[acannell]

Think of a CPLD (or FPGA) as a collection of gates and D-Flops that you can interconnect in any way you want to while MCUs are sequential machines that execute a list of instructions. A problem with 7400 series logic is it's fast becoming obsolete; at one time there was over 500 different logic devices. Now the series has shrunken to less than a dozen unique functions, primarily inverters, buffers and D-Flops. The reason is CPLDs and FPGAs have replaced discrete logic devices.

Mariss

not true. there are quite a bit more than a dozen logic IC's still available. you can see them by just going to digikey. is this one of those cases where its more of a thing you wish were true instead of something that is actually true? i see that alot on here

[grg12]

I'll need pencil and paper to follow that, but looks good after my second reading.

After some more thinking - the device will need some more logic. For example - connecting ouput of binary comparator directly to reset input of counter (like in prescaler) is not a good idea - it can cause unstable state. Let's say comparator triggers when counter hits '101111', curent counter state is '101110', after clock edge counter goes to '101111', comparator activates 'reset'. Flip-flops in counter begin to reset, but some of them are faster so after a short moment output signals looks like this '101000', comparator removes reset and you have a problem... In most cases everything will work ok - flip-flops are faster than comparator, but - there is no guarantee. Next problem - you don't have clean signal to trigger second stage counter. Looks like you need additional flip-flop to synchronise everything.

As for CPLD - i have used Altera CPLDs for hobby project and played a bit with Spartan-3 starter kit from Xilinx. Sadly - i remember mostly my struggle with programming environment. I borrowed some books from work colleague (he's electronics designer i write programs), he helped me with many problems - but still, all i can remeber is the bloated, overcomplicated and supersensitive desing environment. Hope current versions aren't so repulsive

Gregor

[greybeard]

Where does the predetermined constant come from and what is its range?

It comes from the geometry of the m/c build, and I would guess its value will be about 50 +/- 10

Would an 8 bit DIP switch work? (values 0 to 255)

Yes, I see no reason why not.

Where does the 4 bit variable come from? (Keeping in mind that you suggested 8 bits could be desirable.) Maybe the operator turns a knob?

This is the number that, when multiplied by the constant, becomes the number of steps that one axis moves. The constant may be different for each axis, but only by a small amount.
My original thought of 8-bits is now looking more like 12-bits, but this is still made up of a 4-bit word controlling the stepping pulse output !
the other bits would have other control functions, for example a single bit will control direction.

How fast should the k x n pulses be? Adjustable or fixed?

Fixed, but not calculated yet, but ~2khz most likely.

What sort of duty cycle? Adjustable or fixed?

Of the pulses ? 50%

What input signal to get things started?

I might just shout at it.... No, I've not gone to that level of detail yet.

What happens at the end of the pulses? Nothing until a new input signal? Pause and repeat for ever?

Wait for new input.

Depending on where the configuration comes from, something as simple as an 8 pin microcontroller could handle this job.

I'm sure it could, but please read my comments about using the technology that I'm familiar with.
I am not 'died-in-the-wool', but give me a break. I've got a life-time of experience with an older ways of solving problems and I don't have the time left to start yet another exploration of Terra Incognita.

Regards
John

[acannell]

I'm sure it could, but please read my comments about using the technology that I'm familiar with.
I am not 'died-in-the-wool', but give me a break. I've got a life-time of experience with an older ways of solving problems and I don't have the time left to start yet another exploration of Terra Incognita.

Regards
John

welcome to internet forums! where your way is never good enough lol...id say 95% of the people on forums will try to dissuade you or ignore your desire. the other 5% are what make it all worth while..

(*#$ did that rhyme? wow

[Mariss Freimanis]

acannell,

I should have been more clear. There are very few discrete 74-series and 4000-series logic varieties available today in modern CMOS implementations (74HC, 74AC, etc.). I enjoyed designing the G201 and G320 using discrete logic and I felt badly when discrete logic obsolescence became apparent. I had developed many useful design tricks over the years using them and I sincerely wished they would never disappear. I had to drag myself, kicking and screaming, to learn how to use CPLDs and FPGAs and now would never go back.

