PDA

View Full Version : IBM Model F Keyboard voltage issue



PCFreek
August 14th, 2014, 05:59 PM
I have an IBM keyboard model F keyboard that does not work. I have tried swapping cords with a known good board and have verified that the cord is OK.

I have taken voltages on the boards of a known good and the non-working board. Voltages at the caps and resistors is all good... however, voltages at the large chip (shown as 8048 in the Technical Reference) are quite different between the known good and non-working board. The known good board has no voltage on pins 27 through 38. The non-working keyboard has +5v at all of these pins.

I cannot find a datasheet on the 8048, and my chip is not an 8048. I do not know if this voltage can be an internal chip problem, or if the keyboard is energized when it should not be. Any suggestions?

EDIT: datasheet link below.

19866

PCFreek
August 14th, 2014, 06:43 PM
8048 datasheet here (http://home.mnet-online.de/al/mcs-48/mcs-48.pdf)

Unless i am reading this wrong, page 8 of the datasheet says that these pins (port 2) are supposed to be +5v:

"The lines of ports 1 and 2 are called quasi-bidirectional because of a special output circuit structure which allows each line to serve as an input and output or both even though outputs are statically latched. Figure 7 shows the circuit configuration in detail. Each line is continuously pulled to VCC through a resistive device of relatively high impedance."

PCFreek
August 15th, 2014, 05:55 AM
I thought more about this last night and assume that the ports can be pulled high or low. The datasheet is very long and I have not thoroughly digested it yet. I would assume that the good keyboard is pulled low (which must be correct since it works), and the bad keyboard is being pulled high. I will not be able to probe any further until Monday. If this isn't something that is controlled by a smaller chip on the board, then I would suspect the 8048 has a problem.

PCFreek
August 21st, 2014, 06:09 PM
If you have ever wondered what the inside of the keyboard looks like... and this keyboard is still not repaired. any suggestions?

20011200122001420015

modem7
August 21st, 2014, 10:13 PM
I have taken voltages on the boards of a known good and the non-working board.
Are the two keyboards identical? If not, then to be kept in the back of the mind is that 'apples are not being compared with apples'.


... the large chip (shown as 8048 in the Technical Reference)

... and my chip is not an 8048.
This device will have IBM-authored code in its ROM.
The "8493518" on it is possibly an IBM part number.
Although not stamped "8048", the device is possibly a customised 8048 made for IBM by Intel.

PCFreek
August 22nd, 2014, 10:49 AM
The keyboards are the same (as shown in the photos below).
2002220023

I probed all of the IC's for voltage (not data) and found that many pins on chip U2 (14 pin chip in the middle) have 5v present at the pins where 5v is not present on the working board.
The following pins have no voltage on the good board, but have +5v on the non-working board: 4, 5, 6, and 8.

Looking at the keyboard schematics, I see that pin 8 is "data in" for the 8048. This leads me to believe that the problem may lie with the IC in U2. I cannot find a datasheet on it.
Keyboard schematics:
20024

3pcedev
August 22nd, 2014, 04:09 PM
U2 is what is called a Tri-State Buffer with active low control. It appears to be a 4 input-output IC. Let me have a look to see if I can find something pin equivalent...

Edit: This one seems suitable http://www.ti.com/lit/ds/symlink/dm54125.pdf

This is a simple 74125 quad tri-state buffer IC. I quickly checked the pin assignments for the buffers (they all match). The only thing I have NOT checked is the pin assignment for the +5v and gnd. You can check this using a multimeter on your current IC - if it matches you could try replacing the IC with this equivalent one. I would say its 99% certain that it will.

Element14 sell them for about $1.30AUD - http://au.element14.com/texas-instruments/sn74abt125n/universal-bus-function-ic/dp/2371182

3pcedev
August 22nd, 2014, 05:30 PM
Just read your explanation again - Pin 5 on U2 should never have +5V on it. Pin 5 is tied to the ground rail. This does not point to an IC failure - rather a trace on the board / cable failure. I would check that A) this pin isn't dry soldered onto the board and B) that pin has continuity with ground with another point (check while unplugged) i.e. pin A05 on plug 1 and C) the ground is continuous between the cable & PC.

This should fix up your problem. If you ground is floating it will never work.

