Optimising/Modifying Desulfator Design

Discussion in 'The Projects Forum' started by BrainFog, Oct 28, 2011.

Thread Status:
Not open for further replies.
  1. BrainFog

    BrainFog Thread Starter Member

    Joined:
    Jan 24, 2011
    Messages:
    122
    Location:
    England
    Sorry about the topic being so long, I simply wish to be thorough. So many people put up topics with a lack of information making diagnosis difficult. I suspect I may not be one of them. Hopefully I haven't missed anything out.


    I know a few people on here have made these and I thought this forum would benefit from a discussion about desulfators and how to modify them. This topic is also about my experiment to better understand them and how is has not turned out as expected.


    It must be almost a year since I got into electronics through a project on here with the idea of making a 36v desulfator. At the time I didn't have the knowledge (in fact almost almost no electronics understanding) to build a 36v one so had to settle for making a 12v one from a thoroughly tested design which has proven to be quite effective. Recently I made another desulfator with the objective of understanding how to optimise the timing of the pulses to gain the understanding I need in order to design a 36v desulfator. Design wise it seems ok and all evidence shows it is working as it should apart from its ability to desulfate being very poor. I hope to find out where I have gone wrong and how to improve the design for my future project. I will also note that due to me using AGM's not flooded cells an equalisation charge would be damaging.


    http://alton-moore.net/graphics/desulfator.pdf
    [​IMG]


    It is somewhat based on the above design. I used slightly different parts but most are equivalents. The only notable differences are to L1 which I changed to self wound 70uH for experimental purposes (explained later) and used variable resistors and a 1nF capacitor to get the needed timing from the 555. The cables from the circuit to the battery are 1.5mm^2 mains cables that have been doubled up and some thin cables to a charger that are attached across C4. I have attempted to keep everything low resistance. The timing is set to 2kHz with an inductor charge time of 0.000003465 seconds, less than 1/10 of the original design, which is what I found to be the absolute maximum I could set it to without excessive heat from saturation.


    Methods I used to test device (I have not yet gotten hold of an oscilloscope) are: Testing the voltage drop across a resistor in my homemade power supply in order to know exactly how much current is being consumed. This has proven to be very accurate. Using a diode and low value capacitor to test the desulfator's peak voltage output. And the old fashioned method of feeling if any of the components are hot.


    The design itself seems to work, more or less. The output voltage spikes are measured at 61volts and the desulfator consumes about 55mA when the battery is fully saturated and nothing gets any more than warm. I can hear a constant hum coming from the circuit too which I know to be normal from the first one I built which works very well.


    This is what the desulfator looks like. I made quite a mess of the layout in my attempt to be compact and to get short connections:
    [​IMG]


    I gave my design a home wound 70uH inductor, for L1, the idea was to make it work with an inductor that will saturate after an unknown period of time. I also wanted to test if lower inductance and resistance can give more power due to more rapidly charging and discharging the inductor. I have read about the success someone had with a design that used a 22uH inductor for L1. Since I built this I have read into it some more and quite a lot of what I have read conflicts. Could anyone clear this up for me as this has most likely contributed to my desulfator being ineffective?




    Even though L1 is quite small physically and in Henerys it should work more effectively than it does. I read somewhere (can't remember where now) that someone as an experiment put a capacitor across D1 to simulate high diode capacitance and found that it boosted output current but nullified the effects of the desulfator and theorised that it somehow lessened the impact of the pulse on the battery. Could the combination of diode capacitance,130pF according to the datasheet, and relatively high voltage and not massive current be my problem?


    I have two main theories of what the problem may be, remember this circuit is for experimental purposes for a design that will contain unknowns. Any input and ways of improving the design will be appreciated.


    A few more notes and questions:


    I noticed that the mosfet heats up suddenly if I set the pulse width faster than 2 kHz, Why would this be?


    How helpful are fuses in desulfators? When I first plugged it in the mosfet reached well over 100 degrees C ,to touch, within a few seconds and the fuse did not blow. I am concerned that if something goes wrong with the mosfet or timer I could end up with a house fire.


