Tips on using GSF Dyno
Post Reply    Forum -> Not-So-Modern 12Next
Author Message
Hooked
1984 PX(BGM187)EFL
Joined: 14 Apr 2017
Posts: 384
Location: Cornwall UK
Tue Mar 31, 2020 10:01 pm quote
No amusing name – we want the search function to find it easily for others in future. I'll replace this message with a proper How-to if and when we get that far, a second post will follow shortly........
Hooked
1984 PX(BGM187)EFL
Joined: 14 Apr 2017
Posts: 384
Location: Cornwall UK
Tue Mar 31, 2020 10:12 pm quote
charlieman22 wrote:
There is a reason everyone sends you their video...
Yes there is, and this is intended to fix that; lets be honest, I'm doing this thread so as not to have to try to keep up with the impending chaos of your cunning multi set-up test...........

I'm going to try to convey my understanding, and my method, of using GSF Dyno, so that some of you chaps can do it for yourselves, if you want to. If it's too long, then tough; I can't magic a complex subject into something simple. I'm trying to help, but Take it or Leave it is always the option.

The software and some instructions on understanding the GSF front page are here, but we're a good few steps away from using the software anyway, and can go into detail on that later:

http://atom007.heimat.eu/tmt/gsf_dyno.html

I'm going to start jotting thoughts down, and tidy them up later. I'll edit the first post with How-to when everyone's chipped in. I won't be changing my method though because I've been doing the same for years, and want to keep my continuity for comparisons in the future. If you all decide to do it differently, it don't matter a damn to me; this isn't a question of 'make Sime happy'.

Probably best to start with what I think you've been doing wrong so far if you want accurate GSF Dyno results:

My test road is a 750M long, flat, straight road; it's a slip road onto a dual carriageway, so there's no oncoming traffic, and I'm bombing along nicely by the end of it and ready to join the dual carriageway at speed and ride off into the sunset. I do my tests early mornings when there's little traffic about, and I set-off when there's nothing in front or behind me, and can just do what needs doing properly, which is a slow start. You need to find somewhere similar, where you can do proper test runs. The first image below shows my test road; I've been using the same road for years, and all my past Dyno data is taken on it.

Now to what needs doing: We don't want to see how fast you can rev-out in each gear, shift through the gears, and get as fast as you can as quickly as you can; it ain't a quarter-mile competition - we all know we can do that - whoopee. We do want a really slow 1st and 2nd, then into 3rd as low as possible; I'm doing about 2500rpm. You then do a full rev range, I reckon for me that's about 2500 – 9000rpm (try not to blow your engines up though, eh), in 3rd. Hold your max revs in 3rd, say 5 seconds before changing up or slowing down. I'm tending now to go into 4th, get on the dual carriageway, and go round the block. Image 2-5 below compares useful graphs with ones that will just look good and sound good on the road; we want one good long full rev-range 3rd gear curve, not several long peaks for the gears.

It's best to do a few of those runs, in case of gear-misses, recording errors, wandering sheep or cattle (believe it), etc.

What recordings we need: We must have a .wav file, or a video recording with sufficient quality sound to extract one (or convert it). Assuming most will want to do a video (it's not necessary if you can't BTW; we're only going to use the sound anyway), which is probably going to be .mp4 these days, there are some things that will benefit the data, and some that you can keep for Hollywood.
The road ahead, the back of your head, your right elbow, the scenery etc., are useless. What's good to see is throttle position, gear changes, speedo, and if you have it CHT, revs, and GPS speed. I use a headcam that I point at my handlebars; that shows everything that matters. We'll have to spend some time on different kit, and what works – not forgetting Swiss's cunning plan. Anyone heard from Swiss lately; hope he's OK in NY.

Anyway, you get the idea now maybe: it's a boring film of your handlebars, speedo and any gauges, and a boring slow take-off. Low on box-office ratings; good on data.
This is what a boring video, good for data and using low 1st, 2nd, and full 3rd looks like:

https://www.youtube.com/watch?v=W71wNybbRb0

You'll see that I can't use that sound, so I record a .wav at the same time, using the basic voice recorder app on my 'phone, in my pocket on the engine side. The better the quality of the sound, the easier it is to see the curve and gear changes on it. Here's that wav, clipped to 30s:

https://www.youtube.com/watch?v=Ztz6X0bDGh0

You'll note also that's the old DR engine before the swap; I could use my recording of new BGM as an example of not a good method (though it goes off the clock just up the bumpy back lane, I wasn't doing it for dyno), but CM's just provided an example of one of those:

https://youtu.be/aC6IH4BzW3Y

I think I'll pause there, so you can see why, so far, there isn't much in the way of decent-enough quality recordings to do a worked example on. Maybe, instead of a competition to name this thread, there should be a prize for the first person to record something worth using as an example......? CM; I think you need to have another go with a better test-run method if you want yours to be the worked-example, 'cos it's more an example of the don't want category really.

Next I'll spend some time on recording methods/kit, and extracting and trimming a .wav file for GSF Dyno; life's much easier if the wav is short and covering just the run you want, so an hour recording of charging round the streets just makes work for yourself; GSF has some interesting unintelligible error messages if you put big or non-wav files in, and anyway looking for the right curve is a drag if you don't trim the .mp4 or .wav first; I'm going to say ideally less than 60seconds.

When we've got the recording right, we can move on to GSF front page, then selecting the curve, then cleaning the curve, then reading the results – it's gonna be interesting......(You want quick and simple? - you're gonna be disappointed – if it was easy, everyone would be doing it).

That's it for now; here's the images referred to above, which won't mean a thing if you didn't read it! Those who have; Any questions?

