Tabletop Engine Dyno *Build In Progress*

Hi!

I've been thinking about a design for a compact table-top dynamometer suitable for testing ebike-scale motors for a LONG time- as I'm sure many of you have found yourselves, when thinking about motor and controller projects I always end up back at the need for a dyno to get real data about efficiency/power. As Fetcher's sig says "One test is worth a thousand opinions"-- and the "butt-dyno" is really just an opinion. :wink:

So, after going back and forth over ways to build something like this for ages I've come to this method- I finally sat down and roughly modeled the assembly yesterday. There are unfinished parts, and many missing parts, but it should be enough to get the concept across. I'm keen to hear your feedback if any particular aspect of the design makes an impression on you. I have given a lot of thought to the design and most of the choices have some sort of reasoning behind them- but I'm open to suggestions, and would be happy to elaborate on my thoughts as well.





As you can see, an automobile alternator (in pink :lol: ) is used to absorb mechanical energy. The alternator is suspended on large-diameter thin-section bearings from vxb.com. A Love-joy coupler is used to couple to the motor under test. Aluminum structural extrusion from Misumi is used to support the various elements.

Why an alternator?

• Cheap and easily available. This is a Delco Remy alternator used on some GM vehicles- I think maybe Northstar powered Cadillacs. There are tons of them available from junkyards, dismantlers or rebuilders.
• High RPM capable. Since they spin usually something like 2x crank speed they will see 10-12K rpm in service. The rotor is simple, rigid and monolithic.
• Electrical properties- externally controlled field strength. Essentially, I figure if you shorted the armature phases and drive the field it'll behave pretty much like an eddy current brake- but if you attach a fixed resistive load to the armature coils it'll provide a means to get part of the heat production out of the alternator- while the variable field still allows for torque control. I plan to build a 3-phase load bank using nichrome wire- another topic for another day- I think that picking a moderate resistance value for the load and driving maybe 0-20V on the field should allow for plenty of torque load flexibility across the RPM range needed for a given test- added bonus, if a 3-phase load is constructed it could be re-arranged in "star" or "delta" configuration to allow extra flexibility if necessary

What are other options that I thought of?

• Home built eddy-current brake- not a very tough project, but likely to take more effort and cost to build. Also, depending on the size and design will not have the continuous power capability, since all the heat is concentrated within it (will need to be larger for a given power capacity)
• Commercial eddy current brake- I'm sure something industrial surplus would work great- but the supply is not consistant of these sorts of devices
• PM motor- in order to regulate torque the load resistor must change value- much harder to computer-automate than just the 20V/10A purely resistive field load.
• Inertial- An inertial flywheel is great and simple in concept- but can not measure continuous power (no good for thermal capacity tests) and it's harder than it seems to come up with a, say, 12", 25 lb flywheel system that isn't unduly scary, particularly at high speeds
• Water brake- the smallest water brake in current production I found was the Stuska XS-19 which I was quoted about $2k for, as well the torque capability becomes non-linear at speeds under about 2500 RPM- it's just not that well suited to this particular application- though sinking energy into water is indeed a good way to do it
• Friction brake- non-linear, hard to control, requires large mechanical setup to hold all the parts (disc, hub, caliper, shaft...), heat buildup

Exploded view


So- mechanically- the alternator is sandwiched between two plates which are bolted to the front and back using the pre-existing mounting features (with standoffs that are not pictured). These plates have a large center hole and bolt circle, which allows the "bearing boss" bushing piece to mount to it- this part will have a circle of tapped holes to match the mounting plate. The bearing will press over that boss piece and then be captured in another set of mounting plates front and back which are attached via 90 degree plates to the base extrusions. As you can probably guess the motor to test will be mounted on a similar plate and bolted to the extrusion via 90 degree plates. All the parts are aluminum, the outer plates are 1/2" thick, the inner plates are 3/8". The idea is to have the outside contours waterjet cut (I <3 WJ) and then finish-machine bearing bores etc. The rear inner mounting plate mirrors the shape of the rear end of the alternator- it will spaced far enough away so that a shroud could be attached and air forced through the alternator for cooling.

