CVT controller

P

pyrogaz

Member
I'm completely new to Picaxe but was hoping to use it for the following project.

I'm assisting with the build of a small electric vehicle for racing (greenpower, seems to crop up on here quite often) and it will be fitted with a constantly variable transmission. The idea is to use picaxe to monitor the motor current and have it automatically adjust the gear ratio to keep the current as close as possible to a preselected level. We already have a current sensor (Hall Effect I think) in place for an off the shelf datalogger so we could possibly use that to provide the input? The output would need to drive a linear actuator or geared motor powered screw jack to adjust the gearing, drawing up to 2 amps at 24V.

Whilst we think we could get our heads around the code and realise we have to account for times when the motor is switched off, and allow some leeway in the response to stop the actuator running around in circles; we have no ideal where to start when it comes to selecting a picaxe board what we need to drive an actuator from it, any pointers would be very much appreciated. Thanks
 
B

BeanieBots

Moderator
Welcome to the forum.

None of the ready made boards will directly give bi-directional drive at 24v @ 2A.
Depending on how and what you decide to use will determine how and what you need to add to a board's output for your needs.

I'd hazard a guess that you will need quite a fast and frequent change so a solid state solution would be required. Look up H-Bridges.
 
P

pyrogaz

Member
Thanks for the welcome:)

I sort of suspected that 2A @ 24v would be out of the range of the boards but didn't really know what options are available, I'll take a look at H-Bridges

Don't really know at the moment what sort of magnitude of change and rate would be required, but considering that most races are run on fairly flat circuits and average speeds are 25-30mph I shouldn't be too extreme. I think allowing maybe 2-3A either side of the target current would prevent the actuator working overtime and would still be more accurate than an inexperienced driver trying to adjust gears manually.
 
B

boriz

Senior Member
24v@2A=48 Watts!

Of course that’s a peak and the average will be less, but are you sure this system will improve efficiency sufficiently to offset this extra power use?

Maybe you can design a purely mechanical auto-adjust system?

Maybe a manual system, where all the power goes into the motor would be better?

What sort of constantly variable transmission are you using? (I did some work on this problem years ago as an eager teenage engineering pupil. I actually invented the variomatic transmission, despite DAF having used it for years.)
 
B

BeanieBots

Moderator
Do the testing first.
A trap many fall into is designing the solution before they know what the problem is. It quite simply doesn't work that way round.
 
P

pyrogaz

Member
48 watts is the absolute maximum as the actuator starts, with the actuator running for maybe 50% of the time and drawing somewhat less than 48W for the majority of that I reckon average consumption will be around 10 watts which represents a 2% loss of efficiency, but of course that's just an estimate.

We've considered a purely mechanical automatic system but it would inevitably increase weight, require additional bearings, and be difficult to adjust the preset limit so may not be any more efficient.

Definitely want to avoid a manual system, the drivers are inexperienced and there's plenty of cases in the history of Greenpower where the driver hasn't understood the system or got carried away with the excitement of racing resulting in a fried motor.

System is a relatively simple belt drive with just one adjustable pulley and no lay shafts or idlers.
 
I

inglewoodpete

Senior Member
I think that's right on the limit of half the L298 H-Bridge. Have a look at the datasheet. From memory, I think you can parallel up both H-bridges in the package to double the current, though.
 
D

dennis

Member
Have you considered relays ? Much simpler then an H bridge and th respnse might be fast enough. Another alternative that I use for my Autopilot is to use a relay for the direction and a Mosfet to control the current.
 
R

Ruzzz

Member
A similar problem with the L298 came up on another forum I frequent. One of the solutions was a rather nifty surface mount MOSFET pair the IRF7105, Farnell sell them for £0.35 is quantities of 25+, the article is here http://letsmakerobots.com/node/7420 about half way down. The only disadvantage is the max Vdss rating for them is 25V.
 
A

Andrew Cowan

Senior Member
Freshly charged 12v lead acid batteries can be around 14v each - so make sure your H bridge is OK with 28 volts.

Just an interesting question to the forum - can an L298 or L293 be paralled?

