inconsistent speed profile loco Hornby

[jodel]

hello all contributors,

After HRMS has corrected the decoder file for ESU loksound and set the range for Vmax à 255, i 've tried to test my speed again.

First curiously without changing VMax the maximum speed have fallen. 

I set a value for Vmax CV5 to a value :200

the max speed is almost good. but i constat that teh range for CV6 is always lilited to 64. For me it is strange because on my technical guide f loksound CV6 range = CV5 range = 255...

but itdoes not matter.

to diagnose my problem on my loco HJ2211S fitted by Hornby with an ESU loksound decoder

i have set the CV5 : Vmax to 200 and CV 6: Vmid 20, for better profile  !....

you can see here the speed profile : measured actual speed vs speed set point.





the profile is like a logarithmic curve instead, more or less  a straight line of slope 1 ... and it is the best profile i  got. i think it is not good situation.

i have for this moment no other means to precise the diagnostic between either a such curious behavior of Railmaster, or a malfunction of the decoder of the loco.

i think  it's as if the RM sends speed commands when i push the throttle to Vmax with 28 (or 128) Speed steps and the decoder use only 14 speed step... i have tried this test with 14 speed steps and 28 speed steps ( in CV29) with same results.

Or perhaps it is like Vmid CV6 is higher than Vmax ...

Is it possible to trace on RM the number of each speed step that RM send to the decoder ?
With these traces, i can do a chart with the speed steps in abscissa, instead of the set point speed. 

and as after each change of CV we have to update by a click on green tick , the loco database, i suppose RM uses all of these informations to control and steer the decoder. Is RM able to interpret different range for Vmx or Vmid ? 

Finally i need your suggestions  , either for HRMS ( i have opened a request to HRMS,but without reply today) correction of CV6 range, and/or something which could be wrong in RM , or either to determine if i have to send back my loco to the Hornby retailer for, eventually, an exchange of it, under the contractual guarantee... But, i want say again that for this loco all other functions are OK : lights, sounds ... strange not ?

thanks for all your replies

jodel

[96RAF]

You will find the actual loco track speed is different depending upon if you have scale speeds set or not.

One seems to run on some algorithm within RM and the other seems to use standard speed steps (e.g. n/128).

Rob

[jodel]

Hi,

sorry i have not not uploaded well the speeds chart i have done . here is the speeds chart where you can see the measured speeds (in KPH)  are not corresponding with the speeds set points. the conditions for this test was : 

a loco hornby HJ2211S, originally fitted wih a loksound decoder and using scale speeds wtih jauge HO in the configuration window.

it was the best profile ( with the slower slope ) i got with modifications CV5, 6... for Vmax 140 KPH.. And i have also the same problem with digitalised loco fitted with lokpilot decoder .

 /media/tinymce_upload/ab893c35a5278255cb64d61086083f2b.jpg

You can see on this chart that the real speeds are not acceptable : no speed reliability, no reproductibilty between locos speeds  ..

But when reading some topics in this forum : topic "scale speed v speed steps"  i had idea to test with speed steps only. and i have this results for the same loco ( HJ2211S ) : 

 

/media/tinymce_upload/70013034980f8ff7ed2b6433deff7095.jpg

this is almost good results, except the saturation of the curve for high speeds. i have alos the sames results with others test with lokpilot..!

also with scale speeds i never saw any effect ( positive or negative ) on the speeds when setting a speed factor.

the throttle is pushed to a maximum of 127th speed step as i can see it on the top of the throttle cursor in the small box of loco control.. and probably because the loco control are used with speeds steps, there is an inconsistency between the displayed range on the throttle of the  loco control box : sometimes max 200kph , sometimes 140Kph and  for  the small loco control : always 127 ( that is right)   ... but this is another problem

i am going to reactivate my opened request to HRMS. for me it looks like a bug, and of course i am not  very confident with the reliability of RM...

thank you for your opinions.

jodel

[96RAF]

@jodel

In your second graph the red line is what I would expect to see for a linear speed curve with VMax limited to value 200 i.e. the real speed saturates at ss200, whereas yours is showing about 220.

