New heater tuning algorithm

@deckingman said in New heater tuning algorithm:

@oliof said in New heater tuning algorithm:

When doing this with the bed heater, phase 3 also says "fan off". This is likely just a cosmetic issue, as the part cooling fan wasn't on during the first couple cycles.

Are you saying that the bed heater tuning also runs multiple cycles? If so, is that the number of cycles changeable by the user ? If not, that would be huge problem with a big, thick, insulated build plate such as I have because it takes maybe 50 minutes to cool. So one heater tuning cycle using the current algorithm takes about an hour or more and I most certainly wouldn't want to have to repeat that multiple times.

@dc42

Since I also have build as a job a printer with a huge aluminum-plate (I think it was 750x750x8 or 10mm or so) I would also support that multiple cycles can be "optional" limited (which is the same as disabled if you put the limit to 1 cycle), because each cycle just takes too long... If the optional variable is not put, the system would do by default the amount of cycles it needs (the new approach)...
By this those kind of printers with huge thermal-mass-print-plates could benefit by the multiple-cycle approach for the extruder but could avoid its side-effects regarding the heat-plate ? Because of the huge thermal mass of those print-plates if the corridor of the PID algorithm for the temp-sensor-limits are narrow enough and resolution of the temp-sensor is high and it is a fast-reacting sensor, maybe 1 cycle could be enough for the plate? It would be also a fall-back solution for those having trouble with the new approach for whatever reason... This implicates that "the first cycle" (and ONLY this one) would behave like as we are used to it (old style)?

dc42

@gloomyandy said in New heater tuning algorithm:

Hi David,
a quick bit of feedback on the new heater tuning mechanism when it is used with the LPC port.

On the LPC the ADC is quite noisy to the extent that we have a pre-filter on the readings (that uses a median filter). On my test machine I tested with just that filter and with the standard RRF temperature filter disabled. In this configuration I occasionally (about 50% of the time) had a result in which the tuning was not consistent and so ran for the max 30 cycles (with the fan not being used) and produced a warning message.

If I enable the RRF filter (in addition to the pre-filter) then the same printer completes the test in from 5 to 7 cycles (again with the "fan on" test not run). This same hardware ran the old algorithm fine with the RRF filter disabled. I suspect that this may mean that the new algorithm may be slightly more sensitive to noise than the older one. I'm not sure if this is an issue (especially with Duet hardware), but I thought it might be of interest.

The old algorithm calculated the heating rate using the entire time to heat up. from room temperature to target temperature. Likewise it calculated the cooling time by monitoring a long cooldown time at the end. The new algorithm cycles between the target temperature and 5C below it. It measures the heating and cooling rates on the last 3C each cycle, and accumulates the mean and standard deviation. If there is a lot of noise in the ADC readings then that noise will be a greater proportion of that 3C, so the data quality (as shown by the standard deviation) will be less good and it will do more cycles.

You could try increasing that 5C value to reduce the effect of noise, although it will lengthen the tuning process. It's TuningHysteresis in LocalHeater.h.

dc42

@LB, I suggest you try the new algorithm on your bed heater. The new algorithm doesn't need the long cooldown period at the end, so it won't necessarily take longer.

dc42

@oliof said in New heater tuning algorithm:

What Jay said, and then: If you do need to do multiple runs for different target temperatures, the new "A" parameter for ambient temperature allows you to start a new cycle before the bed cooled down.

That's true; but it should not normally be necessary to tune at more than one temperature. An exception might be if you use a heated chamber and the chamber temperature varies a lot depending on the material.

BTW in you use a heated chamber then the tuning for the hot end heaters should be done with the chamber up to temperature, and "ambient temperature" will be the chamber temperature.

oliof

Another use may be multizone heaters like Keenoovo offers for larger beds I guess.

@dc42

Thanks as usual! Just a bit worried.
At the moment can only test with the smaller "renewed" (via the duet-board) anycubic here. With this little 220x220mm plate, I guess it will be fine.
The other printer with the big plate is a few hundred km away...

