Custom Treadmill Project

[charle...@gmail.com]

Arch Reactor!

I'm Charles, the new member who just joined last sunday, and I'm so happy to have found you guys.

The project I'm working on is a custom treadmill. I've bought a treadmill online, and I'd like to weld a smaller frame for it out of lightweight steel or maybe aluminum, and also figure a way out to use a motor controller that I can plug into my computer via USB.

I've taken a welding class before, but if there is someone with some welding experience I could go to for advice that would be great.

Also, I'll eventually need help with the motor controller being USB able (raspberry pi, maybe?).

Hopefully I'll be able to knock this out with a few months of work!

[Chris Weiss]

until we get a proper vent hood, welding inside AR is a no-go.  however, there are members with welders that may be winning to help outside of AR.  

for the motor controller, first you need to know what kind of motor it is (AC, DC, brushed, brushless, induction, synchronous, single phase, three phase).  then finding a controller shouldn't be too hard.  one with USB already on it might be hard to find or ridiculously expensive.  if rPi is what you want ot use ideally you'd look for one with 3.3V compatible controls, but you can always level shift if what you find is 5V or higher (24V is not unlikely).  If you want to plug it into your desktop PC, you'll want an arduino, not a Pi.

[Charles Haine]

thanks for the speedy reply!

Arduino might well be the right move.  From what I can tell most treadmills are DC and PWM, but my worry with arduino (and pi, for that matter) is enough power for the size of DC motors that tend to be used in treadmills.  but maybe something where i use the stock controller board then interfact that with arduino.

Since I want to limit the speed on the treadmill to 2mph (no running, it's for office use), a smaller motor might work out fine and work well with Arduino.

Ch:H

       
                        

                             323-559-9235

On Fri, Apr 22, 2016 at 9:55 AM, Chris Weiss <cwe...@gmail.com> wrote:

until we get a proper vent hood, welding inside AR is a no-go.  however, there are members with welders that may be winning to help outside of AR.  

for the motor controller, first you need to know what kind of motor it is (AC, DC, brushed, brushless, induction, synchronous, single phase, three phase).  then finding a controller shouldn't be too hard.  one with USB already on it might be hard to find or ridiculously expensive.  if rPi is what you want ot use ideally you'd look for one with 3.3V compatible controls, but you can always level shift if what you find is 5V or higher (24V is not unlikely).  If you want to plug it into your desktop PC, you'll want an arduino, not a Pi.

--
You received this message because you are subscribed to a topic in the Google Groups "Arch Reactor" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/arch-reactor/Ft9vNW_DV5s/unsubscribe.
To unsubscribe from this group and all its topics, send an email to arch-reactor...@googlegroups.com.
To post to this group, send email to arch-r...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/arch-reactor/CAEhO07OCZR%2BdAREWqyYXzRFML9rx6uqsOYF8ehPgo_PpfYtpBA%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

[Jim S]

Interfacing an arduino to the existing controller should be pretty straight forward and a LOT easier than replacing the motor control.  Make sure there is isolation from the line voltage.  It may be in place in the controller already but you need to check.  Adding it wouldn't be too complicated. 

[Chris Weiss]

the arduino's "power" is not relevant, only the chosen motor controller.

[Andrew Ricke]

Are you making a tread desk?

On Friday, April 22, 2016, Chris Weiss <cwe...@gmail.com> wrote:

the arduino's "power" is not relevant, only the chosen motor controller.

--
You received this message because you are subscribed to the Google Groups "Arch Reactor" group.
To unsubscribe from this group and stop receiving emails from it, send an email to arch-reactor...@googlegroups.com.

To post to this group, send email to arch-r...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/arch-reactor/CAEhO07OjCcTSJqJUMzxTuDbTsJyMUoZVj4MQvfK6DB%3DWQNcmLA%40mail.gmail.com.

[Charles Haine]

I am making a tread desk!

I have a tread desk right now, but my girlfriend hates it, finds it ugly, and has proposed that I make something small and beautiful.

oh, and light enough she can hide it when her mother comes over :)

Of course, this means I'll now find myself obsessing over building the perfect treadmill for desk use.

Gonna swing by Grainger this weekend and look at new motors.  I have a working treadmill right now I might cut down (and use arduino to interface with the current motor controller), but I have a few friends who have said they might want lightweight treadmills, so it might make sense to find a motor and controller I can purchase over and over rather than hacking up old treadmills and having to figure out how to work with a variety of older treadmill controllers.

but man, new DC motors are like $100 each!  or more.