 

I would be tempted to download Decoder-Pro (free with JMRI), use the speed curve tool to set your choice of response curve, then see what RM thinks of it and if it modifies your DP curve from within RM. Review the graphs of those changes.

 

Rob

[jodel]

Hi RAF96

Thank you for your answer.

of course the red curve has a better profile than the Yellow one. 

When i did this chart, the red curve was done with the hornby  standard settings of the loco . the yellow curve is with the settings i wanted to have a maximum speed of more or less 140KPH. and the flat curve after 140 KPh or 200 kPh is ( it think) because the throttle is at 127 th Step... 

yes i am going to control the curves with JMRI decoder pro.

But for me, what i don't understand is how RM works with speed scales, because i have always bad curves . because it looks like scales speed mode cut the curve at a maximum. it can be correct, but without increasing the slope of the curve....??

 i suppose i had ( or RM .??)  a misunderstanding when i set the speed factor, between KPH, MPH.OO & HO jauge... for me it was not clear ,because when i try to set a time for 6,274m ( which curiously is the same for OO & HO.),  i never saw a real effect on the speed , except Vmax set to the value i entered.

what i would to say is sometimes the difference of displayed speeds beween the small loco control box and the larger one.... i think for this reason it is better to work with scale speeds, but i can't....

so here are my additionnal informations about this charts and this issue.

thank for your reply

jodel

 

 

[96RAF]

@jodel

I always understood the Hornby explanation of restricting Vmax to be that all 128 speed steps would be apportioned over the remaining speed range, thus giving finer overall control, rather than the technique just chopping a bunch of speed steps off the range to give that plateau above the cut-off value. I.e. like your yellow line not the red one.

 

Maybe the scale speed versus speed steps RM thing has some effect on this.

 

I may be wrong in my interpretation of course and would be happy if someone could explain why it is, if some other way.

Rob

[jodel]

Hi RAF, and all contributors,

following the advice of HRMS i did other test with my Hornby Loco , but with the nominal values set in the ESU decoder especially for CV5 CV6.

and i have this results :/media/tinymce_upload/ad4d1c792003fd9942bc66a8e07d97bc.jpg

As you can see , the step speed profile is very nice and correct , to a few percents. HRMS said that the Cv in hornby Loco are set to optimal values according with the real capacity of the motor : it is TRUE. 

But for the scale speed the profile is really horrible . the problem is always the same : the real speed is 5 times greater than the speed set point.. and i tried to reduce this fast speed in setting a scale factor to 3sec/ 6,204 m, for 160KPH of Vmax, with no change on the real speeds.

I can't work with this issue... in fact it's look like,  in scale mode , RM send up to 128 steps to the loco and on the other side, the decoder is programmed with only 28 steps : 28 step for ie 40KPH , and 180KPH actual which would correspond to 128 .... does anbody know how the decoder works with the bit 1 of CV29 ( 14  steps or 28 or 128 steps) . this decoder is set with 28/128 steps  (bit 1 = 1). but how the decoder makes the difference between 28 or 128 steps?. i have never seen another parameter in all ESU Cv to distinguish 28 and 128... how does it work really ?

I think that something is changing (perhaps on my computer, but it could be really amazing ?)  in RM when switching from Step mode to Scale Mode...

And also i hope RM is able to read correctly the real version number of ESU decoder because for a loksound this dispayed number is always 255 !... wich is normal according to  ESU : this number is always hidden, and you can see it only with an ESU lokprogrammer... 

however i have good speeds in step mode. this is why it is really incomprehensible to have so bad speeds in scale mode...

according to the reply of HRMS, scale mode works well with hornby loco decared in RM database..., if this is the limit of RM, it is very inconvenient when you want to use RM with different locos : this means that RM is not an open system...

would anyone have  opinions about this issue ?

Tanks for your replies.

