Mini-ZRacer.com Forums

Mini-ZRacer.com Forums (http://mini-zracer.com/forums/index.php)
-   Batteries, Chargers and Dischargers (http://mini-zracer.com/forums/forumdisplay.php?f=93)
-   -   Building a discharger/test rig (http://mini-zracer.com/forums/showthread.php?t=27095)

lsarccc 2008.01.04 03:18 PM

Building a discharger/test rig
 
I am building an AAA cell discharger, it uses a microprocessor to control the individual discharge of each of the 4 cells. By using power transistors and the PWM feature of the processor the discharge rate can be varied to mimic the actual pattern of current draw experienced on the track. The idea is that I will program the processor to play a set 20second pattern of varying discharge rates that mimic a lap of a circuit. The cells will lap round and round the virtual circuit with a pattern of low to high load mixed with rest periods until each cell reaches a rest voltage of 0.9v when the discharge will stop.

At set points each lap the voltage of each cell will be sampled and sent out to software on a laptop via RS232 (undecided but perhaps each cell will be sampled 10 times each lap).

The device will also have a 3 line 20 character LCD that will display individual cell voltage and calculated indication of internal resistance and the current lap.

I expect that the highest current draw on each cell will cause the cell voltage to drop appreciably, the resolution of the voltage measurement is around 0.002v. At a point in each lap I plan to hit each cell with a big current draw, the voltage will drop - this indicates the internal resistance of each cell. Then I'll stop drawing current and sample the voltage rising back to it's rest state using a rate of 1000 samples a second to try and snapshot the rate the voltage rises back. I think the amount the voltage drops and the rate of recovery might be a useful indication of the performance and status of each cell.

The idea is to have a test rig that can test various cells and care and charging patterns and automatically store the data in a database with reporting and perhaps some php and mysql magic to make the data available on the net.


My questions:
I am assuming that using a varied discharge rate with rest periods to mimic race usage is superior to a static discharge rate; also that using a switched load mimics the car esc more accurately and the test data should match real world usage more accurately. This is based on nothing more than intuition and I wondered if anyone can support or dispel my assumption with experience and hopefully facts?

Does the PWM drive frequency matter? What is the drive frequency of a Mini-Z esc?

In my limited research it seems that no other application of cells cares about the internal resistance of the cells as much as we do? Use of rechargeable cells in cameras seems to be the biggest application that gives data and they just test at a fixed discharge rate that tends to be lower than what we are using. Is my proposed approach to measure the internal resistance and recovery capability of the cells valid and useful?

What current drain range should I emulate? I was thinking 0-4amps - is that reasonable?

I hope this project will take 2-3 months to complete, any suggestions help very very very welcome. :)

a7i20ci7y 2008.01.04 05:45 PM

Which microprocessor are you going to use? I recently tried a PWM project using a PICAXE but found it's abilities to be somewhat limited.

lsarccc 2008.01.04 05:55 PM

PIC 18F4331 - it's well up to the task, I'll run it at 20Mhz, it can do 200,000+ 10bit resolution samples per second and built in 8 PWM channels with 14bit resolution and a EUSART for hi speed comms.

lsarccc 2008.01.09 01:03 PM

Drive frequency
 
OK - not much help so far with my questions, so I decided to help myself :)

Clearly the drive frequency of the esc does matter, the lower the freq. the less efficiently the esc operates and more power the esc/motor will draw from the batteries. I'll be discharging through a resistor which is a completely different kind of load to a motor, but I guess I can only emulate the real world so far. One thing I can do is use the same drive frequency.

http://lsarccc.com/images/ext/6pwmfreq.JPG
I measured the pwm frequency of an iWaver 02 and an AWD...

iWaver 02 - 238Hz
Kyosho AWD - 640Hz

I was surprised at how low the iWaver frequency was - I suspect the iWaver board may be a bit hard on commutators.

So my conclusion is to drive the discharger pwm load at 640Hz.


I have started building a telemetry rig to measure the current drawn as the car goes around my local track, that 'trace' will become the basis for the discharger cycle.

So far I have the basics for the receiver sorted...
http://lsarccc.com/images/ext/6telemetryrx.JPG

lsarccc 2008.01.09 04:06 PM

Intermittent & varied vs constant discharge
 
OK, my core assumption for doing this is that using fixed discharge rates may be flawed when measuring the performance of cells as used in toy car racing. We know that the way we put the charge in can affect the chemistry? of the cell and the performance, so intuitively I think that when we test the cell by varying the discharge rate we may get different results.

So I've been googling away and my findings so far has thrown up some stuff that supports the idea that the result will be different but nothing that suggests that pulsing the discharge will have any other effect.

