A quick question. Back in the day when 74LS was the primary logic family used for microcomputers, the output of 74 bipolar TTL logic typically sinks current much more than it sources (for example can sink 2-3 mA whereas it might only be able to source 0.5 mA). It seems that a lot of manufacturers are using HCT thresholded logic for many inputs as these are compatible with 3.3 and 5 volt logic thresholds for input and output, where the typical low threshold is 0.8 V and high threshold is 2.0 V. This is similar to NMOS, however, as far as I know there are no more NMOS process chips made. The principal difference I am aware of with HCT-like logic is that sinking and sourcing currents both tend to be quite high, 20 mA or more, and the high output threshold can typically be pulled up close to the positive rail at 5 V, whereas typical TTL might achieve only 3.5 to 4 V.
Can HCT logic be substituted and will its greater source and sink capacity at the output and/or its operation closer to the output rails be a problem when placed in circuit with older TTL chips?
Dan
In post #1, "profdc9" asked:
"Can HCT logic be substituted and will its greater source and sink capacity at the output and/or its operation closer to the output rails be a problem when placed in circuit with older TTL chips?"
Uncle Bernie answers:
Spot on. You saw the pitfalls. This higher source and sink capability comes at a price, much higher dI/dt on both the power, the ground, and the signal lines, and many 1970s vintage layouts and their generally weak power supply bypassing (too few / too low performance bypass capacitors) can't take that beating. Sometimes adding additional power supply bypass capacitors of 100nF directly across the power/gnd pins of these HCTs helps. But in general, it's a hit-and-miss and too tedious / not economical to try this replacement across the board. But if you can't find that LSTTL but have the same HCT, it's worth a try.
There have been several posts on this HCT vs. LSTTL replacement topic in the Apple-1 forum.
- Uncle Bernie
Generally these differences are no problem, since they are within the specified margins for LSTTL. There are some circuits that fail to obey digital discipline and would be sensitive to the differences. The above reply mentions parasitics which could also be an issue, depending on circuit topology and layout.
I would generally agree with Uncle Bernie (as usual). It's worth a try if that is what you have or all you can easily get. But I generally prefer to replace with as close to original as possible when I am working on things like ][+, clones or vintage cards. I've used HCT, HC and someting just C parts in new design cards. But my stuff is all uber simplistic because I'm really a software guy and just a hardare dabbler.
If I soldered a 100 nF capacitor between VCC/GND on the back of the chip, would that help offset the power rail current draw spikes? Other than that, does the faster rise/fall time cause problems for TTL inputs?
In Post #5, "profdc9" asks:
"If I soldered a 100 nF capacitor between VCC/GND on the back of the chip, would that help offset the power rail current draw spikes? Other than that, does the faster rise/fall time cause problems for TTL inputs ?"
Uncle Bernie answers:
In general, adding a bypass capacitor as close to the fast CMOS IC can help, but it's not guaranteed. However, if you put it "on the back of the chip" that won't seat in the socket anymore, except if you use SMD capacitors maybe. I usually put extra bypass capacitors on the backside of the PCB, just across the power and ground pins of the offending IC. If you want to avoid soldering on the vintage PCB, you can solder the capacitor on top of the IC.
The faster rise/fall times will not cause trouble with TTL inputs as such but they may cause ringing on signal lines if the PCB traces are long enough. With CMOS that can get worse than with Schottky TTL, where the limit for unterminated traces is ~11 inches. Adding damping resistors of ~390 Ohm 1/8W at the far side of the offending CMOS driver can help. This in no way is a proper transmission line with proper termination but more of a band-aid. You can see the effect in the oscilloscope (if it is fast enough). If you had a proper impedance controlled transmission line with perfect termination then there would be no ringing. But you can't get there. And you don't need that. All you need is to keep the ringing small enough in amplitude that the TTL on the receiving end does not glitch. Would I trust such a kludge with my life ? No way ! But if you can get an old piece of vintage hardware going again with it, why not.
There is a technique called the "magic touch" where you can quickly zero in on offending ringing signals on a digital PCB. Just wet your fingers with spit and touch the IC pins. If the problem goes away you found the culprit. This does not work all to well with modern PCBs but with the 1960s to mid/end 1980s digital PCBs it works more often than not and is a great time saver compared to probing of each and every signal with an oscilloscope. I learned that trick in the early 1980s from a field service technician who worked for DEC.
- Uncle Bernie