PDA

View Full Version : Got my first 595 finished...



random_rodder
08-25-2007, 05:04 PM
Well, at least to where I need it too be. The second is on it's way to be done.

When done, they'll be going into an old 48 port ethernet switch case I have sitting around. This is why there are no RJ-45 jacks on the sides, I'll be hard-wiring the outputs on the 595's, and those will go to RJ-45's on the case. Should be enough room in that for one of my Grinch boards also.

Hopefully more later.


Brian

Ronp
08-25-2007, 05:10 PM
hmm That might work....
http://www.putfile.com/pic.php?img=4302634
http://media.putfile.com/595

random_rodder
08-25-2007, 05:24 PM
Ron, I like how yours turned out. Mine is a little different...

The ports on this exit out of the front and rear. Not entirely sure I like this, but, it's what I've got to work with right now.

I'll be trimming the jacks from the board this week. There's enough room under the PCB to run the CAT-5 from the 595's to the RJ-45's. Gonna look kinda rough I imagine, but, at least I'll be up and running...

Brian

Jeff Millard
08-25-2007, 05:39 PM
That board and case was a nice find Brian. That'll save you a pretty penny on the RJ45 connectors!
jeff

cgarwood
08-25-2007, 11:28 PM
I did something very similar, except i used a dead 24 port cisco switch. Unfortunately i do not have any pictures currently. Will try to get some this winter.

teberle
08-26-2007, 09:46 PM
I have completed all 4 of my 595's this weekend. This is a picture of 2 of them connected in series with all LED's working. Yea!!!!

http://members.cox.net/eberlechristmas/595/Picture%20001.jpg

Macrosill
08-26-2007, 10:17 PM
Nice work T!!!

random_rodder
08-27-2007, 08:41 AM
Very nice!!!

Quick question, guys... I got my second board finished last night. Well, as finished as it can be as I am 2 595's short due to what appears to be a bad batch of chips. I have 6 of the needed 8 for the second board. When connected individually, both boards seem to function right. When daisy chained, they do not.

I set Vixen up for 96 channels, as I didn't want to mess with the partial bank of 16 channels until I get new 595's. When I ran my test sequence in Vixen (basically a chasing sequence) when the LED's on the first board started to fire, the unused bank on board #2 also went. When it came time for the first LED's on the second board, they did complete the sequence. Can having an incomplete board cause problems like this, even if Vixen is set to use a lesser amount of channels? Or could it be the cable between the two units was too long? (only had a 20 foot cable, gonna make a much shorter one today...)

Any thoughts?


Thanks,

Brian

Macrosill
08-27-2007, 09:10 AM
RR,
you say you have some bad chips. Did you have these chips installed in the last 2 positions of the 595? If so the behavior you saw is normal. Any un-programmed channels will repeat what the first channels do. Vixen only sends out 96 channels worth of data so channel 97 just does what channel 1 did and so on. Try taking those chios out of the board, those leds should not light.

Since they worked properly individually and the 96 programmed channels worked properly it appears everything is in order.

random_rodder
08-27-2007, 09:15 AM
Thanks Brian!!!

I kinda suspected that to be the case. Yeah, I originally thought I had seven bad chips out of twenty. Now it appears I didn't have two seated properly. I still have 5 bad 595 chips that will NOT work under any circumstance. Hopefully Jameco will replace them...

Now to start in the SSR's... ;)


Later,

Brian

random_rodder
09-10-2007, 11:47 AM
Well, dammit, now I'm not sure about anything...

I'm in the middle of my first sequence for my display. The 595's were performing as expected. About midway thru the sequence, I notice that a couple of LED's are not lighting as expected and during a chase sequence on board #2 (2 channels chase at a time), sometimes I have three channels other times I have one channel doing the chase.

Next thing I did was use the Channel Test tool. All channels turn on and off with those buttons. Next I tried all channels individually. Starting at channel 1 to 128, they all worked as expected. Starting at channel 128 to 1, they did not. Channel 112 would not do anything. Channel 111 turned both 111 and 112 on. Channel 80 would sometimes work and sometimes not. When it came on, channel 112 would turn off.