Oh, a summary, as requested:

Slow down and concentrate; you might learn something that'll make you happy...


1 Test Road.jpg
Long straight test road

2 Jack's 3rd starts at 5500, quality just OK.jpg
This starts too high in 3rd and the sound is poor - the file is too long as well

3 Charlieman's copying Jack; often wise, but not in this case. Sound OK; thought the mic would be better.jpg
I've clipped this file to a sensible length and the sound is OK, but the revs start too high in 3rd

4 Hibbert's 3rd starts at 2700, but quality is poor.jpg
Here the 3rd gear revs start nice and low, but the sound is poor

5 My DR before removal - 2400-8200 3rd, quality good enough.jpg
Low-start, full rev-range (2500-8200), clipped to keep it short, and good sound. This wav goes with the 'boring' video above.

Jet Eye Master
PX221 Malossi, O tuned PX200 and some motorbikes
Joined: 14 Jun 2017
Posts: 2443
Location: London UK
Wed Apr 01, 2020 1:13 am quote
The run I used on my last one is a slip road too. Slightly uphill for about the first mile but doesn't seem to make any difference anymore. Coming down hill the other side of the road it goes no faster. The wind becomes the biggest factor at speed on a scooter.
With the GSFdyno sound chart I struggle to pick the right bit. Is there an easy way to know which bit is relevant?
Hooked
1984 PX(BGM187)EFL
Joined: 14 Apr 2017
Posts: 384
Location: Cornwall UK
Wed Apr 01, 2020 2:03 am quote
Jack, If you can't start that low, then just start as low as you can, but there's no need to rev-out in 1st and 2nd either. Remember when we first looked at DR pre-detune? You made quite a few interesting comments about low-end losses and mid-range torque gains. If it ain't on the graph, and everyone just wants to know max PS/bhp, I'm not worried about it, but it must be better to give GSF as much of the rev-range as you can. I can't say for certain that a short range is less accurate, but if rate of change varies, which it obviously does, it must be less accurate because it's only looking at part of the curve. A nice smooth climbing curve with no tell-tale bumps, not just the last third...................?

The question of wind; there's some settings on the front page, and some just knowing direction and speed, trying to record on a sheltered road, or on a calm day. Wind at speed? - Maybe you've just reached the usefulness of the GSF Dyno software, and would need a proper indoor dyno to get a decent result? Your near-max revs were about 9500 in 3rd though, and your 4th max don't matter for this.

For the rest of your question, I sort of want to tackle each step in the right order, and we haven't got recording right yet, so your question jumps ahead a bit, but a sneak preview of the next session:

When you trim the .wav or .mp4 it cuts down the length of the recording that you're looking at for your 'best' recording/record.
When you set the rev-range (say 2000 – 10000) in the bottom right of the front sheet, that cuts out lots of irrelevant curves above and below the one you want (you'll see that I didn't do that on CM's recording and it goes up to 16000 as a result).
A good quality sound recording will show the gear changes, as will the video if it's pointed somewhere useful; this helps to highlight 3rd gear (assuming it's 3rd gear you want). I watch the video and make notes about the bits of it I want to trim and look at, then edit the video, then extract the wav.
When you've narrowed it down by trimming and the settings, you then let GSF produce its graph, and look for the 3rd gear curve that covers the correct rev-range (which you'll probably know – best to look for one that goes to your expected max, revs is the 'y' axis, and your revs might also be on your video). On your graph (the first one I posted) it's between 5500-9500, on mine (the bottom one; DR before swap) it's between 2500-8200.

Then comes the cleaning, to make GSF see the curve you want it too; good recording, decent quality sound, sharp trimming, front page settings all make it easier.

That's all to do with quality of recording, data available on the video to help trimming, what to trim, some brief dalliance with the GSF front sheet, and what to look for on the curve; it's where I'm intending to head next....

BTW, my hub is rebooting every five minutes at the moment; about a dozen times while I'm trying to write and post this. I hope it's BT messing about and not my hub packing up, but if I go quiet it's out of my hands. I Just had a grocery delivery, so not important if I lose connection; I can cut-off for weeks now without a care.
Ossessionato
76 Sprint V, 63 GL, 62 VBB, 05 Stella
Joined: 16 Jun 2011
Posts: 2853
Location: Nashville
Wed Apr 01, 2020 6:20 am quote
I'm going to sneak out and do some runs with the newest BGM build once I get a couple more heat cycles on it, so I'll provide some more examples/sound files and updates once I do.

What I've found most useful for selecting the curve is having a tach and knowing what your peak RPM's were so you can edit out the other curves. Also, you need to be sure that your carb is set up pretty much correctly or you can seize a motor in just a few seconds. I learned this the hard way after installing the wrong main jet by accident. Not coincidentally, that was the day before I bought a magnifying glass to keep in the garage for reading jets.

The software's attempt to auto-pick the curve inevitably gives me bad results. I may try building a Line In board to see what that gets me, since I have a lot more project time these days. The closer you can specify the parameters to the auto-picking engine, though, (e.g. rev range, clipping the sound file, etc.) the easier it will be.

I have the sound file from a second gear run I did a couple years ago, so I used that for these screenshots. I stuck my phone in my back right pocket to record it. I seem to have lost the other runs I did at the time, but what I found was that the dip at 7,500 RPM's was less pronounced, the higher the gear I recorded it in.

