In another forum, I'm trying to keep a circuit down to a single chip. I have 1/2 of a dual ff available and need an oscillator. This is a non-standard variation of a classic two-transistor multivibrator. Will this circuit oscillate reliably? It will start up with the Set input clamped high, then (I hope) oscillate when the clamp is removed. I'm kinda fried right now and need a second opinion.

Thanks.

ak

 

My simulation shows a problem due to the small hysteresis between set and reset (at least in the simulation model), causing it to oscillate at a high frequency.


Adding a couple diodes across the resistors to cause a rapid discharge of the caps, allows it to oscillate at the expected frequency.

 

Set will force Q high regardless of the state of reset. If you hold set high long enough, reset will also go high which will force /Q high regardless of the state of set. This becomes a stable state with both S & R asserted and Q & /Q high.

 

A look at the internal schematic shows that the paths from Set to Q and Reset to Q- are significantly different. I knew the diodes would work but hoped I wouldn't have to run up the body count.

Thanks for your help.

ak

 

AK, I breadboarded your design and didn't see any high frequency oscillations without the diodes but the frequency after a few cycles on power up was about twice that of the circuit with the diodes.
SG

 

I vaguely recall raising a query with Helmut at LTS a couple of years ago, re an apparent discrepancy between the 4013 spice model and the datasheet concerning what happens when Set and Reset are held high at the same time. IIRC the model was not revised. Treat it with caution in simulations.

 

I remember someone supposedly got forged CD4013 ofcourse as with all CMOS and TTL theres is obscure and probably not even so much obscure circuit which work with a specific IC (manufacturer) but dont with some other specific kind.

I could imagine why. Theres some fundamental problem as most the circuits are based on simplifications some kind of scheme what the IC should represent but theres always certain variants of one and the same IC, internally they are different, theres new revisions and so forth, buffered and not.

Theres always some glitching in the nanoseconds and subnanoseconds which you may not even see on oscilloscope and even with decoupling, you still have inductive traces (but normally you dont bother some nanohenry)

Mostly one branch of propabilities wins but not neccessarily and especially over time, digital circuits can fail in strange ways and moreover can exihibit weird spurious errors.

I did play around with these IC for some while and found ICs from one manufacturer simply didnt work properly, another manufacturer, worked 100%. The ICs werent fake back then, and the circuits not all that weird but Ive experienced it myself.

I had a Pentium Pro mainboard where the Pentium was a little latched, sometimes when inserting a bus card it didnt start up anymore.
But removing the IC and touching it everywhere, always worked like magic.

Then of course aging PC boards which after a while would just gardually creep more and more, youd see strange windows errors fails and reboots no matter what, and over some time theyd increase.