Roland A-800 Pro � Joystick "calibration"

[dalpozlead]

Hi guys,

Just bought a brand new Roland A-800 Pro Controller and find out the joystick has an annoying deadzone from its rest position, both up and down (confirmed via MIDI-OX).

It takes almost 20% of its range for the pitch start to be affected⦠I wonder if this also occurs in the flagship synths using the same joystick modelâ¦

 

Wonder if it is possible to adjust this (modifying hardware or in any settings).

 

Tks!

[Dr Mike Metlay]

20% in either direction seems a bit much, but a dead zone is pretty much vital to anything of that design; otherwise, tiny mechanical misalignments or vibrations can cause a neverending stream of meaningless MIDI data.

 

I had this problem with the mod paddle on the Oberheim Xk; the dead zone was nearly nonexistent and the only way to fix things was to disassemble the keyboard, reseat the paddle on the mod pot while watching a MIDI monitor, and then use Permatex to try and hold everything in place.

 

However, I have owned a couple of different Roland boxes over the years and I still have my U-50; because the joystick tends to be used hard by less-professional players, the dead zone might in fact be deliberately large. (I remember when they started adding regular pitch and mod wheels to some of their upmarket controllers.)

 

As for mods, I think this is mainly in firmware and I don't know how to get at it; my Rolands have been so reliable that I've never had to dive into that mechanism. Possibly an email to Roland US tech support?

 

I'm sorry if this doesn't help you, but you could always swap it for another unit and see if the problem is just as bad (usually a retailer will ship a spare and wait for you to return the old one before you get credited).

 

Good luck and let us know what you find.

 

mike

[dalpozlead]

Thanks Mike, It seems very much to be a design fault, but don't know if this is also present in their pro models...

I'll try checking internally to see if there are any mechanical solutions....

[J. Dan]

Not sure about the A800 but I have a much older Roland A-70 that needed the bender recalibrated. I found the service manual and there was a procedure. I had to power it up while holding a combination of buttons and it went into a service mode where you could make such adjustments. See if you can get your hands on the service manual, I would assume it has something similar.

[Reezekeys]

I found a service manual here: https://www.manualslib.com/manual/1604967/Roland-A-300pro.html#manual

 

There's instructions for entering a "test mode" but all you can do with the PB is a "bender check", i.e. you move the PB lever and the display shows "OK" if things look good. If not, well you're own your own there, buddy â good luck!

 

I assume you've done a power cycle â that's my first move when I see any weirdness with my A800. A few times a year all my controllers stop sending data. The keys still work. Power cycling fixes that. Step two is a factory reset â it's in the manual. Make sure all your memories are saved, of course. Step three â check to make sure the PB pot isn't loose! That bit me once. You need to take the keyboard's top case off to get to the PB assembly but it's not too hard. While you're at it you may want to adjust the AT trimpot to max (if you have to lean hard on the keys to send AT you have a common issue with these boards, maybe Roland got wise and fixed this).

 

For s&g I filmed my PB's travel to see how far it needed to be off the center to send its first PB â 0 value. I'll assume yours needs more travel than mine. [edit - I wish I knew how to make this u-tube embed smaller!]

 

[video:youtube]

[dalpozlead]

Thanks Reezekeys,

 

Watching your video I think I have the same dead zone length as you do, what makes me think this is not an issue with my unit.

 

i'm much more used to have te Korg Triton stick with very minimal dead zone to bend up or down...

[Reezekeys]

Interesting, I've never felt the dead zone on my A800 to be too big. Then again, I've been playing Roland boards with this exact same PB assembly for almost 14 years now â first an Edirol PCR-M80, then an A800 and now a newer A800. Hope you get used to it! At least it's true 14-bit pitch bend â I've heard some boards toss the LSB of pitchbend messages.

[Markyboard]

I believe this dead zone is built into the firmware and can't be adjusted. Here's the techy explanation for those interested:

 

Let's assume the pitchbend wheel uses a single potentiometer as most do. Let's also assume it's a 10 kOhm pot. By the physical nature of how the pot is connected to the wheel at the extreme negative end the pot is set to 0 Ohms and at the extreme positive end its set to 10 kOhms ...or visa versa.