My technique was the same as others for the recording. I have a relatively quiet, straight, flat section of road that you can get to WOT in third gear. I have a relatively quiet stretch of road near my house, so I get going just enough to get into gear at a little above idle, then go WOT until the RPM's quit climbing or I run out of road. I'd do 4th, but I've never gotten to full WOT in 4th on my more recent builds because I feel unsafe well before that happens. #tunerworldproblems

Lastly, here's my audio file in .wav format, already somewhat trimmed down, if anyone wants it for educational purposes.

GSF Dyno-config.PNG
My config file, with weight, gearing, tire circumference, RPM range, etc. all set. I have no idea what temp & air pressure were at the time, though.

which-curve.png
Once I have the file imported, it shows all the possible curves it detected. This is where knowing your actual RPM's can be a lifesaver. Knowing that I redlined at just over 10k RPM's showed me that the curve I wanted was the dimmest of the three.

draw

before-cleaning.PNG
You can see how it tries to pick the curve, but it misses the curve that I want. To "rub out" the disturbances, you can click or drag on the black line everywhere it's picking the incorrect curve. It will take a little time, but eventually, yo

after-cleaning.png
the noise is now gone, so you can click the Next button and watch the magic happen.

dyno.png
And a dyno chart is born!

Hooked
1984 PX(BGM187)EFL
Joined: 14 Apr 2017
Posts: 384
Location: Cornwall UK
Wed Apr 01, 2020 7:32 am quote
Great, someone speaking the same language; that should make things easier.

Trimming the .wav to 22 seconds cuts all the unwanted curves on the time ('x' axis) – first image below.

Setting the max and min rpm on the front page (bottom right on second image) cuts the unwanted curves above and below the one you want ('y' axis), and as we agree, you should know which one to look at by knowing your revs. I've also noticed it's often the one above the brightest one.

Two extra points to make on the wobbly output curve:
Those little drops right at the end of the 'cleaned' 10s curve (in the red boxes on the bottom image), probably should have been cleaned-up to make a better output curve.
Changing the filter length (try 3) on the front page (bottom left on second image), smooths the output curve, but drops the max a bit.

We've jumped right ahead now, so bollixed my nice orderly plans up, but let it flow.....

1.jpg
.wav trimmed to best part of recording (22s) - it gets another trim (in this example to 10s) when you're cleaning it.

2.jpg
Set max and min revs to remove unwanted curves. change filter length to smooth output curve.

4.jpg
Showing wav trims (time - x), and GSF trims (mainly revs - y) Note: not the same wav, just an example from second post - Jack's wav.

3.jpg
Good sound makes cleaning easier, but also take care with the full length; I think it'll make a difference.

Hooked
1984 PX(BGM187)EFL
Joined: 14 Apr 2017
Posts: 384
Location: Cornwall UK
Wed Apr 01, 2020 8:15 am quote
I've done a little screen recording of editing Chandlerman's wav in GSF.

Chandlerman, I've loaded your wav from your link, used all of your front page except for filter length, (though I've used '3', and would probably try '2' if I did it again). I think also I've clipped the end of 1st gear, but not sure – difficult to catch the whole editing job in one recorded take, so don't want to re-do it just for that.
It shows the process of selecting the curve, making GSF look at the right curve (by erasing them when it looks at the wrong ones), then pressing the magic button. See that the filter '3' makes the curve smoother but reduces the max; as I said, perhaps '2' would be best for this one.

https://youtu.be/yp56TjXN0mc
Ossessionato
76 Sprint V, 63 GL, 62 VBB, 05 Stella
Joined: 16 Jun 2011
Posts: 2853
Location: Nashville
Wed Apr 01, 2020 8:38 am quote
Other than the part where you cost me 1.5HP, that looks great!

I'll have to re-try it with the filter length set to 2 and see what that gets me.
Molto Verboso
2007 Stella 150
Joined: 02 Nov 2019
Posts: 1114
Location: Rochester, Minnesota
Wed Apr 01, 2020 9:26 am quote
We need to see if Swiss can get his direct plug in working.
Ossessionato
76 Sprint V, 63 GL, 62 VBB, 05 Stella
Joined: 16 Jun 2011
Posts: 2853
Location: Nashville
Wed Apr 01, 2020 9:38 am quote
Christopher_55934 wrote:
We need to see if Swiss can get his direct plug in working.
I may order some resistors and gates and build one, too.
Molto Verboso
62 VBB1T Round Tail W/ leaner sidecar
Joined: 26 Jan 2019
Posts: 1778
Location: california
Wed Apr 01, 2020 3:25 pm quote
Sime,
I have three new runs recorded and cut down to just the 3rd gear runs - 3rd run includes small prequel. All are MOV not WAV.

Few questions:
Picture from GSF you posted for key inputs that I wasn't sure about. Can you clarify?
1. Cd value - not sure what this is - GSF pages is extensive - but didn't see.
2. Frontal area - since I have a sidecar - should this be different than the .75 I assume is what you guys are using (are you?). Suggestions to calculate?
3. Rolling Coefficient - I have a third wheel slightly toed in - any thoughts on what value you would use for that one?

Runs 1 & 2 are on the same section.
Run three is on a section with very slight up hill I think.
Here are three links.
Run 1:https://youtu.be/_qnf3kBfmrA
Run 2: https://youtu.be/B2p0LoWhWXU
Run 3: slight uphill https://youtu.be/tKOQIaS_jwE

Data for the boxes:
1009.48 MBAR
73°
Tire Circumference 1300.62
Weight 525 (estimated) +/-
Gear Ratio See pic - Can you clarify - they list a number not a ratio in the red box screen shot from GSF also below.
Tire size 100/80-10
Temps - about 200° until I let off - then up to about 225 as I came to a stop.
120/BE3/132 jest stack - seems to be willing to run WOT all day at low temps (look ma - no hands) may not be optimized yet but solid start.

