Saw Dyno

[MustangMike]

Eric, the graphs look awesome, thanks. For better or worse, I eliminated that 8+ Hp # for 044 from my calculation as an outlier, or my computed Hp for that run would have been 7.73 instead of 7.56.

My thought with taking 7 consecutive data points (or more) was to reduce the effect of inertia on the calculations. It was the fairest method I could come up with. However, it also forces you to use #s that may not be in the peak Hp range. For example, on this run, the first # I used is from 7,500 RPM.

Perhaps that saw really does briefly hit 8 hp at 11,000 RPMs. If you look at the data points right before and after 8.746 Hp they are 7.893 and 7.872.

Don't know if it is feasible, but if we had some type of hydraulic resistance before the saw was accelerated, it would take inertia out of the picture.

I'll continue this stuff later, got to get back to planting my garden while the weather is good!

 

[wcorey]

Dan, if you look back at the pictures in the build thread, originally the lever was longer. I shortened it to make it less sensitive...

Also there already is a screw adjustment with a knurled knob, you had your hand right on it while operating the brake.
Doesn't work particularly well, I think because it's too solid, possibly some rubber in there would improve it.
The pressure applied to the brake caliper feeds back through the master cylinder, I believe a having a light touch on the lever provides a bit of back and forth give that can have a dampening effect against the pulsing created by the crappy rotor.

 

[paragonbuilder]

Dan, if you look back at the pictures in the build thread, originally the lever was longer. I shortened it to make it less sensitive...

Also there already is a screw adjustment with a knurled knob, you had your hand right on it while operating the brake.
Doesn't work particularly well, I think because it's too solid, possibly some rubber in there would improve it.
The pressure applied to the brake caliper feeds back through the master cylinder, I believe a having a light touch on the lever provides a bit of back and forth give that can have a dampening effect against the pulsing created by the crappy rotor.

Ok. I'm thinking backwards I guess. I thought with the longer lever it would need more travel for the same amount of change and while you would also have more leverage, it may be easier controlled??? Maybe I'm wrong.
Did we try controlling the Dyno with the knob instead of the lever?

 

[paragonbuilder]

Oh yeah. I see it now in the pics. I didn't look close enough at it when it was here. I'd like to try a run with that...


Sent from my iPhone using Tapatalk

 

[wcorey]

I found the screw adjustment to be very difficult to articulate to the point of being unusable.

I've thought about introducing some air into the brake fluid to see what some spongy-ness does.

Can't try any of it though until I can get together with Eric or he rigs me up some way to read the load cell...

 

[paragonbuilder]

I found the screw adjustment to be very difficult to articulate to the point of being unusable.

I've thought about introducing some air into the brake fluid to see what some spongy-ness does.

Can't try any of it though until I can get together with Eric or he rigs me up some way to read the load cell...

That stinks. I like the air idea!![emoji106]

 

[MustangMike]

There actually used to be a company that used to make something like that. They were supposed make cars that did not have anti lock breaks perform like they did have anti lock breaks. I didn't like how it worked, and got rid of them. It just gave the pedal a spongy feel.

 

[paragonbuilder]

There actually used to be a company that used to make something like that. They were supposed make cars that did not have anti lock breaks perform like they did have anti lock breaks. I didn't like how it worked, and got rid of them. It just gave the pedal a spongy feel.

Interesting! Never heard of that. It may just be the ticket here though...

 

[malk315]

I found the screw adjustment to be very difficult to articulate to the point of being unusable.

I've thought about introducing some air into the brake fluid to see what some spongy-ness does.

Can't try any of it though until I can get together with Eric or he rigs me up some way to read the load cell...

Gonna work on a rig this weekend. Hopefully won't turn into a field trip

Sent from my XT1080 using Tapatalk

 

[malk315]

I made another improvement to the software -- I turned up the transfer rate of the i2c for the load cell from 100khz to 400khz which allows a load cell read cycle to happen in under 1 ms. Now I can easily read and deal with the load cell in 1/2 the time I was before -- so I've gone from reading it 200 times per second to 400 times per second and have it converting at its maximum rate of 890 symbols per second (each time I read at 400 hz I'm guaranteed to have a new conversion from the AtoD).

