Information taken from .

The links on this page will take you to the appropriate webpage off my site, however the basic information is here in case the original author's website doesn't work.

Lead Acid Battery Desulfation Pulse Generator

Some help and information for builders

Parts kits now available

(Last update Sept. 2002) New "peak reading" kit with digital display so you can see the desulfator making progress on rehabilitating your battery.

This page is intended to provide builders of the battery desulfator circuit, as shown in Home Power Magazine, issue number 77, with additional information. (Here is the original article, PDF format.) The volume of email has made it clear that a few more details are needed. The use of inductors and high frequency pulses makes this circuit require more than a passing DC electrical knowledge to get working properly.

To start with, take a look at this short note on lead acid battery chemistry and the sulfation process.

Here is a patent worth reviewing: It shows that high peak current is essential to overcome electrolyte stratification.

Finished Board

The finished 12 volt version

  • Parts kits now available ! Don Denhardt has taken the initiative to put together some parts kits to make it easier for home builders. He has the original circuit designed at 12, 24, and 36 volts. He has also begun development of a Stamp based, intelligent desulfating circuit. The Stamp module is a microcontroller chip that runs BASIC and hooks directly to the serial port of a PC. This allows for many different possibilites, and anyone wanting to experiment in this area has a great open opportunity availiable. Click here for Don Denhardt's parts kits.
  • Note: The value of C2 is wrong in the article. It should be .0022uF, not .022uF. This explains why some have had problems with frequency/ pulse width being off.
  • By popular demand, here is one alternative schematic:and yet anotheralternative schematic that uses an Nchannel MOSFET:


  • These two circuits use different means to drive the FET, which the P channel circuit does not require. I had no idea that P channel units were hard to come by, but that is what I am being told. This version is slightly more complex, but works identically as the original version, in theory. Check out the bulletin board for further discussion.
  • Parts supply has been a problem for many since Radio Shack doesn't carry the main pieces. I have used parts from, and also from Mouser Electronics. The only item that is remotely critical is the inductor L1. I have used a variety of inductors taken from old switching supplies and TV chassis, and they all work fine as long as they are somewhat close to the value indicated, have low enough series resistance, and that the output of the 555 timer is trimmed accordingly. If you have a larger valued inductor, you would lengthen the pulse width, if you have a smaller value, you would decrease the width. Read below for more about this. L2 is totally noncritical, as long as it doesn't get too warm at the current level you are using. It's purpose is simply to isolate the current pulse from the rest of the circuit, and to control the current going into C4. More on this below.
  • The MOSFET can be any unit that has a sufficiently low "on resistance" rating. Power dissipation is low, but I use a high power unit to lower this resistance, as that is part of what limits the peak pulse current generated. The voltage rating needs to be high enough to stand the highest peak voltage, which might get to be 100 volts. If you want to try for higher peak currents, you can parallel MOSFETS together, as they share current well.
  • The diode D1can be anything that is fast, has a high peak current capacity, and more than 100 volts peak inverse rating. Again, the higher current devices are not required by the circuit, but they will help to increase the peak current by virtue of low resistance.
  • Here are some hints for making higher voltage versions. The same principles apply.
  • NEW: Here is a high power design that I am working on at present. It gives over 70 amp peak pulses, and since it is line powered it is not fixed to any specific battery voltage. Experiments are on going, as I am trying to pin down the transformer core requirements more fully. Use this circuit with caution.
  • Thanks to Glenn Brown, via the desulfator bulletin board, we now have PCB layouts.
    Here is the trace layout: traces.gif:
  • Here is the placement grid grid.gif:
  • And here is the nomenclature: parts.jpg:
  • Here are some hints on reclaiming old gel cell batteries. Thanks to Don Denhardt and Viktor Kroker.

Begin by making sure the 555 is putting out the proper drive waveform to the MOSFET. This is best done with a scope. Using trim pots in place of R1 and R2 will allow for a wide range of conditions and battery types to be accommodated. If you are unfamiliar with 555 operation take a look at this tutorial:, the astable mode equations are about half way down the page. The illustration below shows what our drive signal should look like:

555 output


The frequency of the pulse is close to 1000 Hz. The width of the narrow, negative going part controls how long the MOSFET is turned on. The longer it is turned on, the higher is the peak amperage delivered to the battery, up to a point. At present, it isn't known whether it is better to pulse frequently with a small amperage pulse, or whether a slower, higher peak pulse is better. I am leaning towards the latter at the moment, as a result of the content of this patent.