(If all duet-boards supporting RRF3.x have enough storage-space another option would be to make it optional via M307 the B-parameter there, 0 for PID, 1 for bang-bang, 2 for the "new iterating PID" or similar...? Just thinking out loud)

deckingman

@dc42 said in New heater tuning algorithm:

@LB, I suggest you try the new algorithm on your bed heater. The new algorithm doesn't need the long cooldown period at the end, so it won't necessarily take longer.

But does it still need to "cycle" the temperature by 5 deg C or so? That in itself is likely to be problematic with a large (and insulated) thermal mass.

fractalengineer

Very nice to see betaflight features implemented here

dc42

@deckingman said in New heater tuning algorithm:

@dc42 said in New heater tuning algorithm:

@LB, I suggest you try the new algorithm on your bed heater. The new algorithm doesn't need the long cooldown period at the end, so it won't necessarily take longer.

But does it still need to "cycle" the temperature by 5 deg C or so? That in itself is likely to be problematic with a large (and insulated) thermal mass.

If the overshoot and undershoot is small and the temperature sensors are not noisy, then it would be possible to use a smaller temperature range.

gloomyandy

I originally posted by mistake in 3.2-beta3.2 thread, but I think it is better here...

I've been testing the new algorithm on my printer and have seen some strange results when tuning the bed heater. My bed is a 310x310x6mm aluminium plate with a magnetically attached steel sheet on top. The heater is a mains powered 250x250mm 500W Silicon pad (with the thermistor embedded in it). When tuning my target temperature is 60 degrees. Ambient temperature is around 21 degrees. This combination is probably not ideal as there is a lot of lag between the pad thermistor and the actual bed plate temperature.

If I run the tuning from cold then it always seems to run for the full 30 cycles. However if I run it when it has cooled (from 60) to say 30 degrees (using the A parameter to specify the ambient temperature) then it will complete after 5 cycles. To try and understand what is going on I have added some additional debug to the tuning process. Here is the output from the first case:

