Some help with Power sweep.
I need some help with any form of power sweep(Power sweep of Load Control). I have been tying to get a decent curve and numbers but dont seem to be able to get decent results. I am familiar with the PID concept and have read most of the online docs how this applies to the yourdyno software.
As kP i have tried values from 0.10 until 1 with or without pI. Where pI variable from 0 until 1. 1 results in a stall though. so until 0.35 pI no stall's. With either the Power sweep or Load control it seems like the dyno is braking really late. my run's are from 1500 of 1750 until 4000 rpm. from what i can see in the logs its only breaking from about 2800 rpm which gives really strange curves(like art). I'll attach a set of run's to this post. The car i am running is one with some power and quite some torque. it seems that the "wait for x seconds" it not configured correctly. when i throttle the car at the "start run" rpm it does not seem to be breaking to hold the car at the "start run" rpm. i fly right through to around 3000 rpm. First runs just took shy of 2 seconds because of this.
Maybe someone can help me set up this correctly? or how to approach this.
Thanks in adcance!
Hi, what kind of brake do you have? If Eddy current, what kind of power supply?
Also, make sure to download the Brake and gear ratio logging plugin. Then you can watch the Brake output of YourDyno. If the brake percentage increases quickly as you would expect, but the brake responds slowly, then something is wrong in the brake/power supply setup (outside of YourDyno).
Im am watching the brake % while doing a run. Using the graph display type. I am seeing that the brake is doing somewhat of braking 5-10% an the start of the run and this increases to about 20-35% at the end of the run.
OK good, which power supply is it?
If you do a run with the brake and gear ratio plugin you can also see the various components of the PID, and it is saved in the run. We can then also see how RPM increased vs the brake percentage. You can post a run with this logging enabled.
Yes i have looked at the plugin. seems really usefull. Any good starting points idea's for kP and pI?
I have no idea which power supply this is. its a grey box with 'High voltage' on it. it can be seen on the attached picture.
A good starting point is Kp = 0.3 and Ki = 0.15 and Kd = 0, but depending on the setup you may need higher values. You can add a bit of Kd too, but typically at least 10 times less than the other two.
You need to figure out what power supply it is. Maybe it needs an analog voltage and you control it with a PWM or vice versa. Is there a part number or something? Can you post a picture of the inside of the box?
It does brake pretty well in steady state. Also if we drive the car off the dyno we put the brake on 100% and that works great. It is controlled by 5v. Please see attached screenshot.
I will show you a picture of the inside of the box tomorrow.
I will also do a run tomorrow evening with the plugin and these settings. Im really happy with the help as i feel quite lost right about now(been at this for 3 weeks)
With this a new batch of runs with the plugin enabled. I have used 0.65 kP and 0.25 kI for these runs. it looks better but there is still a weird dip. Maybe you can advice me which way to move the kP or kI based on the saved run files attached? The weird dip is probably where i reach WOT. Maybe this is the wrong approach? Hope to hear from you soon!
Can you describe how you do the run? When you say the dip may be "where I reach WOT", it sounds like you are not holding WOT all the way.
Here is what to do for a Power sweep:
In Power sweep setup, set Start RPM, End RPM, Sweep rate and Wait time. Wait time could be say 3-4 second or so and is to ensure RPM stabilizes before the sweep starts. As the RPM crosses the start RPM, the brake will start to turn on and you ease onto WOT within say 1 sec or so. You should have a stable RPM at WOT before the timeout is over and the sweep starts. Then you keep WOT until you reach the End RPM. All data recording should be at WOT.
Also, for now, turn off the "Enable correction for rolling resistance from retardation data". The sweep will end then at max RPM. Now you are recording retardation data but not doing a retardation run, so the curves get weird. When you are ready for it, do retardation runs. That means clutch in when you reach end RPM. It will then record the drag in the drive train and dyno and compensate for this. Before that you need to determine the Moment of Inertia. There is a video in the YourDyno youtube channel describing it. For now, just turn it off until you get the other things right.
Ki looks to be too low, but it is hard to say until I know you are doing the sweep like described above.
Thanks for your reply! normaly i would do WOT at the start RPM but it does not seem to hold the rpm anywhere near the start rpm. Previous runs on which i did that it ramped up to 3000 rpm quite fast and then hit the brake really hard putting the rpm's down again. this created really weird run curves. So for now i put the wait time at 3 seconds and ease onto wot. But during this the rpm's already increase quite fast. so i hit wot while the rpm's climb. and once the wait time is over the run continues. So thinking of this i might put the wait time a bit longer and easy slower onto wot. maybe it helps. I have no idea why the dyno is not holding the dyno at the start rpm. You have an idea? The current car im running does have a lot of torque. maybe thats the reason?
I will turn the setting off. I did not know it was on. I have done the MOI run's(as stated in the youtube video) to get to the numbers im currently working with. So it should be near close.
thanks a lot!
OK. From the curves you see that the brake %-age is pretty low. If the RPM increases more than it should and the brake is low, then you need higher PID parameters. Higher parameters = more brake actions. You could try with Kp = 1 and Ki = 0.75 and see how it goes. You can also add a bit of Kd, say 0.05 - 0.1, but try without first.
It takes some experimentation to get it right. Make sure you are able to make it stable at the correct RPM before attempting a sweep. You can put it in manual mode when you experiment with that to lock the regulator to a specific RPM.
Hi there again!
I have tried with the new settings and this did give me a way better result. I have tried 1 for kP and started from 0.75 for kI. We lowered the kI a bit until 0.50. Even added some of the kD. we have tried 0.05. When getting to the start rpm it is sensitive on how fast you go to wot. when goin a little to fast on the throttle it will start to oscillate.
We tried to go to 1.2 kP but that resulted in more oscillations so we went back. I will add the run file with this post. Do you have a idea in witch direction the settings need to go? we did some almost excellent runs so we are really close! 😀
Thanks in advance!
Good to see you are getting closer.
I looked at the runs. If you look at one graph at the time, and look at Results vs Time, you can see what is going on.
The best run by far was DynoRun_4. This had an acceptable RPM vs Time, although it had signs of instability. Did you write down the parameters you used?
All the other runs had oscillations. It is easily possible to see that this comes from too high Kp. Just plot the Brake1 and the Kp, Ki and Kd, and you see that it is the Kp that oscillates. It looks like Kp was too high on all runs except possibly DynoRun_4.
You can also see that the Kd you added on later runs helped calm down the brake response, even though Kp was too high.
It looks like a good setup will be same as DynoRun_4 with slightly decreased Kp, slightly increased Ki and Kd.
You need to play around with it, but good to see you are getting closer.
I see you are not reaching the defined end RPM in your runs before you let go the throttle. That is why you also get the graph coming back after max RPM. So either set the end RPM lower or don't let go the throttle before reaching the defined end RPM, and graphs will look more normal.