A
 
B

BeanieBots

Moderator
Andrew Cowan said:
can an L298 or L293 be paralled?
Click to expand...
Sort of, with care, but not a good idea.
If you are hotchit, botchit and scarper type person, then yes.
The general problem with paralleling outputs (or batteries) is getting equal current share. The volt-drop of any given semiconductor (or battery potential) can vary considerably from device to device. This will result in unequal current sharing. The "second" device or output won't start to conduct until the first is already being strained. This can lead to premature failure.

By puting in "current share" resistors, the problem can be significantly reduce, but so is your efficiency.
Much better to use an appropriately spec'd device in the first place.
 
D

demonicpicaxeguy

Senior Member
dennis said:
Have you considered relays ? Much simpler then an H bridge and th respnse might be fast enough. Another alternative that I use for my Autopilot is to use a relay for the direction and a Mosfet to control the current.
Click to expand...
i actaully use the same thing on my cnc mill, it's more effecient and even on a h bridge i think you need to wait 10 to 20 ms before changing direction which is what most relays click in anyway
 
I

inglewoodpete

Senior Member
I checked the L298 datasheet. The 2 halves of the L298 can be paralleled. Each half has an absolute maximum current capacity of 2 Amps.
 
P

pyrogaz

Member
The traction motor is a Fracmo 24v pm brushed, it's fixed by the competition rules so no option to use anything else.
 
B

boriz

Senior Member
“System is a relatively simple belt drive with just one adjustable pulley and no lay shafts or idlers.”

How do you take up the slack when the adjustable pulley gets ‘smaller’?

I can’t add much to the above responses, except to throw some ideas in the air about alternative drive systems.

Chain is much better. Less friction, no slip. You can get a good indexed gear set from a mountain bike and build a Picaxed automatic gear changer, though you are unlikely to need more than three gears. How about a 3-speed Sturmey Archer gear hub?

Best efficiency would be achieved by a direct motor connection (no gears), using the throttle to adjust the PWM duty, with a ‘duty limit’ built into the program that automatically adjusts to the current speed. EG:

Pmax=Speed*FACTOR
PWM=throttle, limited to a maximum of Pmax.

Like maybe... If you’re going 10MPH, then full throttle will only get you (say) 40% PWM, but if you are going 20MPH or more you can get 100% PWM. Standing start might be a special case which can be dealt with in the software.

This should limit the spontaneous current, reducing losses in the wiring and motor coils (as heat), while allowing max speed on the straights. Some analysis of the course would be required for proper calibration of the FACTOR. In fact for absolute best efficiency, the FACTOR could vary according to where you are on the course!

Just some ideas I had.

Good look.
 
fernando_g

fernando_g

Senior Member
"Chain is much better. Less friction, no slip....An indexed gear set....."

You know something Boriz? That is actually an excellent idea.
 
premelec

premelec

Senior Member
variable chain drive OT

There are 7 speed planetary hubs [costly!] that work well on bikes - also if you take two rear gear clusters and put them opposite to each other with large sprockets opposite to small sprockets then the chain length shifted back and forth is close to constant [need some slack for good shifting]. What doesn't seem to have been said here is what ratios are needed... if rubber belts are used toothed belts with different gears are a lot more efficient than 'V' belt and pinch pulley... OT - indeed the question is about current regulation...
 
P

pyrogaz

Member
Chain more efficient than a belt? A single speed chain is marginally more efficient (1 percentage point at most) than a standard V-belt, we've done the experiments and shown this to be the case, a cogged V-belt is the same efficiency as a chain. When you change the gearing with a belt drive (by moving the motor on a pivot a sprung drive pulley automatically adjusts its PCD and maintains the correct tension) the efficiency remains the same. To change gear with a chain drive you have to introduce other mechanisms such as derailleurs (typical 92% efficiency) for a traditional gear cluster, internally geared hubs are notoriously inefficient; the Sturmey Archer 3-speed is around 80% efficient, NuVinci 85-95%, Rohloff 91%, Nexus/Alfine 80-90% and on top of that you usually need to run a layshaft (two bearing and a second chain) to obtain the correct reduction ratio from the motor.

We're going to proceed with the variable belt drive, we actually only need to move the motor about 25mm at a maximum rate of 1mm/sec so the current draw to do this is probably somewhat lower than I initially thought. Apparently the school has a couple of development boards (unused) so when they've been located we'll have a go at entering the world of Picaxe:)
 
P

papaof2

Senior Member
Were your tests run under lab conditions, or in the real world of dirt, grease, water, and wear?

