Turbo Vado 2 4.0 or 5.0 or 6.0?

[mschwett]

A little more context.

Here's the individual chart for 40/100 with all records where Cadence and Pm and Pr are >0
View attachment 205986

And here's the chart with records where cadence > 59.
View attachment 205987

These charts were done in LibreCalc. It did the best fit linear trend lines. For the combined chart above I estimated the Pmax to be 380 since it’s a good assumption that Pmax does not change with Pr.

nice

the 1.6x is a bit higher than you’d expect 40% ease to deliver on an SL2, and it’s quite linear. “2.7x you” at 100% ought to be more like 1.1x you at 40%, but you’re seeing 1.6x. if the numbers are electrical, that’s really 1.3 and not so far off.

 

[mschwett]

Boost ratio and Pm as a function of Pr
View attachment 205988

ok, this starts to explain things. toss all the values where Pr is < 50 or > 200, they’re likely not steady state values in this sample set and there is enough delay in the motor starting and stopping that the values on the edges of transitions are just noise.

i’m betting it’s much closer to the predicted 1.4 ish electrical….

 

[Jeremy McCreary]

It's with some concern that I post this chart as it refutes my whole understanding of how the xx/xx and xx/100 tunes work.

Ditto.

I rode up and down the same street in approximately 1 mile laps in 20/20, 20/100, 40/40, 40/100, 70/70, 70/100 and 100/100. The chart shows Pm as a function of Pr for each segment. The resulting .fit file had 1880 records after I removed each record with cadence < 60. Admittedly these are still very limited rides and I'm not strong enough to pedal at much more than 200 watts for more than a few seconds at a time so there's not any data at higher Pr values. The chart lines end at the highest Pr I achieved at each tune setting.

Nice work. Thanks!

This chart was created using the slope and Y intercept from the linear trend line up to the clear inflection point when Pmax was reached.


View attachment 205980

I had expected XX/XX and XX/100 to have the same slope but the XX/XX curve to go flat at a lower Pm value (the saturation point we've bee talking about). Not so in any of the tunes. In each case XX/100 has a significantly steeper slope than XX/XX.

Tested the four X/X runs by visually estimating the Pm values at Pr = 50, 100, 150, and 200W. As you stated, the slopes of the ramps up to saturation are NOT proportional to X. Above X = 40, slope growth with X is significantly faster than linear at every Pr.

At least it's nice to see that the ramps are straight lines.

I expected the inflection points where Pm=Pmax to occur at the Pr values given the Specialized table showing Pr to get Pmax in each tune.
Tune____________Pr to get Pmax._____
_____________Spec table_____my data____comment
100/100_________85___________70_______pretty close
70/100__________268__________170______way off
70/70___________185__________200______pretty close
40/100__________436__________250______way way off

My simple Pm model based on earlier Spec documents predicts that SL 2 saturation should always occur at 139W in any X/X mode. Nothing like in either your data or Spec's recent SL 2 saturation tables.

Don't see any way to salvage that simple Pm model at this point. Hence, the underlying Spec documentation was pretty misleading. Will have to put up retractions on the old posts where my model was introduced.

I'd love for someone else try a similar exercise and see how your results compare.

Happy to help but have to get my !@#$^&*! bilateral plantar fasciitis under control first. Still riding in moderation with permission from doc and physical therapist, but high pedal forces definitely stir things up.

 

[mcdenny]

ok, this starts to explain things. toss all the values where Pr is < 50 or > 200, they’re likely not steady state values in this sample set and there is enough delay in the motor starting and stopping that the values on the edges of transitions are just noise.

i’m betting it’s much closer to the predicted 1.4 ish electrical….

By eliminating all the cadence < 60 records most of the low Pr records are gone too.

For 100/100 slope is 5.5, cutting 20% to get mechanical power would make it 4.4. Seems unlikely though it would be "4.4 times you."

 

[Jeremy McCreary]

Boost ratio and Pm as a function of Pr
View attachment 205988

The good news about your last few graphs: Electrical Pm is strongly linear in Pr at 40/100 assist — at least before the approach to saturation. And since motor efficiency is roughly flat at cadences > 60 rpm, mechanical Pm is also linear in Pr.