This is a crossection of my reading list so far:

A simplified explanation of Peukert - involving beer!
More on Peukert with a graph
Battery testing for electric vehicles is done with variable pattern load with standards
This suggests that intermittent but constant discharge simply extends the run time, but voltage recovery is interesting

Any more info from people that know about batteries in the RC car context very welcome :)

mk2kompressor 2008.01.09 06:04 PM

ok arthur you are my new best friend :D ;)

are you deffo coming to the hurda challenge?
i have all the paperwork from my 12hr duracell experiments,we'll speak there :)

a7i20ci7y 2008.01.09 06:13 PM

I'm not sure anyone is going to know enough to really answer your questions. Experimentation is in order and I would be very interested in the results, just out of curisosity if not performance gains (all the tweaks in the world can't save my car from my driving :D ).

Perhaps you can make the frequency a variable so you can select a fequency to match the ESC you're interested in using? I realize the test is for the batteries, not the ESC; however, if the frequency matters for real world results, you'd want to simulate your real world equipment, what ever it may be. The best batteries at 640Hz might not be the best at 238Hz? I don't know.

lsarccc 2008.01.10 12:56 AM

:) It's nice to have friends - especially ones that have data!, very interested in the duracell stuff Matt :D Yes, I need to PayPal Jonny - will get that sorted this week. (I will need to tap someone up to borrow a set of the correct offset wheels for the Hurda shell - I'll speak to you/Jonny more on that elsewhere.)


I think you're right Atrocity but there has to be a few Radio Amatures and EEs out there - you know the kind of people that really know what they are doing! I'll put in variable frequency into the discharger - I guess it has more effect in the circuit when you have a load with inductance like a motor. If we compare different frequencies we may find out something - as you say we don't know so let's try.


Here's the tale of reflex charging which I take as a great example of battery myths! (I totally believed in my reflex charger for many years :o)


I'm sure the telemetry bug can be done with a PICAXE easily enough and the receiver could be done with an off the shelf PC/USB experiment board rather than the fancy affair I'm building. I'll post extra detail of the telemetry bug maybe you'll pick it up as a project and improve on my slacker design :)

Wolf 2008.01.10 03:45 AM

It seems it will be a great project and also good thread for battery discharging.

Since we are also in this sector and carrying out a very similar project (infact three projects) at the moment we'd also like add our two cents may be it helps.

First you must be prepared for variable drive frequencies. As you've already pointed our all manufacturers has different drive freqs and I even suspect AWD and MR02 has different drive frequencies as well. MR02 might have (never measured yet) higher freq than a MR02 to increase efficiency and topspeed. On the other hand AWD might have lower freq to increase punch and torque. This is a common issue in larger scale EP racing and we've applied this logic to our new version of Spinner. Very soon new updated Spinner will be announced with programmable drive & brake frequency (it will also have adjustable dead band but it is not very related with this issue).

Emulation seems to be a good idea and your assumptions (especially lap based approach and telemetry) are pretty good. On the other hand what we believe is constant current discharging give more accurate results when it comes to comparing cells.

Last but not the least timing of the motors is an important topic in MiniZ area when we are discussing efficiency. I wonder how many of us aware that we are using motors with reverse timing. Did you ever noticed that your motors runs better and faster when going reverse? Or awd chasis goes almost same speed with MR02 despite of a shorter gear ratio? I guess all manufacturers uses mass produced cans and brushes but this results in such weird conditions.

I'd like to watch this thread closely and we hope we can help.

Regards

byebye 2008.01.10 05:52 AM

I'm gonna sticky this so it stays at the top!

Kris

lsarccc 2008.01.10 07:31 AM

Hey thanks for the sticky Kris, the pressure's on for an interesting result!

My 1/10th touring car speedo can be programmed with varying frequencies across the throttle range from 500Hz to 6.5Khz I suspect that the clever Japanese have decided on that range as the optimum for a racing model car application at 1/10th scale.

@Wolf I notice the published spec for the Spinner is 1.2KHz which sounds good to me - a bit off topic but would you care to comment about the factors that led to that choice? I presume the power being switched and rpm range of the motor and the switching losses of the FETs in use set the best frequency for the esc? Be good to hear it more definitively in the context of Mini-Z style cars :)


I'll test my MR02i frequency tonight and perhaps the iWaver 02M, there's a good chance the frequency changed when they upgraded the board.

So I guess we're looking at variable frequency 200Hz - 10KHz and a fixed current mode now.

Wolf 2008.01.10 11:13 AM

lsarcc,

We've tested Spinner before the release with mostly modified motors since it is designed for open racing. Motors like Anima, Speedy, 3Racing, Chili. We've tested with different track shapes and gear ratios. Unfortunately there is nothing called "best frequency". As I mentioned the effects of the frequency above you are stucked in torque,rpm dilemma. Since typical tracks have 10-12metres of straight and relevant in-track curves we saw that 1.2khz is a nice (or optimum) freq to release the product. As you also pointed out it is higher freq than current PCBs which leads not only better top speeds but also longer run times due to low gate losses of FETs.

New spinner which is under test at the moment will have variable frequency between 0.25 to 3khz. We are trying to optimize its steps at the moment.

