PDA

View Full Version : "TD2" delay line and "C.ADJ" variable capacitor.



Arkady
October 10th, 2013, 12:49 PM
Hello,

I'm trying to fix a 64kb-256kb IBM XT motherboard, it has damaged components :

One is called "TD2" and it seems to be a "delay lines" : where can i find such a component, please ?
The other is a variable capacitor "C.ADJ". It's located between the power connectors P8-P9 and a crystal. I'd need to know it's capacitance to be able to replace it.

Thank you all for your help.

modem7
October 10th, 2013, 07:26 PM
One is called "TD2" and it seems to be a "delay lines" : where can i find such a component, please ?
According to IBM's technical reference for the IBM XT (IBM 5160), TD2 is a 470 ohm resistor (which together with capacitance will be forming a delay).


The other is a variable capacitor "C.ADJ". It's located between the power connectors P8-P9 and a crystal. I'd need to know it's capacitance to be able to replace it.
Variable capacitor, C1, described at http://www.minuszerodegrees.net/5160/misc/5160_motherboard_switch_settings.htm
According to IBM's technical reference for the IBM XT, C1's adjustable value is 5-30 pF.
If you are not going to be using a composite monitor, then you have the option of using say, a 20 pF capacitor, instead of a variable capacitor.

modem7
October 10th, 2013, 10:55 PM
I'm trying to fix a 64kb-256kb IBM XT motherboard,

The other is a variable capacitor "C.ADJ". It's located between the power connectors P8-P9 and a crystal.
I've just reread your post.
The first line I have quoted indicates "XT" (IBM 5160), but the second line indicates PC (IBM 5150).
Do you have an IBM XT 64-256KB motherboard, or an IBM PC 64KB-256KB motherboard, or are you unsure ?

If you in fact have a IBM PC 64KB-256KB motherboard (which I'm very sure is the case), then the information that I provided about the variable capacitor applies to both the IBM PC and IBM XT.

But the information that I provided about TD2 is not. In the IBM PC, TD2 is a delay line unit (part number PE-23088B) which is used together with resistors R11 and R12. In the move to the IBM XT, R11/R12 were removed, and TD2 was changed to a 470 ohm resistor.
That suggests that in the IBM PC, the PE-23088B unit can be substituted by removal of R11/R12 and inserting a 470 ohm resistor between pins 1 and 4 of where the PE-23088B was.

http://www.minuszerodegrees.net/5150_5160/misc/5150_5160_td2_comparison.jpg

Arkady
October 10th, 2013, 11:55 PM
I've just reread your post.
The first line I have quoted indicates "XT" (IBM 5160), but the second line indicates PC (IBM 5150).
Do you have an IBM XT 64-256KB motherboard, or an IBM PC 64KB-256KB motherboard, or are you unsure ?

If you in fact have a IBM PC 64KB-256KB motherboard (which I'm very sure is the case), then the information that I provided about the variable capacitor applies to both the IBM PC and IBM XT.

But the information that I provided about TD2 is not. In the IBM PC, TD2 is a delay line unit (part number PE-23088B) which is used together with resistors R11 and R12. In the move to the IBM XT, R11/R12 were removed, and TD2 was changed to a 470 ohm resistor.
That suggests that in the IBM PC, the PE-23088B unit can be substituted by removal of R11/R12 and inserting a 470 ohm resistor between pins 1 and 4 of where the PE-23088B was.

http://www.minuszerodegrees.net/5150_5160/misc/5150_5160_td2_comparison.jpg

Thank you very much, Modem7.
Everytime i come and read this forum, i have the smashing feeling to be a little boy walking among giants ! :D

I have a 64kb-286kb IBM XT (without TD2) and a 64kb-286kb IBM PC : i noticed that there was a resistor instead of the TD2 and two other missing resistors. But, i didn't dare investigating in this direction.
The other lesson is that i MUST get a "IBM's technical reference" for the IBM PC/XT.

Chuck(G)
October 11th, 2013, 12:34 AM
The fact that IBM would go to the expense of a 7 nsec. delay line suggests that in the original, there was a serious race condition in the inputs to U74 and that someone eventually found out that a 470 ohm resistor, coupled with the input capacitance of the 74LS00 was just sufficient to do the job. A spare gate would have also done the job, but there probably wasn't one in the PCB vicinity. Kludges like that make me uneasy.

Arkady
October 12th, 2013, 05:57 AM
Hello,

My primary goal is to stick to the original design, as far as i can. But, i will unsoldered the two resistors and add a 470 ohm if necessary.

For the moment, i found a variable capacitor that might fit (dimension and capacitance)
But for the delay line, i only found :
- 3ns and 55ohm impedence
- 9ns and 93ohm impedence

Can i substitute the original by one of these two delay lines ?

Thanks

Chuck(G)
October 12th, 2013, 08:14 AM
I'd try the 9 nsec first--the difference (2 nsec) is well within propagation time tolerance of the 74LS04.

modem7
October 12th, 2013, 12:11 PM
A member of these forums, mmruzek, produces the PC-RETRO motherboard kit, a close copy of the IBM PC motherboard.
I wondered what he used.
The advertising website is http://www.mtmscientific.com/pc-retro.html
The motherboard photo there suggests that he used the resistor that the XT has, but kept R11 and R12.

Chuck(G)
October 12th, 2013, 02:55 PM
A good logic analyzer with a glitch detector on the LS00 output will betray any existing race condition.

sergey
November 27th, 2013, 10:37 AM
The purpose of the circuit with the delay line is to change clock's duty cycle so that it can be used by 8237 DMA. 8088 CPU uses a clock with 33% duty cycle (approximately 70ns HIGH / 140ns LOW at 4.77MHz) while DMA needs 50% duty cycle clock (at least 80ns HIGH and 68ns LOW plus at most 10ns transitions).
The circuit works as follows:
- 8284/U11 generates CLK88 with 33% duty cycle ~70ns HIGH and ~140ns LOW.
- CLK88 is buffered by 74LS244/U15. This delays the clock by ~12ns. The output signal is CLK
- Both CLK and CLK88 go through 74LS04 (U51/Gate4 and U99/Gate4 respectively). This inverts the signal (so HIGH and LOW periods change places). Inverters also delay clocks by ~10ns. But this is not really important since both clocks get delayed.
- Inverted CLK next goes through that 7ns delay line that you've mentioned. And so the CLK is delayed by ~19ns in total.
- 74LS08/U52/Gate2 AND gate combines delayed clocks, so that 19ns are subtracted from HIGH (140ns) and added to LOW (70ns) periods, and the resulting DCLK signal has 121ns HIGH / 89ns LOW periods.

IBM XT and Retro PC are using 470 ohm resistor instead of TD2 delay line... IMHO you can just put a wire instead of TD2 and that would work too because 74LS244 provides enough delay :-) Many XT clones use a couple of 74LS04 gates instead of TD2.