“The world moved away from the 68000 too but I still like designing around it” I understand this very well, I sometimes think about building a 8086 or I386 board for my self. That would be fun, but then I remember that nowone pays for it, no one needs it not even me. But I would use 74HC family or some modern 3.3V5V LVC family.
In such case I use BC847 (Vce=45V) driven by voltage divider (for example 4k7+3k3) and with small R in emitter. Voltage divider + small R form a current source to limit current when pulse comes from outside while transistor in ‘on’. If I expect Surges I add SMB (600W) 18V transil. Such my outputs were tested by 1kV surge generator connected to my output via 40 ohm resistor (size close to brick).
In magazine “Elektronika dla Wszystkich” (Electronics for All) one man published an article about made by him calculator whole made of 7400 serie ICs. Don’t remember exactly but his PCB was close to 1/4 of desk. I written to him suggesting to make a DES round PCB and then make combinational (not sequential) DES crypt/decrypt with stack of these PCBs. He said he may be will consider it in future, but now is busy with something else.
That would be fun, but then I remember that nowone pays for it
I’m a software engineer so hardware (and KiCad) are just hobbies for me. Lots of fun. I’ve designed and built quite a few boards (an ATF1508 dev board, 65C02 SBC, etc.). All for fun and no profit.
Speaking of designing purely for fun, check out this channel. This guy is amazing.
We all need hobbies, some people do great things. I wonder where they get the time to do them.
Luckily PCBs are cheap today and even letting them to assemble it is cheap, having fun does not cost much.
Some day you may find out for yourself. Or not. 
I would like to mention a “tangential” issue, and then another which is tangential to that one…
Some years ago I made a lab test fixture. It had a momentary pushbutton switch, followed by a bipolar debounce circuit driving the clock input of a flip flop (I think it was 74HC type) controlling a slew rate controlled power switch. This did not work right. It turned out that the slew rate out of my discrete debounce circuit was too slow for the flip flop clock input, and the flip flop was double triggering. Fortunately I got it working by decreasing the value of the pullup resistor of the debounce logic output. This might have still been too slow per 74HC specification but anyway it worked.
The better design fix would have been to have some sort of 74HC schmitt trigger device between my debounce function and the flip flop clock input. I thought that my debounce circuit had sufficient “snap” but the dV/dt was too slow.
More recently I made a very simple oscillator out of a 74HC inverter. It was running only at some KHz but I was surprised to see that it was drawing about 20 mA from the 5V supply. (Folks, that is a lot for a single inverter.) It turns out that 74HC (and maybe other) schmitt trigger inputs draw high current when the input signal is in the transition region, and my oscillator always had the input voltage in the transition region, oscillating with the hysteresis between the two thresholds.
Lesson 1 is to watch your slew rate when driving non-schmitt input logic with a resistor pullup open collector signal. Using a schmitt trigger input somewhere is probably a good approach.
Lesson 2 is (unless you do not care about logic gates wasting power and running unnecessarily warm) do not use a schmitt gate as an oscillator. Probably better to use a comparator.
I’m always impressed by people who do big replicas. E.g.
Takes a large chunk of devotion to complete such projects.
I think you remember something wrong. May be it was not HC. Looking into:
https://pl.mouser.com/datasheet/2/916/74HC_HCT1G14-1597561.pdf
Fig 10. - about 0.4mA/4.5V. Fig 11. - about 0.7mA/6V so probably 0.5mA/5V.
Higher currents are for HCT - Fig 12 and 13.
Few years ago I also made a simple oscillator (10Hz) out of LVC2G14. I used LVC serie as I needed ns range times (10 times per second the 15ns measurement (test) takes place where 2ns time constant change is what is looked for).
See:
https://pl.mouser.com/datasheet/2/916/74LVC2G14-1597848.pdf
Fig 9 - it looks that about 7mA average current is when VCC=3V. But I had 5V - Fig 12 shows how fast current increases with VCC - for 5V average is about (35+15)/2=25mA.
As my circuit was full CMOS and was active only 15ns per 100ms (0.000015%) I supposed zero, zero, nothing supply current and was surprised seeing 25mA.
I solved that using HC1G14 as a generator an buffering it with LVC. So you have to remember something wrong. HC shmitt negator not takes 20mA in generator configuration.