So, I swap the boards around and have the same exact problem.

Next thing I did was create a test pattern with a knight-rider like sequence, going from the edges to the center of the boards and chasing sequences. I created one 128 channel version and one 64 channel version. The 128 channel test has the timing and number of leds incorrect on both boards. At any point on the test, the LED's for channels 1-32 and 33-64 are not synchronous (they should be according to the pattern in Vixen) during the knight-rider portion. The chasing potion is also off. Channels 1-32 and 65-96 are the same and 33-64 and 97-128 are the same.

Using the 64 channel test, BOTH boards operate flawlessly.

I have checked and rechecked the solder joints on the boards and have examined all of the traces. I have also swapped the positions of the 595 IC's and still get no difference. Does this sound like bad 595 IC's? Or another, BIGGER problem?

I'm kinda at a loss right now. I've got 5 arches put up and am planning on starting my mega tree and mini's next weekend. Don't really want to go there until I'm sure of what this problem is.


Brian

stempile
09-10-2007, 12:00 PM
Have you tried driving the 128 pattern from another computer? Maybe you parallel port isn't fast enough? or the parallel port isn't set in the bios with the optimum setting.

What is the time interval set at? Vixen defaults to 100ms. If you are set lower then that try setting it back to 100 or higher. That might identify a slow parallel port.

I wasted a bunch of time troubleshooting my hardware when it turns out the computer driving it was a to working correctly.


ms

random_rodder
09-10-2007, 12:29 PM
I thought I stated that I had a 128 channel test that worked fine when I first put the 595's together but now does not. Guess I didn't, sorry.

I can try this (another computer) when I get home tonight. I can't imagine the // port being too slow is the problem tho. I'm running on an HP dx5150 series with an Athlon64 3200+. And Vixen is really the only thing running on this machine.

And yes, I am running at 50ms intervals. I'll create two more test sequences for 64 and 128 channels running @ 100ms and see what that does.

Even if this is the case, would a slow // port cause the issues with channels 80 and 111 & 112?

Thanks,
Brian

ErnieHorning
09-10-2007, 12:46 PM
I think Matt’s right. It’s most likely that it’s your parallel port since you’ve swapped the order of the boards and the problem is the same. It’s the speed of the parallel port hardware, not the speed of the processor. The problem is that there was never a documented standard for the parallel port in the way that we use it. It was intended for use as a printer output only. Printers regulate the port speed via a status bit, which we don’t use, and therefore the data is changing at a slower rate. In our case, the data is blasted out as fast as the software runs. Some ports have a bit too much capacitance at very high speeds and tend to round off the normally square signals.

Note: it could also be your parallel cable.

I don’t think that changing the sequence timing will have much effect though, since Vixen will still send the pulses out as fast as it can, even though the bursts will be spread out more.

random_rodder
09-10-2007, 12:56 PM
Yeah, I didn't think about the// port hardware itself...
I have a few other PC's to try it out with.

But the question I still would have is why did it work well before and not now? I cannot find anything that has changed...

The cable I'm using is the 25pin connector from an old 'Check-It' dongle that I removed the jumper wires from. I soldered a CAT5 cable to pins 1,2, 14 and whatever pin I chose for ground, and used an RJ45 at the other end.

Thanks for the thoughts guys... need all the ideas I can get right now...


Brian

ErnieHorning
09-10-2007, 01:22 PM
But the question I still would have is why did it work well before and not now? I cannot find anything that has changed...


You could be on the edge of working and not working. There are a lot of threads over on Computer Christmas on this subject. Some times you can modify the input of your board to make it work. Some people got it to work by adding a pull-up resistor on each of the three pins, lowering the supply voltage a bit or buffering the input with a ULN-2808 IC. Phil even had an oscilloscope picture of a port that he never did get working.

BTW, another way would be to eliminate the parallel port all together by adding the Renard to 595 adaptors. Then you just connect it to one of your serial ports or a USB to serial dongle. You then also have the advantage of dimming all of your channels if you want to.