Here's a useful white paper that compares discrete versus programmable logic and gives a little history lesson as a bonus:
http://www.xilinx.com/support/docume...pers/wp202.pdf

Mariss

[acannell]

acannell,

I should have been more clear. There are very few discrete 74-series and 4000-series logic varieties available today in modern CMOS implementations (74HC, 74AC, etc.). I enjoyed designing the G201 and G320 using discrete logic and I felt badly when discrete logic obsolescence became apparent. I had developed many useful design tricks over the years using them and I sincerely wished they would never disappear. I had to drag myself, kicking and screaming, to learn how to use CPLDs and FPGAs and now would never go back.

Here's a useful white paper that compares discrete versus programmable logic and gives a little history lesson as a bonus:
http://www.xilinx.com/support/docume...pers/wp202.pdf

Mariss

still not clear why you are saying this..ive yet to have trouble finding whatever 74 series logic i need from digikey ready to buy. if you type in 74HC on digikey's search engine you get 12,000+ results, averaging about 500 parts per category and there are more than 20 categories...can you be specific about a 74 series chip that you dont think is available anymore?

[Mariss Freimanis]

OK, for instance try to find a CMOS 74150 or 74147. Both are available only in the obsolete 7400 TTL family (slow, large package, awful noise immunity and power hungry). The first is a 16:1 multiplexer and the second is a 10-line to BCD priority encoder. Digikey will sell you the 74150 for $4.56 and only grey-market distributors have the 74147 for $2. Both are devices I actually used way back when.

My only point is these devices and many, many others are no longer supported (no CMOS versions) and are extremely expensive given the design alternatives available.

Mariss

[acannell]

OK, for instance try to find a CMOS 74150 or 74147. Both are available only in the obsolete 7400 TTL family (slow, large package, awful noise immunity and power hungry). The first is a 16:1 multiplexer and the second is a 10-line to BCD priority encoder. Digikey will sell you the 74150 for $4.56 and only grey-market distributors have the 74147 for $2. Both are devices I actually used way back when.

My only point is these devices and many, many others are no longer supported (no CMOS versions) and are extremely expensive given the design alternatives available.

Mariss

we long ago passed the point where you could get a million times the number of gates in a single FPGA or CPLD that you could in a single logic ic, for the same price and obviously a millionth the size of the equivalent discrete circuit. its not an apples to apples comparison just because they are both "logic". there is alot more going on in circuit design than that. an FPGA or CPLD is not the least expensive, smallest, or simplest solution in many situations, regardless of its incredible density. and the varied uses for discrete logic IC's is practically infinite, only one subset of which is something you might compare to how you would use an FPGA or CPLD, i.e. a relatively complex network of interconnected logic that justifies the coding and cost.

you can create the same functionality of the 74150 or 74147 using a small combination of discrete logic or another type of logic. ive been hearing "discrete logic is quickly going obsolete" for about 15 years, yet its 2013 and you can still get pretty much every ancient 4000 series CMOS logic chip brand new from TI, so obviously that isnt true and isnt going to be true anytime soon. parts get obsoleted all the time, it doesnt mean that the genre of parts in general is obsolete.

[Mariss Freimanis]

Well, in theory you can build any imaginable logic using just 2-input NAND gates. The idea behind the diversity of discrete logic medium-scale functions that existed up to the 90's was to replace this necessity. The family diversity isn't what it used to be and is still shrinking; we get end of life last buy notifications from TI, Fairchild, etc. that reflects this reality. Octal buffers, D-Flops and latches along with simple gate functions will continue to be available, relegated to MCU and FPGA interface applications. In fact, the only new devices I've seen are single gate implementations in SOT-23 and SC-70 6-pin SMT packages.

Mariss

[acannell]

The family diversity isn't what it used to be and is still shrinking; we get end of life last buy notifications from TI, Fairchild, etc. that reflects this reality.

How do you figure this? The vast majority of discrete logic is still available. If the bloated chips at the fringe start being obsoleted, thats not a reason to say discrete logic as a whole is disappearing or not an important part of modern circuit design. And there are still a ton of bloated chips at the fringe. The core is untouched.

Octal buffers, D-Flops and latches along with simple gate functions will continue to be available, relegated to MCU and FPGA interface applications. In fact, the only new devices I've seen are single gate implementations in SOT-23 and SC-70 6-pin SMT packages.