PCFreek
August 23rd, 2014, 05:40 AM
Just read your explanation again - Pin 5 on U2 should never have +5V on it. Pin 5 is tied to the ground rail. This does not point to an IC failure - rather a trace on the board / cable failure. I would check that A) this pin isn't dry soldered onto the board and B) that pin has continuity with ground with another point (check while unplugged) i.e. pin A05 on plug 1 and C) the ground is continuous between the cable & PC.

This should fix up your problem. If you ground is floating it will never work.

Pin 5 of U2 definitely has voltage. Pin 5 resistance to ground is 1.4K ohms between the pin and ground input pin with no cable attached. I also tested resistance between pin 5 and the keyboard shielding and got the same result. Resistance between keyboard shielding and the ground input pin is around 1ohm. the cable has been confirmed as good with a working keyboard. I'll check the solder joint on pin 5 and start checking other ground points for continuity. i appreciate your advice. I also saw this on the schematics, but I am not that good with reading schematics so i was unsure if this was a hard ground or a filter cap etc... I know what a cap looks like on the schematic, but am just never 100% sure.

PCFreek
August 23rd, 2014, 06:54 AM
I am back to thinking it is the 8048... I pulled out the logic probe and went over all 3 IC's.
The good board (verified as working) has pulses on the following pins:
U1 (8048 ) : 2-6, 7-14, 17-20, 22, and 30.
U3 (Chip on the far right): 4
No pulses are found at the keyboard connector unless keys are depressed.

The bad board has no pulses anywhere. Since pulses are not found at the connector, they must all be generated on the keyboard circuit board itself. Pin 4 of U3 does not have continuity with pins on the 8048 (R= around 1M), so it obviously does not feed the signal to the 8048. With all this, I deduce that the 8048 should be generating pulses and is not. Internal failure could also be sending 5v to the ground rail. I am not willing to remove a good 8048 from another board and try this. I did not find a keyboard ROM online. since the 8048 requires a ROM (and I don't even have the knowledge or equipment to burn a chip), this keyboard may be DOA. Bought it on eBay as "worked perfectly when stored, untested at time of sale and sold as-is". I am beginning to think that "untested" means "does not work".

3pcedev
August 23rd, 2014, 04:20 PM
How are you determining which pin is 1,2,3,4 etc on the IC's? The only reason I ask is that based on your last set of results your working keyboard should not work. A lot of those pins you found pulses on are tied to the ground or +5V rails. Where did you find your circuit diagram? It's possible it is the wrong diagram (although unlikely as the keyboard controller chip Vcc,Vdd and Gnd are always in the same place).

Could you have a look at that datasheet I posted up earlier of U2 and make sure your system of pin numbering is correct? I don't mean to be rude or anything by the way; I just don't quite see how your working keyboard works with the results you posted :)

Can you also do the following checks on both the working and non-working keyboard? This will narrow things down even further.

Check the resistance from the ground input pin (connector P1 A05) to:
- Pins 20,6,7 of U1
- Pins 2,13,5,7,10 of U2
- Pins A01,A03,A07,A09 of connector P2

Check the resistance from the +5 input pin (Connector P1 A01) to:
- Pins 26,40 of U1
- Pin 14 of U2
- Pin A08 of connector P2

You can then also plug everything back in and then check the voltages at those pins. All of the ground pins should have 0V at all times and all the +5 pins should have +5 at all times.

PCFreek
August 23rd, 2014, 05:46 PM
How are you determining which pin is 1,2,3,4 etc on the IC's? The only reason I ask is that based on your last set of results your working keyboard should not work. A lot of those pins you found pulses on are tied to the ground or +5V rails. Where did you find your circuit diagram? It's possible it is the wrong diagram (although unlikely as the keyboard controller chip Vcc,Vdd and Gnd are always in the same place).



I follow/count the pins as shown in the sketch below. The diagram that I provided is from the IBM 5150 Technical Reference Manual. I will check all readings again. It is a very fair question to wonder if I am making an error on counting the pins. It would not be the first time I made a mistake. :)

I appreciate your help.
20054

PCFreek
August 23rd, 2014, 06:19 PM
Could you have a look at that datasheet I posted up earlier of U2 and make sure your system of pin numbering is correct? I don't mean to be rude or anything by the way; I just don't quite see how your working keyboard works with the results you posted :)



My pin numbering is correct... perhaps my IC numbers are incorrect? I know the 8048 is U1, but I assumed the other 2 IC's were U2 and U3 in order. See photo below. My voltages for the non-working board are written on the sheet of paper.