Auto tune starting phase 1, heater on
Auto tune starting phase 2, heater settling
Sample 1 dLow 1250.000000 tOn 11250.000000 heatingRate 0.585193
Sample 1 dHigh 1750.000000 tOff 22750.000000 coolingRate 0.268330
  tOn 11250±0, tOff 22750±0, dHigh 1750±0, dLow 1250±0, R 0.585±0.000, C 0.268±0.000, V 0.0±0.0, cycles 1
Sample 2 dLow 1250.000000 tOn 11500.000000 heatingRate 0.581589
Sample 2 dHigh 2500.000000 tOff 25750.000000 coolingRate 0.234478
  tOn 11375±125, tOff 24250±1500, dHigh 2125±375, dLow 1250±0, R 0.583±0.002, C 0.251±0.017, V 0.0±0.0, cycles 2
Auto tune starting phase 3, fan off
Sample 1 dLow 1000.000000 tOn 10750.000000 heatingRate 0.608868
Sample 1 dHigh 2500.000000 tOff 28250.000000 coolingRate 0.222683
  tOn 10750±0, tOff 28250±5, dHigh 2500±0, dLow 1000±0, R 0.609±0.000, C 0.223±0.000, V 0.0±0.0, cycles 1
Sample 2 dLow 1250.000000 tOn 10250.000000 heatingRate 0.629652
Sample 2 dHigh 2750.000000 tOff 30750.000000 coolingRate 0.197948
  tOn 10500±250, tOff 29500±1250, dHigh 2625±125, dLow 1125±125, R 0.619±0.010, C 0.210±0.012, V 0.0±0.0, cycles 2
Sample 3 dLow 1000.000000 tOn 9750.000000 heatingRate 0.676344
Sample 3 dHigh 2250.000000 tOff 34000.000000 coolingRate 0.179575
  tOn 10250±408, tOff 31000±2354, dHigh 2500±204, dLow 1083±118, R 0.638±0.028, C 0.200±0.018, V 0.0±0.0, cycles 3
Sample 4 dLow 1250.000000 tOn 9750.000000 heatingRate 0.678182
Sample 4 dHigh 2250.000000 tOff 37500.000000 coolingRate 0.164589
  tOn 10125±415, tOff 32625±3475, dHigh 2438±207, dLow 1125±125, R 0.648±0.030, C 0.191±0.022, V 0.0±0.0, cycles 4
Sample 5 dLow 750.000000 tOn 9250.000000 heatingRate 0.724365
Sample 5 dHigh 2500.000000 tOff 40000.000000 coolingRate 0.153312
  tOn 9950±510, tOff 34100±4285, dHigh 2450±187, dLow 1050±187, R 0.663±0.041, C 0.184±0.025, V 0.0±0.0, cycles 5
Sample 6 dLow 1000.000000 tOn 9250.000000 heatingRate 0.710471
Sample 6 dHigh 3000.000000 tOff 43500.000000 coolingRate 0.142683
  tOn 9833±534, tOff 35667±5251, dHigh 2542±267, dLow 1042±172, R 0.671±0.041, C 0.177±0.027, V 0.0±0.0, cycles 6
Sample 7 dLow 1000.000000 tOn 9000.000000 heatingRate 0.728104
Sample 7 dHigh 3000.000000 tOff 46500.000000 coolingRate 0.134402
  tOn 9714±574, tOff 37214±6165, dHigh 2607±295, dLow 1036±160, R 0.679±0.043, C 0.171±0.029, V 0.0±0.0, cycles 7
Sample 8 dLow 750.000000 tOn 8750.000000 heatingRate 0.762012
Sample 8 dHigh 2500.000000 tOff 49500.000000 coolingRate 0.128104
  tOn 9594±624, tOff 38750±7054, dHigh 2594±278, dLow 1000±177, R 0.690±0.048, C 0.165±0.031, V 0.0±0.0, cycles 8
Sample 9 dLow 1000.000000 tOn 9000.000000 heatingRate 0.740379
Sample 9 dHigh 3000.000000 tOff 52250.000000 coolingRate 0.125191
  tOn 9528±617, tOff 40250±7889, dHigh 2639±291, dLow 1000±167, R 0.695±0.048, C 0.161±0.032, V 0.0±0.0, cycles 9
Sample 10 dLow 1000.000000 tOn 9000.000000 heatingRate 0.765968
Sample 10 dHigh 3500.000000 tOff 57000.000000 coolingRate 0.110748
  tOn 9475±607, tOff 41925±9015, dHigh 2725±378, dLow 1000±158, R 0.702±0.051, C 0.156±0.034, V 0.0±0.0, cycles 10
Sample 11 dLow 750.000000 tOn 9000.000000 heatingRate 0.739456
Sample 11 dHigh 2500.000000 tOff 58000.000000 coolingRate 0.106328
  tOn 9432±594, tOff 43386±9759, dHigh 2705±366, dLow 977±167, R 0.706±0.049, C 0.151±0.035, V 0.0±0.0, cycles 11
Sample 12 dLow 500.000000 tOn 8750.000000 heatingRate 0.766151
Sample 12 dHigh 3000.000000 tOff 60750.000000 coolingRate 0.104071
  tOn 9375±599, tOff 44833±10504, dHigh 2729±360, dLow 938±207, R 0.711±0.050, C 0.147±0.036, V 0.0±0.0, cycles 12
Sample 13 dLow 750.000000 tOn 8750.000000 heatingRate 0.743942
Sample 13 dHigh 3250.000000 tOff 61500.000000 coolingRate 0.103689
  tOn 9327±600, tOff 46115±11026, dHigh 2769±373, dLow 923±205, R 0.713±0.049, C 0.144±0.037, V 0.0±0.0, cycles 13
Sample 14 dLow 750.000000 tOn 8750.000000 heatingRate 0.793538
Sample 14 dHigh 3500.000000 tOff 62500.000000 coolingRate 0.101916
  tOn 9286±597, tOff 47286±11432, dHigh 2821±406, dLow 911±203, R 0.719±0.051, C 0.141±0.037, V 0.0±0.0, cycles 14
Sample 15 dLow 0.000000 tOn 8750.000000 heatingRate 0.053600
Sample 15 dHigh 3500.000000 tOff 63250.000000 coolingRate 0.098418