Is Pm this linear in Pr in the other assist modes tested? If so, our SL 2 reverse engineering project still has a chance.

 

[mschwett]

By eliminating all the cadence < 60 records most of the low Pr records are gone too.

For 100/100 slope is 5.5, cutting 20% to get mechanical power would make it 4.4. Seems unlikely though it would be "4.4 times you."

according to the 40/100 chart you posted, the average Pm/Pr between 50 and 200 watts Pr is 1.5. 1.2x after efficiency. not too far off from "2.7x you x 40%" which would be 1.08x

how many samples do you have for the 40/100 where cadence > 60 and power > 50 and < 200 or so? ideally there's 5 min or so of steady state data for each setting, so a few hundred samples before smoothing and 50 or more after.

i would caution though that specialized doesn't seem to tou the "Y.Yx you" thing any more, either because their marketing people have moved on or it's no longer as simple as that

i'll repost this forum post from specialized themselves in 2018, which explicitly stated the boost factors for a couple of the systems at the time. and of course support has now been renamed "ease." the SL was sometimes noted as 1.8, sometimes as 2.0, and i recall but can't find reference to the SL2 being a third more, which would track with the one third higher power.

Good question - firstly - here's a link to the User Guide. Secondly, below is more detailed clarification that may help some.

Support = the multiplier of rider power added by the motor
Peak Power = maximum electrical power available to supply the motor

In terms of support, each e-Bike motor is slightly different. 100% on the slider means 4.1x rider power for 2019 Levo (2.1 motor), 3.8x for 2018 Levo (1.3 motor) and 3.2x for older Levo (1.2)

When you talk about motor power, bear in mind that two types of power regularly get confused when discussing e-bikes:

a) Electrical input power (battery power that the motor consumes)
b) Mechanical output power (motor power that is added to rider power at the crank)

Mission Control (+ all other apps using ANT data) measure the electrical input power to the motor (Battery Voltage x Current). This is what you see if you view the stats page whilst riding, and it's what you limit when you reduce peak power - if you reduce peak power to 50% you're effectively limiting the motor to draw up to 50% of maximum permitted battery current. Reducing peak power therefore helps you to preserve battery life/range.

In the real world, what you actually care about/feel is the mechanical output power of the motor. Motors are not 100% efficient, some power is lost to heat so let's use 75% efficiency as an arbitrary estimate. In this example, for every 100W of electrical power the battery supplies to the motor you'd get 75W mechanical power at the crank.

However there are two other things to bear in mind - YOU are not 100% efficient either, often riding in the wrong gear and asking the motor to do all the hard work - PLUS the motor is happiest when you ride in the right gear at a good cadence - let's say 70-90 rpm. So when you're riding in optimum gear and around 80rpm, you're turning 75% of the electrical power consumed into mechanical power - but when you're riding in the wrong gear at 40rpm uphill don't expect the same results!

Below is a graph (data presented by us but measured/validated by an independent lab) showing how power varies with cadence for a number of motors - you can see that aiming for a constant cadence of around 80-90rpm will not only give you optimum power but also efficiency. Whether you are after power or range, pedaling at the right RPM makes a big difference.

To put all of this together, let's talk through the default support settings in Mission Control. You can change these to suit your riding style at any time, but let's use these as an example. (eco 35/35, trail 35/100, turbo 100/100)

In Turbo mode on a 2019 Levo with 4.1x rider support at 100%, you'll only need to put in about 140W rider power for the motor to supply you with the maximum assist of 560W (taken from the graph above), assuming you're spinning at about 80-90 rpm. Lots of fun, also useful to get to the trail but possibly too powerful for steep climbing - especially if the ground covering is loose.

On the same bike with Support set to 35%, you're getting approx 1.5x rider support. This will feel much more natural to ride and you'll use less battery - in Trail mode with 100% peak power the motor will still reach it's maximum assist of 560W but this time you need to put in 390W rider power to get there. You get more of a work out and in practice you'll use far less battery. Aim for this mode to give you the best blend of power and efficiency so that you also have optimum control for trail riding.