A chain drive maintains its efficiency under almost any conditions; a non-cogged belt drive may slip if dirty (or worn or incorrectly tensioned).

If your belt drive is protected from outside contaminants, it may well retain its efficiency under extended use.

John
 
P

papaof2

Senior Member
Is the enclosure easy enough to open/remove that it won't discourage regular inspction of the belt?

John
 
B

boriz

Senior Member
With respect...

“(by moving the motor on a pivot a sprung drive pulley automatically adjusts its PCD and maintains the correct tension)”

Dunno what PCD means, (D = Diameter?), but you seem to be using a dynamic pulley, like the DAF Variomatic. So if you remove the belt, the pulleys cones ‘spring’ together?

That’s NOT a “standard V-belt”. A normal pulley passes most of the force through the bottom flat portion of the belt. Your dynamic pulley forces the sprung cones apart and all the force is transferred by the edges of the belt. The belt is ‘wedged’. This adds extra friction not present in a normal pulley. One of the reasons the Varimatic system never really took off.

And how can you regulate this ‘sprung drive pulley’ in the presence of large torque variations? Surely the sprung cones will be forced apart during periods of high torque?

A spring will be no good. You’ll need to lock the cone spacing somehow between ratio changes.

I could be barking up the wrong tree of course. Maybe I don’t understand your mechanism yet. Please explain further.
 
B

BCJKiwi

Senior Member
PCD = Pitch Circle Diameter
With a V belt (also known as a wedge belt) this is somewhere around the position of the cords in the belt which transmit the power and are much nearer the top than the bottom.

No Boriz a 'V' belt NEVER transmits power 'through the bottom flat portion of the belt'
The bottom of the belt must be clear of the bottom of the pulley (unless the drive is a 'V - Flat' drive i.e. a V-belt pulley one end and a flat pulley the other with a V belt).

So the power transmission on a V belt and a Variomatic is the same - via the wedging action on the sides of the belt which are designed to ensure friction against the pulley. A variomatic belt has a different construction as they are much wider so have extra stiffness across the belt to stop them bowing up in the middle.

However, I would assume the pivot sprung drive pulley means that the drive pulley can move closer to, or further away from the driven pulley to keep the same length belt properly tensioned as the centre distance changes with different pulley combinations.
 
P

pyrogaz

Member
BCJKiwi, you've hit the nail squarely on the head. The motor is on a pivoting mount to allow the distance between the pulleys to change, the tension is maintained by the spring in the drive pulley which also allows the pulley "sheaves" to move apart or together depending on tension thus changing the effective diameter. The aim is to use picaxe to control the actuator which will move the motor depending on the current being drawn, if the current is above the target figure (maybe 23A) the motor is moved away from the driven pulley so lowering the drive ratio and vice versa.

There was a suggestion of using the picaxe to switch relays to drive the motor, I'm currently investigating this one a bit further.
 
Last edited:
B

BCJKiwi

Senior Member
OK, so it is actually a combo standard V belt and pulley at one end and a 'variomatic' style pulley at the other.

As a matter of interest, what pulley dia, centre distance and ratio range does the system achieve?
 
B

boriz

Senior Member
“No Boriz a 'V' belt NEVER transmits power 'through the bottom flat portion of the belt’”

Different radiuses move at different speeds.

A flat bottomed belt (bottom coupled) has only one radius of contact.

Any significant contact with the sides of the pulley groove will cause extra friction.

What speed does the belt move at? The speed of its innermost point of contact or the speed of its outermost point of contact? (Different radiuses. Different speeds.).

There is a small but significant difference in the speed of the pulley at (say) 5cm radius and 6cm radius, at a constant RPM. But the belt can only move at one speed!

And if the belt is ‘edge coupled’ and 1cm thick ...?

The difference is useless friction heating up the belt.

[edit] Or did I dream it?
 
Last edited:
M

MartinM57

Moderator
Never thought about this before - but surely a 1 cm belt 'edge coupled' around a 5/6cm radius pulley has a the inside of the belt on the 5cm radius and the outside on the 6cm radius so the belt/pulley move in perfect unison.