I'll do the checks that you requested now.20055

PCFreek
August 23rd, 2014, 06:51 PM
Can you also do the following checks on both the working and non-working keyboard? This will narrow things down even further.

Check the resistance from the ground input pin (connector P1 A05) to:
- Pins 20,6,7 of U1
- Pins 2,13,5,7,10 of U2
- Pins A01,A03,A07,A09 of connector P2

Check the resistance from the +5 input pin (Connector P1 A01) to:
- Pins 26,40 of U1
- Pin 14 of U2
- Pin A08 of connector P2


Good board:
Check the resistance from the ground input pin (connector P1 A05) to:
- Pins 20 .3ohm ,6 2k ohm,7 1k ohm of U1
- Pins 2 open,13 open,5 3k ohm,7 1k ohm,10 3k ohm of U2
- Pins A01,A03,A07,A09 of connector P2 there is no connector P2... that's the traces to the keys.

Check the resistance from the +5 input pin (Connector P1 A01) to:
- Pins 26 0.1 ohm,40 0.1 ohm of U1
- Pin 14 0.1 ohm of U2
- Pin A08 of connector P2

Bad board:
Check the resistance from the ground input pin (connector P1 A05) to:
- Pins 20 0.1ohm ,6 2k ohm,7 0.1 ohm of U1
- Pins 2 34M ohm,13 open,5 3k ohm,7 0.1 ohm,10 3k ohm of U2
- Pins A01,A03,A07,A09 of connector P2 there is no connector P2... that's the traces to the keys.

Check the resistance from the +5 input pin (Connector P1 A01) to:
- Pins 26 0.3 ohm,40 0.3 ohm of U1
- Pin 14 0.1 ohm of U2
- Pin A08 of connector P2

The only difference in the 2 resistance readings is that pin 7 of both U1 and U2 is grounded on the bad board. I'll check voltages next.

PCFreek
August 23rd, 2014, 07:08 PM
You can then also plug everything back in and then check the voltages at those pins. All of the ground pins should have 0V at all times and all the +5 pins should have +5 at all times.

GOOD BOARD:
Voltages on ground pins:
- Pins 20,6,7 of U1 All 0v
- Pins 2,13,5,7,10 of U2 All 0v except: 10=5v and 13=1.7v

Voltages on the +5 input pins:
- Pins 26,40 of U1 All=+5v
- Pin 14 of U2 =+5v
- Pin A08 of connector P2

BAD BOARD:
Voltages on ground pins:
- Pins 20,6,7 of U1 All 0v except pin 6=5v
- Pins 2,13,5,7,10 of U2 All 0v except: 5= 5v, 10=5v, and 13=1.7v

Voltages on the +5 input pins:
- Pins 26,40 of U1 All=+5v
- Pin 14 of U2 =+5v
- Pin A08 of connector P2

So all the readings are the same except for +5v on U1 pin 6 and U2 pin 5 of the bad board.

modem7
August 23rd, 2014, 07:21 PM
The bad board has no pulses anywhere. Since pulses are not found at the connector, ...
I thought that you would have at least seen a single 20 ms pulse coming in on the clock line from the 5150/5160.
Refer to the diagram [here (http://www.minuszerodegrees.net/5160/keyboard/5160_keyboard_startup.jpg)].

modem7
August 23rd, 2014, 07:29 PM
... perhaps my IC numbers are incorrect? I know the 8048 is U1, but I assumed the other 2 IC's were U2 and U3 in order.
They are in that order according to SAMS Computerfacts, although I see that both the SAMS and the IBM Technical Reference use 'M' rather than 'U' as a chip prefix.

http://www.minuszerodegrees.net/5160/keyboard/5160_kyb_from_sams_computerfacts.jpg

PCFreek
August 23rd, 2014, 08:07 PM
I thought that you would have at least seen a single 20 ms pulse coming in on the clock line from the 5150/5160.
Refer to the diagram [here (http://www.minuszerodegrees.net/5160/keyboard/5160_keyboard_startup.jpg)].

No pulses coming in from the computer on either the good or bad board. a test of the cable (verified as working with the good board) reveals no pulses when not plugged in to P1 on the keyboard. I detect pulses on both data lines on the good board when a key is pressed, but never on the bad board.