 

That means at the center position the pot is set to approximately 5 kOhms. The firmware reads the position of the wheel/pot with an analog to digital converter. It then determines what data to send (or use internally) based on that reading. Based on Rob's video it translates the pot's resistance value to a digital value between -256 and +256 with 0 being the mid point (even though midi allows for much better resolution).

 

Divide 5K by 512 possible values (the resolution) and this equates to approximately 9.8 Ohms resistance difference for each increment or decrement in value. But chances are the firmware is putting padding around the center position of 5 kOhms meaning that not only does a pot value of 5 kOhms give a value of 0 but also any pot value between 4.8K and 5.2K (for example).

 

I don't have specific numbers but the point is there's nothing you can adjust or calibrate here. They would have to modify the firmware.

[dalpozlead]

I believe this dead zone is built into the firmware and can't be adjusted. Here's the techy explanation for those interested:

 

Let's assume the pitchbend wheel uses a single potentiometer as most do. Let's also assume it's a 10 kOhm pot. By the physical nature of how the pot is connected to the wheel at the extreme negative end the pot is set to 0 Ohms and at the extreme positive end its set to 10 kOhms ...or visa versa.

 

That means at the center position the pot is set to approximately 5 kOhms. The firmware reads the position of the wheel/pot with an analog to digital converter. It then determines what data to send (or use internally) based on that reading. Based on Rob's video it translates the pot's resistance value to a digital value between -256 and +256 with 0 being the mid point (even though midi allows for much better resolution).

 

Divide 5K by 512 possible values (the resolution) and this equates to approximately 9.8 Ohms resistance difference for each increment or decrement in value. But chances are the firmware is putting padding around the center position of 5 kOhms meaning that not only does a pot value of 5 kOhms give a value of 0 but also any pot value between 4.8K and 5.2K (for example).

 

I don't have specific numbers but the point is there's nothing you can adjust or calibrate here. They would have to modify the firmware.

 

Thanks for the explanation!!!

[Tusker]

My A500 has a slight dead zone also. It took some adjusting of the pitch bend approach.

 

One question for a performer (particularly someone who has invested in vibrato or pitch bend technique) is ... how do you minimize the disruption of switching between different joystick implementations? The safe (and expensive) answer is to use Nord hardware, which is very constant in it's implementation over time. With the Roland hardware there are a couple of types of short throw paddles and a couple of types of longer throw paddles at different price points. So there can be an adjustment period, if you switch. A case can be made for buying older Roland workstations (fantom's etc.) as controllers, so as to make sure the paddle implementation is identical. I use a Touche, to delink keyboard action from pitch bending device. It's not convenient to carry an extra device, but if Expressive E stays in business (fingers crossed) perhaps I can develop better technique over a longer period of time.

 

What paddle implementation are you switching from?

[Reezekeys]

Since I have a little more time on my hands than usual I looked a little deeper into the A800's pitch/mod paddle's behavior.

 

Moving the PB lever as slowly as I could I got it to send around 35 separate messages in each direction. I say "around" because a few times I got 36. I'm sure there's a routine that looks at the "rate of change" in the pot's value, which translates to how often a message should be generated. Any small hesitations on my part while moving the lever probably account for what I saw. I'm pretty sure about this because when I let go of the paddle and it quickly snaps back to the center position, only four messages are generated. This makes sense as the PB change is more or less instantaneous, so a regular 5-pin midi stream might be overloaded by a blast of 35 separate PB messages in such a short time. This begs the question of whether this behavior is a holdover from midi 1.0/DIN midi; would current USB midi 1.0 streams be affected the same way, or could midi 2.0 handle a greater density of PB messages? (I'm assuming midi 2.0 could do it pretty easily). In the end it probably hinges on our perceptions. There are very few times in my life I've needed to do a wide pitch bend â 99.99% of the time my PB range is 2 semitones each way. I've never been bothered by stairstepping, but for those who need wider PBs it could be a factor.

