I have both Inovelli Red Series VZW31-SN dimmers (~10) and couple Lutron Caséta dimmers installed in my home, controlling the same model of dimmable LED bulbs.
The difference in dimming behavior is substantial.
With the Inovelli, holding the paddle to dim the lights produces relatively little perceived brightness change through much of the upper range. Most of the noticeable change occurs near the bottom, where the lights suddenly become much darker.
The Lutron Caséta provides a much more natural and consistent perceived brightness transition using the same bulbs.
This is not primarily a dimming-speed issue. Slowing down the ramp doesn’t change the distribution of perceived brightness across the dimming range.
It’s also not something I want to solve by reducing the maximum dimming level. While it that can help make it feel like the dimming is more gradual over the range, it reduces that actual amount of illumination even if it “feels” the same and can have an impact on things like photos taken in the room.
Feature Request
I’d love to see an option for configurable nonlinear dimming curves, allowing the perceived brightness to change more evenly throughout the dimming range.
Ideally, this would:
Provide a more consistent perceived brightness change when holding the physical paddle.
Preserve the full 100% light output. I don’t want to sacrifice maximum brightness to improve the dimming experience.
Apply to both physical paddle dimming and Z-Wave brightness commands.
Offer a few selectable dimming curves, or potentially a configurable gamma value, so users can tune the behavior to their bulbs.
Retain the current dimming behavior as the default for backward compatibility.
I really love how configurable these switches are, but the dimming behavior is one of my biggest complaints. If i want to quickly dim a room I regularly find myself having to pull out my phone and open an app to set the switch to a low value because trying to nail the exact right level without having a really slow dimming speed is really painful.
I have seen a couple threads that mention this issue in passing, but dont think i’ve seen an answer on the feasibility of the feature
I use Home Assistant with Z-Wave JS and would be happy to help test a beta implementation.
I completely agree with this suggestion. Based upon readily available descriptions of dimming curves and the dimming behavior described here (which I too perceive) I suspect Inovelli uses a linear model which seems to be a problem shared with other smart dimmer devices. A tunable dimming curve using a gamma value would fit with the maximal customizability paradigm advertised by Inovelli, but may not be necessary as the vast majority of use cases would be satisfied with the standard perceptual correction. Just like we don’t select between curves for volume controls (loudness is also perceptually logarithmic), we probably don’t need to select between brightness dimming curves.
In general, humans seem to perceive all stimuli on a logarithmic scale, known as the Weber–Fechner Law (bonus points to @gpiper for triggering a Wikipedia expedition!).
I’m surprised that Inovelli hasn’t implemented logarithmic/fixed-gamma dimming behavior yet. @EricM_Inovelli, is this something that has been considered?
I think having a few selectable dimming curves would be a nice compromise between simplicity and customization.
One thing I’d be curious about is how consistent the result would be across different LED bulb models. Even with the same dimmer curve, the bulbs themselves might respond quite differently at the lower end.
Maybe a few presets, with an optional advanced setting for people who want to fine-tune things, would make more sense than a single fixed correction.
And keeping the current behavior as the default seems important, especially for people who already have their lighting set up the way they like it.
This is a good point, I don’t know how different brands of bulbs react. For reference, the fancy version of the market leader smart dimmer provides only the low & high level limits and on and off fade rates as relevant parameters. There isn’t an option to select a dimming curve, which leads me to suspect that their research has led them to conclude there is a single “best” curve. However, if we wanted to have the option this would be simple to implement as a single parameter where 0=linear, and any non-zero number would be a tunable gamma.
There is another dimming quirk I’ve noticed that’s probably worth mentioning in this thread: When the switch is in smart-bulb mode controlling two Philips Hue bulbs via Zigbee binding the intensity of the bulb suddenly changes after letting up on the paddle. After dimming partially up the bulb will dim slightly after releasing the paddle and after dimming partially down the bulb will increase in brightness after the paddle is released. It’s almost like the switch dimming has overshot in whichever direction of travel dimming is occurring and then corrects backward. This behavior may be more of a quirk with Philips Hue bulbs as they have their own internal ramp rate for turning on and off which cannot be set by the user.
I can definitely look into some enhancements for dimming levels. Other than just switching over to an optimized or “best” curve, what would the configuration options be like? For example, would you just choose either curve A, B, or C and you would try each out?
I have thought of a configuration option for a “middle” level (gamma correction?). Setting the middle level to 30 would give you a wider range of granularity on levels 31-99. Not sure if that makes sense.
If there is something else then I would love to discuss it.
I have noticed this too and I believe it is when there is a difference between the dimming speed of the switch and the default dimming speed that Philips uses in their firmware. I think if you change the Inovelli dimming speed to 3 seconds it is pretty close and makes it less obvious, but I haven’t messed around with it in a while.
yes please.. would love to see advancements here. What I’ve found with my 25+ white series is I have to customize the min and max one by one and have to do this at each switch which is a pain. I love the Red series I can do this via software.
#2 other LED issues I see is based on how much the load is on the circuit I have to augment a “boost” in HA for white series (red has some boost maybe built in I recall). What I mean by this is if the light is dimmmed to 3% which is common for like a toilet light at night when it goes off and comes back on at 3% it’s not enough juice to light it up, so the boost temp sets to like 10-20% then immediately drops back to the 3%. I have HA automations to do this for me but it’s got to be tweaked switch by switch because of load. a one “led disc” has such little load it needs a higher boost.
Would love to see some optimizations in firmware around all of this along with better dimming curves too…
Spent a bunch of time tooling around with claude + home assistant + Inovelli blue driving led cans. My sense is that most LEDs have a zone of highest sensitivity that’s really what we’re trying to capture. We perceive brightness roughly as the cube root of light output, so with LED loads the bottom of the range is very touchy and the top half barely changes.
What I tried: I compensated upstream, with a rotary knob driving the switch through Home Assistant. The step size grows with the current level (step ∝ 0.08 + 0.002·level), with an extra-fine zone below level 20. Fitted to a simple power curve, my hand-tuned result comes out at γ ≈ 2, and it feels far more even. But it only helps that one knob; the paddle and every other controller still use the linear curve.
Proposal: add one parameter, applied inside the existing min/max trims:
output = MinimumLevel + (MaximumLevel − MinimumLevel) x (level/254)^γ
γ from 1.0 to 3.0 in 0.1 steps. Default 1.0 (today’s behaviour), with 2.0 recommended for LEDs.
Apply it to paddle dimming, remote level commands and ramps alike.
Ideally, keep reporting the requested (pre-curve) level, so hubs and bindings still see a linear 1–254.
It’s a single parameter and easy to explain, and it works alongside existing calibrations. There’s industry precedent, too: DALI specifies a logarithmic curve.
great feedback… hope this can get implemented, you’ve explained what I see and experince, Im just not smart enough on that topic to have gotten there! great job again…