R&D, Please fix the internal dimming speeds!

[Marrt]

I was referred to this forum by the support in order to reach R/D:

Quote

Thank you for contacting us!
Based on the hardware capabilities, it’s possible that the current transition method is intentional.
Unfortunately, our support team is unable to modify the device’s logic. We recommend sharing your inquiry where it can gain visibility from our R&D team - on our official forum and Facebook Group.

 

My problem is that this dimmer seems to have fixed internal dimming-speed (which to my measurement is 10% per second).

The effect is that any dimming transition that is not in line with this dimming-speed will cause stepping or skipping, even the native dimming-speed a.k.a fade_rates (5, 7, 10, 15, 20 percent per second). To better depict this problem, please refer to this image:

The last graph shows that even native transition calls suffer from this stepping or skipping problem.

 

This is really bad. Is this just my unit or is anyone else seeing this?

I bought this dimmer because every button-dimmer on the market has this annoying turn-on delay. Of course, this is because the "hold from off" & "double tap" input gestures introduce a ~1s delay before the light will turn on (since the dimmer has to wait for a possible gesture to complete).
After trying the Shelly2 Dimmer and finding out that implementing my wanted behavior is impossible without scripting, the support referred me to this model. Yes, I can script my behavior and now I finally have dimming and instant-on. But what use is all the scripting and automation if the basic implementation of the dimming itself is flawed? I cannot script my own dimming curves because manual speeds will step or skip making the whole implementation seem like a cheap quick fix (I want slow a dimming-speed from 1-30% then very fast speeds from 30- 100% since there is little perceivable change in this range).

 

If I could make the perfect dimmer, give me the Shelly Pro Dimmer 2PM but with:

• this problem fixed
• transition duration between 0...5s instead of 0.5...5s
• more control parameters on the DimUp/DimDown API (%step, maybe even logarithmic/quadratic/cubic/.. dim curves)
• non-integer sub-percentage brightness levels

 

Also see

 

Or Ticket #130792 regarding the Shelly Dimmer 2 (where I had to talk to an AI I suppose)

 

PS: the "big skip to 100/0%" mentioned in the image is also indication that the internal speeds are inadvertently diverging from the internal program that senses the user input. It seems like the internal dimmer gets periodic calls (1s) to facilitate the diming as well as stop-calls when the user releases the button AND a set-call from the input sensing part once that part thinks 0/100% should have been reached, even though the internal dimmer is still at 10/80%. I would love to view your code for that.

 

Edited October 23, 2024 by Marrt
added PS

[Marrt]

Hmm, I am losing hope that this is of any interest to anyone.

Are there any other programmable ESP32 dimmers on the market that compare to the Shelly Pro Dimmer 2PM?
(ESP32 because there might be a chance that I will program those myself in the future)

[Marrt]

Just noticed that the stepping/skipping is even present in the native transition_duration (e.g. when you want your light to smoothly turn on)

[Heinz]

This is more of interest to the Devs will see if anyone of them has already noticed this post.

This is beyond my level when it comes to the firmware coding

[Marrt]

On 11/4/2024 at 6:45 PM, Heinz said:

This is beyond my level when it comes to the firmware coding

Can't be too hard.

Somewhere in the c++ code of the esp32 in the pro dimmer there is something that sets an Analog (or PWM or whatever) output that controls the the trailing-edge circuit each esp32 cpu cycle. I suppose that that code is sloppy. It does not interpolate smoothly. Only thing I can't say for sure is if this is a hardware limitation of the trailing-edge circuit, but I highly doubt that.

If I could just look at the source code of that portion I could probably fix it myself.

[Heinz]

Hi @Marrt as I dont deal with the coding team on a day-to-day basis.

as to getting the issue fixed all i can do is point make the dev team working on this project look at the issue and or open a ticket for the error. 
getting access to the source code would not be my call to make and don't know if that's even an option 

[Marrt]

Yes, please do that 🙂, Fixing this error would make my life much easier since I won't have to search for another product after sinking a dozen hours into this and the Shelly Dimmer 2 coding and investigation. Thanks!

All they need to do is make sure that the trailing edge pulse width setting hardware is controlled with the correct slope when a transition or dim call is received. Currently the slope is a fixed 10% brightness/second with stops and steps as shown in the first post. In my opinion that is unacceptable for a product that is sold under the prefix "Pro"

[MJMcCaffrey]

Hello, can product management provide an update on this issue?

Many thanks.

 

[Christian Tannheimer]

I rewrote the whole dimming behaviour with HomeAssistant. I was additionally to your problems, that I also had, annoyed, that you have to guess, if the dimming action after the switch on via short click will be up or down 🙂 I met shelly at a convention an told the engeneer-boss directly all your and my problems and he reacted as he can understand why I want these changes.

[Marrt]

Well, Dimmer 3 is out
https://www.shelly.com/products/shelly-dimmer-gen3

I'll get one and report if this thing has acceptable dimming capabilities or suffers from the same oversights.

[Christian Tannheimer]

This dimmer is much better in dimming speed, the first dimming action after switch on is alway up and it has a min and max brigthtnes setting in addition to the others.