  tOn 9250±592, tOff 48350±11740, dHigh 2867±427, dLow 850±300, R 0.675±0.173, C 0.138±0.037, V 0.0±0.0, cycles 15
Sample 16 dLow 500.000000 tOn 8500.000000 heatingRate 0.790129
Sample 16 dHigh 3250.000000 tOff 64000.000000 coolingRate 0.097789
  tOn 9203±601, tOff 49328±11982, dHigh 2891±424, dLow 828±303, R 0.682±0.170, C 0.136±0.037, V 0.0±0.0, cycles 16
Sample 17 dLow 0.000000 tOn 8750.000000 heatingRate 0.056186
Sample 17 dHigh 3250.000000 tOff 64250.000000 coolingRate 0.100125
  tOn 9176±593, tOff 50206±12143, dHigh 2912±420, dLow 779±352, R 0.645±0.221, C 0.134±0.037, V 0.0±0.0, cycles 17
Sample 18 dLow 750.000000 tOn 8250.000000 heatingRate 0.805200
Sample 18 dHigh 3000.000000 tOff 64250.000000 coolingRate 0.096790
  tOn 9125±614, tOff 50986±12231, dHigh 2917±408, dLow 778±342, R 0.654±0.218, C 0.132±0.037, V 0.0±0.0, cycles 18
Sample 19 dLow 750.000000 tOn 8500.000000 heatingRate 0.771688
Sample 19 dHigh 3000.000000 tOff 66000.000000 coolingRate 0.094537
  tOn 9092±614, tOff 51776±12368, dHigh 2921±398, dLow 776±333, R 0.660±0.214, C 0.130±0.037, V 0.0±0.0, cycles 19
Sample 20 dLow 750.000000 tOn 8500.000000 heatingRate 0.797152
Sample 20 dHigh 3750.000000 tOff 65500.000000 coolingRate 0.094485
  tOn 9062±612, tOff 52462±12421, dHigh 2962±428, dLow 775±325, R 0.667±0.211, C 0.128±0.037, V 0.0±0.0, cycles 20
Sample 21 dLow 1000.000000 tOn 8500.000000 heatingRate 0.781730
Sample 21 dHigh 3000.000000 tOff 66250.000000 coolingRate 0.097950
  tOn 9036±609, tOff 53119±12472, dHigh 2964±418, dLow 786±321, R 0.673±0.207, C 0.126±0.037, V 0.0±0.0, cycles 21
Sample 22 dLow 0.000000 tOn 8500.000000 heatingRate 0.056254
Sample 22 dHigh 3500.000000 tOff 67500.000000 coolingRate 0.093961
  tOn 9011±605, tOff 53773±12548, dHigh 2989±423, dLow 750±354, R 0.645±0.239, C 0.125±0.036, V 0.0±0.0, cycles 22
Sample 23 dLow 750.000000 tOn 8500.000000 heatingRate 0.789738
Sample 23 dHigh 3250.000000 tOff 66500.000000 coolingRate 0.094778
  tOn 8989±601, tOff 54326±12544, dHigh 3000±417, dLow 750±346, R 0.651±0.236, C 0.124±0.036, V 0.0±0.0, cycles 23
Sample 24 dLow 750.000000 tOn 8500.000000 heatingRate 0.794132
Sample 24 dHigh 3750.000000 tOff 66250.000000 coolingRate 0.097583
  tOn 8969±596, tOff 54823±12508, dHigh 3031±435, dLow 750±339, R 0.657±0.233, C 0.123±0.036, V 0.0±0.0, cycles 24
Sample 25 dLow 0.000000 tOn 8500.000000 heatingRate 0.055224
Sample 25 dHigh 3500.000000 tOff 66750.000000 coolingRate 0.095105
  tOn 8950±592, tOff 55300±12477, dHigh 3050±436, dLow 720±363, R 0.633±0.257, C 0.121±0.035, V 0.0±0.0, cycles 25
Sample 26 dLow 1000.000000 tOn 8500.000000 heatingRate 0.797152
Sample 26 dHigh 3000.000000 tOff 66500.000000 coolingRate 0.095806
  tOn 8933±587, tOff 55731±12422, dHigh 3048±428, dLow 731±360, R 0.639±0.254, C 0.120±0.035, V 0.0±0.0, cycles 26
Sample 27 dLow 1250.000000 tOn 8500.000000 heatingRate 0.793937
Sample 27 dHigh 3750.000000 tOff 66500.000000 coolingRate 0.093961
  tOn 8917±581, tOff 56130±12359, dHigh 3074±440, dLow 750±366, R 0.645±0.251, C 0.120±0.035, V 0.0±0.0, cycles 27
Sample 28 dLow 1000.000000 tOn 8250.000000 heatingRate 0.805200
Sample 28 dHigh 3000.000000 tOff 65750.000000 coolingRate 0.093893
  tOn 8893±584, tOff 56473±12267, dHigh 3071±432, dLow 759±363, R 0.651±0.248, C 0.119±0.034, V 0.0±0.0, cycles 28
Sample 29 dLow 750.000000 tOn 8500.000000 heatingRate 0.783488
Sample 29 dHigh 2750.000000 tOff 67000.000000 coolingRate 0.095667
  tOn 8879±578, tOff 56836±12205, dHigh 3060±429, dLow 759±356, R 0.655±0.245, C 0.118±0.034, V 0.0±0.0, cycles 29
Sample 30 dLow 1000.000000 tOn 8500.000000 heatingRate 0.808610
Sample 30 dHigh 2750.000000 tOff 66750.000000 coolingRate 0.092288
  tOn 8867±573, tOff 57167±12131, dHigh 3050±425, dLow 767±353, R 0.660±0.242, C 0.117±0.034, V 0.0±0.0, cycles 30
Warning: heater behaviour was not consistent during tuning
tOn 8867±573, tOff 57167±12131, dHigh 3050±425, dLow 767±353, R 0.660±0.242, C 0.117±0.034, V 0.0±0.0, cycles 30
Auto tuning heater 0 completed after 30 cycles in 2137 seconds. This heater needs the following M307 command:
 M307 H0 R0.788 C289.8 D2.74 S1.00 V0.0
Edit the M307 H0 command in config.g to match this.
Heater 0 switched off