Wondering if I can use a smaller DC motor since I'm limiting speed.

Ch:H

       
                        

                             323-559-9235

--
You received this message because you are subscribed to a topic in the Google Groups "Arch Reactor" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/arch-reactor/Ft9vNW_DV5s/unsubscribe.
To unsubscribe from this group and all its topics, send an email to arch-reactor...@googlegroups.com.

To post to this group, send email to arch-r...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/arch-reactor/CABOBYTz4KJhUTig2ywTVEzyGf3kYKTQaEdypjHn7hCi3jj%3DpuA%40mail.gmail.com.

[Chris Weiss]

look into power wheelchair motors.  they tend to be quiet, strong, and the motors almost never go bad, so salvaging from wheelchairs that no longer go is a good option.  

[Andrew Ricke]

I also had built a tread desk but didn't like it for the noise. I also had issues with getting the motor slow enough and quiet enough. 

I first wanted to remake with a self powered treadmill. My current plans is to make a standing desk that can go high enough to use one of these:  http://www.amazon.com/Stamina-InMotion-Compact-Elliptical-Trainer/dp/B008L1TQD2

I already have it and it can be used both standing and with a chair under a desk. The noise is way less and I could peddle on conference calls when I worked from home with none the wiser. 

Just a thought before you go too far down the mechanical path (unless you really want to learn motor controls, that's totally valid). 

On Friday, April 22, 2016, Chris Weiss <cwe...@gmail.com> wrote:

look into power wheelchair motors.  they tend to be quiet, strong, and the motors almost never go bad, so salvaging from wheelchairs that no longer go is a good option.  

--

You received this message because you are subscribed to the Google Groups "Arch Reactor" group.
To unsubscribe from this group and stop receiving emails from it, send an email to arch-reactor...@googlegroups.com.
To post to this group, send email to arch-r...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/arch-reactor/CAEhO07MXLmsWmVpNwH0SvD26nVwLVJ_yjGDq-rRiT5gE7NNS-Q%40mail.gmail.com.

[Charles Haine]

the noise never really bothered me, but bothers others around me sometimes, though folks I'm on calls with claim to never hear it.

I'd both like to learn about motors and also sound dampening, to get something motorized but quiet.  should be possible, though it'll take some doing.

       
                        

                             323-559-9235

--
You received this message because you are subscribed to a topic in the Google Groups "Arch Reactor" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/arch-reactor/Ft9vNW_DV5s/unsubscribe.
To unsubscribe from this group and all its topics, send an email to arch-reactor...@googlegroups.com.

To post to this group, send email to arch-r...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/arch-reactor/CABOBYTzOfnDkYfuw2yr_OrGGxstji_PZYgh0hV%2BCxG%2B2o5xVmw%40mail.gmail.com.

[Jim S]

If you are concerned about price there are better places than Grainger.  I would look at Surplus Center or some of the other surplus places.  They have lots of some models if you want to standardize a design (and it sounds like you do).

[EschewObfuscation]

On Friday, April 22, 2016 at 3:54:49 PM UTC-5, Jim S wrote:

If you are concerned about price there are better places than Grainger...

What he said; Grainger is the place to go if your employer/corporation has an account, doesn't care much about invoice cost, because your 1 hour finding something cheaper costs them more than their cost saving. If you're working out of pocket, places like Surplus Center will serve better, though it's a dice roll what they'll have in stock any given week.

FWIW, you can get chinasport 1/2 hp DC motors from fleabay for $50 or a little less; they're intended for scooter use. Probably noisy, but most options will be. I'd plan on putting motor and chain gearing in an enclosure and sound dampening the whole enclosure, so electricity goes in, force (via a shaft) comes out. And hot air, as you'll need forced air for cooling. Never forget the cooling!

Good luck w your design iterations. "Measure twice, cut once," 

[EschewObfuscation]

On Friday, April 22, 2016 at 5:24:19 PM UTC-5, EschewObfuscation wrote:

...FWIW, you can get chinasport 1/2 hp DC motors from fleabay for $50 or a little less; they're intended for scooter use...