Alternators generally have an annoying semi-stubby, threaded, non-keyed shaft- in this design this can be dealt with once- in order to attach the Love-joy coupler, then any motor to test just needs the mating coupler. I imagine this will probably involve turning the bore of the coupler, keying or pinning the shaft of the alternator. The couplers are cheap, so it's no big deal to machine them to fit any given motor shaft.

A load-cell will be attached to the reaction torque arm- the holes on the arm are 100mm, 125mm, 150mm, 175mm and 200mm from the centerline of the alternator shaft. I have some vauge ideas about a simple data capture system to send torque, speed, voltage and current information to a computer where they can be logged and analyzed.

I think I'll leave it there for the moment- please write if you have questions or comments! I've run this concept over several folks before, but I certainly will value the feedback of the community before I finish detailing the design and start manufacturing the parts!

Thanks!
-Henry

Something similar was discussed earlier. Don't know the outcome tho. http://endless-sphere.com/forums/viewtopic.php?f=30&t=24850&p=359064&hilit=calorimeter#p359064

salty9 said:

Something similar was discussed earlier. Don't know the outcome tho. http://endless-sphere.com/forums/viewtopic.php?f=30&t=24850&p=359064&hilit=calorimeter#p359064

Click to expand...

Indeed! I found that thread after writing this post. That's a nice rig he built- My plan allows for the electrical improvements that were discussed there- but most importantly it measures the reaction torque on the load instead of on the motor so that any motor with any mounting scheme can be tested- you're not locked into the base-mount, belt-driven hobby motor setup. Thanks for linking that, it's a good read.

Have you considered using the electrical energy created by the alternator to feed back into your controller (perhaps through a boost converter if you're only generating 12v)? This way you'd only need to supply energy from a power supply or battery that would be equal to the losses of the system. You could measure output current and voltage, which provides an easy way to calculate power :wink: and you could create a moment arm with a scale to measure the torque generated by the twisting action of the alternator. Just my $0.02

chbaird said:

Have you considered using the electrical energy created by the alternator to feed back into your controller (perhaps through a boost converter if you're only generating 12v)? This way you'd only need to supply energy from a power supply or battery that would be equal to the losses of the system. You could measure output current and voltage, which provides an easy way to calculate power :wink: and you could create a moment arm with a scale to measure the torque generated by the twisting action of the alternator. Just my $0.02

Click to expand...

When I visited Tesla Motors' shop a few years ago they had a dyno set up for testing their powertrain- the Tesla motor was coupled to a large industrial AC induction motor which was operated off a big VFD- they had it set up to regenerate the energy captured back into the DC feed for the motor controller. Of course they were also doing extended thermal tests at 100+KW... so huge amounts of energy were involved.

For this application I don't think the added complexity will be worth the extra functionality--- The emphasis of this design is maximum control over the torque/power load- given the scale of the motors under test, and the test application (ebikes) it's extremely unlikely that any more power than 1kwh in a single test will ever be involved. I think simply burning this amount of power off in a load bank is the easiest way to deal with it.

If you measure the energy coming out of the load then you must also factor in the conversion efficiency of the load- adding another major source of error into the measurement- given that I don't already have a dyno to characterize the alternator on :wink: Torque and RPM speed will give power.

I made some more time to flesh-out the model after work today- check it out:









I realized that it would advantageous to have a longer torque arm to reduce the force on the load cell- the load cell pictured is the 25LB one from eBay seller scale_bargains . Same as Burtie is using on his cool dyno http://endless-sphere.com/forums/viewtopic.php?f=2&t=35176 I also realized it would be more effective use of material to make the arm and the mounting plate separate parts- done and done.

I was hoping to use Hal's cool model of the HXT80100 motor over in the 3d model repository thread http://endless-sphere.com/forums/viewtopic.php?f=28&t=7832#p119900 but it was not immune from the file corruption demons. I think my dummy motor is about the right size, even though it is ugly. The motor mount plate is big and stupid with the idea that any number of bolt-hole patterns could be drilled in it to accommodate different motors. The key is that Lovejoy coupler- to allow for a little fudge factor in positioning and to make a standard easy-swap connection for swapping different motors around.

I also realized that the outside contours of every part really aren't critical, these parts could all be made in a manual mill and then cut to rough shape on the bandsaw. I also might just CNC mill them all out--- because that might just be quickest and cheapest for me.

I have tried to optimize this design for easy and repeatable construction- if I build it and it works I plan to release all the drawings, models and BOMs, and document the construction process. Dynos for everyone!

