Saw Dyno

[paragonbuilder]

I'm anxious to see a particular saw on it.
2 saws actually.

This one?

 

[paragonbuilder]

Any reason the RPM range is lower than I would have expected?

If you mean peak I think that is just where that saw was tuned to. You can see no load at 12,500 when no brake was being applied, than again where it has a load at the same rpm.

 

[malk315]

Any reason the RPM range is lower than I would have expected?

Hey Mike-

Bill's Makita was only taching to about 12500 that day we both thought she was running rich. We start data collection when he's at full throttle and hits the brake to make the torque go up over a threshold you can set in the parameters and then collection stops when the torque drops below the stop threshold. Usually it will stop when the brake is let go which could be at a higher RPM than you want -- but Bill had a good run here on the low side and brought her down to 7000. Most guys probably wouldn't lug a saw much below 7000 when cutting regularly (I think!).

This one had peak HP at 9000 RPM with 6.4 HP.

What will be cool as long things hold together is we could turn the H needle in an 1/8 and run it again, check graph (instantly), turn the needle again, run, check graph... it grabs the runs very easily.

If we ever do a hydraulic setup I'm hoping to use a stepper motor or servo to let the computer control the needle valve -- then the computer could turn the needle valve at a rate that drops the RPMs as evenly as possible over time -- just like a car dyno. Am I right that car dynos usually control the throttle so the RPMs go up at even intervals? I have no experience running any off the shelf dynos.

 

[srcarr52]

Hey Mike-

Bill's Makita was only taching to about 12500 that day we both thought she was running rich. We start data collection when he's at full throttle and hits the brake to make the torque go up over a threshold you can set in the parameters and then collection stops when the torque drops below the stop threshold. Usually it will stop when the brake is let go which could be at a higher RPM than you want -- but Bill had a good run here on the low side and brought her down to 7000. Most guys probably wouldn't lug a saw much below 7000 when cutting regularly (I think!).

This one had peak HP at 9000 RPM with 6.4 HP.

What will be cool as long things hold together is we could turn the H needle in an 1/8 and run it again, check graph (instantly), turn the needle again, run, check graph... it grabs the runs very easily.

If we ever do a hydraulic setup I'm hoping to use a stepper motor or servo to let the computer control the needle valve -- then the computer could turn the needle valve at a rate that drops the RPMs as evenly as possible over time -- just like a car dyno. Am I right that car dynos usually control the throttle so the RPMs go up at even intervals? I have no experience running any off the shelf dynos.

Yes, you are right. Most professional dynos have a PID controller for an "auto load valve". In the software you target a RPM rate of change, usually 200-500 RPM/s, and the software takes over the load valve after you've armed it and hit WOT. It will go to zero load and release control of the valve as soon as you back off the throttle.

 

[jmssaws]

This one?

That one and a 661.

 

[malk315]

Any reason the RPM range is lower than I would have expected?

For RPMs we just have to put Al's insane MS260 on there... Could probably calibrate it at 15000.
I considered that saw when selecting the cutoff frequency for the low pass filter on the HES input

This one -- I think from Saw Dr.'s youtube channel (which I just subscribed today).

 

[malk315]

Yes, you are right. Most professional dynos have a PID controller for an "auto load valve". In the software you target a RPM rate of change, usually 200-500 RPM/s, and the software takes over the load valve after you've armed it and hit WOT. It will go to zero load and release control of the valve as soon as you back off the throttle.

Hey Shaun -- yeah PID control came to mind when I thought about how the computer would need to react to the RPM changing to target an RPM decel rate. Could have a parameter for RPM decel rate (RPMS per second I guess) create a profile and run a PID loop based on RPM feedback and slow down or speed up the needle valve stepper / servo to try and stay on profile. Now we're gettin' complex... but hey why not? we've got the compute power in these cheap lower power toys. This is the type of stuff we're doing with the linear motors that is my day job.

I gotta get off the forum and go get ready to head to Dan's... Yikes.
Software nerd but it's weekend warrior cutting time... with a bunch of great guys at the CT GTG!

 

[MustangMike]

My bad, I'm looking for the high RPM # to be on the right. But I think the # on the right is wrong!

Great job guys! Would love to get something on it.

 

[srcarr52]

Hey Shaun -- yeah PID control came to mind when I thought about how the computer would need to react to the RPM changing to target an RPM decel rate. Could have a parameter for RPM decel rate (RPMS per second I guess) create a profile and run a PID loop based on RPM feedback and slow down or speed up the needle valve stepper / servo to try and stay on profile. Now we're gettin' complex... but hey why not? we've got the compute power in these cheap lower power toys. This is the type of stuff we're doing with the linear motors that is my day job.

I gotta get off the forum and go get ready to head to Dan's... Yikes.
Software nerd but it's weekend warrior cutting time... with a bunch of great guys at the CT GTG!

