weird pot situtation.

[thetimboroni]

so thanks to the troubleshooting tips i finally got "some" pots working.  now i've got a weird situation though.

in the attached picture, the x's represent pots that don't work, and the checks represent pots that do.

as you can see by the blue/brown wires (common/+5v), all of them are wired in columns, not rows.  the weird thing is, the pots only work in certain rows...

what are some things i could be looking for?  has this ever happened to anyone before?

thanks.

tim.

[stryd_one]

what are some things i could be looking for?

Something in the system that is not included in your drawing...

[MRE]

I must assume that you are properly wiring the center pin of the pot as the input, and that all pots are identical.

If you wired power in columns, and input pins as rows, perhaps one or more chips/AIN boards is bad or not sending signal.

By "not working" what do you mean?

Sending 5 volts no matter what?

Sending 0 volts no matter what?

No signal at all for that channel?

Have you disconnected from the AIN board and actually verified the pot itself works down to the connector?  Have you checked that each ADC chip is working?

Have you checked your configuration files for the MB?

[thetimboroni]

By "not working" what do you mean?

Sending 5 volts no matter what?

Sending 0 volts no matter what?

No signal at all for that channel?

No signal at all for that channel.

Have you checked your configuration files for the MB?

What should I be checking as far as the configuration files for the MB?  I'm using the MIDIBox64 editor.

thanks.  i'll check the chips individually.

[TK.]

In order to exclude that this is a configuration problem, please check the pots with the ain64_din128_dout128_v2_0 application. Each pot should send an individual MIDI event: B0 <pot-number> <value>

Best Regards, Thorsten.

[thetimboroni]

Ok so here's some more info after further troubleshooting. 

-Every pot has 5v across it

-I tested the configuration like TK suggested above, and got the same results (no response from the ones in the drawing w/ X's, and response from the ones w/ checks)

-I know the pots themselves are good b/c if I switch the wires of a working pot w/ a non-responding pot, the "non-responding pot" now responds. 

-I tried wiring the non-responding pots directly to their corresponding pin on the AIN chip (thinking maybe the leads on the pcb were bad) and still got nothing.

-A1, A2, A5, and A6 are the only ports responding.  The rest do not respond.

If you wired power in columns, and input pins as rows, perhaps one or more chips/AIN boards is bad or not sending signal.

I'm sorry to have confused you.  I made another sketch to show what I really meant.  Everything is wired in columns, I just meant they were responding in rows... see what I mean?

Any more suggestions of things to check would be great.

Thanks for the input so far.

tim.

[stryd_one]

Something in the system that is not included in your drawing...

Now it's included ;)

OK because the columns are wired in series we can instantly tell it's not there... Your voltage tests and swapping pots confirms this well....But what about the wires going to the AIN jumpers, the AIN board and it's IC's?

You should read MRE's post and check it out... Here's an outline of what kind of thing to do:

Swap over the wires running from the pots to the board Eg D0<->D1

If it changes, there's your problem, otherwise swap the 4051's

If it changes' there's your problem, otherwise look for soldering problems on the board

etc etc

How many pots in total are we talking about here? how many 4051's do you have stuffed in how many AINx4 boards?

[thetimboroni]

You should read MRE's post and check it out... Here's an outline of what kind of thing to do:

what post?

Swap over the wires running from the pots to the board Eg D0<->D1

If it changes, there's your problem, otherwise swap the 4051's

I assume you meant A0 <-> A1.  I did that before and like I said, the pot that responds then doesn't respond, and vice versa for the other pot.  In other words, A0 never responds, and A1 always responds no matter what pot is attached to it.  So what does that mean?

How many pots in total are we talking about here?

I have 63 pots on 2 boards.

thanks again for the help.  i'm still really confused about this.  i'm praying that one day i'll wiggle something and it will work.  haha in my dreams maybe.

tim.

[TK.]

I assume you meant A0 <-> A1.  I did that before and like I said, the pot that responds then doesn't respond, and vice versa for the other pot.  In other words, A0 never responds, and A1 always responds no matter what pot is attached to it.  So what does that mean?

Have you ever checked the multiplexer control lines between CORE:J6 and AIN:J6?

If four analog channels are not working, it sounds like one of the three select lines is not connected, or that it has a short

Best Regards, Thorsten.

[thetimboroni]

Yes, all the multiplexers are connected correctly on J6 of both the core and the two AIN modules.  This was actually the first thing I had to fix b/c it was causing tons of random midi events.

beh... I guess I'll just keep looking at this thing.  I can't lose hope.  This will be conquered!