In Eco mode with support set to 35% you still have approx 1.5x rider support - however you're now limited to 35% peak power (35% of 560W is approx 200W maximum motor assistance). In practice this means that you'll get linear support from the motor up until you put in about 140W rider power (200W added from the motor), but above 140W rider power you'll still only receive 200W support from the motor. This is great to preserve battery life but if you hit that peak support limit whilst climbing something steep, you have to make up all the extra with your legs.

Note: All power mentioned in this Mission Control example is mechanical power output by the motor. If you look in the stats screen of Mission Control whilst riding you will see electrical power consumed by the motor which will be higher. Re-read this article a few times and if still completely unsure...

 

[Jeremy McCreary]

i would caution though that specialized doesn't seem to tou the "Y.Yx you" thing any more, either because their marketing people have moved on or it's no longer as simple as that

Beginning to think it was never that simple. The Spec documentation I pored over in late 2024 — current at the time — strongly suggested that the pre-saturation mechanical Pm you get in mode E/M is proportional to both Pr and E.

Marketing no doubt had a say, but those documents read like the engineers were also involved.

Now we have data showing that pre-saturation Pm is reasonably linear in Pr but more like quadratic in E.

We're on our own.

 

[mschwett]

Beginning to think it was never that simple. The Spec documentation I pored over in late 2024 — current at the time — strongly suggested that the pre-saturation mechanical Pm you get in mode E/M is proportional to both Pr and E.

Marketing no doubt had a say, but those documents read like the engineers were also involved.

Now we have data showing that pre-saturation Pm is reasonably linear in Pr but more like quadratic in E.

We're on our own.

I’ll believe that when I see it documented on a long-ish steady state effort - preferably a shallow climb - at similar cadences across a few different E And Pr. nothing in my experience with the SL1 suggested it was ever as far off from the simple model as that would be. it would also completely throw range estimates to the bin, given how much it would require the model to know the (wildly variable) assist factor.

 

[mcdenny]

Happy to help but have to get my !@#$^&*! bilateral plantar fasciitis under control first. Still riding in moderation with permission from doc and physical therapist, but high pedal forces definitely stir things up.

Getting old sucks!

 

[Jeremy McCreary]

Getting old sucks!

It does indeed — one weird thing after another.

 

[Jeremy McCreary]

I’ll believe that when I see it documented on a long-ish steady state effort - preferably a shallow climb - at similar cadences across a few different E And Pr.

Remind me why on a climb? You also suggested testing on a climb earlier in this thread:

i'd find a hill of around 5%, ideally at least 300' of climb, no stops, keep cadence around 80, heart rate at a steady sustainable value for rider watts in the 100-150 range, do it once at 20/100, once at 40/100, 60/100, 80/100, and 100/100. you won't be in danger of hitting 28mph (heh) and the gearing on your vado should allow 80rpm for all of those scenarios. 20/100 may be a little slow, i'm guessing that'll be around 8mph, more like 65rpm. you could do 25/100, 50/100, 75/100, and 100/100 if you want to keep the cadence and speed ranges a little tighter. of course the results will be more interesting the less time you max out max, so you want to stay below 150 rider watts. a 4% grade would be better but harder to find.

Found and climbed a hill kinda like that yesterday: 305 ft over 1.0 mi, variable 5-8% gradients, 5.8% average. One light that probably wouldn't stop me. That's the closest I'll get to a longish, steady undisturbed climb anywhere close to home.

Wouldn't relish climbing that 4-5 times in one ride given the distance from home, but doable over 2 rides.

Could've done the whole thing in my ECO (35/45) at 80+ rpm if my riding buddy hadn't insisted on checking out an even steeper side street halfway up. (Knees demand 80+ rpm anyway). HR would probably have stayed in your recommended range.

But not sure how 20/100 would've gone.

nothing in my experience with the SL1 suggested it was ever as far off from the simple model as that would be. it would also completely throw range estimates to the bin, given how much it would require the model to know the (wildly variable) assist factor.

Good points. Would love to end up with
a simple Pm model for steady-state riding — especially one that estimates saturation Pr for a given E/M on my SL 1. That can be useful info in the saddle. But the data currently available for the SL 2 doesn't inspire hope.