If not, belts would get exceedingly hot from all the friction caused by differential speeds and be almost impossible to turn - and the fanbelt on my car doesn't/isn't
 
P

papaof2

Senior Member
V belts do use the "friction" of the sides of the belt against the sides of the pulley as the means of power transmission. The sides of the belt and the sides of the pulley have a much greater surface area than the bottom of the belt and the bottom of the pulley grooe. The larger contact area enables transmtting more power than just using the bottom of the belt.

Look at some sample calculations: http://www.mitcalc.com/en/ui/vbelts.pdf
Note in particular the belt thickness (h) and the pulley depth (t) in the diagrams. In item 8, the belt thickness (h) is 0.875 and the pulley depth (t) is 0.99. The belt top width (b0) matches the pulley's top width (b1) - the top of the belt is intended to ride at the top of the pulley groove.

Note the measuement (bw) on both belt and pulley. Would that point would matter if the design called for the belt to run at the bottom of the pulley?

The belt does only move at one speed - in revolutions/minute. The inside circumference and the outside circumference are different sizes, but both manage to make a revolution in the same period of time.

John
 
B

boriz

Senior Member
"The inside circumference and the outside circumference are different sizes, but both manage to make a revolution in the same period of time."

That's why belts are less efficient than chains.
 
B

boriz

Senior Member
"...If not, belts would get exceedingly hot from all the friction caused by differential speeds..."

They do.
 
I

inglewoodpete

Senior Member
Vee belts are an inefficient transmission medium. Very cheap, very forgiving when it comes to shaft alignment but inefficient.

Modern cars don't use Vee belts because direct drive / electric motors / flat or toothed (ie 'thin') belts are more efficient.

Edit: Have a look at Belt efficiency

The typical belt that I think we are all referring to is the "A" section Vee belt. It has an efficiency of about 91% at its rated power and speed. Less efficient at other power levels and speeds.
 
Last edited:
M

MartinM57

Moderator
I don't doubt their (unquantified) 'inefficiencies' but I do doubt that the extra distance that the outside of a belt goes compared with the inside of the belt:
a. allows the whole thing to turn at all
b. is turned into heat

Anyway, I've got real 'work' to do... :)

EDIT:removed potentially dodgy maths...
 
Last edited:
B

BCJKiwi

Senior Member
OK don't really wish to engage in endless circular debates.

You don't need to take my word (or decades of mechanical engineering experience into account) - just check any VBelt dimension and it's matching pulley groove detail (not so easy to find) and you will see there is ALWAYS clearance at the bottom.

The trend for some time has been toward much shallower belts to reduce heating and improve efficiency which has been helped along by new materials and technology.
Latest belts can handle as much as 3 times the power of the same section of belt of just 5 years ago.

The cord in the VBelt (near the top) is what transmits the power. The V sides are there to develop friction to transfer the power to/from the pulleys. The older deeper sections were required to stop the belt rolling over in the groove. Modern multi-belt drives have options with multiple Vs joined across the top to stop roll-over in high power drives.

Have just run a calculation on a modern XPZ section belt - 0.5 Kw 4 pole electric motor at 2:1 ratio. It has (according to the manufacturer's design calculator) an efficiency of 97%.
 
Last edited:
K

kevrus

New Member
Have to agree with BCJ. Anytime i've often worked with 'v' belts, (and I still do on quite large motor/pump/fan combos) without a clearance on the bottom ,which is often the case on worn belts, the belts will slip on the pulleys under load and squeal quite unpleasantly. Inexperienced people often try increasing belt tension to compensate but the only course of action is new belts, tensioned correctly.
The spring loaded pulley system used to be quite common on various types of machinery as a speed control system where there was no requirement for any speed feedback or regulation as such, just a need to either speed up or slow down. Often just a handle or wheel rotating a threaded bar causing the motor to tilt or move was all that was required
 
P

pyrogaz

Member
BCJ, we're running a 200mm driven pulley, and the variable motor pulley is in the range 60-85mm, although we have different sheaves to change the range as required by circuit and weather conditions. Pulley centre is approx 400m.

Do I win a prize for the most non-picaxe/non-electronics relpies on a first thread:eek:
 
You must log in or register to reply here.
Top