    I decided to test to see if it is really worth using such thick wire with these things and the results were that if I use 1.5mm^2 wire to the battery the pulse output is 48v and with 3mm^2 it was 61v. So it does make a difference. It also explains why most commercial desulfators have such a bad reputation as they often use meters of very thin wire in which the resistance would drown out the current pulses. Although with my success I shouldn't be judging. :p


    Thank you
  2. SgtWookie

    SgtWookie Expert

    Joined:
    Jul 17, 2007
    Messages:
    22,044
    Location:
    In the vast midwest of the USA; CST
    I remember that thread.

    Which one did you build? The original Alastair version?
    You should post a schematic and parts list, or at least a link to it.

    Trying to go across three 12v batteries in series would be less effective than desulfating them individually. You could connect one individual desulfator across each battery.

    How did you come to that conclusion?

    That would seem to indicate that the design needs more work.

    A number of AGMs can be charged at a much higher rate than standard lead-acid construction cells. You need to refer to the manufacturers' datasheet for your specific make and model, as specifications can vary even among similar product lines for a given manufacturer.

    The schematic image is the original Alastair Couper design.
    One caveat with Alastairs' design is that C4 is too small. If the shown timing resistor & cap are used, the small value of C4 will allow the polarity across it to reverse, which will destroy an electrolytic cap pretty rapidly. It should be at least 200uF, but larger is better. I used multiple 100uF caps rather than one large cap; this results in a much lower ESR than a single large cap.

    One trouble with just posting the original Couper schematic is that the specifications for a number of components are rather generic; you really need to post the parts list as well.

    You should also post your exact parts list.

    How did you calculate that 70uH would substitute for 220nH, and substituting 1nF for the 22nF timing cap would compensate?

    How did you calculate this amount of ON-time, and the inductor charge time?
    You're saying 3.465uS. With the standard timing resistors and your 1nF timing cap, you would get about 15.3uS ON-time.
    What are you using for the values of R1 (the 470k) and R2 (the 22k) resistors?

    Just measuring across the resistor will leave something to be desired, as your meter will be getting hammered with ~60v peaks that it will have a hard time averaging.

    You need to use an RC lowpass filter to remove the HV spikes.
    A 1N4148 diode works well for this. Don't forget to add the ~0.7v for the diodes' Vf to your reading.

    That works.

    55mA? Are you certain you don't mean 5.5mA? 55mA is quite high. The versions I built used around 8mA.

    It's hard to avoid the hum.


    That looks OK.
    You want to avoid saturation, as that will quickly lead to a burned-up MOSFET. Current through an inductor increases in quite a linear fashion, until the core saturates. At that point, the winding practically becomes a straight piece of wire, and current skyrockets.

    If the inductor has an air core, it won't saturate. However, you would need a LOT more wire to get the same inductance.

    Core manufacturers like Micrometals have plenty of data available for you to peruse to select cores that are adequate for the task at hand.

    Well, you're kind of shooting yourself in the proverbial foot if you use small-gauge wire for winding the core.

    Setting the ratio of ON to OFF time will get more critical at higher PRF's.

    Small core; small gauge wire used; you are saturating the inductor.

    Since you don't have an O'scope, it'll be difficult to tell exactly when your inductor is entering saturation.

    Capacitance across the diode won't help things.

    But you haven't told us WHAT you're using, so it's difficult to speculate.


    How about some documentation? Like, exactly what parts are you using?

    Pulse width is measured in time. So, 1/2kHz is 500uS, or 0.5mS. Since you have not shown us exactly how you have things wired up, all I can do is speculate that when you turn a pot to decrease the PRT, you increase the MOSFET ON-time.

    You should use a 1A or 2A slow-blow fuse to guard against a short in the MOSFET. If you are not using a charger across C4, then you could use a somewhat smaller slow-blow fuse.

    The problem is the length AND diameter of the wires; even straight wires have inductance. The longer the wires, the more inductance they will have; and the parasitic inductance will decrease the effectiveness of the desulfator. If you want the best connection, use some flat, wide copper stock to connect the circuit to the battery terminals. An infinite plane has virtually no inductance; so the wider your conductors are, the less inductance they will have.
  3. BrainFog

    BrainFog Thread Starter Member

    Joined:
    Jan 24, 2011
    Messages:
    122
    Location:
    England
    Sorry, I appear to have not given enough information after all. *tries to recover from shooting self in foot*

    The parts list:
    http://uk.farnell.com/multicomp/her602/diode-fast-6a-100v/dp/1625098
    http://uk.farnell.com/sanyo/2sj655/mosfet-p-ch-100v-12a-to220/dp/1713433
    http://uk.farnell.com/panasonic/eee...ssellid=9695958&crosssell=true&in_merch=true&

    The 555 is an NE555 if I remember correctly. Also C1 is a standard 100uF capacitor, I used a 100uF instead of a 30uF as they are easier and cheaper to obtain.
    The inductors are self wound and both measured using a multimeter which measures inductance. The small one is exactly 1mH (20 turns 2500 AL) and the larger one is 70uH (about 215 AL with 18 turns).