If you can run - and perhaps post screen shot of what values you input with a screen shot - it would probably be good way for me to understand how you interpreted the input above and gear ratio info below.

Video is shot - not of speedo. Less practical for me.
However - used app that posts speed in corner and sound is with little wind noise - so hopefully easy to pluck out.

gsf_12422.jpg

Screen Shot 2020-04-01 at 3.56.06 PM.png
These are my gears. I know cause I put them in.

Screen Shot 2020-04-01 at 4.56.15 PM.png
Edit - having filled in the boxes, when I click "calculate graphs" I get an error message.

Molto Verboso
62 VBB1T Round Tail W/ leaner sidecar
Joined: 26 Jan 2019
Posts: 1778
Location: california
Wed Apr 01, 2020 4:40 pm quote
Ok - bit of playing around - re-saved my WAV file from different converter and GSF liked it better.
Curve seems - well- a bit bumpy.

Unsure still about some of the inputs I used.
Not really accurate - gear ratio question remains.
Sime - see above questions if you don't mind weighing in.

But here is A chart.
Is it a meaningful chart?
Not really - but one step further down the road.
Bait on my hook.
Sime - can you help me to cast?

Screen Shot 2020-04-01 at 5.21.06 PM.png
Curve is 3rd one up - starts around 2500 and goes to about 8500

Screen Shot 2020-04-01 at 5.33.21 PM.png
Dyno chart from above curve. Curve did not look that bumpy when I cleaned it up. Is there a way to zoom in to do it more finely?

Screen Shot 2020-04-01 at 5.20.45 PM.png
Frist attempt. I just liked it so much I felt I had to post it

Ossessionato
76 Sprint V, 63 GL, 62 VBB, 05 Stella
Joined: 16 Jun 2011
Posts: 2853
Location: Nashville
Wed Apr 01, 2020 8:01 pm quote
You'll never get good results if you're using a compressed audio format, hence the specification to use a .wav file.

The compressed audio sounds fine to the human ear, but not to the computer.

interesting dyno's you managed to generate, though
Molto Verboso
62 VBB1T Round Tail W/ leaner sidecar
Joined: 26 Jan 2019
Posts: 1778
Location: california
Wed Apr 01, 2020 8:07 pm quote
Hey - I managed not to blow it up in my garage - or start a fire - so I have that going for me. 🙂

(glass houses... I know).

First time I saved a video in iMovie to WAV file - and it must have compressed it. Got that error message.
Second time I went online and did the conversion - and it created the WAV file that worked - its my lumpy one.

The chart that looks like I drove around the block was created before I figured out how to move the black line with the "eraser" thingy. Thats a little weird - but got hang of it as you said.

Question - can anyone help me understand my gear ratio from third? Doesn't a ratio have to have ":" in it?

Signed
-Confused
Hooked
1984 PX(BGM187)EFL
Joined: 14 Apr 2017
Posts: 384
Location: Cornwall UK
Thu Apr 02, 2020 12:48 am quote
Sorry, I haven't time to wade through that lot for a set of run data, which I can see is still incomplete anyway.
We already did a worked example with Chandlerman's wav yesterday, but if you need another, then please clearly post a run data summary for your single best run, with all the front sheet values in the correct units (or a specific question, on why you can't work it out, rather “can you do it for me”), and the wav for it; I might have time to look at it later.

This is far as I read before realising my time would not be well-spent on it:
charlieman22 wrote:
Few questions:
Picture from GSF you posted for key inputs that I wasn't sure about. Can you clarify?
1. Cd value - not sure what this is - GSF pages is extensive - but didn't see.
2. Frontal area - since I have a sidecar - should this be different than the .75 I assume is what you guys are using (are you?). Suggestions to calculate?
3. Rolling Coefficient - I have a third wheel slightly toed in - any thoughts on what value you would use for that one?
1) Yes, GSF pages extensive, what did you conclude?
2) You want someone 8500km away to tell you the frontal area of the scoot and sidecar in your garage?
3) My thoughts are that your third wheel might well warrant a different rolling coefficient; what are your thoughts?

Looks like you need to brush up your maths on gear ratios too. You've already got Chandlerman's 2nd, my 3rd, and Jack's 3rd and 4th as examples.

Your scoot and sidecar weigh more than half a Tonne? Blimey, it is pulling well.
Wow, it does get hot in California.

I skimmed further, and saw your graphs though; in your final post the image of the wav looks OK for rev-range, but poor for sound – you mic where it is is loud, but not clear (which actually seems appropriate).
Your output graph is not as bad as it might look, but you need to progress and tidy up your data presentation if you want me to look at it some more; you do the donkey work if you want me to spend time on it - or find yourself a donkey.

A little tip: When you've cleaned the wav in GSF, you can save it (see the box at bottom of GSF editing page); that way you can fiddle with the front sheet without having to reclean the wav each time – might help while you're having another stab at those 'too hard' values. Had we worked through the whole process in an orderly fashion, this and other similar little gems would have come to light, but let it flow now.........

Footnote: Internet update: The constant rebooting has stopped this morning, but I have a 'fault' on my line now. I have no landline, and my Internet is at ¼-speed, and there is no possibility of it being looked at for at least a week (and then, only if it's external to the building) – this will probably affect how much time I have to read and reply, and certainly how many files I can easily view, download or upload.
Ossessionato
76 Sprint V, 63 GL, 62 VBB, 05 Stella
Joined: 16 Jun 2011
Posts: 2853
Location: Nashville
Thu Apr 02, 2020 5:37 am quote
I'm going to agree with sime here. Between reading the doc's, google, and the tutorials above, all of your questions are already answered.

