What are the most common solutions to my problems?


Can I use the Parallel Port Cable with my new TI-89?

zeotrope wrote:

The problem is, that it doesnt seem to be any link software for the 89 except texas graphlink, and I dont want to buy a graph link. I tried some 92 link software but the only thing i managed to do was a screendump and remote control.


What do you mean if you say rename *.85p and *.85s to *.85g, is it on the computer or on the calculator?

Use the DOS "ren" command to rename the files on your computer. Otherwise link85x/link85xp is not able to list the files, because of its fixed *.85g file mask.

link85xp and the cable was useable on a 386­16 in the past, but now on a Pentium­133 I get transmission error

I do not really know what the problem is, because I cannot reproduce this using my 486­66. My conjection is, that the i/o port read/write­delay is processor clock dependent. This is not the first time I hear from problems with fast computers. Another possibility is the change of the parallel port hardware. Most of the new computers have three modes for the LPT. I don't know wether this has any influence. I tested my cable with a 286­12 and Hercules Monochrome Graphics Card LPT, as well with my 486­66 PCI and on­board LPT (all three modes are working).

Try a lower clock rate. Switch off the mem­caching. (Just to see wether this is the problem.)

from Steve Peterson: I found that the link will NOT work if in a windows environment. Both version 3.1x and 95. I have to actually "Restart in MS-DOS Mode" to use the link.

BTW: I'm using a Pentium 100 now and I was using a DX2-80. Both work outside of windows

from James Reedy: . . . In the end I fixed the problem on accident, it turns out that your link will work only when the turbo switch is in the off position.

from David Ingham: I've fixed my problem and I think it was due to the fast computer problem. I didn't buy a new I/O card, I just made a boot disk with the old dos 6.22 command.com there by kidding the computer into thinking it was a 16 bit machine NOT a 32 bit, since most pentiums etc come with WIN 95 (32-bit) it might be worth trying makeing a 16 bit boot disk, Just find a 16 bit machine, Format a disk to a system disk (i.e. format a: when thats done, type sys a: ) then make a autoexec.bat and config.sys WITH THE BEAR MINIMUM settings, just enough to get you in dos and run the link85xp program it's worked with me but i don't know if this will be the general case.

In some cases new batteries didn't work, but older batteries work. I use accus with 1.2 volts each (instead of new batteries). It sounds unlikely that new batteries will work not as good as older, but I got this feedback several times.


The link85.exe software, and the software for win from ti is made for serial com ports. The schematic on how to build the cable is for a parellel version. How is this supposed to work?

The problem interfacing the ti­calculator and the serial port is the different voltages. The calculator uses 5 volts and the com port (RS232) uses 12 volts. Therefor a schematic for a serial port interface includes some parts to change the voltage (Z­diodes in the simpliest case, or a MAX232). Why should I use these parts? On my computer is a 5 volts port as well. Unfortunately it is not the serial port. So I have to use the parallel port for the serial i/o. It is only a question of the software. Because I am too lazy to write a new program, I downloaded the freely available link85x.exe from ti and replaced all occurences of serial i/o addresses by lpt i/o addresses using the dos command debug. The result I named link85xp.exe (p for parallel port). link85xp.exe drives the lpt port instead of the serial port.

I'm sorry to bother you but I know absolutely nothing about electronics and I want to build a ti­85 link. I would greatly appreciate it if you could send me info regarding what exactly I need to build it as well as a step by step procedure on how to put all this stuff together. Thank you sooooo much for your time, I really appreciate it.

I have a web page which contains step by step instructions for building your link. This may be of use to people who want to build the link who have had little previous experience of electronics. It also has hints for people whose links do not work.

Nicholas Metcalfe Try the Link Cable Page.


What happens is when I plug the cable into the computer and the calculator, the calculator (TI­85) just freezes up (whether I have the software loaded or not). I have tried numerous software programs with no success. (I even tried the one on your page). I used some extra diodes I found in a spare parts box. When the cable didn't work, I tried using some 5.1v Zener diodes (which was recommended by the guy who built the "$4 TI cable" to use one yours. It still didn't work. I took the whole thing apart and then built it again, making sure there were no shorts. Still doesn't work. I have tried plugging in the cable before, after, and during loading the software programs. I have also tried plugging in the cable before, after, and during initiating a transfer. It ALWAYS freezes up as soon as I plug it in. When I unplug the cable, it proceeds to work flawlessly.