(any other advice?)

tim.

[smashtv]

Have you ever checked the multiplexer control lines between CORE:J6 and AIN:J6?

If four analog channels are not working, it sounds like one of the three select lines is not connected, or that it has a short

Exactly what I was thinking, so I opened the 4051 datasheet to point at a specific pin based on the symptoms.  A0 on the AIN does not = "channel 0" in the datasheet, so I printed the pinout and transposed the AIN numbering to the datasheet "channels".  I -planned- to then use the A-B-C pin truth table to try to call one of those pins suspect....but no such luck.  :-\

There is a pattern when you look at it via the truth table though:

ain	IC pin	IC channel  status	  input states
#	#	#			C	B	A
A6	13	0	    worky!	0	0	0
A5	14	1	    worky!	0	0	1
A4	15	2	    NO WORKY	0	1	0
A7	12	3	    NO WORKY	0	1	1
A3	1	4	    NO WORKY	1	0	0
A0	5	5	    NO WORKY	1	0	1
A2	2	6	    worky!	1	1	0
A1	4	7	    worky!	1	1	1[/code]
It seems the ABC lines are making it to the 4051s just fine.  I cant tie it to a specific pin, and I only included the table showing the pattern in case it rings a bell for anyone else.   :-\
Looking towards the PIC, how does it act when you disconnect CoreJ5, and touch the pins on CoreJ5 with your finger?  Does it seem to be jittering on all 8 analog ins or just 4?  
This [i]will[/i] be conquered!
Without a doubt.  ;)  
Humor me on these three:
Is Core C4 installed in the right direction?
What type of MIDI interface ie gameport, USB, etc.?
Does the regulator run super hot or has it ever?
Best
Smash

[thetimboroni]

Is Core C4 installed in the right direction?

Yes.  The positive terminal is connected to the trace.

What type of MIDI interface ie gameport, USB, etc.?

SB Live MIDI UART gameport.

Does the regulator run super hot or has it ever?

It used to.  Only for short period of time though until I found the problem to be a short on J8.

I'll run the other test you suggested as soon as I get a chance.  Thanks for putting in your effort though w/ that chart!

tim.

[ris8_allo_zen0]

There is a pattern when you look at it via the truth table though:

ain	IC pin	IC channel  status	  input states
#	#	#			C	B	A
A6	13	0	    worky!	0	0	0
A5	14	1	    worky!	0	0	1
A4	15	2	    NO WORKY	0	1	0
A7	12	3	    NO WORKY	0	1	1
A3	1	4	    NO WORKY	1	0	0
A0	5	5	    NO WORKY	1	0	1
A2	2	6	    worky!	1	1	0
A1	4	7	    worky!	1	1	1

It seems the ABC lines are making it to the 4051s just fine.  I cant tie it to a specific pin, and I only included the table showing the pattern in case it rings a bell for anyone else.   :-\

I don't know if this is useful to the cause, but in the pattern I can see the truth table of the expression

W=A XOR B

(where W is 1=working/0=nonworking)

In other words, the pot works only when A and B have different levels... I'd say there's a wiring problem between those lines but I can't explain the nature (short? pins swapped with some other pins?)

[MRE]

Well, Im sort of back posting, in that I should have posted this with my original response. So you may have already 'been there and done that':

Are you using a ribbon cable or a connector block to attach to the AIN boards?

If so,

Have you disconnected the board and done a resistance check for each pot to the connector?

Example: clip one meter lead to the positive line. Then insert a bit of wire into each pot's pin on the connector and clip the other lead to it. Now sweep the pot and verify that it is a nice smooth resistance sweep.

Im guessing this was one of the first things you have tried.

It would still be strange considering your picture showing that entire rows were dead. I only ask to verify it just to make sure its not something silly like a group of wires on the connector block not punched down all the way.

[MRE]

the pattern of 0, 3, 4, and 7 is very strange.. if you are using the current posted AIN module it translates to the first, last and two center pins of each and every connector.

Odd..

I am going to amend something and say that rather than checking at the header pins, pull the chips and check at the chip socket.

So, clip to the postive (or negative) with one lead and stick a wire into the socket hole and clip the other lead, then do a pot sweep.

I once had a problem with a pic where it refused to read my inputs (on another project) I fought with it for days.

Turns out there as a manufacturing defect in the cheap socket and the metal tabs inside the socket were flatened in some pins and never even touched the pic pins!