    As for the timing of the 555 I used a 555 calculator http://freespace.virgin.net/matt.waite/resource/handy/pinouts/555/
    R1 is a 100k variable resistor which is now set to 5k
    R2 is a 1000k variable resistor of which is set to 660k
    C is 1nF for the 555
    I got the resistor values by first measuring the total resistor value then measuring the resistance across the centre and unused pin to obtain the value being used by the 555.

    As a power supply, I am using an L200 as a current and voltage regulator and am using a 1ohm resistor as the current sense which it uses to limit current of which all the output current passes through. I measure the voltage drop across this resistor to inform me of the output current. I assumed that the extra output capacitors on the output of the L200 would be enough to counter any fluctuations in voltage. I used 2 470uF capacitors. I remember being informed that if I place the charger across C4 it would not be subjected to the 60v current spikes. It did indeed measure 55mA not 5.5mA If this is too high the need for a filter may explain it.

    I would not have called it small gauge wire. What gauge of wire would you consider appropriate? Anything thicker and physically winding it becomes very difficult. What about using parallel wires?

    The goal of the 36v desulfator is to maintain already healthy batteries on a regular basis so the 36v desulfator does not need to be that effective compared to the 12v ones which I use to repair sulphated ones. Still currently this is about gaining a better understanding of them.

    This is the closest I have to how a 36v desulfator might be designed (mostly states what I should try): http://home.comcast.net/~ddenhardt201263/desulfator/highvolt.htm

    Just a few concerns bout the 36v version. What would the safety risks be with a 36v desulfator? I am thinking about the high output voltage here. And what should I do to minimise risk? Also I would like to add an off switch to shut down the desulfator until I have connected everything up and am happy that I will not get shocked or create sparks. Where across the 555 would be best to add a switch in order to keep the transistor in its high resistance state without causing problems to the 555 or any other parts?

    The reason I can not use an equalisation charge on AGM's is due to gaussing which is not a problem for flooded cells but with AGM batteries it would irreparably lower their total capacity. It is todo with hydrogen bubbles covering the surface of the plates lowering their usable surface area if I remember correctly.

    Thank you
  4. SgtWookie

    SgtWookie Expert

    Joined:
    Jul 17, 2007
    Messages:
    22,044
    Location:
    In the vast midwest of the USA; CST
    It happens. Particularly when one is dealing with circuits like this, that are somewhat dodgy to begin with.

    The capacitor is a problem.
    "AC Ripple Current: 350mA, impedance: 0.34 Ohms"
    Not good. A quick simulation shows you're placing 700mA ripple current on it, and the impedance is limiting the peak current through L1.

    That's fine for the 555. It should actually have another 0.1uF cap across the power/ground pins.
    Ahh, I see. I thought the smaller toroid was your L1. I wouldn't be too surprised if L2 was starting to saturate somewhat.

    Using your values for R1 ,R2 and C1 (you have R1 & R2 reversed) in the simulation, I get exactly 2kHz for the PRF (500uS), and approximately 4uS for the pulsewidth for an 0.8% duty cycle.

    Again, without a complete schematic of this part of the circuit, it is difficult-to-impossible for me to tell you whether you are doing it right or not.

    I did mention that placing a trickle charger across C4 would be OK. I also mentioned at some length about using multiple caps for C4 to lower the ESR.

    As I already mentioned, you didn't state which toroid was L1, and which was L2. I'd assumed the smaller toroid was L1.

    You can certainly use multiple strands of smaller gauge wire; circular mils are circular mils.

    The goal of the 36v desulfator is to maintain already healthy batteries on a regular basis so the 36v desulfator does not need to be that effective compared to the 12v ones which I use to repair sulphated ones. Still currently this is about gaining a better understanding of them.