P. Short
09-10-2007, 01:38 PM
Depending on the length of the cables involved (and other factors as well), the clock waveform can get to looking pretty lousy at the far end of the universe. And since the data signal between the two controllers changes state on the rising edge, it can end up being mis-clocked. This can be exacerbated if the signals coming out of the parallel port have signal-level issues or other issues to begin with.

This problem is less severe with the Grinch board, since the data changes state on the negative clock edge.

Another possibility is to use the DMX-595 board that RJ was talking about, although I don't know if there was ever sufficient interest for him to document and release it.

--

Phil

random_rodder
09-10-2007, 02:08 PM
You could be on the edge of working and not working. There are a lot of threads over on Computer Christmas on this subject. Some times you can modify the input of your board to make it work. Some people got it to work by adding a pull-up resistor on each of the three pins, lowering the supply voltage a bit or buffering the input with a ULN-2808 IC. Phil even had an oscilloscope picture of a port that he never did get working.

I wondered about this when I first started testing. The // port on this PC I think is at 3.38v at logic high.


BTW, another way would be to eliminate the parallel port all together by adding the Renard to 595 adaptors. Then you just connect it to one of your serial ports or a USB to serial dongle. You then also have the advantage of dimming all of your channels if you want to.

I thought about this as well. In fact it's a direction I really want to go. However, not sure I can afford it this year. I'll need one Ren-C per 595 and Grinch (I have two each, but, the Grinch boards are not built yet) and a Ren-T for all four Ren based boards, right?

Thanks,
Brian

random_rodder
09-10-2007, 02:11 PM
Depending on the length of the cables involved (and other factors as well), the clock waveform can get to looking pretty lousy at the far end of the universe. And since the data signal between the two controllers changes state on the rising edge, it can end up being mis-clocked. This can be exacerbated if the signals coming out of the parallel port have signal-level issues or other issues to begin with.

The cable from the PC is about 3' long. The cable between the two 595 boards is and 18" patch cable.


This problem is less severe with the Grinch board, since the data changes state on the negative clock edge.

Guess I built the wrong controllers first...


Another possibility is to use the DMX-595 board that RJ was talking about, although I don't know if there was ever sufficient interest for him to document and release it.

Maybe when I have a better understanding of all of this... ;)


Thanks,

Brian

ErnieHorning
09-10-2007, 02:28 PM
I'll need one Ren-C per 595 and Grinch (I have two each, but, the Grinch boards are not built yet) and a Ren-T for all four Ren based boards, right?

Not necessarily. The Ren-T has the same spec as a standard Renard. It depends on the serial baud rate.

Check out the ‘Number of Circuits’ section in the Wiki.
http://doityourselfchristmas.com/wiki/index.php?title=Renard

random_rodder
09-10-2007, 03:06 PM
I was going on what I had read either on the wiki or in a thread about the Ren-C/Ren-T boards. I think I remember reading somewhere that the Ren-T could supply the ZC signal for quite a high number of Ren-C boards. Or maybe I misread...

Looking at that chart and other threads, it looks like I'll need two Ren-T's to run my 2 595's and 2 Grinch's with 4 Ren-C's


Brian

Dan Ross
09-10-2007, 05:17 PM
What setting is your parallel port set to? I had a similar problem when I fired mine up, and I had to change mine to "EPP" and it worked fine after that.

random_rodder
09-10-2007, 08:29 PM
What setting is your parallel port set to? I had a similar problem when I fired mine up, and I had to change mine to "EPP" and it worked fine after that.

I tried all of the LPT settings on this PC and the only one that provided results was the SPP setting. Standard, EPP and ECP all would light up the board, but, in a one channel chase sequence, 5 or 6 channels would light up and chase.

random_rodder
09-10-2007, 08:39 PM
Okay, I guess the plot thickens a little here. I haven't had time yet to switch PC's, but, I've worked a little more with the current PC and 595 boards.