 

For grins I looked at the mod wheel. Again, moving the paddle slowly I was able to generate about 54 separate messages from 1 up to 127.

 

Yea, I definitely have some time on my hands!

[Markyboard]

I believe the scanning rate of the pitch bend is a constant rate. But it shares the same scanner with all the pots and sliders and it's time multiplexed so that the A/D and processor can read it and update the register/latch that holds the current value of each parameter. Its these values that are used to generate the midi messages. They could give priority to the pitch bend or sample it twice per scanning cycle giving it added (2x?) resolution. The number of midi messages you see is probably based on the scanning rate vs the physical update rate of the pitch bend wheel. Also the 32 kbps midi rate plays into the timing scheme.

 

Who said midi controllers don't cause aliasing?

[Reezekeys]

Got it. As usual I'm overthinking things - a "routine" that measures how often the pot's value changes, lol.

 

Now I actually have work to do (making the latest intructional video for a [virtual] after-school program I work for), so my exhaustive investigation into the intricacies of pitch benders will have to wait a little longer. Thanks for clearing things up!

[dalpozlead]

Hi, I`ve made a video (inspired by Reezekeys), with clear data:

[video:youtube]

[Markyboard]

Wow - OK much worse than I thought. Not only is there the large dead space around 64 (zero pitchbend) but it's reaching both extremes way too early. Apologies for getting caught up in the 'how stuff works' - It's a sickness

 

I would try the reset that's documented in the service notes that Rob provided. Also have you checked your version #? Maybe there's been an update.

If the reset doesn't improve things I would try the bender test. If it fails then you can contact Roland. Maybe there's a separate calibration procedure? .

 

Btw thanks for doing that video - it's very helpful for understanding the extent of the problem.

[dalpozlead]

Thanks for all the inputs, I really think this is a design definition and not an issue per se.

Unfortunately I don`t have any other Roland to compare...

[Reezekeys]

I might be talking out my butt (once again!) but my understanding is that potentiometers can have rather loose tolerances when it comes to their values and perhaps how their resistance varies during the wiper's travel. Maybe Mark can clarify whether I'm all wet, but I wonder if it might be worth the trouble to replace the pot with a better-quality unit. I recently needed a pot for my bass as the shaft was broken. It was labeled as a 100K but measured about 86K from end to end.

[Markyboard]

I might be talking out my butt (once again!) but my understanding is that potentiometers can have rather loose tolerances when it comes to their values and perhaps how their resistance varies during the wiper's travel. Maybe Mark can clarify whether I'm all wet, but I wonder if it might be worth the trouble to replace the pot with a better-quality unit. I recently needed a pot for my bass as the shaft was broken. It was labeled as a 100K but measured about 86K from end to end.

 

 

It depends on how they implemented the circuit. If you recall our many discussions on expression pedals the pot value really doesn"t matter all that much - FOR MOST INTERFACES. There are exceptions but most of these pots are used as one (of 2) resistors of a simple voltage divider where you get the full or almost full supply voltage feeding into the A/D at one extreme of the pot and ground at the other end. And half the voltage at the middle point.

 

So all this assumes the pitchbend circuit is designed similarly.

 

I can see a bad or out of tolerance pot affecting one extreme but not both. And it definitely wouldn"t explain the middle deadband.

 

Do you have those early arrivals at the extremes as well?

[Reezekeys]

Yes I remember the expression pedal discussion. So the answer is that a tighter-tolerance pot will probably not solve this problem. I guess it all boils down to how annoyed dalpozlead is with the current feel of the PB. I don't give it a second thought though my bender might have the same dead zones as his. Is it that bad that he might want to return the keyboard & get a different one? Stay tuned!

[Markyboard]

Hey Rob- for a non-professional (whatever That is) engineer/tech you are one of the most knowledgeable people I've come across who always tries to understand the inner workings of these keyboards. You ask great questions, make me think and I really enjoy these discussions.

 

We may just have to declare you an honorary geek. Of course we'll need some references from people whose equipment you broke trying to fix it.