NOTE:If you are having trouble with things getting too hot, it is likely that the 50usec pulse width is too long, resulting in L1 saturating. Also, C4 should not get warm, but will if it is a marginal unit with too much ESR (effective series resistance).


Here is the current pulse created by the circuit. This was taken using a .1 ohm resistor in series with the negative lead. The circuit is showing, given the scope setting, a peak current of over 5 amps. There is quite a bit of high frequency ringing on the leading edge of the pulse. This peak current can be reduced or increased by changing the width of the 555 drive pulse to the MOSFET. For small batteries, it might be wise to reduce the pulse width. For larger units, using heavier duty inductors, a longer pulse would give more current. As the 555 goes low for ~50usec, the MOSFET is turned on. Current flows into L1 from the stored charge in C4. The magnetic field around L1 builds up until Q1 turns off. The field now collapses, and as a result the inductive kick back forces a large current spike which goes from the -12 volt terminal, through D1, through L1, through C4, and into the +12 volt terminal. This is all over in less than 100 usec. The rest of the cycle allows C4 to slowly recharge, at a current of around 50 mA, through L2 until the next firing of Q1. Check the DC current drain of the circuit. It should be less than 100mA, preferably less than 50mA. If you increase the peak current output, the efficiency will go down, and so the DC current drain will go up.

The tireless Don Denhardt has supplied the following data which shows, for the Delevan chokes sold through Digikey, the relation of pulse width to peak current pulse. It shows the diminishing returns due to saturation of the inductors:

Tables showing peak amps
at various pulse widths.
(Pulses in microseconds.)

6 Volt battery

L1=(See table)

 Width	220uH	220uH	120uH	120uH

10	.75	.75	0.8	0.8
20	1.2	1.0	1.2	1.2
30	1.6	1.3	1.8	1.6
40	2.0	1.8	2.0	1.8
50	2.2	1.8	2.2	2.0
60	2.4	2.0	2.6	2.2
70	2.8	2.1	2.8	2.5
80	2.9	2.2	3.0	2.8
90	3.0	2.5	3.6	3.1
100	3.2	2.8	4.0	3.8
110	3.6	2.9	4.0	4.0
120	4.0	3.3	4.0	3.9
130	4.6	3.8	---	3.8
140	5.0	4.2	---	---
150	5.1	4.5	---	---
160	5.0	4.6	---	---
170	5.0	4.5	---	---
180	4.9	4.1	---	---

 12 Volt battery


 Width	220uH	220uH	120uH	120uH	330uH

10	1.2	1.0	1.5	1.2	0.8
20	1.9	1.6	2.2	1.8	1.2
30	2.4	2.0	3.0	2.4	1.6
40	3.0	2.5	3.5	3.5	2.0
50	3.5	3.0	3.5	4.0	2.2
60	4.0	4.2	3.2	4.0	2.5
70	4.0	4.8	---	---	3.2
80	4.0	4.8	---	---	3.5
90	4.0	5.0	---	---	3.5
100	4.0	5.0	---	---	---

A few words about the "several magehertz resonance." One of the most frequent questions is about the fact that the drive frequency is at 1 khz, but the resulting vibrations in the battery are at several MHz. Go back over the part in the article about the "plucked string". This is a very common situation in all systems with a resonant frequency. A disturbance of any sort, will tend to create vibrations at the resonance. The megahertz range vibrations are over very quickly, ie they are a damped oscillation. The pulser circuit does not drive these frequencies, any more than a finger nail drives a guitar string at its pitch.