With both boards daisy chained, the system reacts like I described earlier. If I disconnect the cable from the first 595 board, everything works like it is supposed to. If I leave the cable connected to board 1 and disconnect board 2, the timing is off on board 1. I can pause the sequence, disconnect the cable from board 1 and continue the sequence and the sequnce's timing is corrected. If I swap the board positions, the same thing happens.

I can't get to swapping the PC's tonight, hopefully tomorrow...


Brian

Dan Ross
09-10-2007, 08:48 PM
What are you using to power the 595's? If you are using an external power supply, do you have the ground of the power supply tied to the ground from the PC?

random_rodder
09-10-2007, 08:53 PM
Using the PC's power supply.

Dan Ross
09-10-2007, 09:01 PM
This is a long shot, but do you have anything else programmed to use that port, like a printer or something?

I can't remember if you said already but what plugin are you using?

random_rodder
09-10-2007, 09:05 PM
The only other software installed on this PC is TMPGEnc. No printers or anything else.

I'm using the Olsen 595 plugin...



Brian

Dan Ross
09-10-2007, 09:14 PM
If I disconnect the cable from the first 595 board, everything works like it is supposed to. If I leave the cable connected to board 1 and disconnect board 2, the timing is off on board 1. I can pause the sequence, disconnect the cable from board 1 and continue the sequence and the sequnce's timing is corrected. If I swap the board positions, the same thing happens.


You have me puzzled with this, What cable did you disconnect from the first 595 that made everything work?

You should go from the PC to the IN on board #1 then from the OUT on board #1 to the IN on board #2. Do you have something different?

Dan Ross
09-10-2007, 09:16 PM
Also this may be a dumb question, but the power you are getting from the PC, are you getting it directly from the power supply?

random_rodder
09-10-2007, 09:19 PM
I was referring to the cable that connects the two boards.

I am getting power from a molex connector right off of the ATX PS.

Dan Ross
09-10-2007, 09:25 PM
Is there anything else daisy chained with the plug you are getting you power from?

Have you checked the input voltage to make sure you are getting a constant solid 5v from the PC while you run a sequence?

Are you sure you are on the +5v wires and not the -5v?

random_rodder
09-10-2007, 09:36 PM
Is there anything else daisy chained with the plug you are getting you power from?

Nothing.


Have you checked the input voltage to make sure you are getting a constant solid 5v from the PC while you run a sequence?

Yep. 5.05vdc.


Are you sure you are on the +5v wires and not the -5v?

Coming off of a 4 ping Molex connector used for HDD/CD-ROM/DVD-ROM type connector. Has +5vdc, GND, GND, and +12vdc (red, blk, blk, yellow). I'm taking the connection from the red wire...

Dan Ross
09-10-2007, 09:38 PM
Are you bringing the power in via the RJ45 or are you using the external power input?

random_rodder
09-10-2007, 09:45 PM
My cable consists of:

Solder cup style DB25 with CAT-5 cable connected to pins 1, 2, 14 and one of the ground pins. A fifth wire from the CAT-5 goes to a Molex compatible pin which plugs into the 5vdc pin inside the PC case. The other end is an RJ-45 jack.

So in short, yeah, I'm bringing power in thru the RJ-45 jack...


Brian

Wayne J
09-10-2007, 09:56 PM
My cable consists of:

Solder cup style DB25 with CAT-5 cable connected to pins 1, 2, 14 and one of the ground pins. A fifth wire from the CAT-5 goes to a Molex compatible pin which plugs into the 5vdc pin inside the PC case. The other end is an RJ-45 jack.

So in short, yeah, I'm bringing power in thru the RJ-45 jack...


Brian

You need the ground from the molex of the ATX also.

Dan Ross
09-10-2007, 09:57 PM
This is my last thought other than trying a different PC, do you have the +5v going to pin 1 of the RJ45 jack?

random_rodder
09-10-2007, 10:01 PM
Yes...

WayneJ just told me in chat that I need to connect the Molex ground wire to either the DB25 ground or straight to the 595 board...