 

jodel

[96RAF]

I didnt notice it first time around but your column header for V-KPH and V-MPH is reversed. This probably does not make any difference to your overal assumptions.

I love like to try doing 160 MPH down our Highway where the speed limit is 100 KPH and explain it was all a mistake with the maths conversion officer.

[96RAF]

I didnt notice it first time around but your column header for V-KPH and V-MPH is reversed. This probably does not make any difference to your overal assumptions.

 

I love like to try doing 160 MPH down our Highway where the speed limit is 100 KPH and explain it was all a mistake with the maths conversion officer.

 

Now the Techy-Bit:

This is what NMRA says about V-start, accel/decell, V-High, V-Mid. It is not clear as to when V-max (high) is set below 255 if the full range of speed step range is provided pro-rata the reduced motor voltage or not.

 

Configuration Variable 2   Vstart Vstart is used to define the voltage drive level used as the start voltage on the motor.  The voltage drive levels shall correspond linearly to the voltage applied to the motor at speed step one, as a fraction of available rectified supply voltage. When the voltage drive level is equal to zero, there shall be zero voltage applied to the motor. When it is at maximum "11111111", the full available rectified voltage shall be applied. 

 

 Configuration Variable 3  Acceleration Rate  Determines the decoder's acceleration rate.  The formula for the acceleration rate shall be equal to (the contents of CV#3*.896)/(number of speed steps in use).  For example, if the contents of CV#3 =2, then the acceleration is 0.064 sec/step for a decoder currently using 28 speed steps.  If the content of this parameter equals "0" then there is no programmed momentum during acceleration.

 

Configuration Variable 4  Deceleration Rate  Determines a decoders braking rate, in the same fashion as acceleration above (CV #3). 

 

Configuration Variable 5  Vhigh Vhigh is used to specify the motor voltage drive levels at the maximum speed step.  This value shall be specified as a fraction of available rectified supply voltage.  When the contents of CV#5 equal "11111111", the full available rectified voltage shall be applied.  Values of "00000000" or "00000001" shall indicate that Vhigh is not used in the calculation of the speed table.  

 

Configuration Variable 6  Vmid Vmid specifies the voltage drive level at the middle speed step.  Vmid is used to generate a performance curve in the decoder that translate speed step values into motor voltage drive levels and is specified as a fraction of available rectified supply voltage. Values of 00000000 or 00000001 shall indicate that Vmid is not used in the calculation of the speed table.

 

 

NMRA also talk to 128 and 28 speed steps in S-9.2 Para A, but it is not easy to understand in layman's terms relative to what RM may or may not be doing bit-wise to enable scale speeds.

Rob

[Britannia Builder]

Hi Jodel

 

I looked into this question of how Railmaster maps from scale speeds to speed steps last year, in this thread: https://www.hornby.com/uk-en/forum/speed-curves/?p=1/#post-167805

In summary there's an Access database 'resource.mdb' which has a table 'Details' which contains a row for each of your locos. Within this there's a column 'SpeedCurve' which has a list of factors that seem to correspond to speeds of 5,10,20,...80 mph. These factors control the mapping from mph to speed steps. It's not possible to update the database from Microsoft Access, but I wrote a couple of Excel VBA macros to read and update the table from an Excel spreadsheet and I was able to improve the linearity of the speed control. You seem to be skilled in Excel, so perhaps you could try doing the same.

 

Regards, John

[Fishmanoz]

HRMS has confirmed previously that enabling scale speeds does not make any changes to the decoder, it is all done in RM as Bb says.  So the only way 28 sped steps is enabled is if you change CV29 to do so.

 

Also, a return of 255 reading a CV usually means it has not been read, not that it contains 255, consistent with what you have reported from Loksound.

[jodel]

Hi Rob ,John and fishmanoz

First of all, i am sorry with my miles speed calcultation : i mulitplied by 1,609 instead to divid...i'm not accustomed with miles.

your replies are very interesting but i am not to sure  understood correctily. 