I neglected to mention, some of those scooter motors come complete with variable speed controller modules, operated by a simple pot. May be everything you need, barring DC power supply, chain reduction bits and sound dampening enclosure. Chain and sprockets can also be had from surplus center or fleabay etc.

[Charles Haine]

I wonder about chain vs. belt drive, though: I think a belt would be quieter than a chain, no?

Maybe just as simple as swapping a belt wheel for a sprocket?

Charles

       
                        

                             323-559-9235

[Jim S]

Yes.  I haven't ever seen a chain drive treadmill and that's likely one reason. 

Since you want to go slow a two stage reduction might be good and would allow a smaller motor.  But the higher speed motor may be more noisy.

[EschewObfuscation]

Well, you could make a case either way.

I'd be inclined to chain on the grounds the motors already have sprockets in place, so it's a "go with the flow" thing. They also have the advantages a chain can be adjusted in length, is strong, pretty efficient as nothing flexes. They do tend to make some noise, but if it's going into a muffled enclosure, that might not be a problem.

Belts are popular as they're fairly cheap and very common, as near as your auto parts store. Pulleys are common and simple.

V-Belts IMO are overused, just because we're used to them from automotive use. They're good at transferring considerable torque, but inefficient (lots of flexing generates heat and wear).

Flat belts last longer and are more efficient, as they don't rub on their rollers, nor do they flex as much thickness. I think they're underused largely for silly reasons: we think of them as "old fashioned". But there are reasons they were used in the steam era of multiton engines that generated a few horsepower; there wasn't a lot of energy to waste. Notice also how they were used in for example metal lathes before gear drives became fashionable. We tend to assume that everything designers did in the 40's and 50's was because they were stupid, but that's not so. We tend to think that just because they used slide rules, they must not have been very bright. They generally had good reasons for their design decisions.

Neither style belt can be length-adjusted by user, so the shaft spacing has to be adjustable while a bit of slack in a chain drive is desirable anyway. Belts can slip, which makes them very forgiving. It's better for the belt to slip than the motor to stop and maybe start drawing crazy current.

So short form, any of the 3 approaches can work, v-belts maybe easier to source the parts, expect more drag, heat, wear from them than the other 2 possibilities.

A final thought: many treadmills drag the walking belt across a flat plate, which causes them to need a fairly heavy motor to overcome the friction. That friction also makes them noisy. If you want to make yours quiet, a bunch of rollers under the belt will not only quiet it down, but reduce the size motor you need. I'd put some thought into that issue, given your design goals.

[jimski...@gmail.com]

Actually, quite a lot of the scooter motors have cogged belt pulleys on them.

[Charles Haine]

man, this thread is making my day.

are you proposing that instead of a board with a belt (which, of course, has friction with the belt), we look at a series of rollers?  Like, the whole platform is rollers, with a belt on top?

I wonder what that would be like to walk on.  I suppose if it was enough small rollers it would be a reasonably smooth surface, and would have lower friction, for a smaller motor requirement.

I'd be worry about the increased noise from the larger number of bearings, but maybe NSK silent sealed bearings will help.

alright, now I gotta build a prototype to test this "tons of roller' idea.

Ch:h

       
                        

                             323-559-9235

[EschewObfuscation]

On Tuesday, April 26, 2016 at 9:39:26 AM UTC-5, Charles Haine wrote:

man, this thread is making my day.

are you proposing that instead of a board with a belt (which, of course, has friction with the belt), we look at a series of rollers?  Like, the whole platform is rollers, with a belt on top?...

Exactly so. They'd have to be close together, which is why manufacturers of large machines prefer to drag the belt over a flat surface.

Re preceding post, let's be careful with the term "cogged belt", as different people mean different things w that term.

You will find the term used to describe a V belt of fluctuating depth; those perform as V belts are described above.

The other item people apply the term to is a flat belt with transverse ribs molded into its inner surface. These are also referred to as "timing belts". These display all the qualities of flat belts as described above, except that the ribs prevent slipping until catastrophic failure. You will sometimes find motors with timing belt sprockets affixed in the factory, should you wish to go that route.

[Jim S]

I think most people understand "cogged belt" to mean transverse teeth - i.e. timing belt.   If you put "cogged belt" into Wikipedia it goes directly to "toothed belt".  I haven't seen it used to refer to a multi-V belt and Wikipedia seems to make the same distinction.  BTW, the multi V type are what I have seen most often used on treadmills.

https://en.wikipedia.org/wiki/Toothed_belt

https://en.wikipedia.org/wiki/Timing_belt

https://en.wikipedia.org/wiki/Serpentine_belt

Of course Wikipedia isn't the end all authority but it's what was handy.