How do you plan to emulate the brakes? I think (and also afraid :) )this project might lead to a full dyno for ESCs, motors, batteries instead of a discharger :)

lsarccc 2008.01.10 12:53 PM

Wolf - you raise so many good points in your posts I have to focus a little bit (I am too easily distracted!) but perhaps we can revisit some things another time/project :)

Quote:

How do you plan to emulate the brakes? I think (and also afraid :) )this project might lead to a full dyno for ESCs, motors, batteries instead of a discharger :)
Yes - things could loose their clarity, I'd like to go through the thinking again...

We are talking about discharging batteries.

I think an automated way to discharge cells unattended and log the values direct into a database with some kind of frontend to compare discharge curves would be a good thing - for me thats a no-brainer. The easier it is to record the data the more data we can do quicker. If the data is in the same format we can graph and compare easier and the data has more value.

I believe that the way we charge our cells affects the performance, for example a higher charge rate means more punch on the track, so I think it's intuitive to believe that how they discharge might reveal something. So this introduces the idea that we should try in some way to emulate real use.

All the data I have seen is based on a fixed discharge rate with currents that are lower than what we might see in a race. I am open to the idea that this is because fixed rate is the best test for comparison, but I believe that fixed rate is used because it is easy and the main markets are for cameras and other digital devices which focus on capacity more than voltage/Internal Resistance. I think that mixing the discharge rate and letting the batteries have some recovery might reveal effects on IR/punch missed by fixed discharge.

A good measure of performance in RC cars is internal resistance and the speed the cells can recover their voltage - we can measure that with some fidelity using a good pic chip.

Data on Internal Resistance is from manufacturers only. I don't think you can compare numbers. A way of measuring the IR that is comparable and might have some relation to the punch seen on the track would be great.

I recognised from the start that I couldn't emulate everything and so a line must be drawn somewhere...

I guess that the things we can emulate are:
1. Important: Varying discharge rate based on something like a real race - including periods where the cells can rest.

2. Less important: The cells aren't powering a torch bulb, they are in a dynamic circuit with pwm. I assume the cells 'feel' that the discharge has a frequency (Maybe I can see this with an oscilloscope). Maybe the pulsed discharge does something then again maybe it's irrelevant like the reflex charger story. Perhaps the cells don't feel the pwm or effects of the motor inductance so much and this is irrelevant! In any case it's fun to emulate so we will do it and test.

I think the discharger will emulate the current flowing from the cells, regenerative currents will be ignored (not sure Mini-Z escs do that?) and so braking will appear as a rest period with zero discharge.

We'll measure time, instantaneous voltage, rest voltage, voltage under max load and rate of voltage recovery after high load. Because the load is varying, the capacity (as runtime in seconds) is troblesome but I think I can see ways round it.

That is it for this project :)

Does this make sense and limit the scope of the project nicely?



We will do another project where we measure the pulse width of the throttle channel to work out throttle demand and marry that to the rpm of the axle over time (spooling up) and somehow relate the demand of the driver on the throttle with the response of the motor spooling up (and down) to push a car of known mass through a known gearing and tyre diameter (and making more assumptions about more things than we should :rolleyes: ) to try and produce a workable dyno... we'll throw the data out of 433Mhz or 2GHz RF telemetry link and stick that in a database too. For the future, I think it requires a proper collaboration!

(But if you need a little telemetry link to help debug that Spinner drive freq - drop me a pm maybe I can help...)

edits and edits - how many typos :o

mk2kompressor 2008.01.10 01:07 PM

damn no bowing smiley :eek:
how are you working out the internal resistance?

Wolf 2008.01.10 01:29 PM

After a long day (its 9pm and still working) I'm also in a mood for distraction.
Forgive my short post. I'll edit or add tomorrow.

- I aggree with the scope and assumptions.
- One of the reasons why I do like constant discharge rates is; you can compare matched cells between companies. Many matchers has many accronyms. VRS, VDS, XXMAX, bla bla. They all mean same thing Virtual Race Smulation and also indicate something that you can not compare the results of brand X to brand Y. That's why I always sticked to constant ampers. I know, this is not a technical reason but still a valid hobbiest excuse. May be constant ampers would be included as an option.

- Infact variable discharge may be even better considering AAA. The reason we see low constant discharge rates on manufacturer infos for AAAs is they not designed for it. You can easily draw 35-40A from a sub-C for several minutes but you can not even draw 3-4A for several seconds from a AAA

- Yes. We would be interested in a 2.4 telemetry. Infact we have access to those modules and test boards but have no time to work on it to fit our needs. Just send me an email.

- Meanwhile we already have the software you needed. I've written it a year ago for C1. It reads from RS232 according to our protocol and stores detailed data in a mdb and then graphs it. You can compare the graphs of both charge and discharge.

- I totally aggree with IR issue. This how sub-c classified.

Pass for tonight.


All times are GMT -5. The time now is 10:29 AM.

Powered by vBulletin® Version 3.8.3
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Copyright ©2011 Mini-ZRacer.com