My analog channels (some of them anyway) have an offset that send low voltages below where they really are (e.g. 0.5V becomes 0.2V or so). Digging around in the config tool, I found this:
[img]https://twitter.com/AOERacing/status/387668858166722562/photo/1/large[img]
All I can do however is hit cancel, can't edit anything, can't send anything, can't enter calibration mode. I see instructions in the MkII manual for how to do this, but nothing for the MkIII.
I guess I can change the units on my channels, but that seems a bit less than ideal, because the way it is set up now clips low voltages down to zero, so to correct the values in the units config, I'd end up with a non-zero minimum value. E.g. 0V-0.11V would all get mapped to 0.11V...
Calibrate Analog Channels on DL-1 Mk3?
-
Support
Support
Hello,
The calibration on the MK3 is not something can be done by the end user. You will need to connect an accurate variable power supply to all channels. If you are able to set this up we can organize a teamveiwer session to allow me to connect to your computer and enable the calibration mode.
Alternatively you can send the DL1 back for servicing but charges will apply if the unit is older then 1 year.
We have recently improved the analogue channels accuracy substantially and it may correct the offsets at the low voltages.
Please try updating the firmware from the website and checking the offset again.
Kind regards,
Support (K)
The calibration on the MK3 is not something can be done by the end user. You will need to connect an accurate variable power supply to all channels. If you are able to set this up we can organize a teamveiwer session to allow me to connect to your computer and enable the calibration mode.
Alternatively you can send the DL1 back for servicing but charges will apply if the unit is older then 1 year.
We have recently improved the analogue channels accuracy substantially and it may correct the offsets at the low voltages.
Please try updating the firmware from the website and checking the offset again.
Kind regards,
Support (K)
What rev of the firmware changed the A2D process?
I'd seen this behavior change, reported it, and was roundly told that nothing had changed.
Honestly, the #1 thing RT needs to come up from "tolerable" to "good" is a working software change control process. The willy nilly process that's in place now does your excellent engineering no service.
Since 99% of the value of this device is to compare the data from one day to another, changes that result in the same input creating a different output should be very clearly communicated. Like shouted from the rooftops and given a new major release version.
-Kyle
I'd seen this behavior change, reported it, and was roundly told that nothing had changed.
Honestly, the #1 thing RT needs to come up from "tolerable" to "good" is a working software change control process. The willy nilly process that's in place now does your excellent engineering no service.
Since 99% of the value of this device is to compare the data from one day to another, changes that result in the same input creating a different output should be very clearly communicated. Like shouted from the rooftops and given a new major release version.
-Kyle
-
Support
"willy nilly" 
I'm fairly sure we are neither "willy" or "nilly" - I think you are being just a little harsh! All firmware and software comes with a very complete version history, before any firmware is released we go over a very complete check list. For the DL1mk3, this includes just short of 300 tests and covers everything from lap timing to user defined variables. We've got an engineer working full time just on testing firmware. I'm certainly not claiming we are perfect, but our procedures are pretty damn good!
With regards the analogue inputs on the DL1mk3 products we did make a change to reduce the noise from v4.3.1. As part of the 1Khz sampling update the system was changed to oversample the analogue inputs at about 16khz and uses a simple IIR filter. The cut off frequency of this IIR is significantly higher than the hardware filters that are already on the board, so the user wouldn't see any significant difference in the logged data apart from the noise reduction.
There have been no updates to the ADCs on the DL1mk2 products.
I'm fairly sure we are neither "willy" or "nilly" - I think you are being just a little harsh! All firmware and software comes with a very complete version history, before any firmware is released we go over a very complete check list. For the DL1mk3, this includes just short of 300 tests and covers everything from lap timing to user defined variables. We've got an engineer working full time just on testing firmware. I'm certainly not claiming we are perfect, but our procedures are pretty damn good!
With regards the analogue inputs on the DL1mk3 products we did make a change to reduce the noise from v4.3.1. As part of the 1Khz sampling update the system was changed to oversample the analogue inputs at about 16khz and uses a simple IIR filter. The cut off frequency of this IIR is significantly higher than the hardware filters that are already on the board, so the user wouldn't see any significant difference in the logged data apart from the noise reduction.
There have been no updates to the ADCs on the DL1mk2 products.
-
Support
Sorry - the only option to recalibrate the unit is to get it back to us. If you would like to arrange this please email sales@race-technology.com.
Admitting you have a problem is the first step to solving it.
I was sure that my post would result in your response. Sorry to be truthful and thus harsh. I've been an RT user for over a decade. You leave me little choice.