1 : for the CV definitions, ok i have seen it in the NMRA doc. it is OK

2. for the mdb table i succeed to read the mdb base with open office dbase. i got this values for my two loco ( one Honrby , the second for a digitalised loco with ESU lokpilot), and i have the same values for this 2 locos : 

speed curve : 3.134,2.754,2.454,2.274,1.854,1.655,

scale factor : 1... ( however i have changed the speed factor for the loco hornby to try to improve my speed profile : i was expecting the valeu for factor to change.?..)

i compared with a hornby databased loco, i have declared in RM and i have this results : 

speed curve : 2.500,2.250,1.690,1.420,1.270,1.170,1.070,0.968,0.960,

scale factor : 0.99

of course it is not the same range of values.... and i cannot interpret them, but i seems to me that it can be perhaps the origin of my problem ... why there are only 9 or 6 numbers for the steps while i thought there are 14, 28 , even 128??.... 

3. for the CV 29 , ok i understood that RM does not write values in CV in Scale mode , but why when changing bit 1 of CV28 to 1,  i have only 28 speeds step?

 

Thanks for your precisions.

 

Joel

[Gitter]

km to mls     divide by 8 x 5.  80km/hr = 50mls/hr. 

I am trying to follow this thread, but it's away beyond my capabilities.

[Yelrow]

Gitter, not only yours. I find this also heavy going, and wonder if RM, was intended for this. john

[Britannia Builder]

Hi Joel

 

If you look at page 38 of the RM manual, 'Auto Profiling', you'll see how RM intended to allow users to calibrate the speed curve in the resource.mdb database - it involves running the loco over a measured distance at 5, 10, 20, ... mph (or 8, 16, 32 kph) and timing the performance. However, it requires the long-awaited Loco Detection system and so isn't available yet. I don't know if RM have some in-house version of LD to enable them to create the speed curves.

 

In the absence of this, and since RM seem to have 'patched' the resource.mdb database so that it can't be updated from MS Access or the like, you'd have to write some software as I did to update the speed curves.

 

Frankly I don't think it's worth the effort - I now just ignore the speeds on the RM throttle controls and twiddle the mouse wheel to get the speed that I want. I ignore the fact that one loco goes as fast at an indicated 20mph as another does at 40mph.

 

If you really want to try updating the speed curves I can post some screenshots of the code I used in Excel (if the Moderators will allow this - I will quite understand if they won't), but I'm not sure whether this would work in OpenOffice as well as Excel.

 

Regards, John

[96RAF]

If you really want to try updating the speed curves I can post some screenshots of the code I used in Excel (if the Moderators will allow this - I will quite understand if they won't), but I'm not sure whether this would work in OpenOffice as well as Excel.

 

Its not the Mods John,  it seems to be the Hornby forum software prevents code being posted. Mods have little authority other than to act on spam, where we can hide the post and block the spammer.

 

I tried to post some simple code a while back and it was rejected - just white space where it would have been on the page.

Rob

[jodel]

Hi John & RAF,

Thanks for the information about the speed curve in the databse. i understand now where these informations in the database came from.

modifying the database is not clearly a problem for me. but the real question for is which values to choose  for udpdating this column in the table of the base ?

and also  i have another question : i read in loksound datasheet, that the bit 5 of CV49 set the decoder is automatic detection of the speeds Steps ( i think it is for their range :14,28, 128...). in this doc it is also written that , ESU decoders recognize automatically the speeds steps and tested for these controllers : 

ESU Ecos, Bachman E-ZROCO lokmaus2,  3Uhlenbrok intelliboxLenz digital plus V2.3ZIMO MX1...

What about Railmaster with elink or even for Elite stand alone ? perhaps it works well but not tested by ESU ?

Does RM works automatically, or always in 128 steps for ex???

thanks for this precision .

 

Joel

[Fishmanoz]

There is no RM speed step setting, it just works with whatever is set in the decoder.  Same with the Elite.  You can directly select the decoder sped step setting from the Elite menus,  With RM, you need to set in CV29.