PCFreek
August 23rd, 2014, 08:44 PM
I thought that you would have at least seen a single 20 ms pulse coming in on the clock line from the 5150/5160.
Refer to the diagram [here (http://www.minuszerodegrees.net/5160/keyboard/5160_keyboard_startup.jpg)].

I see what you are saying... a pulse during POST... yes, I verify 3 distinct single pulses:
1 almost immediately upon power switch being flipped
2nd pulse is about 2 seconds later
3rd pulse is right before the floppy is accessed during POST (several seconds after power on). this is the pulse you are referring to.

3pcedev
August 23rd, 2014, 11:44 PM
Hold on - is this an 83 key XT keyboard or the 101 key AT style?

The 83 key XT model has a different diagram (attached below courtesy of modem7's site)
20063

That came out a little small... Have a look here at page 117 of the PDF (section 4-21) http://www.minuszerodegrees.net/manuals/IBM_5155_5160_Technical_Reference_6280089_MAR86.pd f

This makes a lot more sense out of the readings you were getting. I'll see if I can come up with some more ideas from this.

3pcedev
August 24th, 2014, 12:07 AM
Ok based on the new diagram your results make sense. What we can check now is if the buffers are working properly (Chip U2 is actually a quad buffer, not a tri state buffer).

You can check these properties on both your working keyboard / broken one:

Pin 17 on U1 should have the same voltage (either 0 or 5V) as:
- Pin 6 of U1
- Pins 3,4,5,6 of U2

Pin 18 on U1 should have the same voltage (either 0 or 5V) as:
- Pin 1 of U1
- Pins 11,10,9,8 of U2

This will check that the logic buffers are doing what they should.

You can also check the reset circuit on the 8048. Pin 4 of U1 should be at +5V about 1 second after power up. You could also try a manual reset of the IC by briefly grounding Pin 4 while the power is on and PC running. This should not blow anything up as the reset circuit seems to be an open collector design. If the keyboard springs to life after this then the reset circuit is faulty.

If you also have an oscilloscope handy it would be worthwhile probing pin 3 of the 8048. It should be oscillating at some relatively high frequency. If it is static at zero then you either have a fault with the 8048 itself or the capacitor/inductor which forms the oscillator.

Good luck!

PCFreek
August 24th, 2014, 05:34 AM
Ok based on the new diagram your results make sense. What we can check now is if the buffers are working properly (Chip U2 is actually a quad buffer, not a tri state buffer).

You can check these properties on both your working keyboard / broken one:

Pin 17 on U1 should have the same voltage (either 0 or 5V) as:
- Pin 6 of U1
- Pins 3,4,5,6 of U2

Pin 18 on U1 should have the same voltage (either 0 or 5V) as:
- Pin 1 of U1
- Pins 11,10,9,8 of U2

This will check that the logic buffers are doing what they should.

You can also check the reset circuit on the 8048. Pin 4 of U1 should be at +5V about 1 second after power up. You could also try a manual reset of the IC by briefly grounding Pin 4 while the power is on and PC running. This should not blow anything up as the reset circuit seems to be an open collector design. If the keyboard springs to life after this then the reset circuit is faulty.

If you also have an oscilloscope handy it would be worthwhile probing pin 3 of the 8048. It should be oscillating at some relatively high frequency. If it is static at zero then you either have a fault with the 8048 itself or the capacitor/inductor which forms the oscillator.

Good luck!

I apologize for having the wrong diagram. I grabbed the early version in the 1981 manual as both keyboards are model F for a PC 5150. I now recall that there is a thread on here somewhere that states that the early keyboards with the metal connector have a different circuit board than the later model F's with the molded plastic connector. I should have known.

GOOD BOARD:
Pin 17 on U1 should have the same voltage (either 0 or 5V) as:
- Pin 6 of U1 =all 0v
- Pins 3,4,5,6 of U2 =all 0v

Pin 18 on U1 should have the same voltage (either 0 or 5V) as:
- Pin 1 of U1 U1 18 = 3.9V, U1 #1 = 5V
- Pins 11,10,9,8 of U2 5V, 5V, 5V, 3.9V


BAD BOARD:
Pin 17 on U1 should have the same voltage (either 0 or 5V) as:
- Pin 6 of U1 #17 = 1.7V, #6=5V
- Pins 3,4,5,6 of U2 =1.7V, 5V, 5V, 5V

Pin 18 on U1 should have the same voltage (either 0 or 5V) as:
- Pin 1 of U1 U1 18 = 1.7V, U1 #1 = 5V
- Pins 11,10,9,8 of U2 5V, 5V, 5V, 1.7V

Please note that the 1.7v readings on the bad board are sometimes double in value (3.4v). When power it up again to take more readings, 1.7v seems to be the most common but I have seen 3.4v at these same pins on several occasions.