A firmware revision resulted in a significant change to the analog output on my MK3. Reverting it put it right back where it was. Reinstall the update, and broken again. Revert - back to previous value. Which was right?
There was nothing in the release notes that would have indicated this was coming.
There are literally countless similar issues that have come and gone over the years.
300 tests for the MK3 is WAY WAY WAY too few. Think of it this way, there are 12 analog inputs, 4 outputs, and 4 freq. Throw in the GPS, accelerometers, etc, and you're at 30 channels (or there about).
10 tests per channel? Not even close to enough. And then consider that you have tested NOTHING ELSE. 3k is more like it.
The best I can say is that this industry is flooded with similarly under tested pieces.
Honestly, I've just nearly given up on reporting these issues. You just deny they exist or simply never fix them.
The single most important channel (speed) isn't immune to these issue. I see speed dropping off when I'm still at WOT and no brakes. This never was before. Maybe it's just wind gusts just at every braking zone at every corner at every track.
Think MOTEC has these sorts of issues with this frequency? They dont.
It's one thing to make mistakes. It's another entirely to deny that they exist.
The simple fact is that you use your customer base as your QA department.
-Kyle
PS: It's been nearly TWO YEARS since I bought my Mk3 for the promise of continuous predictive lap timing (time slip rate). AND IT'S STILL NOT THERE!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
I was sure that my post would result in your response. Sorry to be truthful and thus harsh. I've been an RT user for over a decade. You leave me little choice.
A firmware revision resulted in a significant change to the analog output on my MK3. Reverting it put it right back where it was. Reinstall the update, and broken again. Revert - back to previous value. Which was right?
There was nothing in the release notes that would have indicated this was coming.
There are literally countless similar issues that have come and gone over the years.
300 tests for the MK3 is WAY WAY WAY too few. Think of it this way, there are 12 analog inputs, 4 outputs, and 4 freq. Throw in the GPS, accelerometers, etc, and you're at 30 channels (or there about).
10 tests per channel? Not even close to enough. And then consider that you have tested NOTHING ELSE. 3k is more like it.
The best I can say is that this industry is flooded with similarly under tested pieces.
Honestly, I've just nearly given up on reporting these issues. You just deny they exist or simply never fix them.
The single most important channel (speed) isn't immune to these issue. I see speed dropping off when I'm still at WOT and no brakes. This never was before. Maybe it's just wind gusts just at every braking zone at every corner at every track.
Think MOTEC has these sorts of issues with this frequency? They dont.
It's one thing to make mistakes. It's another entirely to deny that they exist.
The simple fact is that you use your customer base as your QA department.
-Kyle
PS: It's been nearly TWO YEARS since I bought my Mk3 for the promise of continuous predictive lap timing (time slip rate). AND IT'S STILL NOT THERE!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
-
Support
Morning Kyle!
Blimey - WRT to testing, it's clearly never going to be an argument we are going to win. The only thing I would add in is that the 300 tests I mention are individually quite involved - working full time and with a complete test bench set up, it takes several weeks for the engineer to complete the full list. But clearly we are a long way from perfect... and whilst things still get through the net we need to improve.
I'll check on the particular analogue calibration issue and let you know what we find - if that is something we've missed then I sincerely apologise.
If you have anything else that you would like me to check on then as always you can email me and I'll do my best (you have my email address), or if you wish to make a more public demonstration on your depth of feelings, then of course you can post on the forum.
Kind regards,
Andy in Support (and now in need of a strong cup of tea)
Blimey - WRT to testing, it's clearly never going to be an argument we are going to win. The only thing I would add in is that the 300 tests I mention are individually quite involved - working full time and with a complete test bench set up, it takes several weeks for the engineer to complete the full list. But clearly we are a long way from perfect... and whilst things still get through the net we need to improve.
I'll check on the particular analogue calibration issue and let you know what we find - if that is something we've missed then I sincerely apologise.
If you have anything else that you would like me to check on then as always you can email me and I'll do my best (you have my email address), or if you wish to make a more public demonstration on your depth of feelings, then of course you can post on the forum.
Kind regards,
Andy in Support (and now in need of a strong cup of tea)
-
Support
PS: I wonder if the chap at Microsoft is having a similarly pleasant start to his week:
http://www.theregister.co.uk/2013/10/20 ... 81_update/
http://www.theregister.co.uk/2013/10/20 ... 81_update/
-
Support
Support
Hello,
Andy asked me to test the analogue channels are report my findings. I tested with 3 versions of firmware and didn't see any significant offset changes.
