Astable Multivibrator Generates Sub-Ground Voltages

[Jeff]

Hello,

I've been simulating a circuit of the 'classic' astable multivibrator
and analysing the results. This is the old-fashioned two-transistor
design where each device (both NPNs) alternately turns off/on the
other. I'm sure you know the type of thing. Anyway, the collector
voltage of Q2 goes negative to almost the extent of the supply rail! Is
this a Spice anomaly or would a real world circuit do the same? I don't
see how it's possible for a circuit to generate sub-ground voltage
levels. Isn't that theoretically impossible? The clock rate is 400Hz
and Vcc is 5V.

-Jeff.

 

[Jim Thompson]

Hello,

I've been simulating a circuit of the 'classic' astable multivibrator
and analysing the results. This is the old-fashioned two-transistor
design where each device (both NPNs) alternately turns off/on the
other. I'm sure you know the type of thing. Anyway, the collector
voltage of Q2 goes negative to almost the extent of the supply rail! Is
this a Spice anomaly or would a real world circuit do the same? I don't
see how it's possible for a circuit to generate sub-ground voltage
levels. Isn't that theoretically impossible? The clock rate is 400Hz
and Vcc is 5V.

-Jeff.

Which "Spice"? Can you post a schematic, either ASCII, or a real
schematic (preferred) to alt.binaries.schematics.electronic ?

...Jim Thompson

 

[Mike Engelhardt]

Jeff,

I've been simulating a circuit of the 'classic' astable
multivibrator and analysing the results. This is the old-
fashioned two-transistor design where each device (both NPNs)
alternately turns off/on the other. I'm sure you know the type
of thing. Anyway, the collector voltage of Q2 goes negative to
almost the extent of the supply rail! Is this a Spice anomaly
or would a real world circuit do the same? I don't see how
it's possible for a circuit to generate sub-ground voltage
levels. Isn't that theoretically impossible? The clock rate
is 400Hz and Vcc is 5V.

I wouldn't expect the collectors to go beyond the supply
rails, but the bases will go negative. Since the B-E
breakdown isn't usually modeled in SPICE, this can give
some un-realistic waveforms in simulations.

--Mike

 

[Jeff]

Hi and thanks.
I've just double-checked the simulation and it's actually the *bases*
of both devices that go negative to -4.3V rather than the collectors,
so my apologies. But the question still remains about how *any* node in
such a circuit can go below GND. I will post the schematic if the above
correction still doesn't account for what I'm seeing.
-J.

 

[Jeff]

Hi and thanks.
Yes, it's the bases that are going negative. I'm not sure why you
mention "breakdown voltages" since we're only talking about a Vcc of 5V
and I know of no transistor that breaks down at such a low reverse
Vbe., nor why a spice model should fail to take account of it if such
should be the case. It seems like a pretty significant flaw in the
model if so.
BTW, in answer to Mr. Thompson's question, I'm using LT Spice.

 

[Mike Engelhardt]

Jeff,

Hi and thanks.
Yes, it's the bases that are going negative. I'm not sure why you
mention "breakdown voltages" since we're only talking about a Vcc of 5V
and I know of no transistor that breaks down at such a low reverse
Vbe., nor why a spice model should fail to take account of it if such
should be the case. It seems like a pretty significant flaw in the
model if so.

It probably won't breakdown at 5V. But it might around 6. Usually
transistor breakdown isn't modeled because the normal design flow
is to simulate to find what the peak voltages are and then buy/fab
a transistor that won't breakdown in the circuit. If the transistor
breaks down in simulation, then you don't know what kind of transistor
you need to buy/fab. Transistors usually aren't used in a manner that
includes breakdown, because that has a permanent effect on the device
(though the effect from B-E breakdown is not too dramatic). But
basically, if you simulate with transistors that breakdown in simulation,
you might want to think for a second and make sure your not designing
the circuit to fail.

BTW, in answer to Mr. Thompson's question, I'm using LT Spice.

LTspice comes with an example astable circuit typically installed
as C:\Program Files\LTC\SwCADIII\examples\Educational\astable.asc

--Mike

 

[Jeff]

Okay, thanks for that.
So it appears it's either a model deficiency or real-world capacitive
effect. I need to know which, so here's the netlist of the circuit
concerned and I invite comments on it:

"ExpressPCB Netlist"
"SwCAD III Version 2.16i"
1
0
0
""
""
""
"Part IDs Table"
"Q1" "2N3904" ""
"Q2" "2N3904" ""
"V1" "5" ""
"R1" "1k" ""
"R2" "1k" ""
"R3" "15k" ""
"R4" "15k" ""
"C2" ".1µ" ""
"C3" ".1µ" ""
"R5" "100k" ""

"Net Names Table"
"N005" 1
"N003" 5
"0" 8
"N002" 12
"N004" 15
"N001" 18

"Net Connections Table"
1 1 1 2
1 5 2 3
1 9 1 4
1 10 1 0
2 1 2 6
2 6 2 7
2 8 1 0
3 1 3 9
3 2 3 10
3 3 2 11
3 10 2 0
4 2 1 13
4 4 2 14
4 8 2 0
5 2 2 16
5 7 2 17
5 9 2 0
6 3 1 19
6 4 1 20
6 5 1 21
6 6 1 22
6 7 1 0

 

[Tim Williams]

It probably won't breakdown at 5V. But it might around 6.