    This is the closest I have to how a 36v desulfator might be designed (mostly states what I should try): http://home.comcast.net/~ddenhardt201263/desulfator/highvolt.htm

    Just a few concerns bout the 36v version. What would the safety risks be with a 36v desulfator? I am thinking about the high output voltage here. And what should I do to minimise risk? Also I would like to add an off switch to shut down the desulfator until I have connected everything up and am happy that I will not get shocked or create sparks. Where across the 555 would be best to add a switch in order to keep the transistor in its high resistance state without causing problems to the 555 or any other parts?

    The reason I can not use an equalisation charge on AGM's is due to gassing which is not a problem for flooded cells but with AGM batteries it would irreparably lower their total capacity. It is todo with hydrogen bubbles covering the surface of the plates lowering their usable surface area if I remember correctly.[/QUOTE]

    If you simply use three individual desulfator circuits, you avoid the overhead of having to use a regulator (the regulator draws a minimum of 5mA static current by itself) along with the regulator's power dissipation.

    You also avoid the problem of having to generate a very high voltage spike, and the problems that such a high voltage spike may cause with your existing charging system and other connected circuitry.

    I am not interested in designing a 36v desulfator.
  5. BrainFog

    BrainFog Thread Starter Member

    Joined:
    Jan 24, 2011
    Messages:
    122
    Location:
    England
    Do not worry, it is me who will be designing it. This is all about improving my understanding. The purpose of the website is learning after all.

    I must say I am surprised that it is the capacitor, I may never have thought of it as I chose a low ESR one. Thank you. Luckily I bought several of them and will add a couple more to lower impedance.

    Just so I fully understand as I am still a bit unclear. The capacitor issue is due to the rapid discharge of the inductor at which point the circuit is completed through C4 to L1 to D1 discharging the pulse into the battery?
  6. BrainFog

    BrainFog Thread Starter Member

    Joined:
    Jan 24, 2011
    Messages:
    122
    Location:
    England
    I have just updated my circuit by adding 2 additional 100uF low ESR capacitors in parallel with C4. Hopefully it will be enough. I also added a few aditional ceramic capacitors for the 555 and in parallel with C4.

    I have also done some calculations for the RC lowpass filter. Does a resistor of 1ohm and a capacitor of 100uF sound right?

    In about a week I will do a heavy drain test on my test battery to see if the desulfator is now working as it should.

    Thank you for your help.
  7. SgtWookie

    SgtWookie Expert

    Joined:
    Jul 17, 2007
    Messages:
    22,044
    Location:
    In the vast midwest of the USA; CST
    There are caps available which have reasonably low ESR without paying a fortune for them. You just need to look at the specifications. For example, a Nichicon UPH1J101MRH is a 100uF 63v cap with a 0.1 Ohm ESR and an RMS current rating of 900mA. If you place three of those in parallel, you effectively wind up with 300uF, 33.3m Ohms ESR, 2.7A RMS rating; whereas before with that single 100uF cap you had more than 10x that ESR, and less than 13% the RMS current rating. See how that works?

    Well, the 555 needs a 0.1uF ceramic or poly metal film AND a 1.0uF or larger alum. electrolytic nearby. You can add ceramics in parallel with C4 (etcet.) if you like, but they're probably not necessary.

    It occurred to me later that you probably were not taking into account the battery charge current. So, it's likely that your 55mA current flow consisted of ~47mA charge current and ~8mA used by the desulfator circuit.

    Have a look at the attached; it's basically the same circuit with just a few mods. Note R8, C8, R9; R8 provides a voltage drop for the current flow into the timer circuit, C8/R9 are the RC filter. You measure the voltage across C8 (TP2+/TP2-) to get the current flow into the timer portion of the circuit. As you can see on the plot, it's ~7.4mA after C1 charges.

    Same thing with TP1+/TP1-.
    If you subtract the reading across TP2 from TP1, you'll get the battery charge current. If L1 starts getting saturated, you'll see an increase across TP1+/TP1-.


    You should notice the peak voltage decreasing as the sulfation gets removed.