So, any other ideas from the peanut gallery? What do you like about the design? What don't you like about the design? Feedback: Positive or negative- I'll take it

acuteaero said:

I was hoping to use Hal's cool model of the HXT80100 motor over in the 3d model repository thread http://endless-sphere.com/forums/viewtopic.php?f=28&t=7832#p119900 but it was not immune from the file corruption demons.

Click to expand...

These are the files from Hal that I have on my hard drive:

The alternator should make a really good load for lower power. I am not sure how powerful those alternators are, I have heard of people pulling 1kW out of a regular alternator for brief periods, if you find you out grow your alternator, it would be pretty easy to replace it with something bigger.

I imagine the load cell you have found on ebay is pretty cheap, and should work well, When I was looking for a nice load cell I came across this place.

http://www.loadstarsensors.com/

They sell a USB package, so you don't have to build an amplifier / filter and stuff, you just plug it in and it works, but they aren't exactly cheap. With a simple load sell you can probably just read the output with a multimeter and if that isn't good enough, build an amplifier and whatnot to increase the accuracy of the measurement, then incorporate it into a computer reading.

The extruded aluminum pieces also make things easy for a small system. If you are hoping to something larger, than 1-2kW continuous you will really need to think about rigidity and vibration damping.


-ryan

Check post #7 of http://www.rcgroups.com/forums/showthread.php?t=1123956&highlight=reem which claims 3700 W output as a motor. Should produce something similar as a generator. Note the claim of rotor saturation around 2 amps.

Thanks Miles! You're the man!



Thanks for the comments Biff and Salty- I figure the alternator should be good for 1kW for sure, considering 80-100 amp ratings are common. Forced air is clearly a good plan. I am curious to see what the absolute maximum torque/power it can dissipate is- when the field is saturated and the windings are attached to a very low resistance load (or shorted )- Thermal considerations aside... It certainly would overheat very quickly in those conditions. A bigger alternator or a more efficient one, like the Denso SC in the RCGroups link would help that problem. As well- A standard bus alternator is rated something like 24V/300-500A- but they are huge and designed to run at lower speed. Maybe an option for a future bigger version. I have also thought about using a large DC Sep-Ex motor as a load for a bigger dyno- keeps the advantage of the easy torque control through the field. And that I happen to have one. Future project- I'm imagining a heavy welded steel sled for that one... (per your recommendation Biff!)

That's a good link for load measurement stuff- I'll hold on to it for when it's time to graduate up from the Ebay stuff. I'll give this chinese "25 Lb Load Cell Kit" a try- it's inexpensive and Burtie's posted good information about how to use it effectively. Maybe if I ask really nicely I could get him to share his microcontroller firmware and collaborate a bit on the DAQ setup :wink:

I'm going to sit on the design a few more days and then start ordering parts. I'm looking forward to seeing it come together!

I guess the rigidity of the extuded frame could be improved by bolting all the members to a common, fairly substantial, base plate.
The base plate could also serve as a mounting surface for your load cell.

You are welcome to have the relevent bits of C code from my daq setup, but it will take me a while to extract it and comment it up.
Also, the code will probably be subject to a few revisions yet before I am happy with it. Best drop me a pm when you are close to needing it.

Looking forward to seeing this all take shape.

Burtie

I've made a lot of progress on this project but have been quiet about it- Here I'll let the pictures mostly tell the story.

I went back over my design with an eye toward getting all the details just right for "releasing the designs to manufacture"... I decided to CNC mill out the major plate parts and provide drawings for some of the other parts to a friend of mine with whom I am collaborating on the project. I really liked Burtie's idea above of attaching the main aluminum extrusions to a base plate for rigidity- so I added that into the design.

Anyhow- machining the major parts. This is a process I like using for parts made of plate. It works best with plastic, but aluminum can be done just more carefully.

I always start with a good aluminum-cutting endmill- the difference between an aluminum-cutting and a standard grind endmill is very dramatic in my experience.


I set out vacuum gasket around the border of the stock I'm using.


Set the work offsets in the machine based on the position of the fixture- I'm leaving the stock surfaces on the material- flatness is not important- so the offset is calculated from the bottom of the part (top of the fixture)- it's more important to me to not cut through the part into the fixture than it is that the part be a given exact thickness!