This fixes the torque readings so there will always be an update each RPM read at 1/10th second intervals. This is because the 30 point decimation filter is now done in 1/2 the time -- it was 150ms and now it's 75 ms.

What I don't know is if the new timing of the decimation filter will not do as good of a job -- if not it's easy to go to a 60 point filter over the same time and get back to where we were or I can change it back to the timing we had if it becomes worse.

A graph of me pushing on the load cell by hand shows you visually why it's better.
In this graph of how it was -- every 3 points is a dupe because the load cell updates 1.5 times slower than RPM -- it's suboptimal but the best we had right before the GTG. I've had more time now

So here's how the torque lines look with the dupes -- notice the flat spots every 3 points -- you guys can find these "flat lines" in all of your graphs if you look closely at the torque lines:




The one below doesn't have any duplicate points -- every 100ms update (nothing spinning so RPMs always zero) we have new torque data that has gone through a 30 point filter. Maybe with the vibration frequencies we're dealing with the shorter time on the filter won't matter -- we'll have to try some different values and I'll collect 400hz data for Shaun (without holes this time!) to try and derive an optimal filter setting if a single frequency can be targeted and it's not varying all over the place (Shaun knows better the me if I'm mis-speaking there on frequency variance w/ saw RPMs for instance).



We have a smoother mountain up there (that's what she said!). Should help things... next time we run we'll do a bunch of testing and see if things improve. We can try filter settings from zero all the way up to 60 and still have at least as good of timing as we had before.

Bill I have a scout campout that's going to kill most of the weekend -- I'll try to get that raspberry Pi doing something but not a lot of time.
Won't be able to work on it too much next week due to work obligations, but after that I'll have more time to spend.

Will be very interested to see what we get next time electronics and mechanical are in the same place.
Once I get Bill a good way to analyze the load cell for mechanical improvements he has in mind it will advance things. Maybe I'll just build a whole second box as I'd like to have more than one in case the prototype gets zorched.

 

[malk315]

Here's what 6 seconds of data collection can look like -- can be very smooth.
When there's no vibration going on this thing could be used like a postal scale (w/ 10 kg cell I could easily get 1/2 oz. resolution).

 

[drf256]

Looks really great Eric.

So does this improvement only work going forward, or will the prior data be recalculated?

 

[malk315]

Hey Al. It only works for new runs. Just gotta make some more dynos and keep improving. Enjoying it.

Sent from my XT1080 using Tapatalk

 

[Deets066]

How big of a saw you figure your dyno can handle? Biggest I've seen on there was J's 64. Can you even mount an 88 or 90 on it?

 

[Mattyo]

the rotor can stop a motorcycle from 140 mph + .... so that part is plenty strong to handle bigger motors... i think bigger saws mounting is no problem physically, the question is how much can the load cell handle. if Bill hits the brake hard the load cell will snap

 

[wcorey]

How big of a saw you figure your dyno can handle? Biggest I've seen on there was J's 64. Can you even mount an 88 or 90 on it?

The brake is designed to handle a 500+ lb bike with two passengers at triple digit speeds, the drive shaft is 1" mounted on 1/2" aluminum, so I doubt saw at a hundred and something cc's (or a 250cc bike saw) would give it any difficulties.
The limiting factor at present would be the 3/8" chain and that the largest bar I have made has a 12.3mm stud dia.
Oh, and I guess the capacity of the load cell but they're cheap and easy to swap out...

Oops, you're too quick for me Matty, lol...

 

[huskihl]

In response to the quick deceleration, and the rotor feedback felt at the brake handle/ tension screw...how about an expansion chamber of sorts located mid-brakeline? Like a 50-100cc chamber. Might act like a shock absorber.

 

[wcorey]

Eric, I can't imagine how that's not a big improvement, at least for future builds with smoother output.
So now if we triple the number of magnets for the HEL...

 

[wcorey]

In response to the quick deceleration, and the rotor feedback felt at the brake handle/ tension screw...how about an expansion chamber of sorts located mid-brakeline? Like a 50-100cc chamber. Might act like a shock absorber.

An 'expansion chamber' would only work with air (or other gas) in it, the brake fluid doesn't compress/expand (enough) on it's own to matter, regardless of the volume.
The line/caliper has plenty of capacity to introduce a bit of air to get the desired effect.

 

[RI Chevy]

How about some sort of brake booster?