Hi All,
I have lurked here for years reading and learning constantly. I work during the day at a hospital, and am good friends with one of the researchers where we sometimes collaborate. In this case we want to time a very tiny event, and since this is on the side, we have no real funds. I do have a pretty well stocked project corner.

I have already started assembling the project, and some photos will be forthwith when I have a bit more time. I figure this is a great learning thing and so others can learn from the thread as I progress.

I have done lots of small one offs. Timers, one shots, some pic stuff, but do not want to use pic for this.

I am shooting for a 10mhz clock speed, and do not want to use a pic micro.
So think of it as a very accurate stop watch. I have Vishay clock timers in various speeds, one being 10mhz. (I also have max high frequency waveform generator, and if I get the counting block working well, I may turn to MAX038 to give a wide range of selectable time bands).

Ultimately I am trying not to buy any parts as part of the challenge. J

And the main impetus on this is that I am trying Manhattan Construction, and so far, for me it’s great!!! That is, I can spend a lot more of my time at a work table rather than with calculator, simulator, looking up stuff, etc. I solder very well and fast. I cut my nodes with a jewlers saw. Use a touch of sandpaper on the end of an eraser to prepare the board, and use cyano acrylate. I am a bit lazy when it comes to flux, so I use what stained glass people use, and that’s a solution of zinc chloride. A jar from stained glass supply, and your set for life (just make sure to clean your board when your done soldering). By using this one can solder a joint in under a second. Great soldered joints every time. Been using it for years, never had circuit fail on me. Saves me time cleaning the board (will work on oxidized).

[I am still new to Manhattan construction, and am developing my own style. How new? How about one week new. But I have been able to build more circuits in this one week than I usually do in a half a year or more given time constraints and effort required. Its turning out so far to be Manhattan construction with a mix of all different kinds of parts, including 1206 capacitors. Dead bug too, with a bit of using a wire wrap tool to get wire tightly on the pins to be soldered]

Power for the device can be a wall wart ranging from 6 to 24 volts (or in a pinch I can use a lantern battery). I have already assembled a standard power jack to regulated supply using LM317M. I can add another regulated stages if I have to (and increase the supply).

Right now I have selected SN74LS160S synchronous decimal counters for the main part.
The counter will have seven places to start with as that’s what fits on the board.
Though using Manhattan method I can just glue an extension to the board and add a few chips (which is another reason to regulate each row of chips)

Their output will end up connecting to DM74LS47 decoders (with open collector outputs). I have both common cathode and anode numbers.

For a clock I am using a vf150 I removed from a router. It’s a ttl compatible clock oscillator in a can. I have a bunch of them, but I am eying a 1 mhz, and 10 mhz (hopefully being able to drive my event timer with that). given I can change construction,
I was going to start with a 555 timer board I have and use that as timer input to test the circuits operation, then make a timing block.. using two octal D flipflops as divide by.

Using different outputs of the flip flops will give the stop watch different bands of timing operation. (of course with declining accuracy).

The other part I will have to work out is how to make detectors to allow for accurate trigger and release of events…but I will worry bout that when I get to it.

Right now I am trying to figure out if I need to add another regulator for the ls47, or can I power all 14 chips with the one LM317, and use a different regulator (higher) for LED voltages…

The other thing is trying to see what the circuit for trailing 0 removal taking certain pins low by connecting them from highest digit to lowest.

Whats a good way to do quick calculations as to how much mA the chips are going to require. Whats a good rule of thumb?

Boy is this FUN!!!
and yes I want the counter to be counting up, it’s a stop watch, not a alarm timer
we ultimately want to time an event that happens at about a 10000th a second and just have to show that the idea is good enough (knowing that there are much better electronics people and chips available).


We figure 10,000,000 cycles should give us a close enough measure for 10,000th of a second… if not.. then I will have to make something better using other chips.

the above is how the voltage regulator turned out...

and below is how far i got on wiring the power...

 

Interesting. Do you have a schematic (I'm betting you do)?

A bit ugly, but it does look good so far.

Look for 30 gauge wire wrap wire to do the wiring. It will be a rats nest when you're finished, but that isn't a big deal.