Verify the CAM programming for feeds and speeds, then post-process it to generate the NC G-Code file to load into the machine. I use GibbsCAM.


Before running the part for real I like to "dry run" with the tool offset 2" above the final location. This allows me to stop the machine during the deepest cuts and check to make sure the tool is more than 2" from the surface of the fixture- indicating that in the final run it will not cut through into the vacuum fixture.


Engage the vacuum!


And Go! The coolant doesn't cause a problem as long as the vacuum gasket is sealing well. If something goes wrong the vacuum system will suck up the coolant and things get a bit messy.


Twenty minutes later- a part! About .010 of material is left around the part to allow the vacuum to continue to seal. This is removed later with the belt sander/die grinder. It's a little bit of a drag, but it allows these sorts of parts to be made easily without building fixturing, and in only one operation. There are some tricks- that .010 of material is fragile and if the cutter picks it up (the foam gasket can sometimes push the thin material into the cutter and cause the vaccum to be broken) things can go bad! One tip is that if you plan to cut say, five depth steps and then a finish step- you might instead want to cut four depth steps, leaving .100 at the bottom, then in the finish pass cut the last .100, as well as finishing the walls- instead of running the cutter on and off the part over the thin foil at the bottom- where it's most likely to pick it up and break the vacuum. If the foil is captured entirely between two pieces of part- as it is in the center bore of this part there should be no issues. But in situations like the outside contour of this part, where the cutter moves on and off the part- give opportunities for the cutter to pick up the foil and break the vacuum- at which point the part is pretty much ruined. Also, mind the machining forces caused by excessive speed or depth. There was only around 500lbs of force exerted by the vacuum on this part- if I cut in depth steps that are too large or too fast the part would shift around during the cut. Another way to scrap parts!


Drill the side-hole for the clamping screw.


Saw cut a slot for the clamping action.


I turned and tapped these insert-nuts for the clamp. It's way overkill, I know, but it works very nicely. If you take this approach everything must be located very accurately- the through hole, the side hole, the tapped hole in the nut- otherwise the assembly will bind and not work.


A quick test fit with the aluminum extrusions-


I bought 6-series (30mm base) extrusions from Misumi USA. I was curious to evaluate the difference between the "heavy" style and regular style of extrusion so I bought a set of each.


I specified a bunch of fasteners and extra parts to order from Mcmaster in order to try a first assembly. The load cells seem to be on the slow route over from China.

My order of parts arrived on Wednesday and my friend and collaborator finished a batch of assorted parts yesterday- so we spent some time today attending to last minute details and doing a test-fitting assembly. Time to see just how effective my solid modeling and tolerancing was!

These bars are attached to the main base plate and allow for leveling feet to sit outboard.


All the screws and T-nuts are preassembled through the plate so the extrusions can be slid right over. I realize that this is massive overkill, but I figure distributing the clamping along the length of the extrusion will help make it more rigid.


The 90 degree brackets from Misumi have little protrusions that drop into the slots in the extrusion- since once side is bolting to the mounting plates I made instead my friend used an endmill to spot-face the protrusions flush with the surface.


I installed 3/8-16 threaded inserts into the back holes of the alternator.


Here's the assembly- test fit together. It looks really fantastic! The alternator rotates on the thin section bearings along its axis in a very satisfying way! My measurements of the mounting holes in relation to the shaft axis were not perfect- but we turned some alignment bushings that fit along the inside of the main bearing supports and fit with shaft-concentric features on the alternator that will force the plate into center with the shaft. This way the holes in the plate can be drilled extra large and the alignment is achieved during assembly using the bushing. I'll post a photo later to make this clearer.







My friend and I figured that I'm probably somewhere around $650 in at this point. There is still a significant amount that remains to be done- hooking up the load cells once they arrive, setting up the lovejoy and a motor mounting plate, as well as the electrical work of setting up the DAQ system and tachometer/encoder.

The plan is probably to set aside work on this project until the big push to prepare for the April Socal motor-bicycle race is over. My friend builds combat robots and has a big event coming up as well. We should be picking the project back up sometime in April. Still- encouraging progress! I am really pleased with how it's coming together!

Looking great ! Wish I had access to CNC machines like you. Very jealous.