Gonna give that a try...

Dan Ross
09-10-2007, 10:07 PM
You need the ground from the molex of the ATX also.

I was thinking that also but in the WIKI it says if you use the PC power supply you don't need to.

Wayne J
09-10-2007, 10:08 PM
You need the ground from the molex of the ATX also.

I was thinking that also but in the WIKI it says if you use the PC power supply you don't need to.

Never hurts to make sure you have a good ground when troubleshooting.

P. Short
09-10-2007, 10:09 PM
Dan,

Where does it say that? It's not good practice, may or may not work.

--

Phil

Dan Ross
09-10-2007, 10:09 PM
Never hurts to make sure you have a good ground when troubleshooting.

words to live by...

Dan Ross
09-10-2007, 10:11 PM
Where does it say that? It's not good practice, may or may not work.

--

Phil

http://doityourselfchristmas.com/wiki/index.php?title=Olsen_595

step 5 second paragraph, unless I'm reading it wrong.

P. Short
09-10-2007, 10:17 PM
OK, I changed the wiki.

--

Phil

random_rodder
09-10-2007, 10:25 PM
I added the ground and I still have the same issue :( (never hurts to hope...)

RavingLunatic
09-10-2007, 10:39 PM
It appears that you are having the parallel port problems that have plagued many others so far.

You appear to be using a strong 5.05 for your VCC but your data in is only at 3.38 as a HIGH. Looking at the data sheet I have for the 595 chip you are close to the edge of the minimum for the chip to see as a HIGH with at that VCC. And when you connect the interconnect cable for the next board you are loading the CLOCK line down to a point that it probably isn't being seen as a high all of the time when it should.

What others have done in the past to correct this is to lower the VCC. In you case probably lower it to around 3.5-4.0.

OR

You could use pull-up resistors to increase your data lines ( I believe Dave F prefers that method when used with an inverter).

P. Short
09-10-2007, 10:54 PM
Dave F was using two Schmitt-trigger 'inverters' to form a Schmitt-trigger buffer... which I think is the key element to his method.

--

Phil

random_rodder
09-11-2007, 08:39 AM
Okay, I took a few voltage measurements this morning on the voltages present on the input and output of the first 595 controller.

All measurements taken directly from the RJ-45 jack with a DMM (the O'scope is at work).

Input connector/all off mode:
Pin 1: 5vdc even (down from 5.05 when I first put these together in late August)
Pin 8: 0vdc

Input connector/all on mode:
Pin 1: 4.81vdc
Pin 8: 3.25 (down from 3.38vdc...)

Ouput connector/all off mode:
Pin 1: 5vdc
Pin 8: -0.06vdc

Output connector/all on mode:
Pin 1: 4.79vdc
Pin 8: 4.78vdc
(I'm assuming these are normal for this connector since the output level is Vcc)

I'm not entirely sure I'm reading the data right on the 595's data sheet, but, here's what I see:

------------- Vcc--------------Guaranteed Limit

VIH--------- 3.0 ----------------- 2.1
-------------- 4.5 ----------------- 3.15
-------------- 6.0 ----------------- 4.2

Does this mean that with the Vcc values shown, I need the input voltage (from the // port) at those values under the guaranteed limit in order to make the 595 work? Basically 70% of Vcc? If so, I'm at about 67.5%...

Sorry if it's a dumb question... just trying to get a handle on what's going on...


Brian

P. Short
09-11-2007, 10:41 AM
That means that if the power supply on the 595 chip is 4.5V, the data, clock and latch signals must go above 3.15V in order for the the chip to be guaranteed to see the signals as '1', (and below 1.35V to see them as '0'). I think this is what you were saying...

--

Phil

random_rodder
09-11-2007, 11:29 AM
That means that if the power supply on the 595 chip is 4.5V, the data, clock and latch signals must go above 3.15V in order for the the chip to be guaranteed to see the signals as '1', (and below 1.35V to see them as '0'). I think this is what you were saying...

--

Phil

Thanks, Phil, that's what I was asking.