• 4_4_9 (Current)
• 4_3_4 (Randomly picked version)
• 1_0_60 (Pre analogue upgrade)
I connected all the analogue inputs to the DL1's 5v reference output which the multi-meter measured 5.02v.
If you know of the firmware version that has been causing the problem i will retest and confirm.
Kind regards,
Support (K)
Andy asked me to test the analogue channels are report my findings. I tested with 3 versions of firmware and didn't see any significant offset changes.
• 4_4_9 (Current)
• 4_3_4 (Randomly picked version)
• 1_0_60 (Pre analogue upgrade)
I connected all the analogue inputs to the DL1's 5v reference output which the multi-meter measured 5.02v.
If you know of the firmware version that has been causing the problem i will retest and confirm.
Kind regards,
Support (K)
My tests were between the 3.x (not sure of x) and 4.4.9.
Specifically, it affected A1.
Test at lower voltages (under 1) and understand that 1/10th can be a big change for some sensors.
0 and 5v should never be the only values tested.
This was VERY repeatable for me. I'm sure I could get it to happen again.
I dont want to get into an online debate about your release process. You have my email too. If you want to open a dialog offline, please feel free.
I was trying to, yet again, point out that you need to address software quality. I thought a little public kick in the backside might motivate you.
What's the status of time slip from a Mk3 to a Dash2?
-Kyle
Specifically, it affected A1.
Test at lower voltages (under 1) and understand that 1/10th can be a big change for some sensors.
0 and 5v should never be the only values tested.
This was VERY repeatable for me. I'm sure I could get it to happen again.
I dont want to get into an online debate about your release process. You have my email too. If you want to open a dialog offline, please feel free.
I was trying to, yet again, point out that you need to address software quality. I thought a little public kick in the backside might motivate you.
What's the status of time slip from a Mk3 to a Dash2?
-Kyle
-
Support
Hi,
Kieran did some more work on this, we tested all published firmware revisions that have been made available. We tested at a range of voltages and our findings are that at mid range voltages (5v as above) there is no noticeable changes in DC levels, at lower voltages (0.5v and under) there is a very small DC voltage shift of IRO a few mV, however the new filtered voltage (v4.x.x) is well within the noise floor of the older nosier data (v3.x.x and before).
So where does that get us?
From the very extensive testing, I don't think that there is an error or change in our firmware that shifts the DC characteristics of the analogue inputs. However, I do entirely believe you are getting exactly what you say. What I would suggest is that perhaps it's something to do with the characteristics of the noise of your system. For the tests we are doing on the bench we've got a low noise voltage reference, and so the noise is probably close to Gaussian, so averages to 0.
In your case I wonder if the noise is in the form of spikes (from ignition etc). With the previous firmware with the lower sampling perhaps these were "missed" or aliased out to a large extent, however, now the inputs are oversampled at 16kHz and filtered with the IIR filter they are all "caught" and affecting the apparent DC levels.
I certainly don't know that this is the case, but it would explain what we are both seeing. If this is the case then it might be worth looking at cable routing and or additional shielding on any low level sensor signals.
With regards which firmware is "right" the later IIR filter revisions improve the noise performance by a order of magnitude, and is a better engineering solution - so I completely stand by these.
I wanted to wrap this up on-line so other users are aware in case they seem the same issue - hopefully you don't take the above to be "point scoring", but a full and fair response.
Can we hug and make up now please!
RT Support - Andy
Kieran did some more work on this, we tested all published firmware revisions that have been made available. We tested at a range of voltages and our findings are that at mid range voltages (5v as above) there is no noticeable changes in DC levels, at lower voltages (0.5v and under) there is a very small DC voltage shift of IRO a few mV, however the new filtered voltage (v4.x.x) is well within the noise floor of the older nosier data (v3.x.x and before).
So where does that get us?
From the very extensive testing, I don't think that there is an error or change in our firmware that shifts the DC characteristics of the analogue inputs. However, I do entirely believe you are getting exactly what you say. What I would suggest is that perhaps it's something to do with the characteristics of the noise of your system. For the tests we are doing on the bench we've got a low noise voltage reference, and so the noise is probably close to Gaussian, so averages to 0.
In your case I wonder if the noise is in the form of spikes (from ignition etc). With the previous firmware with the lower sampling perhaps these were "missed" or aliased out to a large extent, however, now the inputs are oversampled at 16kHz and filtered with the IIR filter they are all "caught" and affecting the apparent DC levels.
I certainly don't know that this is the case, but it would explain what we are both seeing. If this is the case then it might be worth looking at cable routing and or additional shielding on any low level sensor signals.