If you want to see the math behind gear ratios, it's all in my gearing spreadsheet. It uses LibreOffice with some custom macro's for the more interesting functionality, but all the basic stuff that you need will import into excel cleanly.

Past that, ensuring that you have a truly uncompressed recording with decent quality is key.

Anything else is Garbage In, Garbage Out.
Hooked
1984 PX(BGM187)EFL
Joined: 14 Apr 2017
Posts: 384
Location: Cornwall UK
Thu Apr 02, 2020 7:28 am quote
chandlerman wrote:
I'm going to agree with sime here. Between reading the doc's, google, and the tutorials above, all of your questions are already answered.

If you want to see the math behind gear ratios, it's all in my gearing spreadsheet. It uses LibreOffice with some custom macro's for the more interesting functionality, but all the basic stuff that you need will import into excel cleanly.

Past that, ensuring that you have a truly uncompressed recording with decent quality is key.

Anything else is Garbage In, Garbage Out.
Hallelujah! - and thanks for the reinforcement, Chandlerman.

I redid yours with Filter Length = '2'; paid more attention to where the gear change is (remembered to save the trimmed, cleaned wav this time as well), and you got half yesterday's losses back (20.8PS/bhp now). Posted it below 'cos it kinda shows what you can get out of GSF with decent data input, and taking a bit of care with it.



Internet update: Totally fecked – won't even reboot now, I'm on public unsecured Wifi on my laptop, no hub and no 'phone - proper social isolation!

Chandlerman's Output Curves with Filter - '2'.jpg
Chandlerman's GSF Output with Filter = '2'

Molto Verboso
62 VBB1T Round Tail W/ leaner sidecar
Joined: 26 Jan 2019
Posts: 1778
Location: california
Thu Apr 02, 2020 9:25 am quote
Chandlerman,
Used the gear calc to get my ratios.
Very useful to any who wants to determine.
Pic below is screen shot.
Makes my 3rd gear ratio to crank shaft turns 6.73.
Much appreciate the link.

WAV compression:
First WAV file I created gave me the error message - I posted thinking others may have gotten that error message - thought might point me in right direction. Suspect from Chandlerman's comments that the software is compressing.

I then went online and tried another software for converting and made a second WAV file. GSF software took that - no error message - used that to post my results.

Sounds like you guys seeing something that would suggest the second software I used - and the GSF software liked - was also a compressed WAV file? Can you clarify?

Front page also posted below.
Not sure what gives the impression the info going in isn't good - but happy for any constructive feedback.

Sime - Your posts have helped me sort my way through the software. Helpful.
You specifically asked if anyone had any questions - my request for insights on a few highly technical details - rolling friction coefficient and frontal area calc method - seemed inline. Are therer any insights or approach you might suggest for that?

Thanks
-CM

Screen Shot 2020-04-02 at 9.47.36 AM.png
Chandlerman's tool - link above in his last post. Great for sorting out the needed info for "gear ratio" box on software

Screen Shot 2020-04-02 at 9.56.33 AM.png
Front page - questions posed on rolling coefficient and frontal area calc approach were to see if I could improve data

Screen Shot 2020-04-02 at 10.00.55 AM.png
Curve after clean up. Agreed I can get a darker curve if I play with recording method. This was first effort - in process of refining.

Ossessionato
76 Sprint V, 63 GL, 62 VBB, 05 Stella
Joined: 16 Jun 2011
Posts: 2853
Location: Nashville
Thu Apr 02, 2020 11:09 am quote
From our friends at Google...
Rolling Resistance Calculation

Frontal Area and Drag calculation

Frontal area calculation is trivial and how to do it with a sidecar vs. a motorcycle or scooter, is also trivial extrapolation from there.

Drag Coefficients are a little trickier and best handled by either using a "typical" value and comparing expected calculations to actual observations in the field or using specialized simulation software.

Rolling resistance either needs to be calculated very specifically or just go with a "standard" value and maintain consistency of where you run your tests so even if your value is wrong, it's the same amount wrong over time.

The real value of a Dyno is for evaluating how a particular change impacted performance. Right now, you're worrying about precision when you don't even have a baseline yet. Get a good sound file, get a good curve built off that, compare it to your personal observations.

We're all standing on the shoulders of giants here. Asking the giants to stand on tiptoes because you're still not tall enough to see isn't a good look on anyone.
Hooked
1984 PX(BGM187)EFL
Joined: 14 Apr 2017
Posts: 384
Location: Cornwall UK
Thu Apr 02, 2020 11:11 am quote
charlieman22 wrote:
Makes my 3rd gear ratio to crank shaft turns 6.73.
Much appreciate the link.
Nope.
Chandlerman's spreadsheet is correct, your use of it is incorrect, so your result is incorrect; I bet Chandlerman can see why in a jiffy – I'll maybe look in the morning to see if you've taken a bit more care with it, or had someone give you the answer. In any case your front sheet was still not even changed to what you thought was correct; it still says 6.32, not even the 6.73 you stated, and certainly not the correct ratio. Rushed again; and I've just wasted more time checking it and finding the careless error.

Gotta say I've pretty much lost interest in helping with this.
Molto Verboso
62 VBB1T Round Tail W/ leaner sidecar
Joined: 26 Jan 2019
Posts: 1778
Location: california
Thu Apr 02, 2020 11:28 am quote
Sime - the image posted was from my first attempt to use the software - as I have noted.
I did not have an understanding at the time on calculating the gear ratios - and I asked for some help doing so.