Which reminds me, I've ran such a circuit at 18V before with no
problems. .

Tim

 

[none]

Hello,

I've been simulating a circuit of the 'classic' astable multivibrator
and analysing the results. This is the old-fashioned two-transistor
design where each device (both NPNs) alternately turns off/on the
other. I'm sure you know the type of thing. Anyway, the collector
voltage of Q2 goes negative to almost the extent of the supply rail! Is
this a Spice anomaly or would a real world circuit do the same? I don't
see how it's possible for a circuit to generate sub-ground voltage
levels. Isn't that theoretically impossible? The clock rate is 400Hz
and Vcc is 5V.

-Jeff.

When you capacitively couple two stages, the coupling capacitor will
charge up to the power supply minus the base voltage. When the
transistor collector goes to ground, you now have a charged capacitor
whose positive end is near zero volts. Thus the other end will be
nearly the supply voltage below ground. The capacitor will then
discharge and charge up in the other direction to the difference of
the collector saturation voltage and the base voltage.

 

[Jim Thompson]

Hi and thanks.
I've just double-checked the simulation and it's actually the *bases*
of both devices that go negative to -4.3V rather than the collectors,
so my apologies. But the question still remains about how *any* node in
such a circuit can go below GND. I will post the schematic if the above
correction still doesn't account for what I'm seeing.
-J.

Capacitive coupling?

...Jim Thompson

 

[Jamie]

Hello,

I've been simulating a circuit of the 'classic' astable multivibrator
and analysing the results. This is the old-fashioned two-transistor
design where each device (both NPNs) alternately turns off/on the
other. I'm sure you know the type of thing. Anyway, the collector
voltage of Q2 goes negative to almost the extent of the supply rail! Is
this a Spice anomaly or would a real world circuit do the same? I don't
see how it's possible for a circuit to generate sub-ground voltage
levels. Isn't that theoretically impossible? The clock rate is 400Hz
and Vcc is 5V.

-Jeff.

i guess that would depend on where the test points are?
you maybe getting a - reading but maybe your testpoints
are between to area's that are reverting the polarity ?
the only other thing i could think of is that you
are not testing with DC but maybe via a cap that could
generate - voltages.

 

[Jamie]

Hi and thanks.
I've just double-checked the simulation and it's actually the *bases*
of both devices that go negative to -4.3V rather than the collectors,
so my apologies. But the question still remains about how *any* node in
such a circuit can go below GND. I will post the schematic if the above
correction still doesn't account for what I'm seeing.
-J.

sounds like the Cap Discharge effect.

 

[Jasen Betts]

["Followup-To:" header set to sci.electronics.cad.]

Hi and thanks.
I've just double-checked the simulation and it's actually the *bases*
of both devices that go negative to -4.3V rather than the collectors,
so my apologies. But the question still remains about how *any* node in
such a circuit can go below GND. I will post the schematic if the above
correction still doesn't account for what I'm seeing.
-J.

That's normal and does happen in the real world.


The culpret is the capacitor attached to the base, when Q1 switches on
the other end of the capacitor does from somewhere close to the supply
voltage down to near gound, this voltage change is mirrored at the other
end of the capacitor pushing the Q2 base below ground.


Bye.
Jasen

 

[Fred Abse]

But the question still remains about how *any* node in
such a circuit can go below GND.

Consider the circuit below:




o------o------. +V
|
|
|
| o
S1 |=|>
| o
|
----------o------------
| | |
| | |
| .-. |
| | | |
| | |R1 |
| '-' |
| | |
| | |
| | o |
.-. | |=| S2
| | | o |
| |R2 --- |
'-' --- C |
| | |
| | |
'---------o |
| |
| |
V D |
- |
| |
o--------------o-----------o--. 0V


R2 >> R1

Close S1 until C has charged to +V, minus the forward drop of D


Open S1 and close S2


C cannot discharge instantly, only at a rate set by C(R1+R2). Its upper
terminal is now connected to 0V, hence its lower terminal must be negative
wrt 0V. D is now reverse biased.

We now have a decaying negative voltage at the junction of C and D anode.
The rate of decay is set by C(R1+R2)


That's how you get nodes going below 0V

 

[kell]

Okay, thanks for that.
So it appears it's either a model deficiency or real-world capacitive
effect.

It's real.
Stop thinking about your simulator for a minute.
If you look at an astable multivibrator and analyze it, you
will see.
You have a charged capacitor and suddenly you pull the high
end of the capacitor to ground.
Well, obviously the low end of the capacitor will go to a potential
that's negative with respect to what you are considering "ground,"
which by the way is only a reference, not a barrier.
You could analyze the circuit with a positive "ground" at Vcc,
all the potentials would be negative with respect to that.
I suspect that it wouldn't bother you that the negative terminal
of the cap flies up and down, since it never crosses Vcc "ground"
in doing so.

 

[Jeffers]

Okay, Fred and Kell. Some really eye-opening and helpful ways of
re-examining the 'problem' (such as it was) for which I'm most
appreciative.
Thanks!