I really don't care what you call them.  Just don't call me late for dinner.  :)

[Charles Haine]

I've mostly seen multi-v belts used as well, but I'd be curious to figure out why not use a toothed belt.  it would slip less, so tightness would be less of a problem.

so, if I'm going to make a platform out of all rollers to reduce drag, where would be a good place to source rollers?

       
                        

                             323-559-9235

[Jim S]

I don't see why a toothed belt couldn't be used.  The pulleys and belts are more expensive - that's probably the main reason the other type are used.

[EschewObfuscation]

On Tuesday, April 26, 2016 at 12:13:47 PM UTC-5, Charles Haine wrote:

I've mostly seen multi-v belts used as well, but I'd be curious to figure out why not use a toothed belt.  it would slip less, so tightness would be less of a problem.

so, if I'm going to make a platform out of all rollers to reduce drag, where would be a good place to source rollers?

No idea where to source something suitable, but if you wanted to build them...

I don't recall whether the Taig can pass work through the chuck, or if so, how big. Kieth would be a good source for that data.

If it can, or if you have access to a full sized lathe, I'd consider the following approach: Choose a size, let's pull 1" out of thin air. Cut some lengths of 1" cold rolled steel to your desired belt width plus an inch. Find some conventional bearings 1" OD. Note whatever the ID is. Let's pull 5/8" from thin air. Okay, turn down the ends of the cold rolled bar to 5/8", maybe 1/2" or so to fit into the bearings. Now you have rollers with bearings, and which can be spaced closely together (maybe 1mm or so apart) in a row, with no need to stagger them to cope with bearings bulkier than their rollers.

If you wanted a proof of concept prototype, you might try this with wooden dowel instead of cold rolled? 

[Charles Haine]

now here comes a fascinating question: can we get away with tube steel instead of solid steel, if the tube walls are thick enough?

then we could run a threaded rod through the whole thing as an axle for the bearings (to bolt onto the frame), and save on weight.

if there are enough rollers, it should support the weight of the walker.

I mean, one tube should be able to support weight if the wall is thick enough.

if we reduce friction enough, maybe no motor is needed, though a small motor would be great to regulate speed.

hmm.

Ch:H

       
                        

                             323-559-9235

[EschewObfuscation]

On Wednesday, April 27, 2016 at 7:56:18 AM UTC-5, Charles Haine wrote:

now here comes a fascinating question: can we get away with tube steel instead of solid steel, if the tube walls are thick enough?

I wouldn't want to run threaded rod through the tubing, because... if the rod is large enough diameter to fill the tube, there's no weight savings. If it's smaller than the tube, then it'll either rattle, or need spacers of some sort anyway. But the concept of tubing isn't bad per se...

If you turned fittings for the ends of the tubing, a shoulder to slide inside to be welded or glued, and a stub to slide into the bearing, then *that* little lathe turning definitely *would* be within the capability of the Taig. Hm? 

[Charles Haine]

the stub that went on the inside of the bearing would be long enough to attach to the frame?

I was picturing the threaded road would be much smaller than the tube: let's say it's a 1" steel tube, with bearings press-fit into either end, then the threaded road serving as an axle would be .5" at most, since it'll need to sit right up on the bearings.

but if there is another way to connect the frame of the treadmill to the bearings of the rollers, I'm all ears.

I guess it could be some sort of axle that comes in front the frame but only far enough for the bearing race to sit on it, but doesn't have to run the length of the tube.

I just wonder if mounting only on the outside will be strong enough.

ch:H

       
                        

                             323-559-9235

[Chris Weiss]

don't need the rod to go all the way through, you just need a stub on the side rails to go into the bearings, the frame would hold it all together.

[Charles Haine]

ah, so, weld into the frame some threaded nuts, then use flat headed bolts coming in from outside the frame that are flush when you are done.

the bearings side on the threaded part of the nut, but if you need to replace a roller, you can back them out with a hex wrench.

use tubes, and freeze the bearings to shrink them, press them in to the tubes and let them expand to fit.

tube steel for the frame (or maybe even wood for the test), and then at one end a moveable roller for tension.