I'm in this to learn from the group - teach something if I can - have a laugh - and share a laugh.

Let's just stay focused on the good stuff.
-CM
Ossessionato
76 Sprint V, 63 GL, 62 VBB, 05 Stella
Joined: 16 Jun 2011
Posts: 2853
Location: Nashville
Thu Apr 02, 2020 11:33 am quote
Sounds good, CM. I took your screenshot to be the config you were using currently, too. Specificity helps a lot on the Internet

Also, make sure your primary & secondary tooth counts are correct when calculating ratio's. I'm not running stock gearing, so it's probably throwing your calculation off if you didn't correct that.
Veni, Vidi, Posti
74 Super, 75 Super, PX project, LML off-roader and new to 2018, '66 Blue Badge Smallframe
Joined: 30 Nov 2011
Posts: 7955
Location: Victoria, Australia
Thu Apr 02, 2020 6:19 pm quote
Charlieman - don't know what you're like on 3D applications, but if it helps I can build a partial model in Sketchup and measure it for you. There's a useful bit of info here -
https://info.simuleon.com/blog/how-to-calculate-drag-coefficient-for-motorcycle

Happy to help any time mate, I know in a short time you'll be all over it and passing on your knowledge to others.
Molto Verboso
62 VBB1T Round Tail W/ leaner sidecar
Joined: 26 Jan 2019
Posts: 1778
Location: california
Thu Apr 02, 2020 9:04 pm quote
Ginch - Modeling offer is awesome - thanks.
I was chasing the Dyno test so I had something to measure the Great Cylinder Run Off (working title - needs work) between the Polini on crack, the blemished but nearly free BGM, and the shinny new Malossi Sport.

Now that I've played around with the software a bit - I think we can get a pretty tight comparison of the cylinders in relative terms by simply keeping the inputs constant.

As to what those constants should be set at - if we do a decent job using SWAGs - I think we can get decent absolute numbers also - without going full modeling. Keep in mind - there is a mud guard, and an open cockpit that I should probably have a tonneau cover over if I wanna hit top speed. Not sure we will be able to more accurately predict with measures than we could with approximations.

For example. Sime and Chandlerman are using .75 for "Frontal Area". I have about 20%-30% more (guessing but can measure) so we could make that .9 - .95. Is it perfect? Meh - but its probably not bad.

Rolling coefficient is tougher. Chandlerman posted a link (thanks!) and it has some charts - but with the toe in and scrub caused by my leaner set up - I am in my own stratosphere. When I push my scoot - and I lean - I can feel and hear some scrub (inherent in a leaner). So with the third wheel and the scrub effect - I am proposing we take the std. .015 and add 75% = .026 - .03

Are those too conservative?
Are they too aggressive?
Will they overstate my HP?
F*4# I hope so.
🙂
Open to some SWAG inputs as well.

Screen Shot 2020-04-02 at 9.59.33 PM.png
Inputs might look something like this? Gear ratio still in question?

Hooked
1984 PX(BGM187)EFL
Joined: 14 Apr 2017
Posts: 384
Location: Cornwall UK
Thu Apr 02, 2020 9:22 pm quote
Note: this is not a reply to your latest, which I haven't read; we posted at the same time.
charlieman22 wrote:
Sime - Your posts have helped me sort my way through the software. Helpful.
You specifically asked if anyone had any questions - my request for insights on a few highly technical details - rolling friction coefficient and frontal area calc method - seemed inline. Are therer any insights or approach you might suggest for that?
The GSF literature, which you have studied, says:



Cd Value:
*“Most of us has to guess”, implies it isn't crucial.
*“between 0.3 & 1” gives the range within which to guess (experiment); a sidecar might suggest the higher end – I have claimed no knowledge of sidecars.
You'll remember me stating previously that consistency within your own settings is the most important point; I certainly remember doing so.

Frontal Area:
*“together with the Cd value”, so it's combined with a 'guessed' value, therefore implication still applies that this isn't critical.
*My reminder that I've already stated consistency within your own settings is the most important point still applies.
*How about just measuring it?

Rolling Friction Coefficient:
*Doesn't even warrant a mention in the literature; Google for sidecar clues if concerned, or use GSF forum/email, or use the same as everyone else.

There is an email address at the bottom of the GSF literature, and the GSF Forum is also available for those few "highly technical details"; tried either of those?

For all of the above, as I've said a few times in preceding posts, saving a cleaned wav within GSF Dyno, enables you to see your output curves, then fiddle with the front sheet afterwards. As with weather boxes, these are not crucial, certainly not as crucial as good test technique on the road, and decent quality wav recordings.

There is nothing preventing you from finishing this yourself now, applying it to all future engine set-ups and test runs on your build thread, and leaving me to tidy this thread up in my own time.

Just for interest, Chandlerman, I ran this with low and high values for Cd value and Frontal Area, and it does change the power, but that's extreme ends of range, and not the shape of the curve anyway; this could be investigated further by anyone for whom it mattered, but as we've both said, after good testing and recording technique, consistency within the individual's method is the most important point. In any case, that's sidecar stuff, not 'Tips on using GSF Dyno' stuff.

Below is the the result of incorporating everything discussed above for the front sheet; collated, corrected and converted where necessary; using a wav I extracted from your Ride 1 Charlieman – Output Graph only is shown; the rest is already here above in preceding posts.

Fin

CM Output Graph after front sheet details added but without drag etc adjustment..jpg
GSF Dyno from CM's 'Run 1', with front page info gathered from preceding posts, but Cd, FA & RFC left unchanged for sidecar.