    Attached Files:

  8. BrainFog

    BrainFog Thread Starter Member

    Joined:
    Jan 24, 2011
    Messages:
    122
    Location:
    England
    For some odd reason the importance of ESR during discharge did not twig until you mentioned it was most likely the capacitor. It seems to obvious now. I was only thinking about charging the inductor.

    When fully saturated and only plugged into the charger the batteries only draw about 5-10ma each. It seems more likely to me that it is due to inconsistent drain as the charger is powering the desulfator. It may be a combination however, time to experiment! I will test out a few variations of RC lowpass filters and see what the results are.

    I was looking for just such a schematic as that on the LTSpice yahoo group earlier today. I would really appreciate it if you could send me the .asc file or put it up on the LTSpice yahoo group.

    Thanks
  9. SgtWookie

    SgtWookie Expert

    Joined:
    Jul 17, 2007
    Messages:
    22,044
    Location:
    In the vast midwest of the USA; CST
    Here you go.

    You'll need to un-zip the zipped files and put the contents into the appropriate folders.

    Make sure you rename your standard.mos to standardmos.bak before you un-zip standard.mos.zip into the cmp folder.

    .asy files go in c:\Program Files\LTC\LTSpice\lib\sym\misc
    .mos file goes in c:\Program Files\LTC\LTSpice\lib\cmp
    .sub files goes in c:\Program Files\LTC\LTSpice\lib\sub

    Attached Files:

  10. BrainFog

    BrainFog Thread Starter Member

    Joined:
    Jan 24, 2011
    Messages:
    122
    Location:
    England
    Thank you for the upload, unfortunately there appears to be a slight problem. I have double checked and all files are in the correct places. When I open up the .asc I get a popup saying "Couldn't find symbol(s): LM555dip". I have also attempted to replace the empty space on the schematic where the 555 goes with an NE555, I got the message "Singular matrix: check node m1#gate Iteration No.1"

    There is an LM555 in the parts library however its pins are differently arranged.

    Google did not give any usable results for the search LM555dip. What has gone wrong?
  11. SgtWookie

    SgtWookie Expert

    Joined:
    Jul 17, 2007
    Messages:
    22,044
    Location:
    In the vast midwest of the USA; CST
    Oh boy, I must've given you an older version that didn't include all the .asy's I have now.

    Here, look in the attached .zip; I've added the other .asy files I have; also threw in TLC55x. TLC555 is a CMOS timer that Radio Shack carries. They operate similarly to the standard BJT 555 timers, but the output goes all the way to +V instead of being limited to ~ +V -1.3v. They are limited as how much current they can source (10mA) or sink (100mA) though.

    [eta]
    Wouldn't you know it - I forgot to attach the .zip file..

    Now it is.

    Attached Files:

    Last edited: Nov 1, 2011
  12. t06afre

    t06afre AAC Fanatic!

    Joined:
    May 11, 2009
    Messages:
    5,939
  13. SgtWookie

    SgtWookie Expert

    Joined:
    Jul 17, 2007
    Messages:
    22,044
    Location:
    In the vast midwest of the USA; CST
    He referenced an entry on D. Denhardt's site in post #3. ;)

    I think just about anyone searching for info on desulfators stumbles across Denhardt's site at some point in time.
    t06afre likes this.
  14. BrainFog

    BrainFog Thread Starter Member

    Joined:
    Jan 24, 2011
    Messages:
    122
    Location:
    England
    SgtWookie is correct, however thank you for the link t06afre.

    I have tested my updated desulfator with a battery over the last 5 or so days. Unfortunately there is little difference in the battery performance under a heavy load test. :(

    My next thought is that even though I can feel no detectable heat from L2 (the 1mH inductor) it may simply not be up to the task. I believe this was suggested and it is therefore the next logical step. If that fails I will replace L2 with a 220uH inductor. After that all it could really be is a faulty semiconductor.

    Just a note for anyone who downloads the LTSpice simulation. When I used it the MOSFET was upside down in the schematic. This may simply be because I need to update LTSpice as it was not upside down in the screenshot. It was very informative as to how the desulfator worked. I was also surprised at how much the current varied when I entered a peak current of 2A as peak current for L1 (70uH).