The calculator freezes if you short­circuit its interface, or if you switch it to ground (0 volts).
  1. Try the plug alone (without any wires or cables). It should not freeze. Otherwise you cannot use this plug. Or grind the socket down. Look at the original plug what could be done.
  2. You have to start the link program first. If the lpt is not initialized by the link85xp.exe, it may initialized that you switch the calculator to ground.
  3. Check the direction of the diodes (look at the schematic).
  4. Check the diodes, perhaps they are gone.
Don't use Z­diodes, it is unnecessary (they are more expensive than normal diodes).

I am willing to pay $?? if you could build me the TI­85 cable.

Try this link: List of Link Makers

May I sell parallel links at school? (ozone@clinet.fi)

You may. Please do not sell the idea, but only the hardware, i. e. tell everybody, who is asking you for, that he can get the info on how to build this cable for free.

Where can I get the plug, or a socket (preferable), to make the parallel cable for my TI­85?

This is a question to you all. My isolating tube (see my web page) is certainly not the safest way to connect the plug. Makeshifts have a long life.

I got the diodes and stuck all the wires in the port. I ran the test program and all the should be values were correct. but the lc85 and link85xp programs did not seem to work. I am sure that all the connections are correct and can't seem to figure out the problem. I used tiny alligator clips to clip onto the cable that came with the TI­85. I checked the cable and it worked fine.

I think either this is a problem of the plug (the TI plug is different from a 2.5 mm stereo plug), or your parallel port is not compatible (try to use different computers/parallel ports).

I used your test program on it and it said this:
LPT1 at 0278
send 00 ­ read 20  should be 00
send 01 ­ read 20  should be 20
send 02 ­ read 30  should be 10
send 03 ­ read 30  should be 30
The values you measure with the computer have, in general, nothing to do with the calculator. It is only a loop back test. These values tell you that one diode is working, the other one is not. This could be a problem of the diode, or it is a loose connection. (Or your LPT is not working correctly.) The diodes are used to drag down the voltage of the anode by switch the cathode to 0 volts. You may check using debug and a voltmeter. In the following example your port is at (hex) 278. The calculator is not attached to the cable. The cable is plugged to the LPT at (hex) 278.
C:\>debug
-o 278 0	; both lines low
-o 278 1	; only one line is low
-o 278 2	; only the other line is low
-o 278 3	; both lines high
-q
C:\>

I use 1kV, 2.5A silicon diodes. Could they not be the right ones?

It is better to use Si­diodes in the range from 50 . . . 150 (400?) volts. Diodes (diode devices) with very high voltages are often realized by an internal chain of several diodes with lower voltages. The problem is, that such a chain of diodes leaves more than 0.7 volts from anode to cathode (0.7 volts is the forward voltage for a single silicon diode), which may cause difficulties. We need a low forward voltage.

Can I use Zener diodes, 5.1 volts

It is better not to use Z­diodes. You should prefer fast switching diodes. But Z-diodes with 5.1 volts or higher should work, I think.

I just made the graph link as you specified. I have a Ti­82. I was wondering if there were any software for the graph link with the Ti­82. The only software that I could find did not work.

Magnus Hagander (mha@algonet.se) told me, that he will implement the parallel port driver into his CONNECT software. I don't know wether it is finished.

All the test programs you can find on my web page have source code included. It should be possible to get the screen dump program working if you change all "0x05" to "0x02" in the C source code. There is also a link program for the TI­92 on my web page. If you want to write a link program this is perhaps the best start. Ask the authors for the source code.

I've tried both "LINK85XP 0x378" and just "LINK85XP" (my lpt 1 is 0x378 and is the 'correct' port) and still i get Transmission Error when trying to send to the 85 while the 85 just sits there saying "Waiting . . . " and when trying to send to the computer, the 85 immidiatly says, "done" but the computer says "waiting . . . " as if the 85 didn't do anything . . .

1. Try the test software you can find on my web page. 2. It is not necessary to pass command line args to link85xp. 3. Read about the "fast computer problem" above. Perhaps you can try another computer. 4. There is a setup menu in link85xp, select the right LPT there.

Do you have any instructions on how to make a computer link for a macintosh? Or where I can find the information?

Perhaps some readers of this can help.

I built your parallel port link cable for the TI­85 but it doesn't work completely. I have no problem capturing the display from the calculator, nor do I have any difficulties backing up all of the data on the calculator, but if I try to send a program I get a Link 35 transmission error from the calculator. The funny thing is when I try to send the program for instance . . . Tetris.85g . . . the calculator will say Transmission error Tetris PRGM. It would seem that if it can tell me the name of the program I'm trying to send, it should be able to store and execute it, why does it produce an error?