Mariss

You're applying your personal circuit design experience to the big picture. In my experience I see plenty of discrete logic being used in places very far away from any processor or data pathway. Analog signal chains frequently use logic to create a myriad of functions. If I were to try to start listing the zillion places you can find all varieties of discrete logic it would go on forever, and thats just my personal experience. They arent the vacuum lines of cars from the 80's that got replaced by better engine control and intrinsic design, they arent _anything_ that can be specifically nailed down or categorized, thats why they have such incredible longevity, its because they arent any one thing, they can be made into anything and can fit into unique circuit design niches and do a fantastic job at it.

Heres an example, the 4046 PLL chip is still widely available in CMOS. You can go on digikey and find it in a variety of topologies and available from 5 different manufacturers, the cheapest being 60 cents at single quantity. This isnt the kind of part you are going to find shuffling data along a logic line. This is a specialized, unusual component thats only going to be used in analog circuitry or very clever and special digital circuitry. And its 40 years old. Thats even a surprise to me. But just because it seems like it should be that discrete logic is dying out, its not. And that really isnt contradictory to anything because if I look at all the circuitry I have designed, and the circuitry other engineers I've worked with have designed, in my experience, I see discrete logic all over the place in all kinds of roles.

[Mariss Freimanis]

The 4046 is a quasi-analog IC and not discrete logic in the conventional (nothing but gates) sense. I'm very familiar with it and implement its Type II phase comparator in Verilog. We probably agree more than disagree; I'm not a fanboy of any fixed point of view. If logic is called for, I'll try to use discrete logic up to about 5 packages. If more are necessary to achieve an objective, I'll weigh it against other alternatives. Past 10 or so packages, programmable logic becomes overwhelmingly attractive. For simple glue-logic requirements (< 64 D-Flops), CPLDs are a good choice; for anything bigger, an FPGA is the better choice. Other tasks by their nature may indicate an MCU solution. Also let's not leave out my personal favorite, analog circuits; a good example is an an op-amp implemented PID filter.

My main point was I don't have as many choices as I once used to and that limits my flexibility in deciding for a discrete logic solution. In the 70's minicomputers like the PDP-11 were built using small-scale 7400 logic ICs; I don't think any sane person would try that today.

Mariss

[greybeard]

OK gents, I think I'll call it a day.
Thanks all for your contributions.

Regards,
John

[vegipete]

Acannell observed (roughtly) that "95% of forum denizens tell you you are wrong. 5% actually get enthusiastic and help."
Where do I stand?

Well, I still say an MCU is the way to go. Indeed, a problem like this got me into microcontrollers in the first place. I wanted a circuit to control the brightness of the light on my mountainbike, with buttons for brighter, dimmer, max and off. The logic chips were getting hairy but a simple PIC12C508 (yes, a C version, I've still got a couple of windowed ones around 15+ years later) solved it and even added more features like flash mode and an illuminated switch lever when the light was off.

However, to put my money where my mouth is, I cobbled a circuit together using 75LS191 chips. I fired up CircuitMaker (you can download a student version here) and bodged in the attached circuit. It basically works although it needs cleaning around the edges, such as proper counter setup when the reset signal arrives.

(Hopefully obvious but the blue wires are logic 0 and the red ones are logic 1. :-)

Top left is a clock generator. The switch banks on the left set the input values. The semi-connected resistors on the right are just so there is a wire connected to various pins in order to see the state of said pins. The NAND gate top right should be an AND gate instead, the output of which is the N x K pulses.

The upper 'LS191 continuously counts down from the load value N and resets itself back to the load value at zero (or -1?) Each time it hits zero, it sends a pulse to the lower 'LS191. Thus the lower one counts K times the upper one. However, it does not reset itself. Instead, the extra gates inhibit the output pulses once the lower counter hits zero, and prevent the lower counter from seeing more clocks until reset.

Cascading the counters will get you as many bits for N and K as you desire.

I can post the CircuitMaker file if someone wants it.

[greybeard]

Vegipete, I am deeply grateful. You have given me an answer to my question that a) I can understand, and b) that I am perfectly capable of building and expanding as necessary.
Current price of the 191 <£5 for two

PM on its way.

John