One can see them readily using an oscilliscope, but some care is required to get the triggering just right. A better way to see the fact that batteries have a high frequency resonance is to use a sweep generator, as in the diagram:

battery resonance

In the original article, I put forth an idea of what might be happening in the battery to cause this resonant frequency, and guessed that it was occuring in the electrolyte itself. A recent email (Mar. '01) from battery expert Heinz Wenzl, in Germany, said this:


The next question in my quest to understand this is the following: Given the same battery type there are small deviations of resonant frequency which are no measurement artefact. Now under normal conditions (i.e. the battery not being deep-deep discharged with an acid density close to one), there are always a lot of hydrogen and sulfate ions around which can create a plasma type charge distribution. A change of the frequency could be linked to the electrolyte density which is related to the state-of-charge (I have found no real correlation here), the viscosity (this I would imagine would be linked to the IR spectrum of molecular vibrations), the current exchange density (linked to state-of-charge and surface area and catalytic properties, etc.) and to .... some others. But gut feeling would tell me, that all these effects should be small compared to the plasma properties themselves. In which case, all lead acid batteries with flooded electrolyte should have the same resonance and NiCd a different one.

What makes you think that the plasma condition is associated with the electrolyte and not with the solid material? Some lead minerals, e.g. PZT are piezoelectric and the few Mhz are really in that range of effects!

So it is likely that my original idea about what may be going on needs changing. Nevertheless, the resonance still creates conditions of high peak voltage (ringing) that are favorable to the process of desulfation.

Does this really work ?
The results are coming in. Here are a couple of comments:
  • Hello Alastair, I would like to thank you for such a neat product. I have reclaimed several batteries now that were junk. I have gathered up as many more as I can find and have them connected to an Air 403 for charging and running of the desulfator. Free batteries and free power, doesn't get much better than this. I have built several other desulfators for other people to use ..... thanks again for such a fantastic project.  Ed Goddard Castle Dale, Utah.
  • I am pleased with the performance. Yours works faster than the $90 Pulse Tech unit I've had 3 years. George Ficklen Newport News, Va.
  • I can actually see etching into the sulfate crystals on top of the cells. Eric Wiggins. Thames, N.Z.
A number of comments on the message board have been about battery testing. Here is a response from Geoge Aumann:
Using a resisitive load to measure battery condition is a standard method. For each battery type a standard load is defined, and if the voltage under load drops below a certain level, the battery is bad or in need of recharging. For small batteries this load can be typical, like a 5 Ohm load for a AA Alkaline drops the voltage at end-of-life to 1.0 Volts. Using a load across the battery (for a few milliseconds) is used by laptop computers to assess the charge status of the battery.

For big batteries the "standard" load resistor may get to be very small. However, given the availability of good and fairly cheap (under $40) 3 1/2 digit digital volt meters, it is not necessary (or safe) to draw a big current spike out of the battery to measure its internal impedance. For my Dynasty UPS12-310 High output battery I use a 1 Ohm 1% 20 watt resistor shunted with a bar directly to the battery terminals. The resulting drop across the 1 Ohm resistor is easy to measure with the voltmeter set to the 200 mV scale.

A fully charged 12.6 volt lead-acid battery will have an internal resistance of about 0.01 ohms. My Dynasty UPS12-310 high output battery is specified at 0.0033 Ohm. Determine the internal resistance of the battery by measuring the terminal voltage with open circuit, V, and then the voltage drop across an accurately known resistive load R, voltage DV. The internal impedance of the battery, Ri, is then given by Ri= DV * R / V.

Example: V=12.60 volts and DV= 81 mV Volt using a 1 Ohm 1% 20 watt Ohmite resistor. Ri= 0.081* 1.0 / 12.6 = 0.0064 Ohm.

The power dissipated in the resisitor is V*V / R = 12.52 * 12.52 = 158 watt. The resistor will get warm very quicky. If this experiment is not finished quickly, the temperaure increase will change the resistance. This will make the measurement inaccurate and will burn your fingers).

Somebody on the email suggested pulling 200A, presumably using a 0.062 Ohm resistor. Pulling that much power (200 * 12.6 = 2.5KW!) has to be done fast indeed. Batteries of this size can be very dangerous.

If you would like to communicate with others in this project, or to ask questions, please try the
desulfator bulletin board.


Some relevant links:

Technical details on why pulse charging is good.
This shows that Ni Cads are similarly benefitted
Here is an email exchange on battery testing techniques.
A French desulfator, not sure about what it does. (machine translation)

Back to Aquarian Homepage Back to Aquarian Homepage


Copyright 2001-2003  Peter Ferlow