Page 2 of 3

Re: Blown RAM card?

PostPosted: January 17th, 2022, 1:51 am
by Ttpilot
TronDD wrote:Double check you have exactly the right chips in the right places.


Looking back over the CPU board I found that ICQ should have been a 74123 but I have a 74LS123. Maybe that's the problem?

Re: Blown RAM card?

PostPosted: January 17th, 2022, 11:35 am
by TronDD
Not 100% sure. I think I had a 74LS04 in IC P which rendered the clock useless. Both IC P and Q are in the clock circuit so I suspect having the 74LS123 instead of a 74123 to be a likely problem.

Re: Blown RAM card?

PostPosted: January 17th, 2022, 3:51 pm
by AltairClone
A 74LS123 gives slightly different one-shot timing than a 74123. However, the timing typically still needs an adjustment even with a 74123 in place. However, the difference in timing between the two chips might be enough to at least make your system work (even if the clock timing isn't perfect), so if you have a 123 around, you might as well try it.

Mike

Re: Blown RAM card?

PostPosted: January 23rd, 2022, 10:44 pm
by Ttpilot
Well, my oscilloscope arrived, but they made a mistake and only included one probe, so I'm waiting for the second one to get here. I've been using the time to study up. I haven't used an oscilloscope since high school in the late 60s!

I've also been reading up on the 74123 chip and the 8080, and studying the schematic for the cpu card. What I'm getting is that the phi 1 and phi 2 signals are generated on the 74123, and the resistors, R 41 for the phi 1 signal and R 42 for phi 2, control the discharge rate, which affects the length of the signal. The 74123 spec sheet includes formulae and tables to determine what level of resistor is needed to adjust the signal to the desired length.

Hopefully I'm on the right track

Re: Blown RAM card?

PostPosted: January 24th, 2022, 7:11 pm
by Ttpilot
I took my first oscilloscope readings of the cpu card timing. Actually, I measured the phi 1 and phi 2 bus signals--it was easier than trying to hook little pins on the cpu card. Here's what I got

phi 1 channel 1 phi 2 channel 2.jpg
Phi 1 on channel 1, phi 2 on channel 2


Voltages look okay. Phi 1 runs high for about 130 ns and low for about 370 ns. Phi 2 starts when phi 1 goes to low. It runs high for 276 ns and low for 224 ns.

If I've read things right, this is close to the specified timing. Perhaps someone could correct me if I've got it wrong

Re: Blown RAM card?

PostPosted: January 25th, 2022, 12:26 am
by AltairClone
Switch to 1v per division and put 0v of each trace at the same horizontal position on the display so that the waveforms overlap each other. Also set the time scale to 50ns and let’s take another look. The biggest issue is generally the phi2 high period which affects the critical phi2 falling edge to phi1 rising edge spec.

Mike

Re: Blown RAM card?

PostPosted: January 25th, 2022, 10:35 am
by Ttpilot
AltairClone wrote:Switch to 1v per division and put 0v of each trace at the same horizontal position on the display so that the waveforms overlap each other. Also set the time scale to 50ns and let’s take another look. The biggest issue is generally the phi2 high period which affects the critical phi2 falling edge to phi1 rising edge spec.

Mike


Here is the result

AB8E9F51-ADC8-4ED9-8A25-769CF41D7B0D.jpeg
Phi1 + phi2

Re: Blown RAM card?

PostPosted: January 26th, 2022, 7:22 pm
by AltairClone
Perfect capture. Biggest issue I see is the falling edge of phi2 to the rising edge of phi1 is too short. In your case, it’s primarily caused by the phi1 pulse being too long. To fix it, let’s shorten the phi1 pulse first and then see if we need to shorten the phi2 pulse as well.

Are you using an LS or regular 123 and what are the values of R41 and R42?

Re: Blown RAM card?

PostPosted: January 26th, 2022, 7:59 pm
by Ttpilot
AltairClone wrote:Are you using an LS or regular 123 and what are the values of R41 and R42?


It’s a regular 74123. R41 is 13k and R42 is 6.2k. I double checked those on my meter. The color bands on R42 looked a bit wonky

Re: Blown RAM card?

PostPosted: January 26th, 2022, 8:04 pm
by Ttpilot
Maybe go to an 11k for R41?