This is the PSpice model scheme for the original Alastair Couper lead
acid battery low power desulfator, a simple but greatly performing
circuit: it allows for heavily sulfated batteries reclamation.
For more details on the original Alastair Couper desulfator, please
visit the Low
Power Desulfator Info Home Page (there is a power version too):
you'll find all sorts of schematics, tips and tricks to build your
My PSpice model for this circuit is freely downloadable: try it, modify
it, check it with PSpice simulator.
You can see the PSpice model at top of this page: node numbers are also
reported (in red colour) for easy printout readings.
But... why simulation? This is a simple circuit, derived from well
known DC-DC converters, and it behaves exactly as described in Low
Power Desulfator Info Home Page.
Well. IMHO, simulating a circuit may be the best manner to go "deeply
inside" it, this because:
Some voltages and currents may be improperly measured (scopes or
meters disturbing normal circuit operation, erroneous measurement
techniques giving wrong results etc.).
You may think to make some experimentations with the real circuit,
but this may toast expensive devices; making changes in a model won't
harm the PC, but will show the results (of course, you have to decide
if that current flowing through that device will damage it or not...
to decide, simply take a look at the device datasheet, and you're done).
You may try to improve circuit performance, substituting devices,
timing, output waveforms and so on... but some circuits (AC desulfator
being a perfect example) are operated with "allfruit" power supplies
and loads to check for performance.
How will you judge if your modified circuit is performing better than
the unmodified one? For AC pulsers, you'll use sulfated batteries to
make tests, and damaged battery electrical parameters are strongly
variable. You'll measure desulfation process speed, and this is ALWAYS
the conclusive test, but... how can you find two identical batteries,
having identical sulfation degree? Not that simple!
Using the model, you set a guess load/power supply, then by signal
analysis (once understood how and why a pulser desulfates batteries)
you may change values to enhance model desulfating capabilities, and
then apply improvements to the real circuit.
You may be simply curious, and would "take a look" at circuit
waveforms for educational purposes only, without building it.
In a few words, models are the starting point to evaluate, dimensionate
and improve real circuits: many IC builders won't even diffuse a
new chip without having its running and reliable model, becouse
diffusing silicon chips is rather an expensive process!
They first simulate the chip, then (when maximum model performance is
reached) they diffuse a few units, test these units to find out "real
world" limitations, if out of spec they adjust the model to better
reflect chip behaviour, again tune model to maximum performance, again
diffuse a few units, again test and compare with the model... finally,
they have a working chip and its reliable model as well, and series
production is started for that chip.
Even in the future, should "strange bahaviour" occurr for the chip,
they FIRST take its model and see if they forgot something; when this
happens, normally the model shows this "strange behaviour" (it's
really difficult to measure currents inside a chip, you rely on the
Same procedure may be applied for discrete circuits, of course: this
is exactly what I'm trying to do... and playing with the model, who
knows? I may be able to achieve improvements for the pulser, sometimes
in the future.
For those not wanting to simulate the circuit, here's a little
collection of waveforms obtained for this simulation, with useful
Fig.1: battery input current waveform.
This waveform flows into the pulsed battery: please compare it with
the one found in
Low Power Desulfator Info Home Page, and note that we've less
overshoot here. This because:
A simple RL model is used for simulating the battery: real
units will have a more complicate behaviour, depending on their
dimensions, chemistry and state of charge.
Measurement errors are not present here, but may be present in
real scope current measurements, performed using series shunt
techniques (see also Fig.10 for details).
Fig.2: battery input current frequency spectrum, calculated with FFT
Probably, the most important finding for this PSpice simulation: it
shows HOW pulser desulfates.
This waveform contains relevant energy for all frequencies ranging
from f0 (T=785us) to several MHz; it would be more correct to say
that, being this waveform periodical, power is "quantized" in current
rows placed at f0, 2*f0, 3*f0... etc., being this spectrum their
But, considering that f0=1274Hz, we may assume that at least one row
will "hit" sulfate crystal resonant frequency fs (depending on crystal
physical dimensions), disgregating it and returning it to electrolyte.
Fig.3: voltages measured at pulser output and at battery leads.
Note the overshoot (due to overall inductances), and note how cables
stray inductance reduces overshoot for the pulse traveling from pulser
These two waveforms are too close in this picture: see also Fig.4 for
Fig.4: same as Fig.3, but using a smaller time scale.
Overshoot reduction is dramatic for the waveform at battery leads, due
to connection cables stray inductance: FOR BEST RESULTS, KEEP CABLES AS
SHORT AS POSSIBLE!
Fig.5: NE555 output and MFT1 gate voltages and currents.
It should be noted that:
NE555 output and MFT1 gate waveforms are almost identical: this
makes an evidence that no RC network is needed between NE555 and MFT1
for correct circuit operation (things would be different using a PNP
BJT as a switch: in this case, RC network would effectively work as a
speed enhancer, due to "diode" BE junction effect provided by switching
NE555 output current exceeds 200mA for a few nanoseconds: this may
not happen for a real world 555, whose output current is internally
limited (555 has output totem pole stage, with SOA protection).
Should 555 output current try to exceed 200mA, SOA protection will
take place, and lower slewrate will be observed at the output, due to
current limitation, depending on your luck (some 555 may even deliver
over 300mA, others may not; however, output exceeding 200mA is NOT
guaranteed: consult 555 datasheets for details).
Fig.6: flyback diode and mosfet currents.
D1 goes ON state when MFT1 goes OFF state: this is flyback operation,
widely used on DC-DC converters.
D1 should be fast type (you may use a Schottky diode for even better
results): a slow diode would allow some energy to flow from L1 to MFT1
at the beginning of MFT1 shutoff, instead of "flying" this energy back
to the battery.
Therefore, using a slow diode would cause MFT1 heating up, less power
returned to battery, i.e. lower circuit performance.
Fig.7: L1, L2 and C4 current waveforms.
L1 sustains a strong transient current, as well as C4; L2 is required
to conduct a moderate current, therefore may be chosen smaller and
cheaper than L1.
C4 and L1 ESR (Equivalent Series Resistance) DOES MATTER: choosing L1
with too small current rating will degrade circuit performance,
choosing for C4 the "low cost" type will cause C4 premature death (it
will heat up, due to high ESR and therefore high wasted power).
Fig.8: other interesting voltage waveforms, that will help you to
fully understand AC pulser operation.
Note V(10) ringing when D1 goes OFF: this is due to diode and mosfet
parasitic capacitances, resonating with L1 in LC series mode, through
battery low impedance path.
Fig.9: same as Fig.8, using a smaller time scale; note D1 delay.
Fig.10: real current flowing into battery, compared to scope readings
obtained using a low ESL shunt.
The shunt used is a simple carbon film, 1/4W resistor, with an average
ESL in the 10nH range: this ESL is negligible when measuring low
frequency sinusoidal waveforms, but will generate ringing when
measuring fast slewrate pulses.
Sometimes, electronic measures need to be evaluated with a grain of
salt: once we know this ringing effect, we'll simply ignore it,
considering scope readings only AFTER ringing has gone.
To complete this work, I'll give you average power waste for the most
important devices in this circuit: