LEDs do not light up via LED test program, but do manually

[m00dawg]

So I finally hooked everything up between the mainboard and CS. However, even though testing the LEDs manually (using J2) yields good results, none of the LEDs work when I run the LED test program (see http://www.midibox.org/forum/index.php/topic,10390.0.html). I tested the continuity with the cables and that seems to be good; I think I'm missing something basic here but I'm not really sure what it could be or really where to start testing since I can't get any of the LEDs to illuminate.

Anyone have any ideas?

[Wilba]

Did you try the second test app?

http://www.midibox.org/forum/index.php/topic,10390.msg78577.html#msg78577

Do the encoders work?

Do the switches work?

by "work" I mean really do what they should, not just do something. You'll need an LCD connected for that.

Looking at the base PCB, there are shift registers above where the cables connect.

The left three are outputs, the right five are inputs.

The FAR left one is the current sink for the LED/switch matrix.

In terms of a DOUT chain, the chain is going to the left, so the far left one is last in a chain of three.

Now if you take out the far left IC, you can ground the pins on the cable (at JD8) and thus simulate one "column" of the LED/switch matrix being activated.  If you run the LED matrix test and get LEDs working (even if they are "wrong"), then that's the problem - the IC or connections to that IC or the transistors are all dead (very unlikely)... etc..

Testing the other outputs... you'd probably need to run a DOUT test program and test voltages on each pin.... but shift registers tend to be pretty solid little things and don't break easily... most likely it's a bad solder joint stopping one of them from working, and if one doesn't work, others further along the chain can't work either, hence no LEDs at all.

Take out all those ICs, check continuity of 5V/ground and the shift register connections (see J8_CORE1/J9_CORE1 and refer to Core schematic PDF... RC,SC should be common to all the ICs).

I'll do more thinking, get back to me with more info :)

[m00dawg]

As always Wilba you are absolutely most helpful! The LED test program I tried was the one directly linked off the Wiki for that forum post so I thought it was the second one but I will double check to make sure.

At least some of the knobs are working. The buttons seem like they might be having issues as well although I am able to test button presses etc. I think I will need to load up a test app for those to be sure.

On the note of the current sinking, is there an explanation on how that works exactly? I tried to do some searching on the forums and Wiki and didn't come up with how that's actually doing its magic. Might help me understand what I'm doing wrong :)

In fact, at first I didn't even have the transistors populated because the parts list said their were optional and I was being stupid :)

Either way I will try your recommendations this evening hopefully with some good results! Thanks again for your great help and advice!

[Wilba]

I don't know exactly what you want to know about current sinking... I'll just ramble then:

How a switch or LED matrix works is explained elsewhere...

briefly though... instead of connecting the anodes of 8 LEDs to one 74HC595 (serial in, parallel out, shift register) and connecting the cathodes to ground, you can connect all those cathodes together into one "column", and then add more columns (i.e. with the anodes connected into "rows" and each row connected to the output pin). All the common cathode "columns" can then be multiplexed, by another 74HC595 for example, so one pin will be 0V while all the others are 5V. The 0V pin thus will "sink the current" of one column of LEDs, so whatever outputs on the other 74HC595 will show up on that column of LEDs and not the others, because the other cathodes are at 5V so no current will flow.

The change I made to the MB-6582 is instead of the 74HC595 sinking columns directly with a 0V (and not with a 5V), I used a transistor so a 5V output on the output pin will drive the transistor to sink the current, and a 0V output to not sink the current. It's a subtle difference which really only means that more current can be sinked (up to 100mA per column).

So you can't really test the voltages on JD8 the same way as you would a normal DOUT module output pin.... you can test if it works by using the current meter mode on a multimeter, testing current FLOW between a 5V supply and the transistor's collector... I never really needed to so I can't say if it would work as I expect, it should tell you a difference between zero and something, which will validate if that 74HC595 is working... but more likely is the transistors are really fine (if they're not backwards or you didn't cook em at 400 deg C. for 20 seconds) and you can just test the VOLTAGES on the 74HC595 pin instead.

Maybe you could try a DOUT test program, take out the last 74HC595 (above the transistors), ground all the pins on JD8, then ALL the columns will be sinked, and whatever "pin" being tested by the DOUT test program should light up to 16 LEDs on the CS... at least 8 on the 8x8 mod matrix, and up to 8 somewhere else.

[m00dawg]

Ah ok. I think that will take a while to "sink" in (rimshot, queue wah wah noise) but I think I get the basics. I'll go ahead and try grounding those pins and see what I get.

It makes a bit more sense though as to why I was able to illuminate a test LED by jumping off one of the other LEDs. All the LEDs I tested were reading 3.4V or so (I didn't test them all while running the test program) but no current as best as I could tell. So it sounds like all of the LEDs are being sinked for some reason?

I guess the test will help me figure out for sure what's going on.

Thanks again!

[m00dawg]

So after all my testing, I've determined that there is indeed something up with, most likely, the current sinking. Removing the '595 from the last socket and grounding those pins causes the matrix to light up (I was using the first LED test) on whichever pins were grounded.

I've checked and rechecked the continuity and ran the SRIO tester. Everything appears to be as it should be. I also tried swapping chips but all the chips seem to work as they should. That leaves the transistors but I have a hard time believing they could be the cuplrit?

They are installed the correct way according to the silkscreen (flat part to the right) and they pass a continuity test as well. Could one bad transistor mess up all the LEDs? I wouldn't have thought so?

Can a short on the control surface possibly cause this? I haven't done an extensive test of the switches but the rotary encoders seem to be working at least.

Hmmm...

[Wilba]

If you remove the last 74HC595 and sink JD8 pins and LEDs light up (i.e. from the other two 595s working) then there probably is no problems like shorts or breaks on the CS.

One bad transistor can only mess up those LEDs on that "column".

I suggest swapping 74HC595 to see if they all are working properly. Label them and put the last one (above the transistors) into one of the other slots and try the "grounding JD8" test again.

If they're all working, then perhaps there's a problem with the last 74HC595's socket... ie. not getting 5V/gnd, no connection to the SC/RC or the previous 74HC595 (i.e. how they are "chained"), etc.

You can also try an extension of the "grounding JD8" test you've done... take out the 74HC595 and instead of grounding JD8 pins, apply 5V directly to the 1K resistor where it connects to the IC pin... i.e. test both putting 5V into the socket pin and also to the resistor pin FURTHEST from the transistor, not the one NEAREST (you don't want to put 5V directly into the transistor pin!). This will test if the transistors are OK and also test if the socket pins are OK but they should be IMHO... the problem now is probably the IC is dead or a connection to it is faulty.

[m00dawg]

Sounds like it's a connection is my bet. I played swap the IC chips a few times already and re-ran the grounding test with the same results, so I don't think it's a bad IC (or doesn't seem to be). I believe I have some extra ICs I can try if I just want to eliminate the whole lot.

I did check the socket but I didn't check the chaining or +5V all that much. I'll do that again and see what comes of it. I DID find a cold solder joint for one of the resistors on the 2nd IC though so I fixed that.

[Wilba]

Sorry I didn't read all the details of your last post and suggested stuff you have done already.

When testing SRIO... did you test high (5V) outputs on the last 74HC595? i.e. DOUT outputs 16-23?

Try the LED test again with 5V going into the resistor, as I suggested... i.e. simulate a 5V output from the 74HC595. That will test the transistor (but then if the SRIO test on that 74HC595 passes, it's sort of testing the same thing). SOMETHING is wrong so test everything and work out what's wrong later (don't skip a test and assume anything!)

Working encoders just means the first four 74HC165 are working. You need to test if the switches work, in order to test if the current sinking is working as expected.

The good news is, you installed all the LEDs the right way ;)

[m00dawg]

Looks like the transistors are working. I removed the chip and supplied 5V as suggested and it lit up the lines on the matrix. I ran through testing the IC socket and that, too, appears to be working. I tested continuity across the chain from the first to the last IC and it seems to be working correctly. The other pins are working as normal too it looks like (I'm getting +5 and GND where I should be, things like that).

If the transistors and ICs work, though, it seems like it has to be something in the communication between ICs? I'll take another close look at the bottom of the board too just in case there is a short anywhere around the socket. I checked the transistors, resistors and cabling and didn't see one. I can try re-running some tests too just to make sure.

As for the buttons, I assume you mean current sinking specific to DIN? The buttons do not relate in any way to the LEDs electrically do they?

So close yet so far!

Thanks again for all the help Wilba! Gave me some food for thought so we'll see what happens!

[Wilba]

If you ran the SRIO tester and got the right outputs on the last 74HC595.... then that indicates no problem with communication between that 595 and the previous ones, the whole chain is "good" and the PIC is outputting correctly.... which means I'm confused how that can work and the transistor sinking works yet the matrix as a whole doesn't.

The problem now seems to be that IC is not really outputting into the transistors, yet the other two are... you could try resoldering all the joints on that socket and see what happens ;)

There's one track that connects each 595 with the previous one on the chain. See DOUT schematic or MB-6582 PCB PDF (I forget which pins it is). Check that for continuity BETWEEN THE IC PINS not the pads.

I need more thinking time... ;) bbl

[m00dawg]

Ah good point. Yes I did get the right outputs, but one thing I didn't test for was if I was getting the wrong signals on the test, only the right ones. So I am likely going to rerun it and check all pins at each sequence of the test to make sure they are all what they should be.

When you say check for continuity between the pins can you elaborate a bit? Having trouble producing a clear picture there. You mean the soldered pins on the bottom of the board?

Re-soldering is a good idea. I did a few reflows on the resistors but didn't do anything on the socket so I'll go back and check again.

Thanks for being so helpful and patient Wilba!

[Wilba]

Referring to the PCB PDF... http://www.mb6582.org/plans/MB-6582_Base_PCB_R2_Color.pdf

U22 pin 9 connects to U23 pin 14... i.e. as you're shifting bits into the shift registers, they shift along across this track to the next shift register in the chain, in the direction U21->U22->U23... so the pin 9 of one connects to the pin 14 of the next.

now you COULD do something really wacky to test stuff...

take out the IC in U22 (the 2nd one)

apply 5V to U23 pin 14 (with IC in!)... i.e. you're making this shift register always output high values.

test with matrix text... you should see only the mod matrix LEDs light up

try applying 5V to U22 pin 9 instead... which connects to U23 pin 14. Should get same results. If not, broken track or bad joint.

BTW be careful when testing IC socket pins from the top side, don't push the probe in, this can widen the contacts.

[m00dawg]

Ah yes that's a very good idea! I think I'm going to reflow the solder on the sockets to be sure as well. If that doesn't work, I'll do this as the next step.

[m00dawg]

Ok so I ran +5V to pin 14 of U23 (with chip in) and pin 9 of U22. Both cause the matrix to light up as advertised...well, the 7th column didn't light up so that's an issue but I think I should worry about that later.

So it looks like the shift register stuff is working. I'm going to go back and try the SRIO test again to see if I missed anything but other than that I'm still kinda at a loss...

[m00dawg]

Ah ok here's something interesting (maybe?). When I run the Serial Interconnection Test the data pints of U23 show up as -0.40V to ground whereas the data pins on U22 and U21 show up as 0 volts. I also noticed that the voltage from the SC and RC pins is slightly lower (4.94V) than testing the +5V on one of the CORE sockets (4.98V).

[m00dawg]

AHA! I found the problem. It was that the capacitor for U21 was touching the track that went to pin 14 for U22 via a bad solder point. Fixed that and now the board lights up! I do have a few LEDs that aren't lighting up that I need to investigate but at least the damn thing lights up now, w00t!

Thanks again for all your help Wilba!

[m00dawg]

Hmm...well I fixed the biggest problem, but now I have a column (column 8 on the matrix) that won't illuminate. I've pretty much ruled out everything but the transistor. I get continuity from the IC to the resistor, from the resistor to transistor. The transistor also appears to be properly grounded and the output to the CS works as well. If I tie the LEDs to ground, they light up.

This column, I thought, was working yesterday so my thought is that something caused the transistor to go bad. Is there any way I can test this without unsoldering the thing and getting a new one (which unfortunately I do not have and will have to order - damn :/)?

[Wilba]

Try and identify if the fault is with the transistor not sinking, or a break after the transistor (i.e. through JD8 pin, through the cable, in tracks on the CS PCB).

So... LED matrix test again, put in U21 and U22, take out U23, apply ground to the collector pin of the transistor (the one at the bottom, which connects directly to JD8). One "column" of LEDs in the mod matrix AND other LEDs should light up. If column 8 lights up, then it's not a fault from JD8 onwards. Could be the transistor. Try 5V at the resistor/U23 junction (i.e. drive the transistor to sink). If that lights up, it's not the transistor, possibly a break/short between that point and the IC socket.

BC547 transistors are perhaps the most common transistor on the planet... except for Americans :)

If desperate for a quick fix, use 2N3904 (like from your local electronics shop  here ;) ) but you need to invert the orientation because the pinout is different to the BC547. That will at least get you going while you wait for a BC547.

[m00dawg]

Hahah mad props on the local electronics shop search :) I have been wanting to go there for quite some time. Trouble is, they aren't open on weekends so I've had a hard time getting there. If it is the transistor though, I can get it from Mouser or Allied Electronics within just a few days so it's not a huge deal.

I tried some of the tests you described but didn't try to force the sink (should've thought of that :/) so I'll give that a go.

[m00dawg]

Well I was able to get things fixed. As best as I can tell it was a cold solder joint somewhere - it didn't seem to be in the cable though (I found quite a few bad solder joints for the cable connections, but this wasn't one of them). Nonetheless, I reflowed some solder and now it seems to be in good shape!

You are a saint, Wilba!

[m00dawg]

Aha! Let this be a word of warning to others! Turns out, the method I was using to test the connectivity of the cables didn't account for the orientation of the control surface. When I was testing it, the boards were folded, more or less, together. But when I actually supply power to the board, I stand the control surface up so it doesn't short out anything.

In that orientation, there were  shorts both with the 8th column and a few of the encoders (which explains why I was having issues in another post). Trouble is, I'm not sure how best to test things easily when in this mode but at least I know what the problem is.

So, for anyone else building an MB-6582, be sure to take into account board positions when checking for continuity!

[Wilba]

Good to hear you got things working!