I would love to have a little dyno too. At the moment I have to rely on on road testing to check out how different motors hold up. But something like this would be a dream come true. Any chance you would do testing for others? I have a bunch of 50mm and 63mm sized outrunners that I would love to get some real data from.

Wow, look at that art! Jealous 8)

Thanks guys! I'm really happy with how it's coming together. It sure is a shiny and sexy pile of aluminum! It's got a lot of... presence :lol:

Once I get it up and running (probably will be a little while yet...) I'd be happy to test motors for other folks- and publish the data I get from tests for everyone to see as well. I'm also thinking about maybe making another whole unit using only basic non-CNC machine tools and posting that build process to try to encourage others to build their own dyno. Like I said on the OP I think the more people that have access to tools like this the better- it's such an important way to be able to collect data on and tune up motors and motor controllers.

I have also been imagining an add-on with a traction drum with pillow block bearings that slides onto the extruded rails and hooks up with the alternator using a lovejoy coupler- so that it could be used as a chassis dyno as well. That would be certainly be useful for testing hub-motors or entire non-hub drives on the bench. I'll have to play around with modeling that when I get a chance.

I'll also have to knock on Burtie's door in a bit about a little bit of software collaboration :wink: - it's going to be a little while though, my first priority for the next couple weeks is the April motorized bicycle race.

Acu,

You should make 10 of these and sell them to ES guys like me who would really love to get my hands on one. I am sure you would find 9 others here on ES. Every electric motor factory should have one of these. The home made dyno that Astro Flight built wasnt near this cool and it looks like they spent a lot of $ to make it with water brake, torque sensors etc.

A dyno is real nicety for anyone trying to build a race ebike....especially with the tendency for electrics not to finish races due to overheating etc.

Nice CNC work. Looking forward to the results.

Subscribed!

I'm waiting for some 13awg wire to show up to do a double 6-turn wye wind for my 80-100 that will run on 59.9v 16s Lipo. I'm interested in both performance and efficiency, with an emphasis leaning towards efficiency.

I've broken my collar bone into 3 pieces, so I'm in no hurry now and would be interested to see some real data come in on these motors, maybe even before I do the wind.

Anyone have any ~opinions~ on my winding plans?

Looks great! Good job! How's it coming along now?

Later,
Jay

GITech said:

I've broken my collar bone into 3 pieces

Click to expand...

Ouch! Sorry to hear it.

Thanks for the positive comments! Since the Grange race I have been working on rebuilding and tuning my hub-motored ebike so that I have something to ride around (keep me happy!) Now that that's about done I will be picking this project back up- there are several tasks that need doing in short order- I will hopefully start making some progress this week. I would like to be able to test this small dyno in the next couple week or two. I have a handful of 6364 motors to (rewind, and) test with. I have backordered some 80100s, I hope they come soon. I also have an Astro 3210 which I am very curious to compare.

Back in march I said:

acuteaero said:

There is still a significant amount that remains to be done- hooking up the load cells once they arrive, setting up the lovejoy and a motor mounting plate, as well as the electrical work of setting up the DAQ system and tachometer/encoder.

Click to expand...

The load cells have arrived now... first up is hooking the Lovejoy coupler to the alternator shaft. I think I have a plan for that, we'll see if it works. Add to that list constructing a resistive load bank. I'm thinking light-bulbs at this point.

acuteaero said:

I am going to finish my dyno up in the next few weeks-ish... in order to do some RC motor testing

Click to expand...

Any progress on this, Henry?

I have gotten some good work in on the dyno project over my winter break- it's almost mechanically complete now!



New since last time:

-Motor mounting plate: This I ended up designing as a two-part system:



The plate that bolts to the skid is used for any different sort of motor- the plate that the motor bolts to is bolted to the face of the fixed plate with the three socket head screws that fit in the circular slots. The screws have plastic bushings that are a nice snug fit in the slots. This allows the motor to be rotated over 90 degrees by loosening the screws and swiveling the plates with respect to one another. The action is actually very nice and smooth, does not bind at all. The purpose of building it this way is so that the load rotor can be locked and the motor can be moved through its electrical cycle and positioned at any arbitrary position, say a position of peak torque- stall torque can then be tested. Or something- it seemed like a useful way to set it up.