Hi, Piotr
I am 100% certain of my memory regarding the 20 mA current measurement. This because it impressed me and was within the last year. Looking at my lab stock list; the precise part number was definitely not an HCT type (I do not use TTL or LSTTL and have never purchased any HCT) and most likely was TI SN74LVC1G14DCKR. Even if the current level were 7 mA instead of 20 mA, it does not change my point:
When using a schmitt trigger CMOS gate as an oscillator, you are likely to get supply current drain that is much higher than expected based simply upon static supply current ratings and operating frequency. This might be OK for a lab test circuit, but is probably not good for most mass production applications. This is because minimizing power consumption and/or heat generation are usually important in a mass produced product.
I have previously found a datasheet graph which would indicate what I have measured but cannot find that graph right now.
If I can find the datasheet graph that I am looking for, I will add it to this post.
Why work with newfangled stuff like the 8086 when you can work with a 4004?
http://4004.com/
https://insanity4004.blogspot.com/
1 Like
Nah, build a computer from discrete transistors:
There are also a lot of communities enjoying their retro-computer hobby, and because of this there are several differen Z80 based systems available in kit form, and there are also other projects which are made in in KiCad such as for example: https://github.com/rosco-m68k/rosco_m68k, which is both availabe as Open Source and as kit.
LVC is the another fairy tale than HC (LVCs are 10 times faster).
Nexperia LVC1G14 datasheet shows less current then LVC2G14. Do they specify that current if you drive both inputs together ?
In TI:LVC1G14 I don’t see that current specified other than for input voltage = VCC-0.6V.
That’s what I’m doing: a replica 4004 CPU using discrete components:
1 Like
Ah, yes, I had not followed your link…
but the “MegaProcessor” has more LED’s:
https://duckduckgo.com/?q="megaprocessor"&t=hy&va=g&iar=images&iax=images&ia=images
Not sure what you mean by “fairy tale” in this context. I think that speed to me is like current is to you. To me 1 MHz is very fast and almost any CMOS gate is fast enough. Frankly I got lost with the numerous parameter variations between vendors and types. I look for 5V capability and have also learned to look for 24 mA or 32 mA output current capability which is more useful for general work with analog/power interface. I think SN74AHC1G14 is specified only to 8 mA output.
I am not sure why this matters. I do not have any of the 2G types, and connecting two schmitt inputs together in an oscillator sounds like a worse idea. It would be even worse yet if the outputs were also connected together with the likelihood that the input thresholds are not identical. So the outputs might fight each other and smoke could result.
Me too 
I used Google translator to translate the Polish word ‘bajka’ meaning ‘story you read children’ and that word can be used as ‘another bajka’ meaning something belonging to completely another world.
Not matters but simply is weird (also dictionary). For me LVC14 gate should behave similarly in LVC1G14 and LVC2G14. And in specifying supply current when input is driven by some voltage
I would expect only one input is driven. I am just surprised that form datasheets for LVC1G14 and LVC2G14 of the same manufacturer I see very different values.
Of course.
I just was wondering why they got much higher current for LVC2G then for LVC1G. And I jokingly suggested that may be they just driven both inputs together during measurement of supply current as a function of input voltage (not connecting outputs together of course).
Thanks
It is difficult to completely know what is going on inside the ICs. In many cases two different ICs are 80% identical but have different interconnecting metal layers (applied later in the process). I suspect this may be true of 74HC4020 and 74HC4060. But perhaps the single and dual schmitt inverters are made from completely different silicon “mask sets”.
BTW I see that I also have Fairchild NC7SZ14 so the device I tested might have been that, although this is less likely.
Have you considered it was just a faulty IC?
For example, ESD damage often does not destroy a chip, or even a single pin. It can also manifest itself in decreased performance (a half destroyed transistor) or increased power consumption.
I read once that 80% of ESD damage is not noticed by quality control, but can still lead to early failure.
Next time I’m placing an order I’ll try to buy some extra HC chips, both with and without schmitt triggers. Putting some DIP IC’s on a breadboard is a quick enough test. for differences in power consumption.
Well, no…but I find that happens pretty seldom. I remember finding some explanation on a DS and right now I can find it only qualitatively but not quantitatively. As I say the device in question was a 5V capable CMOS chip but probably not a 74HC series device.