Member
'13 LML '70 Sprint Veloce
Joined: 01 Aug 2019
Posts: 47
Location: London
Fri Apr 03, 2020 12:16 am quote
I'm not sure that compressed audio would harm the results too much- the software probably can't deal with some file types because it doesn't have access to the codec libraries required to decode it, hence it needs to be an uncompressed WAV file.

This could be easily verified by converting a wav to mp3 and back to wav again and seeing if the introduced compression artefacts create any differences in the measurement.

This is an important point because some of the audio recorders that most people will have (a smartphone or video camera) could potentially use hardware accelerated compression (ie they compress as part of the recording process) and I'm not sure that they should be written off just yet.

I'm reasonably confident that the voltage divider method recorded as an Mp3 and on a phone and transcoded to WAV would give more accurate results than a microphone recorded natively as wav, for example.

If anyone wants to send me some files I'll happily do some analysis or conversions to help verify!
Ossessionato
76 Sprint V, 63 GL, 62 VBB, 05 Stella
Joined: 16 Jun 2011
Posts: 2853
Location: Nashville
Fri Apr 03, 2020 4:35 am quote
sime, again you make an excellent point. To estimate the impact of (in)accuracy of Cd & Frontal Area, you can use the same sound file and vary those parameters to see how much the change the results. I also posted a link on how to estimate frontal area, but you could perform the same experimentation with that.

If the goal is to compare three different motor configurations, then holding those variables constant eliminates them from that equation. If the goal is to get a truly accurate measure of horsepower, spend the cash to go get an hour of dyno time at the local cycle shop. You can also then use that information compared to the GSF readings to get an estimate of how much the two differ.

It's also worth remembering that even a true dyno is only accurate to within 5-10%, so much of this is chasing after precision that simply doesn't exist.

As to rolling resistance, as it states in the link I shared, paraphrasing: "rolling resistance is difficult to calculate because it is such a small value. For a bicycle on asphalt, it is about 10 grams."

If you think of this problem in terms of Significant Figures and the limits of dyno accuracy in general, that makes it pretty much irrelevant.
Molto Verboso
62 VBB1T Round Tail W/ leaner sidecar
Joined: 26 Jan 2019
Posts: 1778
Location: california
Fri Apr 03, 2020 7:23 am quote
Quote:
go get an hour of dyno time at the local cycle shop. You can also then use that information compared to the GSF readings to get an estimate of how much the two differ.
Below is a sample graph of what happens each time you add .015 (the default amount) to your rolling coefficient - all other things held equal.

Graph inputs for rolling coefficient:
Bottom is .015
Middle is .030
Top is .045

It's linear - and exactly 1HP.
I suppose that also means the .015 assumes you use 1HP of lost energy to rolling coefficient
As I noted above - I can actually feel and hear the friction on mine - but its also possible that it falls away or is not linear at 65MPH - and more prevalent at 1MPH. Hard to know.
The key for mine is the toe-in that creates scrub.

Chandlerman - couldn't you find a link with the numbers for a 62VBB with a leaner side car to send me? Come on man!
🙂

Love idea of calibrating my numbers with Dyno - and could always do retro-actively if I ever get the chance. Good suggestion.

Screen Shot 2020-04-03 at 8.12.18 AM.png

Veni, Vidi, Posti
74 Super, 75 Super, PX project, LML off-roader and new to 2018, '66 Blue Badge Smallframe
Joined: 30 Nov 2011
Posts: 7955
Location: Victoria, Australia
Sat Apr 04, 2020 2:58 am quote
Here is a much closer calculation of frontal area, not sure about cd as yet.
http://www.vespalabs.org/projects/vespa-cfd-3d-model/openfoam-vespa-tutorial
Molto Verboso
62 VBB1T Round Tail W/ leaner sidecar
Joined: 26 Jan 2019
Posts: 1778
Location: california
Sat Apr 04, 2020 9:51 pm quote
Ginch - good stuff.
My quick take on the link - he shows how different means of modeling give different calc for frontal area.
Copied from his page:
Quote:
(I calculated an area 0.660839 compared to 0.64438 for Rhino3D below and 0.506266 for the method above).
The .64 is pretty much what the GSF uses - and since I think its been calibrated vs dyno - probably a good number to go with as starting point.
I've added about 50% for my side car - its a SWAG - but probably reasonable assesment.

Below is my set up.
I liked Chandlerman's suggestion - of calibrating to actual dyno at some point
I could always go back and do that.
For time being - want to lock in on some settings and methods.
Wish my curve looked prettier - is there any means to zoom in to fine tune black line smoothness?

Questions for the crowd.
1. What would you guess my HP is?
Generally Jackified motor
No hole left unmolested.
Long duration inlet/crank
Unreasonably large ex. port
123/177 ish timings.
24/24 SI
Etc.
Does below seem about where you would expect?


2. Why do I get that weird blip about 6K RPM - can't see it on my blue graphs. Bump in the road?!

CM's #s.jpg
My inputs

Screen Shot 2020-04-04 at 9.06.00 PM.png
Setting RPM to fairly tight window of 2000-9000 magnifies the curves making it easier to erase and place the black line correctly

Screen Shot 2020-04-04 at 9.07.43 PM.png
Here its erased and on the line. I can't see the blip at 6KRPM that my curve seems to always have.