I think we're getting there.

where should I go steel tube shopping, or bearing shopping?  skateboard shop is where I bought my last bearings, but I bet they can be found cheaper.

Ch:H

       
                        

                             323-559-9235

On Wed, Apr 27, 2016 at 8:21 AM, Chris Weiss <cwe...@gmail.com> wrote:

don't need the rod to go all the way through, you just need a stub on the side rails to go into the bearings, the frame would hold it all together.

--

You received this message because you are subscribed to a topic in the Google Groups "Arch Reactor" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/arch-reactor/Ft9vNW_DV5s/unsubscribe.
To unsubscribe from this group and all its topics, send an email to arch-reactor...@googlegroups.com.
To post to this group, send email to arch-r...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/arch-reactor/CAEhO07PbA8P4ot2A%3DJV4R15cVsFu6CTLAHqcPLcc4Z167gsgDg%40mail.gmail.com.

[EschewObfuscation]

On Wednesday, April 27, 2016 at 8:25:46 AM UTC-5, Charles Haine wrote:

...

where should I go steel tube shopping, or bearing shopping?  skateboard shop is where I bought my last bearings, but I bet they can be found cheaper....

Either from fleabay.

Ppl here like shapiro supply because it's local and you can go looksie to see what they have. If you have a bulk order, other suppliers *might* offer a better price.

One option for bearinigs is vxb.com

[EschewObfuscation]

FWIW, I'm not crazy about the notion of threaded section being in contact with the bearing, if that's what you're considering. Lateral forces should be minimal; even drilling a modest hole in the end of your center element, whether threaded rod, smooth rod, end fitting etc, then tapping it should be more than sufficient for side to side forces. The bearing supports the major forces, in up/down axis after all.

[EschewObfuscation]

My mad skillz w win paint aren't all that awesome, but maybe this will be comprehensible; here's what I'd suggest: main scalloped support could be lasered from thin plywood, 2 or 3 layers glued together. That would support the main forces (your weight). Angle and bolts would simply prevent sliding side to side, or jumping out of the scalloped support. Bolt in the end of the assembly would spin freely in the angle; a drop of loctite would be required to keep it from backing out of the roller/bearing assembly.

[Charles Haine]

Oh, so the bearings OUTSIDE the tube.  I had been picturing them inside the roller tube steel.

then I could man something that fits on the outside with smooth posts for the inner wheel of the bearing to sit on.  that seems mechanically simpler.

and those are great PAINT skills in my book :)

       
                        

                             323-559-9235

On Wed, Apr 27, 2016 at 9:11 AM, EschewObfuscation <google...@mailfilter.33mail.com> wrote:

My mad skillz w win paint aren't all that awesome, but maybe this will be comprehensible; here's what I'd suggest: main scalloped support could be lasered from thin plywood, 2 or 3 layers glued together. That would support the main forces (your weight). Angle and bolts would simply prevent sliding side to side, or jumping out of the scalloped support. Bolt in the end of the assembly would spin freely in the angle; a drop of loctite would be required to keep it from backing out of the roller/bearing assembly.

--

You received this message because you are subscribed to a topic in the Google Groups "Arch Reactor" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/arch-reactor/Ft9vNW_DV5s/unsubscribe.
To unsubscribe from this group and all its topics, send an email to arch-reactor...@googlegroups.com.
To post to this group, send email to arch-r...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/arch-reactor/465dc735-054a-4b01-8b86-d63a6ac08cde%40googlegroups.com.

[EschewObfuscation]

Sorry for the delayed response, yesterday's storm knocked out the power, so I was busy today catching up on things.

Okay, so we've discussed two possible approaches. The lathe turning I proposed should properly be called a trunnion, so let's call the exposed bearing approach the trunnion approach. The other uses a rod through the center of the tube, so let's call that the center rod approach.

I'll mention I'm not crazy about running *threaded* rod through a bearing. In the final analysis, the races are stamped and rather thin sheet metal. If the threaded rod is an easy fit, it would be applying its force on the inner race in very narrow, diagonal stripes. If it's not an easy fit, you'll have to turn or file the tops of the threaded rod to make it fit. And if it peens over any in use, it will loosen over time. So if you went with the center rod approach, I'd say use round rod, and either cut outside threads on the ends for a nut, or drill and tap a hole to put a bolt in the end. Either way would work.