 

I didnt schematic the whole thing... kind of want to learn to do a bit more without that. what i did lay out was the counters in workbench as those were where the trickiest thing would be first.

the simulation went well... but had to add extra circuitry to put numbers in so that i could check roll overs. i have an old copy of workbench and if it runs too long on xp it fails... part program problem part circuit problem. that is if i get it right, it goes for a long time.

previously i had modeled the lm317m circuit. and doing it this way i can look at it on the scope.. i only have two voltmeters (one digital, one old timer with a sweep (has uses digital doesnt)), LC meter, and a 25 meg used scope... but i do ok.

i do a lot of playing with coils and magnets so magnet wire is not a problem.. one reel i have techically can run a line around the world twice... (thin as a hair though).

i didnt think of using the magnet wire... but then again, its not like the coating on either plays a bit part in mitgating magnetics...

i thought about the ugly part...
but didnt think it was too bad given i am mixing everything in but the kitchen sink.

is it normal to be planning version 2 in the middle of version 1?

in programming i dont do that much... been doing that too long and so i just move to the thing... and there isnt much i can do about parts 2..

thanks for answering my post on the level.

i will post more when i have a bit of time..
its fashion week, and that takes my energy
[i am one of those odd eclectic ny kind of guys that has his day job and all kinds of other things outside the norm outside of that. and i tend to do them serious enough to be published, paid, etc... ]

i was also posting to see if someone else catches a big faux pas or something odd right out of the gate... dont have a lot of these chips, so this is pretty much a one shot deal...

i think its pretty straight forward though...its the comparator circuit to turn it off and on that i think will be more of a problem... it needs a high slew rate to be as accurate as possible given the tiny block of time we are going to attempt to get near with our twiddlings.

thanks again! will post more in a day or so... including a better white balanced image. [maybe that will make it a bit less ugly ]

 

The problem with not having a schematic is that it makes it virtually impossible for any of us to assist you with any problems you may encounter with your circuit. Hopefully you will not postpone creating a schematic for much longer.

hgmjr

 

A schematic would have been a major aid in laying it out. You have put the cart before the horse. If you forgot something it will be much harder to adjust and fix.

 

we ultimately want to time an event that happens at about a 10000th a second and just have to show that the idea is good enough (knowing that there are much better electronics people and chips available).

Is the event(s) a single one off or repeatable at regular or random interval?

What exactly do you want to time? The duration of the event or the duration between happening of the event(s)?

So far you have mentioned 100uS (1/10000th a second) but how accurate do you want the result be, ±1uS or what?

 

Hi,
I think some here are missing the point of the exercise. I AM working off a loose plan (that is different areas are modeled). I had been just too busy this week to get it into an image and put it up (right now i am working in the Honey Do patch. )

as an example. i had previously created a spice part in workbench to do the lm317m circuit. now i use that as a circuit block, which you can see in my photos. when doing a logic circuit in workbench it leaves out all the power handling, grounding, etc... so you really only work with the timing, logic, clock, connections. so in order to get the clock down right, i modeled the decade counters (and i will put that up later tonight i hope).

once i had that wiring working, with the extra circuitry to load a number so that i could get to rollovers/carries and make sure that is working the way that they should as far as proper wiring.

i did NOT model the binary output to the ls74, as that's straight forward. if my other chips are counting, then that assembly will work too. (given the caveats of course)

so while there is no full plan, its not a super complicated thing

i am used to working on much more complicated things in software, design, and physics. i have been lucky as almost every circuit i have ever tried has worked (and that was before i got the L/C meter, and scope!). but these are tiny things, and this is a bit more involved (and for a project at the med school i work at), but only in moving me to digital. have used basic stamp and other versions, am going to probably move to pic chips next.

ok.. as to the last posters questions...