So, check at the socket with the chip out. Alternatively you could check with the chip in by touching on top of the pin lead, but you need some dexterity to manage both holding the probe and turning the knob.

you need to look for something that could be repeated for each chip and on each AIN board module. Solder joints is unlikely since you have the same set not working on each input chip, and on each board.  So, what is the repeated mistake or (more likely) defect?

Is it possible that you got a bad batch of input chips where a sequence of bad ADCs was repeated for several chips? Highly unlikely, but ceck the make dates on the chips. If they are all different, then you are not using parts all from one batch.

Is it possible you have a batch of bad sockets? You can confim that by checking the pot resistance with the module unpluged while the chips are installed, as described above. Note that if you check the socket chip out, you are not testing that the pot is ACTUALLY connected to the chip. You can only verify this chip in.

Is it possible that there is a miss connection between the boards and core?  Well, not with the chip select, since you get 'some' data from ALL chips. The other possability is with the ABC lines. These set various combinations to form an 'address' to read each input. The work with the chart proves that there is no logical correlation of one wire not working while the other two are fine. Further, if it was not, you would get all sorts of bad data. Exampled by not just "pot works or not" but missread values. A bad address line would cause it to 'read the wrong pot'

Is it possible that the PIC itself has something going on with the outputs for the ABC lines or analog in lines? Well, a problem with the Ainput line would mean that EVERY pot on the assoicated chip would not read. So, if you had whole banks of pots go out, then yes. In your case you get some pot values from each chip, so no on that question. What about the ABC lines? Same situation as above.

[MRE]

bad pots:

0, 3, 4, 7

ABC table:

CBA chanel

000 0*

001 1

010 2

011 3*

100 4*

101 5

110 6

111 7*

Standard binary. Oddly symetrical.  When both A and B are high, OR when all three are high or low. In any case, the distribution shows that all three pins are functioning from the PIC.

Divide and conqure:

Verify that the supect pots give a full resistance sweep ALL the way to the chip input pin.

That is your division point. If you get good resistance sweep on each pot with your probe on the chip pin, then your hardware from that point on is perfect. The problem is from there to the PIC. If not, then do the same test, point by point, up the wiring chain till you find the break.

1: Pot test software has been tested hundreds of times, and is known to work. So Softare check.

2: Based on what we have just shown, your PIC is addressing and reading values from each analog chip fine. Check.

3: In the 'I have a better chance at winning the lotto' chance that you have several chips with the same pattern defect, verify that they have different lot numbers. If they dont, buy a few spares and play chip swap for a bit just to make sure nothing suddenly changes. Check.

One final thought:

The 4051 gives no feedback to the PIC what chanel its on. The PIC simply sets and address and then reads the analog input.

You could play some tricks on it:

Disconnect J6, and tie all the pins on the connector low. This sets all the chips to read pot 0 ( a known bad pot).

as the PIC scans the pots, it will THINK that it is reading each one but its only reading pot 0 7 times.

If pot 0 is bad in the wiring, then suddenly all of your pots will be in failure mode.

If you still have the pattern 0,3,4,7 result you had before, there is something wrong in software or the PIC. Bad registers? Keep in mind that if this is the case, your PIC is reading 1256 properly, and failing on 0347, when in fact it is reading the EXACT same hardware each time.

You could double check by setting the ABC lines to an address for a good pot. All 8 pot values for a chip should reflect the pot value of the address you set. If they do, then software is good, and your hadware is flawed.

[thetimboroni]

MRE,

Thanks a bunch for puttin' so much effort into this.  It's alot for me to take in, still being new to this, but I will look over your posts several times and keep checking stuff.

Thanks.

tim.

[stryd_one]

Wow nice one MRE!

[MRE]

Chock it up to a combination of insomnia (it was 4 am Japan time) and a bit of OCD.

Now I want to hear the end results!  What does it turn out to be?!

Some people enjoy the sunday word puzzle. I enjoy the troubleshooting problem that refuses to die.

Also, I have noted that while this situation is certainly unique.. many people here have some of the same generalized questions: "how do I go about solving this riddle?"

Heh..what gets me is when its something simple in the end. That little 'gotcha!' happens all the time on professional tech benches all over the world, every day.

"Aww.. crap!  I spent two days troubleshooting this friggin thing, and all it was was a loose solder joint!"  But there is still the pride in finding the problem and fixing it.  One half of you your brain (the self concious side) says "I cant BELIEVE it took me so long to find something so simple!"  The other half (the pridefull half ;)  says "I cant belive I FINALLY solved the problem!  Im a friggin HERO!"

Then imaginary milk-maids prance around your head and flirtatiously tell you how special you are...

oops.. went too far there didnt I? heh.

[stryd_one]

Yeh I spent a yew hours tracking down a 'w' that should have been an 'm' last night.  :-[

I also spent a whole 8 hours a few days ago finding out where my C syntax was wrong, only to discover that I'd found an SDCC bug...

But this morning I saw the program running, and dancing milkmaids, and it was all worth it ;D

oops! </highjack>

[smashtv]

"Thinking sometimes produces results more effectively than swearing.

Nevertheless, on a project like this, profanity, judiciously applied, seems to help."

;D

I still want to know if all 8 jitter when you disconnect J5 and touch the pins.....That will (mostly) rule out the PIC as cause.

After that we -should- be able to at least figure out the best angle and force amount with which to apply the precision metric adjustment hammer.  ;)

Best

Smash

[thetimboroni]

I still want to know if all 8 jitter when you disconnect J5 and touch the pins.....That will (mostly) rule out the PIC as cause.

alright, I'm confused there.  If I disconnect COREJ5 I'm just gonna get random MIDI events, so how and I supposed to do this test?

[smashtv]

alright, I'm confused there.  If I disconnect COREJ5 I'm just gonna get random MIDI events, so how and I supposed to do this test?

Sorry!  I should have explained this way better.....

The idea is to keep all but one of the analog input pins at J5 grounded, and watch the ungrounded pin jitter and cause random events (with a touch on the pin to cause worse jitter if needed).

Grounding that one open pin should stop the random events.  Then move to the next pin and test the same way.  Jitter=test signal.    ;)

One of two outcomes:

Jitter on all eight pins when ungrounded as expected.  Cause is -probably- on the AIN or associated wiring.

or

Jitter only on four of the pins.  PIC is cause. (assuming no issue with the traces from PIC to J5, or issues with J5) 

This test is super easy if you are using the DIL pinheaders at J5a and J5b vs. J5, since you can put hard drive master/slave type jumpers on these to ground them.  Obviously you only need four jumpers per header, you don't want to put one on the Vd pins. (a jumper here would short +5 to ground!)

The MB64 app is handy for this test, but anything that reads at least 8 pots and goes nuts with open inputs making jitter should do fine.  ;)

I have seen analog input issues after a pair of PICs had been run on noisy power, like they were stuck off or something.  I describe it that way because I thought the analog ins were fried/dead, but erasing the PICs (even though checksums were good) and reloading everything cured it, analog ins working fine.  Can't call it a fluke since both failed the same way and came back the same way after exposure to the funky power. 

One more random direction:

Sometimes the soundcard MIDI cable ties your MIDIbox ground directly to your PCs ground, and depending on the supply in your PC -and- the supply on the MIDIbox this can be a bad thing.  A large ground differential between the two can be enough to get into the funky power scenario mentioned above.  In other words if you disconnect your MIDI cables from the Core, power up the Core and the PC, put one test lead on ground on the Core, the other on the ground pin in the PC MIDI IN cable, you should get a near 0v reading.  Anything past a few volts is trouble, as the weaker supply will start acting more like load on the stronger supply.

 

We -will- beat this issue, with a hammer if needed.  ;D

Best

Smash

[MRE]

Ahh... good point Smash..  Often people get in the mode of thinking that ground is absolute zero, as if it were some point on a calibrated scale.

Ground is just a reference. Voltage is not absolute, it is relative.

[thetimboroni]

So after soldering and resoldering time and time again, I got to the point where I was actually worse off than I started.  I was just getting random MIDI events.  Finally today I got something to respond again.  Oddly enough, though, now I'm just getting response from A5 and A6.  Having said that here's another piece of info that might help troubleshoot:  When I turn the pots on A5 and A6, the "midi channel" box in Serge's MIDIbox program constantly changes until I stop moving the pot.  There is no pattern to what "channel" it starts and ends on either. 

Ex. -  It will intially be on MIDI channel 10.  While I turn it random numbers appear in the box.  When I stop turning it, it lands on 16.  Then when I turn it back, it lands on 14.  Turn it again, it lands on 12...

Having said that, what exactly does "MIDI channel" have to do w/ it?

Thanks,

tim.

(And yea, as soon as I have a chance I'll do your test Smash!) ;)