[Reezekeys]

Thanks Mark, you're making me blush! I'm not ready for the honorary geek title yet I'm afraid â does helping someone install a laptop hard drive really count as "fixing" anything? Other than that and a few solder jobs my record is intact - so far. This can only be attributed to very good luck! PS - I always have questions. Answers don't come as easily!

 

(I just remembered I fried the output transistors of my McIntosh integrated amp trying to use it as a keyboard amplifier many years ago - I had hooked up my Rhodes through a "bucket brigade" chorus I built on perfboard from a schematic somewhere!)

 

PPS - just checked Ebay and saw a BIN price on my McIntosh model for.... $2700! I think I have my next restoration project mapped out! Prepare for MORE QUESTIONS!

[dalpozlead]

Is it that bad that he might want to return the keyboard & get a different one? Stay tuned!

 

LoL

The answer is? No, I wont return this gear... don`t have the budget to get another one and I might get used to the issue... I`ll definitely check Roland`s top notch boards like the new Fantom or Jupiter X to see if the bender has the same behaviour.

I doubt it has....

[Reezekeys]

I'm not shocked that you're gonna keep it. I could come up with better reasons to return it - or not buy it in the first place! The crappy little LCD display reminiscent of my DX7 from 1985 (even worse, it's located where your viewing angle makes one of the displayed digits unreadable). The brain-dead semitone transpose function that requires menu-diving several levels every time you want to transpose to a new key (oh please 'tranpose police', I'm not looking to start that debate â just commenting on a very clunky implementation!). Despite these issues I'm on my second A800. I live with its deficiencies mostly because I like the synth action, it speaks Roland sysex easily (a feature I don't need as much anymore, admittedly), and... it weighs only 10 lbs (4.5 kg)! That's super important for doing gigs where you need to carry two powered speakers plus stand & all accessories in one trip, like I have to sometimes. Except for the USB jack on my first A800, I've found the build quality to be decent â my first A800 flew several hundred thousand miles on the road with me and the second has been fine. Here's hoping you get used to that bender paddle!

[dalpozlead]

Great updates from a fellow youtuber... this shows Roland doesn't treat the "budget" gear as the premium ones...

[video:youtube]

I'm sad...

[Reezekeys]

Yes that data stream sure looks different than what you get from the A800. It's hard to tell if the paddle assembly on the Fantom is the same as what we have. The top case of the Fantom is cut differently but it might be the same internally, and the increased precision could come from an A-to-D that uses more bits than what ours does. It seems weird since the A800 does send true 14-bit values - not 14-bit numbers with the LSB zeroed. Well, boo Roland! Luckily for me I've never needed more than 2 semitones of bend either way so this shortcoming hasn't really impacted me. For those doing steeper bends I can see where it might be a problem.

[dalpozlead]

I guess I"ll keep searching for a good controller with pitch stick...

[Markyboard]

Not surprised here. They're using the midi spec 14 bit messages instead of 7 bit in the A800-Pro. Since the Fantom generates sounds internally there are other parts of the design architecture (Analog-Digital Converter for one) that require higher quantization levels. The better resolution pitch bend sort of comes for free.

[drawback]

I keep looking at the pitch stick thinking it would make a great fast/slow Leslie control for B3X (spring back to center is livable). But the rest of the controller seems overly complex to set up, plus I'm not positive you can set the sliders in drawbars mode anyway. I haven't seen another controller with a stick that would retain all the other upscale inclusions.

[brenner13]

Nearly every Roland keyboard I"ve ever owned, the pitch sticks have gone wonky. I don"t buy Rolands anymore because they no longer sell sell the replacement parts directly to the public. They expect us to ship off our instruments to some shop a hundred miles away or more and spend a third of the cost to replace the whole synth to have a simple $50 part installed when their poor quality designs fail.

I"ve been biting my lip for a while about this but the thread all but forced the airing of my grievances.

[Markyboard]

I"ve been biting my lip for a while about this but the thread all but forced the airing of my grievances.

 

Air away- I"m totally with you on this bs.