and this is the output from the "pre-heated" case:

Auto tuning heater 0 using target temperature 60.0°C and PWM 1.00 - do not leave printer unattended
ok
Auto tune starting phase 2, heater settling
Sample 1 dLow 750.000000 tOn 9250.000000 heatingRate 0.753922
Sample 1 dHigh 2250.000000 tOff 36000.000000 coolingRate 0.165439
  tOn 9250±0, tOff 36000±0, dHigh 2250±0, dLow 750±0, R 0.754±0.000, C 0.165±0.000, V 0.0±0.0, cycles 1
Sample 2 dLow 1000.000000 tOn 9250.000000 heatingRate 0.727826
Sample 2 dHigh 2250.000000 tOff 40250.000000 coolingRate 0.155488
  tOn 9250±0, tOff 38125±2125, dHigh 2250±0, dLow 875±125, R 0.741±0.013, C 0.160±0.005, V 0.0±0.0, cycles 2
Auto tune starting phase 3, fan off
Sample 1 dLow 1000.000000 tOn 9500.000000 heatingRate 0.699097
Sample 1 dHigh 2500.000000 tOff 43500.000000 coolingRate 0.141769
  tOn 9500±0, tOff 43500±0, dHigh 2500±0, dLow 1000±0, R 0.699±0.000, C 0.142±0.000, V 0.0±0.0, cycles 1
Sample 2 dLow 1250.000000 tOn 8750.000000 heatingRate 0.769238
Sample 2 dHigh 2500.000000 tOff 46250.000000 coolingRate 0.132351
  tOn 9125±375, tOff 44875±1375, dHigh 2500±0, dLow 1125±125, R 0.734±0.035, C 0.137±0.005, V 0.0±0.0, cycles 2
Sample 3 dLow 750.000000 tOn 8750.000000 heatingRate 0.763718
Sample 3 dHigh 2750.000000 tOff 50750.000000 coolingRate 0.126651
  tOn 9000±354, tOff 46833±2988, dHigh 2583±118, dLow 1000±204, R 0.744±0.032, C 0.134±0.006, V 0.0±0.0, cycles 3
Sample 4 dLow 1250.000000 tOn 8500.000000 heatingRate 0.800977
Sample 4 dHigh 3500.000000 tOff 53500.000000 coolingRate 0.115505
  tOn 8875±375, tOff 48500±3877, dHigh 2812±410, dLow 1062±207, R 0.758±0.037, C 0.129±0.010, V 0.0±0.0, cycles 4
Sample 5 dLow 1250.000000 tOn 8500.000000 heatingRate 0.761442
Sample 5 dHigh 3500.000000 tOff 56250.000000 coolingRate 0.110297
  tOn 8800±367, tOff 50050±4651, dHigh 2950±458, dLow 1100±200, R 0.759±0.033, C 0.125±0.011, V 0.0±0.0, cycles 5
tOn 8800±367, tOff 50050±4651, dHigh 2950±458, dLow 1100±200, R 0.759±0.033, C 0.125±0.011, V 0.0±0.0, cycles 5
Auto tuning heater 0 completed after 5 cycles in 460 seconds. This heater needs the following M307 command:
 M307 H0 R0.895 C267.3 D2.67 S1.00 V0.0
Edit the M307 H0 command in config.g to match this.
Heater 0 switched off