This sounds like a clobbered file. The are also so­called "edit locked" programs. These programs include a few hand­changed bytes. It is not possible to edit the code from within the TI­85, but you can execute the code. Perhaps there are newer, smarter ROM versions detecting this as an error. I think your link is "in Ordnung".

I can send data from the TI­85 to the Laptop with no problem, but when I try to send something from the laptop to the TI, it seems to work for a split second, then I get an error 35 and my memory gets cleared!

For both, sending and receiving, you need to have bits exchanged in both directions (hand­shaking protocol). The difference is only in the size of the packets. The sender transmits larger packets than the receiver. I only have conjectures. Perhaps the batteries of the calculator are weak and not able to hold the voltage necessary to transmit a large package. But I use 4 accus, each of it with 1.2 volts, and it is working.

You should try several parallel ports, there are sometimes tiny differences in voltage.


In the diagram of the pins 1–25 are you looking at them from the front or the back of the serial port?

It is the side of the parallel port plug you solder.

Is your link the only one that works with the original TI GraphLink software, it would seem that would be the case. And lastly does the TI link cable have all of those diodes and resistors inside of it, for some reason I doubt it, are your zeners used as a protective measure for the calculator?


I have both a TI­85 and a TI­92, but I can get the link to work only on the 92!

Does the TI­92 use a different plug?

I can understand how the diodes work on pins 2 and 3, but what do you do with pins 12, 13, and 22?

Pin 22 is ground (0 volts). Pins 2 and 3 are to set the lines either to high or low, pins 12 and 13 are to read in the level of the lines set by the computer (this is used for the loop back test linktest.com), or set by the calculator. In two cases it was necessary to connect the pins 18 ... 25 all together (they are all specified to be ground, so it is only necessary with incompatible parallel ports).

. . . should I do more than simply sticking the ends in the port?

Please use a parallel port plug and solder the connections.

I have a TI­83 (uses the same programs as TI­82). Does the program work for my calculator?

See the note on the TI­82. It should be possible to get the screen dump program working if you change all "0x05" to "0x03" in the C source code.

I'd like to ask you if you will consider releasing to me the source code necessary to communicate to a TI­85 using your "$5 Cable".

All source code I wrote is included in linktest.zip (see the web page). There is also a link92xp with source code available by the authors.

How do I convert 92p, 92g to 92@?

Good question, unfortunately I only have a TI­85. Ask the authors of link92xp for the source code. So you have a description of the 92@ format.

What diodes did you use?

As they came out of my rejects box: one is SAY 17, the other one is SAY 30 (old silicon diodes made in GDR).
			SAY 30	SAY 17
forward current		50 mA	175 mA
reverse voltage		25 V	50 V
total power loss	150 mW	300 mW
total capacity		< 8 pF	< 3 pF
			65 ns	2 ns
Note: There is no special reason to use different diodes. Just use the cheapest switching diodes. They should be good enough.

Here some successfully used diode types: 1N914 glass switching diodes, 1N4148, . . .


Does Connect­85 now work with your cable­design?

Yes, it does, and it works just great. (This info is from Oskari Ojala, okko@mtv3.fi)

My problem is that when I use your test program, the readout for the third line (sends 10 out) comes back as 0 most of the time, but about 1/5 of the time will return the correct number 10.

This sounds like a loose connection. Use the solder iron again. Look at the 2.5 mm plug, is it an original TI? If you would like to be sure wether you can use particular stereo plug, try to build a TI-to-TI link cable with two such plugs. If it works, you can use this type of plug.

How are the data transmitted?

One line is ground. The other two lines are two i/o lines, I will call them A and B. Initial state: both lines high (1). If you want to know it exactly, please have a look at the source code of the test software.

To send a 0:
sender: switches line A to low (0).
receiver (detects the change on line A): switches line B to 0.
sender (detects the change on line B): switches line A back to 1.
receiver (detects the change on line A): switches line B back to 1.
sender waits for line B = 1
end

To send a 1:
sender: switches line B to 0.
receiver (detects the change on line B): switches line A to 0.
sender (detects the change on line A): switches line B back to 1.
receiver (detects the change on line B): switches line A back to 1.
sender waits for line A = 1
end

Andreas Westfeld, Tue Feb 16 13:19:43 MET 1999