[Marrt]

On 1/22/2025 at 5:33 AM, Marrt said:

I'll get one and report if this thing has acceptable dimming capabilities or suffers from the same oversights.

The unit arrived, I assembled it side by side with my Pro Dimmer for a quick test of the native dimming:
So far the obvious stepping is gone, that is a huge plus for me.

I still see the jumps to Min% and 100% when dimming over a longer range, but I have to investigate how these relate to the Min/max Brightness settings. I also need to test how I can manipulate the dimming speeds to create manual dimming curves since I want to dimm slow at the ranges from min-20% but super fast from 20-100%

[Marrt]

Probably the first bug:

• Dimstop (https://shelly-api-docs.shelly.cloud/gen2/ComponentsAndServices/Light#lightdimstop) causes Dimmer 3 to got to 100%?
  
  
  Same thing while dimming down via scripting, DimStop causes jump to minvalue
  

 

• Transition Duration will still not take values between 0.0 and 0.5s
  Only way to get transition duration 0 by scripting is to ommit it in command and set it to zero in the light settings.

Edited January 25 by Marrt

[Christian Tannheimer]

There are many more. As i deactivated and activated the blu gateway in this device, the turn on an off time turned from 0 seconds to 3 in reality, but in the app it stayed at 0. I had to choose another value than 0 and after that 0 in the app to reset it.

[Marrt]

2 hours ago, Christian Tannheimer said:

I had to choose another value than 0 and after that 0 in the app to reset it.

Did you reboot the shelly? I tend to always do that now before i take all settings for granted

[Christian Tannheimer]

Just try it. If you disable or enable blutooth, the transitional period resets to 3 seconds. No matter what you chose in the app. A reboot doesn't help, you have to change the time to another value press ok and set it back to 0 and press ok.

Edited January 25 by Christian Tannheimer

[Marrt]

... it sucks, always working around needless limitations.

Shelly 3 status report:

- DimStop is bugged

- still slight skipping to 0% 100% when dimming for more than a few seconds

- Transition duration cannot be set lower than 0.5 in script (transition duration 0 requires above workaround)

- Programming better curves that don't jump/jitter feels lika a chore... but i managed to get is working (see below)

 

I mean, the simple thing I want is instant-on + useable dimming curves, and a useable dimming curve (strangely, found in most other dimmers) looks like that

and not like this

 

 

Anyway, here is my investigation code, it just dimms so I could isolate how that behaves. So no turn on/off today and a lot of code is from the PRO dimmer script und from a time where we haven't got "Shelly.getUptimeMs()", thanks for adding that feature BTW, was really needed to make duration measurements easier

EDIT: Script removed, better script further down

Edited January 25 by Marrt
Removed Script

[Marrt]

4 hours ago, Christian Tannheimer said:

transitional period resets to 3 seconds

I noticed that behaviour too, but not due to bluetooth, it was something else, like a simple reboot after editing a script 😑

[Marrt]

Anyway, I finished my first shelly 3 script that checks all my boxes (except for smoother curve control, but it is good enough now)

Features:

• Single or Dual input (set at the top of the script: isDualInput = false)
• Instant-On (normally you have to wait 1s)
• Faster Dimming Initiation (~dimming starts after 450 ms)
• Better Dimming "Curve" (not really a curve but the dimming skips faster across the 20-100% range)
• Double Tap to go to 100% brightness
• Moonlight Mode (Hold Down Button from off to start at min_brightness_on_toggle)

Have fun:

Script removed, check new one:

Last words: I worked around the DimStop Bug so i don't care about native dimming, but the Transition Duration Setting is still reset on reboot. @Shelly, please take care of that and i can finally buy 10 of those dimmers.

Edited February 9 by Marrt

[Marrt]

On 1/25/2025 at 8:09 PM, Christian Tannheimer said:

If you disable or enable blutooth, the transitional period resets to 3 seconds

I found a workaround, just set the transition duration on startup via script:

Shelly.call("Light.SetConfig", { 'id': idOut, "config":{'transition_duration': 0}});

 

[alexza]

On 1/31/2025 at 4:27 AM, Marrt said:

I found a workaround, just set the transition duration on startup via script:

Shelly.call("Light.SetConfig", { 'id': idOut, "config":{'transition_duration': 0}});

 

Hi @Marrt - thanks for sharing your feedback here with regards to the dimmers. Could you advise the current state of the dimmers and whether you would recommend them?

I was looking at purchasing 20 x Pro Dimmer 2PM for a project as I wanted dimming, but I have been reading about your issues, as well as reports of flickering etc with Dimmer 2 which im not sure if Pro 2PM suffers from the same issue.

What is your current recommendation now? 

Should I buy 40 x Dimmer Gen 3 rather than the 20 Pro 2PM dimmer? I will lose LAN and DIN mount which was a big plus for me but I can still get a DIN adaptor for them.

I don't want to go to all the effort and expense and still have issues.

Also not sure if Gen4 Dimmer will be released soon.

Thanks for any advice

Edited June 17 by alexza

[muddy]

@alexza I can't answer your question, yet, but I'm committed to what is probably a similar project. Attached is 1 of 2 boxes I have as part of a large remodel. I too hope the dimming doesn't require any workarounds to be ideal. Happy to chat more about the project, if desired.