I suspect that the relatively large thermal mass means that as the full bed heats up it is impacting the heating/cooling times of the test cycle. But there also seems to be something a little odd going on with the "dLow" values as these are sometimes zero.

It may well be that I need to add an additional thermistor (perhaps embedded in the edge of the bed) to get more accurate readings of the actual bed temperature. However I don't think the setup I have is that unusual so I thought the data might be of interest to you.

Zhang Jianyu

The new auto-tune is working well for me on Duet 2 board. Thanks for your work on this

Exerqtor

Working great here to! Took good over an hour to tune a 330x330x4mm 24v alu heated bed with a 4mm glass surface and cork insulation underneith for 60C tho

gloomyandy

@Exerqtor How many cycles did it take to tune your 330x300x4mm bed?

oliof

My 300x300x4mm bed took seven or eight cycles. 220V AC bed to 80C

Exerqtor

@gloomyandy said in New heater tuning algorithm:

@Exerqtor How many cycles did it take to tune your 330x300x4mm bed?

I didn't look at it tbh! After 45minutes i went downstairs and watched a movie lmao. Just sent a M500 before i went to bed when i saw it was done, then turned off the printer without thinking more about it i just saw it said 4700 something seconds.

gloomyandy

No problem, sounds like it ran for a fair number of cycles!

Exerqtor

Don't quote me on it, but I THINK it was 11 for the bed. And 7 for the hotend (std Mosquito with a silicone sock, 40mm fan and 50w/12v heater).

dc42

@gloomyandy said in New heater tuning algorithm:

I originally posted by mistake in 3.2-beta3.2 thread, but I think it is better here...

