I've a problem that should be simple to solve, but I must be missing the
glaringly obvious, as the solution eludes me.

This is a discrete monostable divide by 5 countdown circuit being driven by
pulses every 1mS on the input, and should drive a pulse every 5mS at the
output.

I've checked the transistors (and subbed them just in case), the diode, and
the resistors, and replaced the 0.1uF timing capacitor, but still the darn
thing triggers almost randomly (though there is some preference for 5mS).

Here's the circuit (resistors using : characters) - use a monospace font to
view.

+12V
|
:
:3K6
:
|
------------------------o
| |
| |
| |
| +12V +12V |
| | | |
| : : o--||--o--o-- Output
| : 2K2 : 61K9 | .2uF | |
: : : | : |
39K0 : | | |/ 9K1 : --|--
: o--||--o-o--|--o--| : |
| | 0.1uF | | |\- | ___
| |/ --||-- | ___ .
-|-:::::-o-| 22pF ___ .
10K0 |\- .
|
___

Wow, been awhile. It sounds as if you need a 5 stage ring counter.
(pausing remembering just how to do it)

Using 5 J K flip flops you hook all the Ts together and in a ring (thus
the name) you tie the J to the Q and the K to the not Q of the previous
flip flop.Output at any of the five phases can be tapped at any of the
Q terminals for a 20% on time, or the not Q for 80% on time.

How far in the cowbwebs of my mind did I have to go ? The last time I
thought about ring counters was when I was builing a bootleg cable
decoder for the old SSAVI2 scrambling method.

There are also programmable dividers if you want a one chip solution.
With 7 terminals per FF, you only can get two in a 16 pin DIP so this
means three chips minimum. If you go with the ring counter you don't
use the other two, C and cD I think, you should consult the datasheet
for whatever chips you use to see whether you should ground them or
leave them float.

To sum it up, digital division is the only way to get reliable
operation. If you wind up with three dual flip flops you can also use
the last one to get on times other than 20% if desired.

Good luck with it.

JURB

If I were starting from a "blank sheet", I probably would use a digital
divider chip or several. BUT, this is an attempt to repair to a venerable
piece of test equipment (Tektronix 184 time mark generator), which I'd like
to keep pretty close to original ...

TIA
Dave

"David C. Partridge" wrote in message
...
I've a problem that should be simple to solve, but I must be missing the
glaringly obvious, as the solution eludes me.


You've probably checked the obvious but is the supply to the divider the
right voltage, well decoupled and free of glitches/ripple?

Dave

"David C. Partridge" bravely wrote to "All" (14 Nov 05 20:38:19)
--- on the heady topic of " Monostable divide by 5 not working."

DCP Reply-To: "David C. Partridge"
DCP Xref: core-easynews sci.electronics.repair:348583

DCP If I were starting from a "blank sheet", I probably would use a
DCP digital divider chip or several. BUT, this is an attempt to repair to
DCP a venerable piece of test equipment (Tektronix 184 time mark
DCP generator), which I'd like to keep pretty close to original ...

DCP TIA
DCP Dave


Geez, you guys, this student just wants us to do his homework for him!
Actually, the circuit is simply a monostable which is not retriggerable.
Basically, it's a pulse stretcher. It starts by discharging the 0.1uF
cap through the base resistor of Q2. Because the base is driven hard
negative, a series diode has been added to protect Q2 from reverse
bias breakdown.

The trouble with the circuit is that the input pulses might not be
properly biasing Q1 when high or low and then the trigger point is
constantly changing. Also if the input is driven through a capacitor
then its time constant may be further randomizing the trigger point.

A*s*i*m*o*v

.... When I was your age, we carved transistors out of wood.

Actually, no, not a student.

D.

Thanks, will check that again, may well be relevant. All the other
countdowns are working (at least all the way from .1uS to 1mS). It's the
5mS that's all over the floor.

Dave
"Dave D" wrote in message
...

"David C. Partridge" wrote in message
...
I've a problem that should be simple to solve, but I must be missing the
glaringly obvious, as the solution eludes me.


You've probably checked the obvious but is the supply to the divider the
right voltage, well decoupled and free of glitches/ripple?