Could they have changed the Pm algorithm since the SL 1?

 

[mschwett]

Remind me why on a climb? You also suggested testing on a climb earlier in this …. especially one that estimates saturation Pr for a given E/M on my SL 1. That can be useful info in the saddle. But the data currently available isn't looking good.

i think testing a fairly powerful bike like the SL2 with a range of E from 40 to 100 will likely bump up against the edge cases of speed and rider power without a little gravity thrown in the mix. a steady hill is a consistent known load unlike the mystical vagaries of aero drag at > 20mph, making for varying load and by association varying rider powder, cadence and then motor power.

as an example the SL2 at 60/100 with a nice steady 150w rider power “should” be putting around 200 watts of mechanical power. at 350w we’re talking about 24-25mph probably, even the slightest tailwind or decline and we’re at 28, and we all know the power feathers out at least 1-2 mph before that. even without the 28mph limiter the 44x11 top gear of the SL2 4.0 gives us 85rpm at 28mph. practically speaking this test can’t really be done beyond 60/100 on level ground. much easier to do it at 20mph up a hill with gears on either side to maintain cadence if you slow down or speed up a bit, no risk of hitting the limiter, and half the kinetic energy if you crash because you’re paying too much attention to your phone

 

[mschwett]

i also have to say … realizing how much more powerful this bike is than my creo was, i’m kind of wondering if the ideal replacement for my commuter would actually be a vado SL2 4, with the drivetrain replaced by an electronically shifted rohloff and veer belt kit! a fairly light 320w mid-drive bike with the gear range of a rohloff could probably climb my hills and be carried up and down subway and train stairs way easier than a full power bike!

 

[Jeremy McCreary]

i also have to say … realizing how much more powerful this bike is than my creo was, i’m kind of wondering if the ideal replacement for my commuter would actually be a vado SL2 4, with the drivetrain replaced by an electronically shifted rohloff and veer belt kit! a fairly light 320w mid-drive bike with the gear range of a rohloff could probably climb my hills and be carried up and down subway and train stairs way easier than a full power bike!

I feel a test ride coming. Let us know what you think!

 

[mcdenny]

mschwett said
"how many samples do you have for the 40/100 where cadence > 60 and power > 50 and < 200 or so? ideally there's 5 min or so of steady state data for each setting, so a few hundred samples before smoothing and 50 or more after."

the curve above has 236 data points (seconds) fitting cadence > 59, Pr > 40, Pr < 201. 3.9 minutes of riding.

 

[mcdenny]

Here's the 100/100 curve, the instance that fits the Spec table data the best.

Blue dots and lines are my data from .fit file. The red line is defined by Pm = 0 at Pr = 0 and Pr = 85 at Pm = 380. In other words rider effort of 85 watts gets max motor watts. Looking at the dots and lines it is very plausible that the Spec defined red line is the correct one.

The red line slope = 4.50, adjusting for mechanical power would give slope = 3.6. 3.6 times you???

The 70-70 curve fits the Spec table data pretty well too but the others are way off.

 

[Jeremy McCreary]

Here's the 100/100 curve, the instance that fits the Spec table data the best.

View attachment 206006

Blue dots and lines are my data from .fit file. The red line is defined by Pm = 0 at Pr = 0 and Pr = 85 at Pm = 380. In other words rider effort of 85 watts gets max motor watts. Looking at the dots and lines it is very plausible that the Spec defined red line is the correct one.

Strong work!

The red line slope = 4.50, adjusting for mechanical power would give slope = 3.6. 3.6 times you???

Before saturation in a given assist mode E/M, my simple Pm model used

Pm = B (E / 100) Pr,

where Pm is mechanical, and B is a constant boost factor supposedly baked into the motor. For the 1.2 SL motor in the SL 2, Spec gives B = 2.3.

Since you've confirmed a linear pre-saturation ramp passing through (Pm,Pr) = (0,0), the ramp slope S is just

S = Pm / Pr = B E / 100,

where Pr and Pm can be taken at any single point on the ramp. This slope should be constant at constant E — just as your data shows.