If you went with the center rod approach, I'd not recommend playing with heating/cooling parts and hammering them to achieve an interference fit, on the ground that would be a nuisance, could damage the bearing during installation, and would apply a lot of compression to the outer race if all goes well. To place the bearing into the tube, I'd suggest sizing the tube and bearing for a reasonably close fit, and finish with epoxy or another adhesive instead. But the bearing could go inside the tube.

So the center rod approach has the benefits of simpler machining and the full length of the tube is available for use by the walking belt. It has the down side of being heavier, because of the rod up the center of each tube. Both could use the scalloped main support as in my drawing; center rod would use smaller scallops with more concentrated forces, so you might want to make the main support a layer or two thicker and the rod a little longer, but that's not a big deal IMO.

The trunnion approach requires more machining (but it would be within the capabilities if the Taig), but is lighter. It's also less sensitive to the ID of the tube, as you just turn the large side of the trunnion to whatever is required, and likewise it's insensitive to the ID of the bearing as again, you just turn to whatever diameter is required. Since the full outer surface of the bearing can rest on the main support, that support need be no thicker than the bearing itself.

So either approach could work, with different tradeoffs as described above: mainly weight and machining effort.

Seem like a reasonable summary?

[Jim S]

The races on bearings, at least real ball bearings and not the cheaper things you find on consumer stuff sometimes, are NOT stamped sheet metal.  The are substantially thick, VERY hard metal.  That said, those are valid points about the points of the threads wearing down and making the fit loose.  Skate bearings (608 is the common designation) are probably the most common bearings in the world and would be your cheapest ones.  The on line sources mentioned earlier are the way to go.  These are metric (as is more common than inch sizes in bearings) and have an 8 mm ID.

The way a lot of  things with bearings are assembled is that the bearing is pressed on the shaft with a few 10s of thousandths of interference - i.e. the shaft is very slightly larger than the hole in the bearing.  It has to be done right and not put the pressure on the balls by pressing on the inner race and the fit is critical so it isn't too tight as it will decrease the internal clearances of the bearing.  Unless your machining skills are very good that's probably not an option. 

The other thing I would avoid is putting uneven side loads on the bearings.  They can handle forces directly in line with the shaft if they aren't too big so that shouldn't be a problem.  But a load in one direction on one side of the outer race and the other direction on the other side I think should be avoided.  If you don't have a through shaft and the shaft goes down with the weight on the roller the shaft will be trying to go at an angle to the outer race of the bearing.  If the shaft is anchored solidly enough that would avoid the issue.

The other potential issue I see with putting the bearings outside the tube is that the piece between the bearing and the tube has a lot of leverage on it trying to pry the part from the end of the tube so the part needs to extend into the tube some distance or be welded or some other strong connection.

[EschewObfuscation]

If the bearings chosen are more robust than I expect, then that's a good thing.

It sounds as if we are agreeing that interference fit isn't a great idea for a hand tools project like this. That said, the alternatives would be looser fit plus adhesive (my preference), set screws, spot welds, something like that. The same applies to the trunnion approach, though if the trunnion intrudes a diameter into the tube (which it about what I drew, and is more contact area than the outer rim of a bearing), then I'd think that plenty of contact area.

I'd be a bit nervous if bearings with ID 8 mm were chosen; that seems a bit light to me regardless how that ID is supported. I'd be more comfortable with 10 or better yet, 12mm. IMO shaving to minimum is value engineering, which should happen after the conservatively built prototype survives its torture tests. 12 is still too tight for common 1/2" rod, so 12mm rod would have to be found. That shouldn't be too difficult, but one more item on the check list.

I'd be greatly surprised if a trunnion turned from mild steel, with a 12mm cylinder protruding maybe 15mm or so, should bend or shear off. Consider that the load is dynamic load of one person's weight, what, maybe 1000#? Spread over 3 or 4 rollers at any given moment, with 2 ends per roller? What would it take to bend 1/2" diam of even mild steel using a lever arm maybe 3/4" long max? More than that, I think.

So basically, IMO either approach would work. With a trunnion, we accept the chore of turning the OD of each end, receiving in exchange lighter weight (if the central rod had any diameter at all), and the ability to accommodate any tubing ID and bearing ID necessary. If we go with a central rod and bearing fit inside the tube, then the machining is less (just tap and die work), but the ID of the tube will have to match the OD of the bearing rather closely. That might take a little doing.