So, my 5vdc drops to 4.81vdc when I switch all channels on due to the load of the 595 controllers? Doing the math, I would need a minimum of 3.37v from the // port to to be seen as a logic 1 @ 4.81v Vcc , and I'm not getting it. 5vdc Vcc is even worse needing 3.5vdc. Guess we've found the problem...

Thanks,

Brian

P. Short
09-11-2007, 01:08 PM
No, I don't think that's the problem. Lowering the power supply voltage will probably improve the situation, not make it worse (since a lot of parallel ports have trouble generating a sufficiently high output signal voltage). In fact, that was my last suggestion last night...try lowering the power supply voltage.

--

Phil

random_rodder
09-11-2007, 03:18 PM
That was what I meant. My system's // port not generating a sufficient amount of voltage for the 595 to work properly with the Vcc I'm supplying it with. I'm not reaching that 70% threshold (// port signal strength vs. Vcc level) of where the 595 is guaranteed to see a logic 1.

Maybe I'm not explaining my thoughts clearly... wouldn't be the first time...

I plan on trying another PC tonight. If it's // port is not any better than the current system, is there a 'How To...' somewhere on how best to accomplish this feat of either lowering the 5vdc output (not sure how to do that other than using another power supply) or how Dave made his work?

This is my first foray back into the world of electronics since 1995 and I'm a little behind the bell curve :wink:


Thanks,

Brian

P. Short
09-11-2007, 03:43 PM
One trick would be to use a diode (1N4001) between your power supply and the 595 board to lower the VCC by ~ .7V

--

Phil

random_rodder
09-11-2007, 07:09 PM
I might be mistaken , but, based on my last test I did a few minutes ago... I do not hink the voltage level was the issue.

I switched PC's and noted the voltage level on the // port for this one was 4.22v and 5vdc Vcc. The 64 channel test again looked great. the 128 channel test, again, did not... same problems. I even created a 100ms, 128 channel test just to be sure i wasn't trying to go too fast... no change, timing problems and wrong number of LED's lit on the first and second controller during the sequence... only it went slower (100ms instead of 50ms)

Is there any way possible it could be defective 595 chips? (Yeah, i know, grasping at straws...)

Wayne J
09-11-2007, 07:18 PM
Well, I would not think bad IC's. If both boards work fine alone, then I would be looking at the connection between the two.

random_rodder
09-11-2007, 07:29 PM
I have. The cables (both I have used) test good with a cable tester. The only thing I can do regarding that is make a shorter cable. What's confusing is that when just the cable is plugged into the 'OUT' jack, I still have the same problems I do if the second controller were actually hooked up as well...

klanger
09-11-2007, 07:45 PM
RR, if you are bringing power into the boards via the RJ45 cable, have you installed the shunt on J1 as per the instructions on the Wiki?

Board Options
There are a few options you need to consider.

Daisy Chain Power
If you are running 2 or more of these controllers you will have to decide if you want to “daisy chain” the power from board to board or if you will provide separate external supplies for each board. If you want to “daisy chain” the power through the “OUT” RJ45 connector you will need to install a shunt at J1.

Just a thought.

Kev

random_rodder
09-11-2007, 08:10 PM
Kev,

I do have a shunt on J1, per the instructions. And as of right now, I am bringing power thru the RJ-45... for now.

RavingLunatic
09-11-2007, 08:37 PM
Check the voltage of the signals on pin 5 (Strobe) & pin 7 (Clock). These signals are shared thru the board without any conditioning and could still be dropping too low with the extra load of the interconnect cable.

random_rodder
09-11-2007, 08:50 PM
what should I be seeing there?

Dan Ross
09-11-2007, 08:54 PM
Have you checked the pins of the out jack, maybe when you plug in the cable it's pushing one pin into another.

