Worlds Worst Dyno - Version 2

[crossbreak]

want to built a dyno...hope this thread is not dead :x I found this document http://www.engr.sjsu.edu/bjfurman/courses/ME106/ME106pdf/electronic_scale_arduino.doc
maybe i'll buy a scale from the same vendor I use the magic pie2 as a brake and an ebike as a load.

got 11 Fets from a brunded 12-fet controller (TO-220) to drive the bulbs. Got 12atm, will be 4x10 in the end, I use 50W ones, 2000W total.

How do you drive your bulbs? are you also using FETs? I think of using a TC4420 gate driver for one FET and drive the 12Fets by it. So a cascade of arduino->4420 Fet driver-> 1Fet->12Fets-> 40x50W bulbs. Dont know if it works, i'll try. Have to lower PWN frequency to ~1kHz i guess.

Have a 6 step plan:

1. Built a 600Wmax load for testing battery/BMS, using 12 Fets, 12bulbs, a DIY gate driver + FET +PWM of Arduino may work like this one:

from here http://arduinodiy.wordpress.com/2012/05/02/using-mosfets-with-ttl-levels/

then, I can test my BMS wihtout sparks ->great advantage

2. connect the MP2 generator to this, using some flat bar as a mount on a box for the MP2, will use a 20" tire as a roller dyno, use 3-phase AC to drive the bulbs

3. wire the whole thing so i can measure AC voltage and current of the MagicPie2 motor as a brake, place my bike on it, use PID feedback control for bulb power

4. measure dyno speed, using the internal hall sensors of the MagicPie2

5. measure torque using a load cell, use ball bearings for a torque-free mount, connect a torque arm to the load cell

6. make a virtual drive cycle, test the whole stuff

7 wire some better ADCs for more accuracy: got a 12-bit STM32 Hantek Oszi that is perfect for this purpose

Use the arduino leonardo/micro/labview software to measure the force/troque in the first place. Leonardo/micro has built in 100x op-amp/gain factor for the load cell, shunt etc.

3. wire the whole thing so i can measure AC voltage and current of the MagicPie2 motor as a brake, place my bike on it, use PID feedback control for bulb power

maybe this is not so clever...just use the torque value of the load cell for feed back loop PID control...I should really make my own thread, but got inspired so much from yours

Regen Brake Load

Quick update of the variation in load I am able to apply via a simple coupling to a motor using regen.

Setup
Regen Motor: C8085-250kv
Regen Controller: EB306 Sensorless
Drive Motor: SK3 6364-190kv
Drive ESC: Tz85A
Power Supply Voltage: 22.2v
RPM prior to braking: ~4260rpm
RPM under regen: ~4000rpm

Results:
Drive Motor Power / Regen Power
46w / 0w - Free spinning motor
170w / 0w - Coupled to load motor, no regen
297w / 87w - "Slip Throttle" regen, when applying a tiny amount of throttle on the regen controller, I see some load
480w / 257w - Regen (EBS Level = Unlimited)
550w / 315w - Regen (EBS Level = Moderate)
610w / 350w - Regen (EBS Level = Strong)

Summary:
This gives me various loads up to 1.35 Nm with this load motor.
Might be interesting to try with a lower kV regen motor to get a higher load.

You cannot control torque that is a great drawback. How can we manipulate the controller so it has variable regen? maybe feed a "self desigend" shunt voltage to the regen controller, using an µC like arduino leonardo? This could make your controller use "variable regen" by PID control of the Arduino

My approach, a DIY 3-phase PWM (all 3 phases conduct at the same time, no determination) with just some bulbs, but variable regen would help us with ebikes regen, too, just as load cells for pedal torque measurement

 

[adrian_sm]

For my purposes of stress testing controllers and motor, I just needed a repeatable load in the torque and speed range I was interested in. So I abandoned a lot of the bells and whistle nice to haves like fine tuning the torque level, or real time measurement. It was just going to take more time to set-up than the time it would have saved me.

I drive my bulbs by having a string of bulbs between each phase of the generating motor.
My torque adjustment was either reprogramming the regen controller for a different regen level, or using different number of bulbs. Both very easy, and both gave me the level of adjustment I needed.

For variable regen, cutting the current trace on the regen controller from the shunt, and feeding a fraction of the shunt current sensing voltage using a pot across the shunt to the uC would work. This is assuming you don't need real-time closed loop adjustment, based on something inputs from outside the controller.
But the controller wastes a lot of heat during regen (due to the heavy PWM and FET body diode losses), so you will be limitted in how far you can go. I had the lowest RDson FETs I could find in my 6fet, and it was already at thermal limit with the modest regen levels I was working with.

Best advice I have is first understand what you want to get out of testing, and only build in the features you really need to the dyno. Otherwise you'll waste a heap of time designing the worlds best dyno, when something much simplier may have got you the data you need.

 

[crossbreak]

thanks for the tips. I definitely need a battery test stand at first. Have to use FETs to connect the batteries otherwise there are lots of heavy sparks, since the bulbs have really low resistance when cold, around 200mOhm. So this has to be the first step. No loop back control, just connect a poti to arduino for PWM duty cycle adjustment. Keep it simple

With three phases connected to bulbs, I could just use the complete 3-phase full bridge of a controller and swap just the PWM control :? Lets see if I find right traces.

 

[adrian_sm]

Watch the heat on the FETs from the collapsing current through the diodes.

 

[crossbreak]

i'll use the 12Fet controller simply as rectifier, use 12 extra fets for PWM.