It's a PID controller of the valve position depending on the error of the RPM rate of change (engine acceleration). You hit WOT, the computer takes over to load the engine to a start RPM, holds it for a short time and then lets it accelerate at the target rate until you lift off the throttle (micro switch on the console throttle control).

Have fun at the GTG. It's race weekend here, my Dirt Late Model team will be racing tonight at Davenport, IA.

 

[malk315]

I see -- so load it to an RPM and let it come back up. I was thinking decel gradually from WOT by applying valve and watch for RPM decel rate of change with PID targeting error off a profile setup in RPMs per second decel rate. What you mention gets us accelerating RPMs from a start point that lugs it down with a load and then let the RPMs go up... Would be cool to operate that way since that is how car dynos go. Will think about this for dyno 2.0!

Sent from my XT1080 using Tapatalk

 

[cnice_37]

I can't build a saw but I can build some circuits. I appreciate both.

Nice build, I've got a background in ee/ rf and dsp so let me know if you need any armchair quarterbacking from the southeastern end of the state. Looks like @wcorey has come a long ways since I met him

 

[malk315]

I wanted to share with everyone how we're able to get some usable data from the dyno even though the raw data is really noisy from what we suspect is pulsations from the brake as it grabs variably which creates a frequency up and down as it pulls on the load cell. Over all it's a pulling action that we're trying to measure which is what you get when you filter it. This is why those of you who receive graphs from the runs we did at the CT GTG will see some variability -- but it's quite usable and you can pretty easily see a trend line for your power band.

For one of the runs I recorded both the raw unfiltered data and the filtered data so they could be compared which I will show here.
It's pretty amazing how much a 30 point decimation filter cleans up the torque curve that was taken.
Here are the points from the raw data where a point is taken every 5 millisecond or 200 Hz collection rate. I believe this was a 5.4 second run:



And here is the data recorded for the torque for that run which is a 30 sample decimation filter -- i.e. 30 x 5ms points contribute 1/30th each to a datapoint we actually use which yields torque changing every 150ms as we go along in the run:



When I saw this last weekend when I first did a run w/ Bills Makita saw I knew we finally had something that could be used to some extent. This cleans up that mess above really well.
What I don't know and I was hoping Shaun you could take look is what is the optimal number of samples to use for the decimation filter given the frequency if that can be figured out from the 200hz data that I took.
Maybe 200hz isn't fast enough to know what's going on. Shaun, I could send you the csv files for these if you think you can identify a frequency from the 200hz data above and run the FFT method you mentioned to pick the best decimation filter setting.
Any help / advice would be appreciated -- my expertise in trying to characterize it is limited. If the 200hz graph isn't enough, then I can look into going faster but I'm already close to the limit. We can also put a scope on the load cell lines and possibly film the load cell arm movement with a high speed camera I have access to that we use to analyze vibrations and the like.

Thanks!

 

[MustangMike]

Now I see what you were working on last night, wanted to stay on till you signed off, but I bowed out some time after midnight!

Eric, that graph looks fantastic, nothing more is needed except RPMs.

Hope you got a good night sleep last night afterwards.

 

[srcarr52]

@malk315 Go ahead and send me the data via PM. I'll run some analysis on it and place some graphs in this thread. From the graph I would guess the noise is around 25Hz.

 

[David Young]

Very exciting and great info.

Thank you for all your hard work.

 

[David Young]

I suggest a disclaimer, you put your saw on this mowshin. its public info. We are trying to figure out how to tune these motors. they are getting stronger because we discuss the good bad and ugly. Thanks to mopar, al and jason and anyone I may have forgotten for being open to share their runs and their build specs.

 

[jmssaws]

I suggest a disclaimer, you put your saw on this mowshin. its public info. We are trying to figure out how to tune these motors. they are getting stronger because we discuss the good bad and ugly. Thanks to mopar, al and jason and anyone I may have forgotten for being open to share their runs and their build specs.

A tool like this would be very handy to have in the shop, it takes the "it feels faster " out.
I need one!

 

[Moparmyway]

I suggest a disclaimer, you put your saw on this mowshin. its public info. We are trying to figure out how to tune these motors. they are getting stronger because we discuss the good bad and ugly. Thanks to mopar, al and jason and anyone I may have forgotten for being open to share their runs and their build specs.

My pleasure Sir.
I've gotten a ton of help from "The Offline Forum" men ............. They all know who they are, and I wouldn't have been able to get this 661 where it is without them. My thanks go out to them too, as well as to Al for all of his work transporting and wrenching in Connecticut too.

 

[drf256]

I just wanna see one of my 026's crush Jason's 064, and I wanna know why, and I want an oompa loompa now.

 

[jmssaws]

I just wanna see one of my 026's crush Jason's 064, and I wanna know why, and I want an oompa loompa now.

I wanna see that to al. Seriously