Help with jfet switching...

[david1991ross]

I'm trying to figure out this jfet switching and I'm having some issues. I'm not sure where my issue is so I put my voltage measurements throughout the circuit. My main issue is that there is no flip flop. The voltages don't change regardless of pressing the switch. I have my transistors all socketed and when I pull Q4 I get the LED to come on, otherwise it's off. I do get a very faint flash from the LED when it press the switch but it's next to nothing. I can confirm that the SPST footswitch does work.

Does it matter what NPN transistors I use? I'm using a combination of 2N4123's for the input and output buffers and 2N5088's for the flip flop and LED.

Does anyone see any mistakes?

Where should I begin my troubleshooting?

Thanks!

[Rob Strand]

I think you are missing a resistor on the base of Q2.  The transistor driving the LED.

(BTW, the low 2.4V on the base of the input buffer could be due to the meter loading, *but* the low emitter voltages means it  actually is low.  Check you haven't got E and C swapped on the pinouts.)

[PRR]

Or Q2 shorted B-E?

[Vivek]

Is this MicroCAP ?

I have just downloaded the software and could possibly assist with the simulation shortly

[Rob Strand]

I have just downloaded the software and could possibly assist with the simulation shortly

I'm fairly sure the missing Q2 base resistor will cause problems from the start.

You can see the correct circuit here,

http://4.bp.blogspot.com/-W90Rsg7YA9Y/T75MYyT9szI/AAAAAAAABZE/SrJ-EsvSkNg/s1600/Ibanez+STL.gif

[Vivek]

My very first circuit that I built when I was 14, was a similar flipflop. I made it with AC128 on vero. It had 2 little filament bulbs. I think I kept it on for about 6 months, always in amazement at what electronics could do !!

So this circuit brings back happy memories !


Yes it does appear that the missing part made the circuit into a flop,
instead of a flip-flop !!!

[Rob Strand]

My very first circuit that I built when I was 14, was a similar flipflop. I made it with AC128 on vero. It had 2 little filament bulbs. I think I kept it on for about 6 months, always in amazement at what electronics could do !!

I think the astable versions were some of my earliest projects.   Flashing lights and high-voltage zappers (using a transformer).   The self-oscillating zappers were less troublesome but you had to scrounge around for a suitable transformer.

[david1991ross]

I added a 1Meg resistor to the base of Q2 and the LED turned on, unfortunately now it won't turn off when I press the switch. I'm going through the circuit and I see one mistake, a 10k resistor was in with my 22k resistors and I placed that for R7. I'm going through the rest of the circuit now for any more mistakes.

Edit: I can confirm that that's the only stray component.

[Vivek]

Do you have same value for R14 and R20 ?
The schematic that Rob posted has different values

[david1991ross]

Do you have same value for R14 and R20 ?
The schematic that Rob posted has different values

I do. I've looked at various schematics and I'm not sure if that value makes a difference. I suppose I could give it a shot.

[R.G.]

I've always found the two transistor flipflop a little touchy to make work perfectly. There's always some little thing to be tinkered. I find that it helps me to think of it as two common emitter amplifiers in series, with a wire connecting the output of one of them back to the input of the previous one.

You may already have read it, but if not this might help: http://www.geofex.com/Article_Folders/bosstech.pdf

[david1991ross]

I've always found the two transistor flipflop a little touchy to make work perfectly. There's always some little thing to be tinkered. I find that it helps me to think of it as two common emitter amplifiers in series, with a wire connecting the output of one of them back to the input of the previous one.

You may already have read it, but if not this might help: http://www.geofex.com/Article_Folders/bosstech.pdf

I have read that article and found it informative. I've decided to go back to the breadboard and figure out the issues. It came to my attention that the pinout of the flip-flop transistors in the original schematic were ECB and not EBC which seems to account for some of the issues I was having. With that said, I breadboarded both and got the following results:




I believe the top one is correct, however, when I press the switch there doesn't appear to be any change in voltage when measuring at the outputs. I'm not sure why this is. I've yet to add the other NPN transistor for the LED or the jfets until I have this part figured out. Any idea on why I'm not seeing a change in voltage?

[R.G.]

I got frustrated with staring at the circuit and went off to the circuits machine and dumped it into the simulator.
You're right - with values as shown, it doesn't work in simulation. Your results are correct.