But at E = 100, the simple model predicts

S = B = 2.3,

which is WAY OFF your empirical mechanical S of 3.6.

Clearly, something's amiss. In this 100/100 case, a constant fudge factor F = 3.6 / 2.3 = 1.57 would fix this discrepancy like so:

Pm = F B (E / 100) Pr

Later today, I'll try to find the F factors implied by the empirical ramp slopes in your other test runs.

 

[mschwett]

here's what i think is happening.

when specialized changed to "ease" from "assist" they changed one very significant thing - the rider/motor power ratio for any given "ease." i strongly believe the relationship was linear on the SL 1 motors with the mission control system and associated era firmware. no matter how many of my old rides - thousands and thousands of miles - i don't find any where the relationship of rider and motor power falls outside of the predictions, including the tests where i changed assist in 10% incremements from 10 to 100.

i have two theories here. 1: we have observed that on the old system, motor performance at very very low assist levels was very bad. i think it just doesn't work well at such low power levels, so they've adjusted those "ease" settings to actually do something rather than just waste power. 2: when people use turbo, they want turbo. they want to feel the boost and the specialized bikes were suffering by comparison to similarly powered bikes, in both reviews and on the road, so the settings at the higher end are now boosted.

the great documentation they've produced on the system gives us 9 good datapoints at the 10% intervals plus 35, 55, and 75 from their "examples" on the effect of different settings. lay those 12 data points out and you get a fairly rational curve. it makes sense that they wouldn't just randomly fool around with the values, they would want changing the settings a notch or two at a time to gradually change the experience, especially with micro tune. @mcdenny has arrived at a value for the assist ratio at ease 100 which comes damn close to matching what specialized states.

a minor note on power and efficiency : let's remember that max motor mechanical power on the SL2 is 320 watts (not 300) and we know it is achieved with 400 watts of electrical power at 95rpm. we also know that at 80rpm, only 290w of mechanical power is achieved - but we don't actually know if that still takes 400 watts of electrical power. @mcdenny what peak motor electrical power are you seeing at 80rpm? if it's still 400 watts, i think we can say (from specialized's power chart) that efficiency is more like 75% at best at 80rpm and full blast than 80%....

 

[Jeremy McCreary]

the great documentation they've produced on the system gives us 9 good datapoints at the 10% intervals plus 35, 55, and 75 from their "examples" on the effect of different settings.

Could you specify this data, just so we're all on the same page?

lay those 12 data points out and you get a fairly rational curve.

Could you show us exactly how you got that curve — presumably the one below?

View attachment 206036

Based on all the SL 2 data we have now — from both Spec and @mcdenny — totally agree that the slope S of the pre-saturation ramp in the SL 2 MUST be a nonlinear function of ease E. Then we have

Pm = S(E) Pr

Now it's just a matter of reverse-engineering the function S(E).

Your graph is effectively a graph of S(E). The flattened convex-downward shape looks roughly quartic — i.e., dependent on the 4th power of E. My own analysis of Spec's saturation Pr data suggested a monotonic (constantly rising) and roughly quadratic S(E), but I'll revisit it before sharing.

 

[mcdenny]

A new trial, a little longer and with more rider effort to more clearly define Pm saturation values. I just did 40/40 and 40/100.

Summary from the Spec app:
____________________Distance______time_______avg MPH_______Adj Rider power______avg motor power
Ride one 40/40_______1.35_________6:26________12.6_____________126_______________80
Ride two 40/100______1.30_________5:42________13.7_____________130_______________125

.fit file data

BTW, 40/40 slope of 0.70 compares to 0.66 from last weeks experiment shown on page 10
_____40/100________1.51___________1.63


and my comparison plotting the functions together:

If you squinted you might conclude Pr saturation is the same for both 40/40 and 40/100.
40/40 Pm max is about half of xx/100 Pm max.
Slope of 40/100 is about double 40/40 slope so Pr saturation is the same in either case.
Neat - check; Logical - check; Correct - ?????

I'm going to repeat the route in 100/40. I expect to see a much steeper slope to a Pm max = 40/40 Pm max.