I've been testing the new algorithm on my printer and have seen some strange results when tuning the bed heater. My bed is a 310x310x6mm aluminium plate with a magnetically attached steel sheet on top. The heater is a mains powered 250x250mm 500W Silicon pad (with the thermistor embedded in it). When tuning my target temperature is 60 degrees. Ambient temperature is around 21 degrees. This combination is probably not ideal as there is a lot of lag between the pad thermistor and the actual bed plate temperature.

If I run the tuning from cold then it always seems to run for the full 30 cycles. However if I run it when it has cooled (from 60) to say 30 degrees (using the A parameter to specify the ambient temperature) then it will complete after 5 cycles. To try and understand what is going on I have added some additional debug to the tuning process. Here is the output from the first case:
I suspect that the relatively large thermal mass means that as the full bed heats up it is impacting the heating/cooling times of the test cycle. But there also seems to be something a little odd going on with the "dLow" values as these are sometimes zero.
...
It may well be that I need to add an additional thermistor (perhaps embedded in the edge of the bed) to get more accurate readings of the actual bed temperature. However I don't think the setup I have is that unusual so I thought the data might be of interest to you.

I think I know what is going on. My theory is that systems such as yours behave somewhat like a FOPDT (first order process plus dead time) plus a thermal reservoir that is somewhat weakly coupled to it. When heating from cold, the temperature of the reservoir lags the temperature of the heating element and temperature sensor. It is only when it has heated up to their average temperatures that the behaviour during the tuning cycles becomes stable.

This phenomenon also occurs with hot ends to some extent, which is why the tuning algorithm does two heating/cooling cycles before is starts collecting data. It's clear that with bed heaters like yours, more of these dummy cycles are needed.

I too have a bed heater that behaves in this way, however in my case the thermistor readings have become noisy, so it is not surprising that it always goes the full 30 cycles.

Any solution must not only solve the problem you encountered but also handle noisy thermistors reasonably well, by continuing to average over a reasonably large number of cycles. One option would be to check the consistency after 5 cycles, and if it is poor then throw the data away and start again. So instead of doing 2 dummy cycles and up to 30 real cycles, it would do 7 dummy cycles and up to 25 real cycles. Do you think this would work for your heater?

gloomyandy

@dc42 said in New heater tuning algorithm:

Any solution must not only solve the problem you encountered but also handle noisy thermistors reasonably well, by continuing to average over a reasonably large number of cycles. One option would be to check the consistency after 5 cycles, and if it is poor then throw the data away and start again. So instead of doing 2 dummy cycles and up to 30 real cycles, it would do 7 dummy cycles and up to 25 real cycles. Do you think this would work for your heater?

On my test build I cleared the data every 5 cycles and continued on up to a max of 30 cycles in total. So basically I tuned in groups of 5 cycles. In my case this resulted in the tuning completing after 10 cycles total. So yes I'm pretty sure that your proposal would work in my case (though I'm not sure if it would in the case of even bigger/thicker beds as they might take even longer to heat up). I wonder if there is any way to differentiate between a noisy reading (which will benefit from more readings) and a lagging reading (which probably needs a restart)? But yes I'm pretty sure that being able to extend the dummy cycles would help.

dc42

Some other solutions come to mind:

We could store the last 5 readings instead of just the mean and standard deviation; then we could check whether the last 5 readings are good enough. I am reluctant to do this because of the additional RAM that would be needed. One of the goals of the new algorithm was to use minimal RAM so that it can be implemented on the Duet 3 tool board.
We could monitor the ratio of heating time to cycle time, and continue doing dummy cycles until this stabilises or has the appearance of being noise rather than a trend. For example, wait for the ratio of heating time to cycle time to be at least 95% of the value in the previous cycle.

Your thoughts?