I did mention that it was always some value or other that needed tinkering, right? The amount of base drive looked excessive, so I tinkered with that. Try changing the 47K base drive resistors to 220K and the base pull-down resistors to 1M. That got it working in my simulator. I used 2N3904 models for the transistors.

One comment. I don't know if you have the transistors hooked up wrong or mislabeled. Irrespective of where the pins are on the package of the transistor, the one that is the collector has to go to the 56K resistor, the one that is the base has to go to the was-47K-but-now-220K resistor. The labeling on the schemos seems wrong.

The too-low base resistors lead to a 4.6-ish collector voltage on the "off" transistor, as you measured. That rises to over 6V with the bigger base feed resistors.

[david1991ross]

So I've still been working on this and made some progress with the suggestions in this thread and from elsewhere. Here is what I have:



It's working reasonably well on the breadboard however it's not perfect. I'm using a soft touch SPST footswitch in a 1590LB enclosure with a standard cable connected to the board. The switching works (as in the LED turns on and off) between 50% and 75% of the time. I'm aware that the switching for a soft touch SPST footswitch is not going to be 100% accurate, however, it is significantly less accurate then I would hope and I'm not quite sure why. I've played with a number of values, increasing and decreasing, but the accuracy remains where it is for the bistable multivibrator portion, and when I add the jfet switching. I should add that the switch is practically brand new and I have no reason to suspect that it's faulty. Does anyone have suggestions on where to proceed?

[Rob Strand]

I got frustrated with staring at the circuit and went off to the circuits machine and dumped it into the simulator.
You're right - with values as shown, it doesn't work in simulation. Your results are correct.

Spice often biases these things up with equal collector voltages.   There's no asymmetry for it to decide to go one way or the other (except perhaps the 1M output load).  You need force an initial state where one of the transistors is on.   After that button presses will toggle the flip-flop.

One way to start-up correctly is to set some initial conditions on one or more of the caps.  Another is to create a start-up circuit which pulses some current into one of the bases of the transistors, for example a spice current source set to narrow pulse say 10us.

The fix will only show-up on a transient analysis, not a DC bias.

If you it to start-up on a DC bias you will have to make some of the resistors unequal to create some asymmetry.   Off-hand, by a reasonable amount.

I might have posted a sim for the Boss unit in the past, it's probably got the current start-up pulse method.

[Vivek]

http://4.bp.blogspot.com/-W90Rsg7YA9Y/T75MYyT9szI/AAAAAAAABZE/SrJ-EsvSkNg/s1600/Ibanez+STL.gif

R21 and R22 are different values here, possibly to pre-determine a start condition ?

[R.G.]

Spice often biases these things up with equal collector voltages.   The no asymmetry for it to decide to go one way or the other (except perhaps the 1M output load).  You need force an initial state where one of the transistors is on.   After that button presses will toggle the flip-flop.

Yeah, I'm familiar with the oddities of SPICE startup. The "nothing starts" problem was even more infuriating back when we had to submit text jobs and get results printed on watermelon paper an hour or so later.   

However, in my runs, the circuit did in fact start up with one collector down at about 0.2V, one at between 4 and 5 volts. It just would not switch until I raised the base resistor values.

I also messed with the relative values of the trigger and speed-up caps to see if there was some race condition that happened to be just perfectly balanced and prevent switching because the values were exactly the same. That didn't help either, although slight offsets in the specified component values is another way that can trick SPICE into an initial transient, as well as using close-but-not-quite transistor or diode models in digital simulations to stop the too-perfect condition. These dodges were published in the (archaic!) instructions for the older SPICE versions we used inside the corp.

[Vivek]

Could it possibly have anything to do with the SPICE option "Skip initial Operating Point Solution" ?

Have you got this ticked or unticked ?

[duck_arse]

your last shown version is missing the cap to ground from the C8//C10 node. it might be switch bounce. on the breadboard anyway, does spice include bouncy switches?

[Vivek]

LTSpice hardly has any physical switches built in.

One has to make models for all physical switches. It's cumbersome to make and use switches in LTSpice.

The way LTspice works, it seems impossible to "press a switch" in the middle of a simulation run.

Spice has Current controlled and Voltage controlled switches though


Microcap might be different, I have not gotten up to speed on it yet.