-Lovejoy couplers: The motor under test is attached to the load via a L-75 size Lovejoy coupler, which is completely encased in the tube support in front of the load. The coupler on the load is a standard off the shelf steel Lovejoy that I bored to fit the (17mm?) shaft. Torque is transmitted by a setscrew onto a flat on the shaft- not 100% happy with it, but I think it will be OK, if it's not I'll redo it. I made up custom couplers for the motor side out of 6061 aluminum. The custom couplers are a bit longer than standard and have bores set up to work with several different sizes of trantorque bushings. With the two couplers I made I can accomodate 8mm, 10mm and 3/8" shaft motors. There is just enough room to sneak a socket in around the trantorque and tighten it down. It works really nicely. The collar on the load-side of the front load mounting plate has two 1/4" holes through which 1/4" dowel pins can be inserted into holes in the load's lovejoy coupler, locking the load's rotor position.





-RPM Sensor: I added a spindle to the rear shaft of the alternator with 12 alternating Neo magnets on it, and placed a latching hall effect sensor next to it. Works fine!

-Load-cell Mounting: The solidworks screengrabs show the load cell mounted off the side of a piece of extrusion. I tried this and found that the torsion exertited on the extrusion by this setup caused a lot of movement in the business end of the load cell, allowing the whole load to rotate a fair amount. This eliminates the benefit of being able to lock the load's rotor, and throws the geometry out. I redid it with the extrusion mounted lower and a bracket added to eliminate the torsion effect. The rig is nice and stiff now. The load cell appears to return to a consistent zero position as long as torque is applied in one direction only- maybe stiction in the swivel bearings causes the zero for opposite torque to rest in a different place...

And then I spent a few hours and built the resistive load: Three 28v/600w aircraft landing lightbulbs wired in Delta. This allowed me to do a real honest-to-goodness test! The results are encouraging!

[youtube]ZLHhCNCsB_4[/youtube]

o

So Where Next?

Unfortunately I start school again next week :|

But I will try to continue to make some progress. It should be pretty straight-forward to get a very basic level of functionality working and get some numbers from it- I just have to try to turn off my perfectionist impulse and take that pragmatic approach... at the same time, nobody is paying me to do this, I get to do it however I like :wink: On that note, I've been thinking about making a neat-o controller box thing. (Isn't this pretty obviously a further ploy to avoid working on the DAQ software, since I'm very out-of-practice on it, and very in-practice on making things in the machine shop?)



Still, it would be fun.

I have a blower that I intend to use to cool the load through a air-shroud that I haven't got entirely figured out yet...

I've been working on detailing out my plan for electronics/software, and buying all the odds and ends I will need. The key component is the LabJack U6 pictured above, along with a little load-cell amplifier. After reading the U6's book I'm pretty sure I can use its differential inputs and PGAs to read the load cell directly. Same with K-type thermocouples for temperature measurement. The internal hardware counter will be used for RPM (frequency) measurement. Where I have gotten a little snagged in my thinking has been when it comes to throttle command output to the motor controller under test, field strength control, and current and voltage measurement. The LabJack is not isolated, and does not isolate the USB connection-- it is reading some small signals with some large amplification- it seems like isolating any connection to any power electronics is a really good idea.

Anyone have thoughts about this? I'm interested in suggestions about the importance of this. I think I have a pretty good plan for how to isolate all the signals to/from the LabJack in question, and then I was thinking moving V/I measurement off to a cycle analyst, having the computer program read the data from it via a USB/serial cable, potentially also through a USB isolator since the CA has zero isolation...

I've also thought about some tests I'd like to run once it's up and running:
-freewheeling drag torque as a function of speed (use alternator to spin motor under test)
-freewheeling drag with metal object in proximity to spinning outrunner can
-stall torque (verify torque constant)
-saturation amps/turn for various motors
-compare 6364 to Astro 3210, etc.

A typical alternator can easily do way more than 12v. Most have diodes rated for 50-60v, so don't exceed that unless you know the exact ratings. If you run a higher voltage into the load, you can get much more power out of it for a given amount of heating. Pumping the power back into the pack driving the motor is not such a bad idea (as long as you stay below 48v or so). Since there are always losses in the system, you could never overcharge the pack running. If you wanted to run over 48v, you could get a different set of diodes for the alternator.

Your load cell setup should make for very accurate measurements.