Is the event(s) a single one off or repeatable at regular or random interval?

its measurements of a single event of a duration.
so in this case, the circuit will turn on, tick off the clock counts, and turn off.
it will count up. AND i have a long list of where all the different timing errors will collect. like what part of timing tick did it start in before the uptick was hit is one margin of error. and there are many more that will add up and create uncertainty.

however in this case good enough is good enough and rather than fret over all that, we are just going to slap it together and test time the event. its actually taking less of our time to make it and test it than sit and compute every little thing. sometimes back tracking is easier than anticipating. (and if i have to make another one, i am going to buy some binary counters to say 64bits and then translate their final state by pic chip, and use two 4 character LED modules with a lot less leads... so there is always version II )

.What exactly do you want to time? The duration of the event or the duration between happening of the event(s)?

duration of the event... so basically timing the on portion.

thats the next stage to work out and THAT will be the hard one for me
not so much the digital. but i wanted to put the whole process up, rather than just a finished thing, that way good ideas can come up, and others who are around this level can learn with me.
[been reading here for a while, and in a way this is my give back]

So far you have mentioned 100uS (1/10000th a second) but how accurate do you want the result be, ±1uS or what?

now we are getting to brass tacks.

i have worked the duration of the event out to float around 35,000ths of a second... so technically the more decimal places that i can add the better it is. (thanks nyquist)

I took the time this morning to rework the numbers and post the results in another post down below
its actually a better case than the above range, however maximal decimal places makes for better average accuracy
and so will shoot for 10mhz (maybe 20mhz) clock if i can.



however, the more decimal places i add, the harder it is to get it right.

my scope is limited to 25 mhz...
so i was going to test the clock stuff with a 555 circuit i slap together to get the numbers to roll... and THEN replace that with a more accurate clock. i have a nice 10mhz TTL clock can i intend to use, AND a 1mhz clock can (i even have a 145mhz one.. but thats out of my league)

the LS chips have a max around 30mhz... so i was fudging to a good speed under the max that was doable with limited equipment. and just test it.

given that this is also the first time using manhattan construction, and i like it as i was able to do a lot, i am not too worried if i have to build another better one and spend the money. right now its from lots of spare parts i have... pretty much zero cost...

so what i am intending to do is use a comparator to zero out a detector signal against itself, and so any disturbance above the noise floor will bring it high, and so create a time event. (which is why the clock will have a reset, but no load numbers).

i will probably have to buy a comparator, as the 339 (?) i have is too slow a slew rate taking too much accuracy away.

but its an adventure too!!! no?

 

Counter Wiring Schematic



oh.. and before anyone wastes time (please don't!)
there will not be a load number circuit in the real device, so the values of things connected to that dont matter. (1k i know, but i had a lot of 4.7k)...

Also the termination of the clock line is wrong (any hints?)

I usually have lots of purposes in the things i do (when i am not working and using my time)
and so its usually a mix of learning, pushing a limit, being useful, making others happy....
I would appreciate not letting that get around

in this case i work where there are researchers, and believe it or not their
resourcefulness in making tools is quite, shall we say, limited.
this is not to say they are not smart able people, its just what mix they generally have.
so they can think of things they want to examine, but often do not know how to
build something to help them even test whether its worth examining (if thats not too hard to understand)

think of this as my personal crusade to lower the costs of (med) research by making rather than consuming

I do lots of interesting things despite no one believing me...
[not really my problem, is it?]

thanks for the hand holding!!!

Will post more later...

 

Results of reworking the duration range that is being measured.


the total duration of measure (longest) is lower than .006

the smallest duration of interest is .001 wide, so that about 10k clock ticks
[basically the part that will vary between tests]

and i am hoping to use 2 decimal places (100 divisions) as being fairly accurate
(and as a free test is worth then justifying doing more with a better measure/circuit with a better EE (which i am not))

 

It wasn't a problem in disbelief, more of an experience issue. I have a lot of incomplete projects behind me because of planning. I find if I draw it down first it is much more likely to be completed. I am also more likely to find problems before I make them (no guarantees though).

 

i am NOT disagreeing with you bill, i think your right!
but i also want to get a more intuitive feel for what i am doing without having to have the other stuff around. i want to be able to work on a peice of paper and know it will be close enough once i work it out... and seat of my pants has always been my best way to get the feel for what i am doing. i do ok that way most of the time. and when i dont, i just have to work harder longer.

 

oh, and the disbelief thing wasnt to you or any one in particular. its tends to happen if i say what i spend my time doing. more of a general thing (like having to finish up the photog at fashion week before i could get back to the project).