Dave

"Asimov" schreef in bericht
...
"David C. Partridge" bravely wrote to "All" (14 Nov 05 20:38:19)
--- on the heady topic of " Monostable divide by 5 not working."

DCP Reply-To: "David C. Partridge"
DCP Xref: core-easynews sci.electronics.repair:348583

DCP If I were starting from a "blank sheet", I probably would use a
DCP digital divider chip or several. BUT, this is an attempt to repair
to
DCP a venerable piece of test equipment (Tektronix 184 time mark
DCP generator), which I'd like to keep pretty close to original ...

DCP TIA
DCP Dave


Geez, you guys, this student just wants us to do his homework for him!
Actually, the circuit is simply a monostable which is not retriggerable.
Basically, it's a pulse stretcher. It starts by discharging the 0.1uF
cap through the base resistor of Q2. Because the base is driven hard
negative, a series diode has been added to protect Q2 from reverse
bias breakdown.

The trouble with the circuit is that the input pulses might not be
properly biasing Q1 when high or low and then the trigger point is
constantly changing. Also if the input is driven through a capacitor
then its time constant may be further randomizing the trigger point.

A*s*i*m*o*v

... When I was your age, we carved transistors out of wood.



Homework? It's the most primitive divide by five counter I ever saw. Can't
imagine a master to invent such a thing for a question. Actually, as this
group is about repair, I guess it's an existing circuit. This type of
circuit is very sensible. Noise and/or disturbances on input or power make
it fail easily. It really needs uniform inputpulses. As I see no possibility
for adjustment *and* about all parameters of all components influence
working, I guess the components need to have high accuracy. The original
transistors might as well be specially selected. Aging of components,
especially the transistors, might be caused the problems. As an aside, you
may need to check or replace the decoupling capacitors of the power supply
near the circuit.

petrus bitbyter

Agreed it is primitive, but it is an existing circuit put there by
Tektronix, and yes, the component tolerances are critical on the 0.1uF
capacitor and the 31.9K resistor - if they are off, you get a divide by four
or divide by 6 (typically). Here though it looks like anything from 3 to 7
based on scoping the output, and trying to scope the 0.1uF cap results in
nothing much that I can interpret as it won't trigger stably.

In this case, based on what folks are saying, is seems plausible that the
local supply decoupling caps (which are after all quite old by now) may be
dying/dead. Alternatively the 1mS input pulses my not be as regular as I
think (even though they 'scope up very clean).

Dave

"petrus bitbyter" wrote in message
l...

"Asimov" schreef in
bericht ...
"David C. Partridge" bravely wrote to "All" (14 Nov 05 20:38:19)
--- on the heady topic of " Monostable divide by 5 not working."

DCP Reply-To: "David C. Partridge"
DCP Xref: core-easynews sci.electronics.repair:348583

DCP If I were starting from a "blank sheet", I probably would use a
DCP digital divider chip or several. BUT, this is an attempt to repair
to
DCP a venerable piece of test equipment (Tektronix 184 time mark
DCP generator), which I'd like to keep pretty close to original ...

DCP TIA
DCP Dave


Geez, you guys, this student just wants us to do his homework for him!
Actually, the circuit is simply a monostable which is not retriggerable.
Basically, it's a pulse stretcher. It starts by discharging the 0.1uF
cap through the base resistor of Q2. Because the base is driven hard
negative, a series diode has been added to protect Q2 from reverse
bias breakdown.

The trouble with the circuit is that the input pulses might not be
properly biasing Q1 when high or low and then the trigger point is
constantly changing. Also if the input is driven through a capacitor
then its time constant may be further randomizing the trigger point.

A*s*i*m*o*v

... When I was your age, we carved transistors out of wood.



Homework? It's the most primitive divide by five counter I ever saw. Can't
imagine a master to invent such a thing for a question. Actually, as this
group is about repair, I guess it's an existing circuit. This type of
circuit is very sensible. Noise and/or disturbances on input or power make
it fail easily. It really needs uniform inputpulses. As I see no
possibility for adjustment *and* about all parameters of all components
influence working, I guess the components need to have high accuracy. The
original transistors might as well be specially selected. Aging of
components, especially the transistors, might be caused the problems. As
an aside, you may need to check or replace the decoupling capacitors of
the power supply near the circuit.

petrus bitbyter