, what a guessing game🤣😢

 

**** End of Note ****

[rookie]

40 years  .... 😭

[Ian_Grant]

I wrote a little test program to move the servos to static positions, according to numbers I type in and which the Arduino uses to set their control pulse widths in microseconds. I was then able to inch the servo arms along to decide on their extreme positions given the physical setup.

 

Arduino provides a "map" instruction which is used to scale a variable's range into a new range with all readings adjusted proportionally. The guts of the program assumes full movement of the servos, and at the end they are "mapped" into the limits found as above just before the pulses are output to the servos. When I have a ship I will use the test program and again "map" to its physical limitations, on a per-side basis.

 

I also changed the software to have a better stroke shape. Originally the stroke was simply rectangular, with the lift servo moving abruptly up or down at the ends of each sweep. I added a "trapezoid" stroke shape. The entire return stroke is flat; the power stroke consists of a ramp down into the water to full immersion, a central flat portion underwater, then a symmetrical ramp up out of the water. The ramps, or "catches", are defined by a new variable "catchFraction" which is the ratio of the two ramps to the overall sweep length. "catchFraction" is entered at the start of the program before compiling.

 

I use one of the DIP switches on my daughter board to dynamically select between these strokes.

 

Enough words, here is a video. The oars are run with "catchFraction" =1.0, then with a value of 0.4 to demonstrate.

 

 

[rookie]

you are getting good at that Arduino coding 

[Ian_Grant]

You might not say that if you had seen me trying to run a little test program to check out my code for the "lift" values during the trapezoidal power stroke......wanted to make sure it worked before sending it to a servo.  Basically my lift code placed in a "for" loop to run it through a cycle. Wasn't working as expected at all, thought the "for" loop wasn't set up right (first time trying one), changed to a "while" loop with no improvement, then finally realized I had messed up the syntax in the "conditions" of three "if" statements which sort between the catch and steady portions of the stroke (forgot "&&" between two comparisons in each set of conditions).  Stupid error, however I can't understand why the compiler didn't bitch at the strange input. In any event it all worked out.

Edited December 30, 2022 by Ian_Grant

[Bedford]

I admire your determination and commitment to this, for what it's worth I think the 0.4 (last) stroke style is the best

 

[gjdale]

This is just fascinating to watch Ian. 😊 I’m looking forward to the day you put this in an actual hull on the water.

[Glen McGuire]

Ditto what Grant said above!

[Ian_Grant]

9 hours ago, Bedford said:

I admire your determination and commitment to this, for what it's worth I think the 0.4 (last) stroke style is the best

 

Thanks Bedford for following and your helpful comments e.g. your prescient one about beefing up the u/w hull for the weight. As to commitment, I look upon this as a design challenge, something unusual, and it brings out the stubborn engineer in me. I spent my career "making stuff work". 😉

Not sure about the value for catch fraction.....don't want to waste too much of the stroke only partially in....need to see it in water. Might have to change program to run in slow motion for observation.

 

2 hours ago, gjdale said:

This is just fascinating to watch Ian. 😊 I’m looking forward to the day you put this in an actual hull on the water.

Me too! I will get going on the hull this winter. I'm hoping to use the library's laser machine to cut out all the rib profiles and keel etc from baltic plywood. More software to blunder through...😬

 

43 minutes ago, Glen McGuire said:

Ditto what Grant said above!

Thanks Glen!

 

I appreciate all those following along with this. 👍

[rookie]

If I understand teh "&&" comment does that denote variable names?

The "if" may interpret the name without those as a constant so the If can never be satisfied but the compiler may not think it is a mistake

 

just a guess

[Ian_Grant]

No, I had for example:

 

if (minsweep <= sweep <= catchend) {

etc etc

which would have been fine on a math test, I just wasn't thinking.

 

whereas it should have been:

 

if (minsweep <= sweep  &&  sweep <= catchend) {

etc etc

[rookie]

ok

so the && is boolean "AND" 

[mtaylor]

Compilers don't catch everything as you found out.   Doing a series of "for" and "while" loops early in my programming days, things sometimes just went bananas followed by a lot time tracking down the problem and explaining to the boss why I locked up the test main frame.  I'm glad you sorted it out.

[Ian_Grant]

55 minutes ago, mtaylor said:

Compilers don't catch everything as you found out.   Doing a series of "for" and "while" loops early in my programming days, things sometimes just went bananas followed by a lot time tracking down the problem and explaining to the boss why I locked up the test main frame.  I'm glad you sorted it out.

HAHA!!  Yes, I've had problems with nested IF's so I try to avoid them.

 

In this case I found through judicious insertion of Serial.print statements that the program was running all three IF statements every time it looped. Compiler decided that my garbled conditions were true, every time. 🤪

[Kevin-the-lubber]

Clever stuff! Not just the coding, but the manufacturing.

[gsdpic]

Just catching up a bit on this cool project.  A nice mix of mechanical engineering and software.

 

As for the "if (minsweep <= sweep <= catchend)", I am not an expert on the arduino language but have done C programming for more years (more decades actually) than I care to admit.   That is syntactically valid.   A condition evaluates to either 0 or 1, so in the second comparison, it is using either 0 or 1 instead of sweep, assuming it evaluates left to right.  In other words, it replaces the first "minsweep <= sweep" with either a 0 or a 1, then checks if the 0 or 1 is <= catchend.  So both of the <= have a number on both the left and right side, and no reason for the compiler to complain about syntax.

 

Ok, sorry for the pedantry....carry on!

 

 

[Ian_Grant]

Thanks gsdpic for the explanation......problems with hardware/software are all perfectly logical in the end, when you find them at last. I spent my career on hardware. I used to think to myself when debugging a problematic PCP "There's always a logical reason...". 

 

This is a great digression but it's something I never forgot:

 

Early on in my career, in the 80's, I was assigned to design a prototype Viterbi coding system to provide anti-jamming capability in a military radio system. Nobody in the engineering group knew anything about Viterbi's scheme so I started by reading his original IEEE paper. I designed a larger prototype according to his example small design, but worryingly I did not quite fully understand the math behind it all. My colleague meanwhile designed a companion gadget to inject random errors in a data stream, at a desired rate, in order to test the Viterbi prototype. We found that the prototype was correcting errors in incoming data, but not as many as it ought to have been. I was about tearing my hair out fruitlessly probing around for problems, could not find any, and finally as a last ditch attempt went through the math board chip by chip (this thing consisted of five PCP's entirely built from MSI TTL) with IC clips and a logic analyzer until I found an octal flip-flop one of whose output pins went high if clocking a "1" through but then went low again at the falling edge of clock, for some reason. One quarter of the decoder's calculations went through this register and were thus corrupted, spoiling the error-correcting performance. The powers that be were threatening to cancel the project even though I was ahead of schedule (probably convinced I knew not what I was doing 😆) so it was a particular triumph to replace this chip and see the error correction performing as expected literally minutes before a departmental lunch with visiting suits. In fact, my manager poked his head into the lab to tell us we were leaving for lunch just as I threw my arms in the air, shouting "YES!". 😃  One of the great moments in my career.....😊

 

 

Edited February 16, 2023 by Ian_Grant

[rookie]

Wow!

You are very smart, Bro.

 

Thank goodness I spent my career in software

🤔