If you can't see anything wrong try pluging in both cables and check each pin to all the other pins with an ohm meter ( 1-2, 1-3, 1-4, 1-5, 1-6, 1-7, 1-8, 2-1, 2-3, 2-4, and so on till you end with 8-7)

random_rodder
09-11-2007, 09:00 PM
just tested again with 128channel doing a knight-rider type sequence... both controllers start from the center channel (16&17 and 48 & 49) and do a three channel chase out to the edges and back. Channels 1-32 work right, starting at 16&17, go out then back in and meet at 16&17. Channels 33-64 start at 48&49, go out, then back in and meet at 50&51, and there are 4 LED's lit on the 48-64 side of the board...

The second controller is off even further from the center and each bank has more LED's lit (this is progressive from chans 33-64, 65-96 and 97-128, each bank has more LED's lit than the previous, when the sequnce is supposed to have 3)...

RavingLunatic
09-11-2007, 09:02 PM
Thinking about it, it would be best to look at them on an oscilloscope. And of course you need to be running a sequence and not just the all on or all off function.

But if your DMM has a max hold function you should be able to see if you have enough of a signal to register as a high.

random_rodder
09-11-2007, 09:03 PM
Have you checked the pins of the out jack, maybe when you plug in the cable it's pushing one pin into another.

If you can't see anything wrong try pluging in both cables and check each pin to all the other pins with an ohm meter ( 1-2, 1-3, 1-4, 1-5, 1-6, 1-7, 1-8, 2-1, 2-3, 2-4, and so on till you end with 8-7)


Nope, no problems there...everything looks good...

Dan Ross
09-11-2007, 09:05 PM
Keep in mind if you test the RJ45 jacks with an ohm meter pins 1, 3, 5, and 7 are the only ones that should show a short from the in RJ45 to the out RJ45 everything else should show an open.

random_rodder
09-11-2007, 09:37 PM
Just for S&G, would the logic high be the full 4.2vdc? All I'm seeing with my DMM is mV readings. My scope is at work...

Okay, stupid question time;

Pins 1, 2 & 14 are used for data and such from the // port to the RJ-45 pins 5, 8 & 7. I'm using pin 23 for my ground, which should go to pin 3 on the RJ-45, right? And pin 1 on the RJ-45 is Vcc...

random_rodder
09-11-2007, 09:38 PM
Keep in mind if you test the RJ45 jacks with an ohm meter pins 1, 3, 5, and 7 are the only ones that should show a short from the in RJ45 to the out RJ45 everything else should show an open.

This is correct. Pins 8 on the 2 RJ-45's are not shorted.

Dan Ross
09-11-2007, 10:21 PM
here is the pin out looking at the bottom of the pcb with the open part of the jack on the left

random_rodder
09-12-2007, 08:10 AM
Thanks, Dan. That's how I'm set up.

I just made a 6" cable to go between the two controllers. Hopefully I can eliminate the 30" cable as a fault. If that doesn't help, I'm not quite sure where to go...


Thanks,
Brian

P. Short
09-12-2007, 09:58 AM
There are more steps after verifying that all of cables and shunts are OK, that the power and ground is OK, etc (the things that Dan, Raving, and Kevin mentioned). These next steps are more complicated, and address some design flaws in the Olsen scheme that are not talked about very much. These flaws involve how the parallel port is used, and the unsuitability of the 595 chips for driving cables.

First, try using pullup resistors and 74HC14 chip (or HCT14) to buffer the signals coming out of the parallel port, preferably as close to the port as possible. Some parallel ports do not have a sufficiently strong pullup resistor on pins 1 and 14, and some of them do not drive the signals on any of the pins high enough to reliably active the inputs of the HC595 chips (especially when the HC595 chips are powered with +5V). You can see a schematic of this change on David Fansler's site (at least, it was there the last that I looked).

Second, consider using the method that Robert Stark describes here (http://computerchristmas.com/christmas/link-how_to/HowToId-30/How_to_build_a_High_Speed_Differential_Line_Transm itterReceiver), which addresses a whole slew of issues beyond what Stark identifies (mostly distance related). This would be used between the PC and the first controller (possibly after the first change above), as well as in between controllers. You might want to use either a 26C31/26C32 pair or RS485 chips instead of the ones that he uses, and you should make sure that the outputs of each section go onto one twisted pair in the cable.