    After looking around it would appear that this may have been a better capacitor choice: http://uk.farnell.com/panasonic/eehza1h220p/capacitor-hybrid-polymer-50v-22uf/dp/1907245

    Thank you
  15. SgtWookie

    SgtWookie Expert

    Joined:
    Jul 17, 2007
    Messages:
    22,044
    Location:
    In the vast midwest of the USA; CST
    Don't expect things to happen overnight.
    I used a desulfator on a very badly sulfated battery that came from a riding lawn mower. Three of the cells in the middle had a specific gravity so low that it wouldn't register on a very expensive optical specific gravity tester.

    It took about five weeks for the desulfator to work its' magic. The specific gravity eventually equalized across all cells; measuring ~1.260-1.265. I was duly impressed.

    Don't decrease the value of L2. If you wish, get a larger toroid to wind the wire on. L2 is supposed to keep the current flow constant. It will wind up having a small DC bias, which will be roughly the average current through L1; however it won't really have much in the way of peak currents - unless you reduce its' size significantly.

    No, that's my fault. I had modified pmos.asy for my IRF9Z32 .subckt; I didn't like the default LTSpice pmos.asy with the drain up instead of source up. Every time I use a P-ch MOSFET, I have to rotate it 180° and then flip it horizontally, so I did that to some of my P-ch symbols. When LTSpice does an update from the website, it makes certain that the default symbols like pmos.asy and nmos.asy are the LTSpice conventions, but it does not check the customized versions; it doesn't know anything about them.

    How did you do that? Do you mean changing the ON-time of the 555 timer?

    Keep in mind that the inductors used in the simulation are ideal. The wiring has no resistance, there is no parasitic capacitance, etc. and they will not saturate; in that regards they are like ideal air-core coils, as air core coils don't saturate either.

    Ferrite toroids, OTOH, will saturate when the flux density gets too high for the material and size of the toroid in use. This is not easy to simulate, and becomes even more difficult when the material characteristics are unknown (as in salvaged toroids) and one has minimal test equipment available.

    Right-click on one of the caps that I used in the simulation. You will notice that the ESR of each cap is 0.1 Ohms. Since the three caps have equal ESR, and they are wired in parallel, the ESR is 1/3, or 33.333...mOhms, which is three times better than 100m Ohms.

    When you are looking to reduce your ESR, you are generally better off to use multiple smaller-sized caps than one large cap. Also, note that the voltage rating of the cap can and will impact the ESR; as for the voltage rating to increase, the dielectric must increase in thickness, which means that the plate area must also increase to maintain the same capacitance, which means that there is more parasitic inductance and resistance, which increases the ESR.

    Did you follow that? It's an important point.
  16. BrainFog

    BrainFog Thread Starter Member

    Joined:
    Jan 24, 2011
    Messages:
    122
    Location:
    England
    Just like my project, my replies are anything but fast. The excuse “I have been studying” is good.


    For a test I have been using a 12v 7ah AGM I was given for free a while ago. It is at about 1/3 capacity I would say. I updated the design again by shortening the cables and directly soldering connectors onto them rather than connecting through a terminal block. Probably should have done that sooner. I will test them again in about a week. I would expect to notice at least slight results by then.


    In fact I was considering upping L2 from 1mH to 2mH with a larger toroid and more windings.


    On the simulation by 2A peak current I am talking about entering into a setting which was left blank when I right clicked on L1. I assumed it was there to help make the simulation more realistic as many inductors are provided with such information. Am I wrong?


    Also I have been reading a physics book and having a look through the relevant areas of magnetics and inductors, still getting to grips with much of it. Inductors are a component that I am struggling to understand. Most likely because I intend on making them which is odd as they are so simple in design.


    When you say size, do you mean physical size or inductance? This is something I found ambiguous in the book I am reading.


    During research of my SMPS I came across an article regarding the skin effect. Are desulfators affected by this?


    Yes I am gaining a much better understanding of capacitors.


    Thank you
Similar Threads: Optimising/Modifying Desulfator
Forum Title Date
The Projects Forum Battery Charger/ Desulfator Problem Apr 30, 2011
The Projects Forum 36 volt desulfator Jan 26, 2011
The Projects Forum Lead acid battery desulfator.. Mar 26, 2010
The Projects Forum Battery desulfator Nov 27, 2009
The Projects Forum Battery Desulfator (with resonant frequency detector and locker) May 19, 2009

Thread Status:
Not open for further replies.

Share This Page