With regards which firmware is "right" the later IIR filter revisions improve the noise performance by a order of magnitude, and is a better engineering solution - so I completely stand by these.
I wanted to wrap this up on-line so other users are aware in case they seem the same issue - hopefully you don't take the above to be "point scoring", but a full and fair response.
Can we hug and make up now please!
RT Support - Andy
You buy the first round and we're square.
Oh ... and get time slip to my dash 2.
I'm just trying to lead the horse to water. I appreciate that you're trying to reproduce it. I know that I can.
-Kyle
BTW: the issue was tested with the car not running and the signal is from a Kavlico pressure transducer on a brake pressure line. So I'm not sure about noise as a factor. I did see a notable improvement in noise handling in the steering position channel. That multiturn pot is noisy.
Oh ... and get time slip to my dash 2.
I'm just trying to lead the horse to water. I appreciate that you're trying to reproduce it. I know that I can.
-Kyle
BTW: the issue was tested with the car not running and the signal is from a Kavlico pressure transducer on a brake pressure line. So I'm not sure about noise as a factor. I did see a notable improvement in noise handling in the steering position channel. That multiturn pot is noisy.
Support wrote:"willy nilly"
before any firmware is released we go over a very complete check list. For the DL1mk3, this includes just short of 300 tests and covers everything from lap timing to user defined variables.
First of all sorry for the thread jack, but honestly disquek isn't being harsh. I'm sure you guys are trying your best, but lately I cringe any time I have to make a configuration change or flash firmware wondering what will break.
I spent more than a year trying to chase down problems with my system that Al Seim finally tracked down to your config program writing out a "bad" file that the DASH4Pro accepted. Taking the exact same years-old layout file and rewriting out a hex config file with a newer version of the software fixed it.
I still don't have real-time prediction working between my Mk3 and DASH4Pro. (The whole reason I updated as well.) Neither the lights, nor the time slip variable (displayed as text or a bar graph) work. Is this a feature that is supposed to be working?
Most recently I had an alternator (suspected of) failing so I changed my main screen to display the car's voltage. Imagine my surprise when I had no lap times, nor shift lights after updating the config. Turns out this was a DIMM that had gone bad on my laptop, but you'd hope that there would be some validation of files the DASH4 accepted.
Then you add this to the long list of feature requests that have never been done. (Relatively simple things like having a clock variable that would be super useful for enduros.)
I realize you guys have to have a revenue stream to exist, and I wouldn't even be opposed to a subscription fee after X amount of time for the software if it actually meant getting features implemented/fixed, but it is frustrating to see new products released constantly, while requested/promised features that would make your current products better/functional never get addressed.
I still think you have the best product on the market, having used them since the Mk1 and before you even had a dash having to pair it to an AIM system and have sold 10+ systems for you, but please listen to disquek and others that the software engineering could be better.
P.S. I also saw weirdness in our speed channel on an enduro car we just built where I took one of my older Mk2's (it is on V39..see above about dreading flashing it to V43, not to mention that I can't get your serial port to work with serial<->usb adapters so it is a PITA to flash) Notice how the red trace has surges in the speed at the end of straights despite the acceleration decreasing. If accel is decreasing, the slope of the speed should not be increasing. (Blue is a different driver, same car, same day, different session. Doesn't really show the same behavior.)
-
Support
I don't think I was involved in these support requests, so I can't comment. If you still have outstanding issues then let me know directly at andrew@race....com.
With regards charging for updates - for a while asked users to start registering software with an email address (note: no suggestion of charging, only registering so we could email about updates) and we got everything short of death threats from customers
With regards the bump on the speeds, this is due to a timing-misalignment of accelerations and GPS speeds. It can also happen when there is lots of vertical vibration, I've seen it a couple of times on formula cars going over a curb - and in this case the only fix is to use GPS speeds only (not with accelerometers) or add a little vibration isolation into the installation. If you think this doesn't help email me the run and I'll check it over.
RT support (A)
With regards charging for updates - for a while asked users to start registering software with an email address (note: no suggestion of charging, only registering so we could email about updates) and we got everything short of death threats from customers
With regards the bump on the speeds, this is due to a timing-misalignment of accelerations and GPS speeds. It can also happen when there is lots of vertical vibration, I've seen it a couple of times on formula cars going over a curb - and in this case the only fix is to use GPS speeds only (not with accelerometers) or add a little vibration isolation into the installation. If you think this doesn't help email me the run and I'll check it over.
RT support (A)
Return to “General support questions”
Who is online
Users browsing this forum: No registered users and 39 guests