Screen Shot 2020-04-04 at 9.46.03 PM.png
Result. dip at 6K - not visible in my blue screen - which is strange. Is this where you would expect the power band? How can I smooth

Veni, Vidi, Posti
74 Super, 75 Super, PX project, LML off-roader and new to 2018, '66 Blue Badge Smallframe
Joined: 30 Nov 2011
Posts: 7955
Location: Victoria, Australia
Sun Apr 05, 2020 3:38 am quote
Set the filter length higher?
Ossessionato
1980 P200E , 2005 Stella 177
Joined: 20 Jul 2018
Posts: 2189
Location: Staten Island, NY
Sun Apr 05, 2020 3:43 am quote
Jack221 wrote:
The run I used on my last one is a slip road too. Slightly uphill for about the first mile but doesn't seem to make any difference anymore. Coming down hill the other side of the road it goes no faster. The wind becomes the biggest factor at speed on a scooter.
With the GSFdyno sound chart I struggle to pick the right bit. Is there an easy way to know which bit is relevant?
I just make a mental note of what rpm i start my run at and what rpm i peak at the stop the recording between runs. Bring it into the app and find the curve that closest matches where i start and and peak and wipe everything else away
Ossessionato
1980 P200E , 2005 Stella 177
Joined: 20 Jul 2018
Posts: 2189
Location: Staten Island, NY
Sun Apr 05, 2020 3:45 am quote
You should also measure your tire circumference or find the tech specs for your tire to enter that info as i noticed it changes the results a decent amount. And estimate the air temp. Try to fill in as many of those fields as you can.
Molto Verboso
62 VBB1T Round Tail W/ leaner sidecar
Joined: 26 Jan 2019
Posts: 1778
Location: california
Sun Apr 05, 2020 7:56 am quote
Morning ya'll.
Continue to tighten up inputs.
Here it is rerun as per Ginch's suggestion - and sure enough - smooths out.
Funny - had played with making it smaller to no positive effect.
Nice call Ginch.
Tks!

Also adjusted tire circumference based on Swiss's insight - previously used looked up specs - found flexible tape measure from wife's sewing drawer and did actual measure this am. Now updated.


All other things remained unchanged from last set of graphs above - and same input data - now give a nicer looking curve. Thanks all.

Screen Shot 2020-04-05 at 8.42.01 AM.png
Tire circumference and filter length changed

Screen Shot 2020-04-05 at 8.41.17 AM.png
Blue screen looks exactly the same. Means software wants the gross curve- but is listening to all the subtle details - not shown when you identify curve

Screen Shot 2020-04-05 at 8.41.45 AM.png
A more elegant dyno chart! Doesn't look like I was driving on a gravel road any more

Ossessionato
76 Sprint V, 63 GL, 62 VBB, 05 Stella
Joined: 16 Jun 2011
Posts: 2853
Location: Nashville
Sun Apr 05, 2020 9:57 am quote
That looks much better, and probably is fairly accurate for your build, CM
Molto Verboso
62 VBB1T Round Tail W/ leaner sidecar
Joined: 26 Jan 2019
Posts: 1778
Location: california
Sun Apr 05, 2020 10:10 am quote
Thanks CM - from CM.
Hat tip to Ginch on filter adjustment.

For others that might experiment with this software: I have been using a plug in Mic to my iPhone as means of trying to get a nice bright curve on that blue screen - makes for quicker alignment of the black line on top of it with the software.

The Mic is likely picking up so well vs straight iPhone - that it is overwhelming the software with information.
The filter adjustment of 3 is likely the right move if you use a remote mic as I did.

Also - Chandlerman - appreciate the comment on likely range of HP results.
Assume I am in the ball park as well.
Still like the idea of finding a Dyno at some point to calibrate against.
Then we can have even better version of data.
All my iterations fo cylinders will be done with these settings - which I saved on home input screen with "Save" button at bottom.

To pull them back up - I just hit "Load" when I open the software - and choose the file I named "Speed Run Final" (or something like that... Might be speed run final final. or speed run final final final.)
Veni, Vidi, Posti
74 Super, 75 Super, PX project, LML off-roader and new to 2018, '66 Blue Badge Smallframe
Joined: 30 Nov 2011
Posts: 7955
Location: Victoria, Australia
Sun Apr 05, 2020 7:13 pm quote
charlieman22 wrote:
Thanks CM - from CM.
Hat tip to Ginch on filter adjustment.
That was Sime's tip to Chandlerman, I was just repeating the message... but glad you got a nice curve. Looking good!
Ossessionato
1980 P200E , 2005 Stella 177
Joined: 20 Jul 2018
Posts: 2189
Location: Staten Island, NY
Sun Apr 05, 2020 7:15 pm quote
Not bad results!
Molto Verboso
62 VBB1T Round Tail W/ leaner sidecar
Joined: 26 Jan 2019
Posts: 1778
Location: california
Sun Apr 05, 2020 7:17 pm quote
Quote:
That was Sime's tip to Chandlerman, I was just repeating the message... but glad you got a nice curve. Looking good!
Takes a village.
getting there.
Molto Verboso
62 VBB1T Round Tail W/ leaner sidecar
Joined: 26 Jan 2019
Posts: 1778
Location: california
Tue May 12, 2020 5:32 pm quote
Gents - some discussion about recording directly.
Have converted to bellows on my carb - and recording seems to be quite different now.

Has anyone successfully built some form of direct line in capture machine?
Swiss?
Craig?
Others?

Interested to understand what's in it.
Soldering gun warmed up and at the ready.
🙂

-CM
Missouri Loves Company Rally   Vespa Wasp Pin Badges   Yelcome Leather Top Cases and Roll Bags for Piaggio Vespa PX LX LXV GTS GTV
Post Reply    Forum -> Not-So-Modern 12Next
[ Time: 0.2130s ][ Queries: 25 (0.0511s) ][ Debug on ]