Grounding pin 4 to reset the 8048 does nothing. I do not have an oscilloscope... only a logic probe. logic probe picks up nothing on pin 3.

3pcedev
August 25th, 2014, 01:28 AM
No problems re the diagram - I have done the same things many times. IBM really should have changed part numbers (significantly) for their different keyboards.

Based on your numbers above the good board looks like its working as I would expect. Some of the voltages are a little low due but that is to be expected as it uses resistive pull-up's.

The 1.7V on pin 17 & 18 of U1 is a problem. This would be ok if the output of pin 17/18 was pulsing (which it will when a key is depressed & it sends data back to the PC) as the average voltage will be < 5V due to the logic 1's and 0's. Since you tested for pulsing outputs earlier (and found none) it seems like the output drivers are floating. This indicated the 8048 is basically dead in the water... as for why:

On balance of probability I would say that most likely the 8048 is non-functional, followed by some problem with the C5,C4 and L1 clock oscillator network which results in no clock signal (and thus no-go). Its also possible that U2 is faulty and loading up the outputs of the 8048 too much; but in my experience this is a 1 in a million.

Could you just check one last thing... when the keyboard is powered can you measure the voltage on Pin 4? It should be around 5V. It's unlikely that the reset circuit is broken but it's definitely worth a shot.

If it were me I would double check the +5V and Gnd connections with the new diagram. I would then try replacing C5,C4 and if that didnt work I would then replace L1. Finally I would replace U2 as they are cheap (note it is not the part I posted up earlier, ask me if you want exact details); but at that point it's very unlikely to make any difference. That just leaves the 8048 which I am guessing is now unavailable.

If you lucky C5, C4 and L1 might just fix it up. It will only cost a few dollars or so to try.

PCFreek
August 26th, 2014, 02:47 PM
Could you just check one last thing... when the keyboard is powered can you measure the voltage on Pin 4? It should be around 5V. It's unlikely that the reset circuit is broken but it's definitely worth a shot.

If it were me I would double check the +5V and Gnd connections with the new diagram. I would then try replacing C5,C4 and if that didnt work I would then replace L1. Finally I would replace U2 as they are cheap (note it is not the part I posted up earlier, ask me if you want exact details); but at that point it's very unlikely to make any difference. That just leaves the 8048 which I am guessing is now unavailable.

If you lucky C5, C4 and L1 might just fix it up. It will only cost a few dollars or so to try.

Pin 4 input is 4.9v. The voltages at C4, C5 and L1 are as seen in the pic below:
20131

PCFreek
August 26th, 2014, 05:50 PM
C7 has a resistance of only 50ohms on the non-working keyboard, and over 1k ohms on the good keyboard. Could this cause the described problems with voltage readings?

EDIT: C1 and C7 have identical values of 50 (rising to 80) ohms on the non-working board, while the same C1 and C7 on the good board have identical values of 1+k ohms. there is a slight difference between the boards. My non-working board looks exactly like the graphic that modem7 posted in #17 of this thread. My good board has 2 resistors above U1. See photos in post #6 of this thread. I do not think I can compare resistance values between the two.

3pcedev
August 29th, 2014, 04:13 AM
The rising resistance is somewhat normal around C1 and C7. That happens because the multimeter is 'charging' them up slightly due to the small voltage present across the probes.

Unfortunately you will need a CRO to measure the voltages across C4,C5 and L1. They actually form an oscillator circuit which is then used to create a square wave clock signal for the 8048. As such there should be some reasonably high frequency AC signal on pin 3 of the 8048.

If your still keen I would replace C4 and C5. I admit it's a long shot but for a couple of dollars it's worth a try: http://au.mouser.com/ProductDetail/Vishay-Cera-Mite/561RU2JQA102EC200K/?qs=sGAEpiMZZMsh%252b1woXyUXj7XQ5AMFgLJenKAVI%2f11 3%2fo%3d