Third, instead of the above changes, get another parallel port for your PC to drive the second board.

Methods 1 and 3 are the most likely to address your immediate problem, while the second step is more likely to address problems that would show up with longer cables. And yes, I know that these are not the easiest things to do.

--

Phil

random_rodder
09-12-2007, 11:04 AM
Thanks for all the tips, Phil. I like the idea of the second // port. One question; if I bite the bullet and go with the Ren-C, will that erase the problem? (of course this is assuming the 595 boards are constructed right in the first place, which I believe they are).

The thing that bugs me most is the damn things were working fine for about a week and half...

Brian

P. Short
09-12-2007, 01:15 PM
Yes, it should...the Ren-C uses chips that are meant to be used with cables.

--

Phil

random_rodder
09-12-2007, 01:30 PM
Phil,

Looking over Dave's design, it looks pretty straight forward. A couple of 1.8k resistors, one side of each attached to Vcc, then the other end of one to the strobe and the other to the clock. These signal go to a 74HC14A for the buffering. Actually looks like the pull up resistors and Robert Stark's ideas could be incorporated fairly easily into one small perfboard or stripboard design... I may do this over the weekend...

I am still going to look at the signal thru the o'scope tho, that'll at least tell me whether the pull up resistors are necessary.

Thanks,

Brian

P. Short
09-12-2007, 02:35 PM
Brian,

I've got a small PCB designed for the differential drivers and receivers to go with the 595 or Grinch board, as attached. It shows how I think it should be designed, especially as far as the connections to the RJ45 serial connectors are converned. It doesn't have the HC14 circuit on it, which I think would be nice to add.

Also, I think that it would be a good idea as well to order HCT14 as well as the HC14, just to have something to fiddle with. The signals coming out the parallel port are spec'd to be TTL level rather than CMOS (at least the original specs), so the HCT might actually be a better fit than the HC.

--

Phil

random_rodder
09-13-2007, 05:05 PM
Brian,

I've got a small PCB designed for the differential drivers and receivers to go with the 595 or Grinch board, as attached. It shows how I think it should be designed, especially as far as the connections to the RJ45 serial connectors are converned. It doesn't have the HC14 circuit on it, which I think would be nice to add.
--

Phil

Phil,

Thanks, I'm not entirely sure how the connections would go with this board. Too me, it looks like the top right RJ-45 is the input from the PC, bottom right output to the 595/Grinch. The connectors on the left have me puzzled... are they for a CAT5 cable to loop the signals to the receiver?

I was playing a little today, just to see if I understood the data sheets and the schematics from David F., Robert Stark and yourself. Here's what I came up with...

U1 is a 74HC14 Schmitt Triger
U2 and 3 are the DS26C31 & 32 Line Driver and Receiver.

Am I anywhere in the ballpark for designing a combo board that has all of the discussed previously?

P. Short
09-13-2007, 06:43 PM
Top left is the output connection to the next board in the chain, top right is output from the 595 board. Bottom left is input from the previous board in the chain, bottom right was output to the 595 board.

--

Phil

Edit...the upper left connector is to have a long cable attached, going to the lower-left connector of a similar board (if there is a downstream board). The upper right connector should have the shortest possible cable, going to the output of a 595 board. The lower right connector should also be as short as possible.

Alternatively, the board that I designed could be split into two boards. The upper part would become the transmitter board, the lower one the receiver board.

--

Phil

random_rodder
09-13-2007, 07:58 PM
Okay.

Interesting thing, I tried the 6" cable I made yesterday, and the 595's work correctly with that. Switch back to the long cables and it goes apey again.

Next, I stuck one of those 1N4001 diodes inline of the Vcc wire, which lowered Vcc to 4.37vdc. With this mod, both 595's worked like a charm, regardless of what cable I used to daisy them with. I guess I now need to make that diode more permanent... ;)

Thanks for all the help guys!!!

Brian