[Britannia Builder]

modifying the database is not clearly a problem for me. but the real question for is which values to choose  for udpdating this column in the table of the base ?

 

Hi Joel

 

When I was playing with the speed curves I just changed the values by trial and error to get the correct lap times - increase the factors to make the loco go faster. If you already know what speed step value you need for each speed you may be able to work backwards to get the required speed curve factor. I've just done a few timings both with and without scale speeds, and I think the formula that RM uses must be something like

 

speed step = speed in mph * scale factor * speed curve factor * constant, where the constant appears to be about 1.6 - in fact this 1.6 may be the conversion from mph to kph, so perhaps the kph calculation works without a constant multiplier.

 

Regards, John

[jodel]

Hi Fishmanoz

you wrote "With RM, you need to set in CV29." of course but when setting bit 1 of CV 29, is the decoder in 28 steps or is it in 128 steps. i have never seen other parameters in CV to distinguish between 28 or 128?

 

Hi Britanniabuiler

your hypothesis about the formula is very interesting. about the constant, that' s what i thought: a confusion between MPH and KPH in scale factor setting.

But, on the values in my database i noticed that the values for the factor  are always 1 . Factor is  the name of the column : i think in fact this column contains the input paramter for scale factor.

if it is , i think it is not correct to have always 1 for scale factor because i have tested it with different values for scale factor time : 3 sec( the minimum to hope to slow down the speeds), or 30 sec perhaps? and i never saw any change on the measured speeds !... even if it could not be a perfect speed curve, i should have noticed a modification in the measured speeds, even at a minimum. 

in fact i think this is not correct in my RM with the input scale factor parameter  ::it looks like i never commit the data in the base, even if i cliked on each green tick.....

what do you think about that ?

Thanks for your interesting replies.

Joel

[Chrissaf]

CV29 Bit 1 just tells the decoder how to handle the incoming DCC packet containing speed step information. Thus the speed steps being 28 or 128 is a function of 'what controller am I using'. These days, most modern controllers use the 128 step DCC packet format by default. Some controllers give the user a choice for compatibility with older decoders that do not support the 128 steps.

.

Want to know more - see this link. Mark Gurries 14-28-and-128-speed-steps

[Britannia Builder]

Hi Joel

 

I've just tried entering a speed factor time for one of my locos in RM (by clicking on the stopwatch icon in the loco setup screen and entering an arbitrary 5 seconds) and this did update the database with a new value in the Factor column. It wrote to the database after clicking the green buttons - I didn't need to exit Railmaster. I'm surprised that Factor is set to 1 for all your locos - my 6 locos range between 0.406 and 0.8, and some if not all of these are the original values as installed by RM. Changing the speed factor does change the speed for my locos.

 

Regards, John

 

Edit: is it possible that your spreadsheet is rounding the Factor column to the nearest integer?

[jodel]

Hello John ,

Thanks. you confirm what i was feeling about the factor in the database. i  always thought that when changing the scale factor,  the measured speed must change also, even if perhaps it could not be efficient at 100.% to have an accurate result.

for your question , i dont understand what you mean when you write "the factor on my spreadsheet"... i have not a column factor in my spreadsheet.. 

Can you precise more your question.?

Hello Chrissaf

thank you for this web site : after a first glance on it , it seems to me to be very detailed and complete. i have to read it carefully...

thanks for you

JOel

[Britannia Builder]

Hi Joel

i dont understand what you mean when you write "the factor on my spreadsheet"... i have not a column factor in my spreadsheet.

Sorry, I meant the Factor column in the database - I assumed that you were using a spreadsheet to view it, but perhaps you are using MS Access or the OpenOffice equivalent.

 

Regards, John

[jodel]

hello John,

you can see on the loaded image how i got this datas. these result was always the same for the HJ2211S with different input time for scale factor. ( 3sec, 9,5sec; 30sec).

 

/media/tinymce_upload/0c7bd0504f8e9caeb5097d1b03e65577.jpg

Is it what you want?

joel