Sciencemadness Discussion Board - MOT supply for electrolysis

MOT supply for electrolysis

FrankRizzo - 1-6-2009 at 16:59

After seeing quite a few of those MOT-powered low voltage spot-welder projects, I was thinking that maybe it would be possible to use that beefy transformer as a low-tech supply for an electrolytic cell. If I were to current-limit the primary with some ballast resistance, and rewind the secondary such that I get ~7VAC on some 4-AWG wire, could I use a car alternator bridge rectifier to get 4-5V DC at 70-80A without too much loss?

[Edited on 6/2/2009 by FrankRizzo]

hissingnoise - 2-6-2009 at 09:37

70-80 seems a lot of amps for an ordinary cell---do you need several Kg. in a big hurry, or something?
The most I ever used was ~30A and even then, the electrolyte fairly boiled. . .
In any case I'd be suprised if alternator diodes would handle that current density!

[Edited on 2-6-2009 by hissingnoise]

merrlin - 2-6-2009 at 16:15

Quote: Originally posted by hissingnoise

70-80 seems a lot of amps for an ordinary cell---do you need several Kg. in a big hurry, or something?
The most I ever used was ~30A and even then, the electrolyte fairly boiled. . .
In any case I'd be suprised if alternator diodes would handle that current density!

[Edited on 2-6-2009 by hissingnoise]

I tend to agree. Unless you have a cell geometry that gives you a low resistance and a reasonably large volume, the joule heating of the electrolyte is going to be considerable. If you are willing to work with a lower current limit you can get a 50 amp bridge rectifier for $5 at http://www.mpja.com/viewallpict.asp?dept=171. They also have some inexpensive transformers, although the current is less than what you are looking for.

FrankRizzo - 3-6-2009 at 12:34

Thanks guys. I'm not going to need that much current but it would be nice to know that I have the overhead if needed. I'm mostly just looking for a confirmation of theory, that the beefy transformer coupled with some rectification, will give me a cheap low voltage, high current source.

woelen - 3-6-2009 at 12:43

Another idea might be to use several parallel rectifiers, each capable of handling e.g. max. 10 A, with its own filter capacitor. Then you can put multiple cells in parallel. That seems more feasible to me, than having a current of 50+ A in a single superhot cell.

For electrolysis purposes, you do not need a really stiff voltage regulation. AC ripple suppression, however, is something which I certainly want to have. At 10 A currents you need big filter capacitors (33000 uF or so).

12AX7 - 4-6-2009 at 03:16

I took this route, filtering ripple with another MOT: cut out the windings and add as many turns of 4 AWG as possible. Reassemble with a 1/8" air gap between the "E" and "I" sections. Connect this filter choke in series with the rectifier output. Above about 20A load, voltage dropped from about 9V O/C to a steady 6V (maybe down to 5V up at 50A). I used a hunk of steel wire to take up the remaining 2-3V between supply and cell. (Typical starting current was 20A, up to 50-60A once everything warmed up.)

Unfortunately, the cell had very low efficiency, probably due to the large cathode area. But since I ran it for a couple months, I still collected a total of about 50 pounds of chlorate and perchlorate.

Tim

higher turns density

jimwig - 6-6-2009 at 10:04

for higher output amp try parralleling bridge rectifiers.....

and a question -- in transformers one can dual wind - oh say two #12 copper wires to achieve a total of 40amps... this is a question.
secondary, of course......this is certainly easier than one #4......

also probably able to achieve a higher turns density. ?

[Edited on 6-6-2009 by jimwig]

watson.fawkes - 6-6-2009 at 15:52

Quote: Originally posted by jimwig

and a question -- in transformers one can dual wind - oh say two #12 copper wires to achieve a total of 40amps... this is a question.
secondary, of course......this is certainly easier than one #4......also probably able to achieve a higher turns density. ?

Yes, this works, but beware of the gauge you're using and the total resistive heat load you're putting into it. I've tried putting three windings of #4 AWG onto an MOT, and it just didn't work. #8 AWG, though, is flexible enough. As for current rating, the US electrical code puts, for a 60°C level, a rating of 40 A on that wire. If 40 A isn't enough, use #6 AWG or double wind with #8 AWG. If this isn't enough information, you need more than a recipe. It's easy enough to compute resistivity, heat load, etc. and actually engineer the conductor.

dann2 - 7-6-2009 at 09:33

Hello,

It is difficuly to seperate the 'E' and 'I' parts of an MOT to give an airgap as the core is usually welded. Another MOT simply winded as per what you have already done and wire connected as in the picture will do the job too (I believe).
There is some info on LC smoothing circuits here.

In my experience the input winding on MOT's have a very skimpy amount of turns compared to a properly designed power transformer (about half of what is very comfortable). If you wire them up and leave them sitting on the bench without any load connected whey will get very hot. It is better to use two transformers and wire the primaries in series. This will also in effect give you twice the core area and potentially twice as 'powerful' a set up (naturally!). This is by no means critical though and the high current drawn by a single primary is not a very big problem.

Use lots of taps on the secondary. If you can use a center tapped output winding (or windings if you are using two transformers) then you need only two big diodes to give full wave rectification.
If you have a Variac on the input that would be even better. to give smooth output voltage. Since it's a cell you are going to run you would be better thinking in terms of having a controllable constant current out of this thing.

Anyone suggest how to modify an MOT so that it can be used as an output current controlling inductor?
Would the following work:
Winding on a few turns (similar to what would be done when making a regular high current output modified MOT transformer) of heavy wire onto the MOT core. Wiring this secondary in series with the load/supply.
With a variable resistor, feeding in a given DC current (into the primary of this MOT) to saturate the core and limit the current at out chosen value.

This total set up will only take 4 MOT's, one large capacitor, two large diodes, heavy wire.
The Mother of all cell supplies????????

I often wondered (as one does) about what shape the current waveform going into a cell has, when the power comes from a raw DC, unsmoothed, Voltage source. It would be interesting to put a scope (using a resistor ) on a cell to see some time. I think smooth I current is the best to work with if you are trying to figure out efficiency and times etc.

Dann2

Don't forget some meters!
Whopper of a diode on ebay here

[Edited on 7-6-2009 by dann2]

Taoiseach - 9-6-2009 at 02:47

I've done electrolysis with a modified MOT from a broken microwave.

I only cut out the secondary winding, being extra careful not to damage the primary. I sawed off the protuding part of the winding, then drilled holes into the remaining chunk of copper wire and drove it out with hammer blows.

It worked well for electrolysis but IMHO you will never reach the 1000W of power which are theoretically possible. These transformers are not designed for constant operation and they get considerably hot.

10A is feasible tough and thats more than enough for most electrolysis projects.

watson.fawkes - 9-6-2009 at 05:13

Quote: Originally posted by Taoiseach

It worked well for electrolysis but IMHO you will never reach the 1000W of power which are theoretically possible. These transformers are not designed for constant operation and they get considerably hot.

Microwave ovens will operate continuously for an hour or two because the transformers are cooled with forced air. Take away the active cooling system and they overheat. So if you are willing to deal with active cooling and, as to its potential failures, either (1) put in an overheat detection + shutdown subsystem or (2) deal with the consequences of meltdown, you can drive these transformers harder.

hissingnoise - 9-6-2009 at 07:02

A heatsink can be improvised from 3mm aluminium sheet (good contact between trans and sink), add a directed cooling fan and thermal cutout and continuous use is possible. . .
NSTs, I've found work best, but removing their secondaries takes work.

dann2 - 11-6-2009 at 09:08

Hello,

Quote: Originally posted by 12AX7

I took this route, filtering ripple with another MOT: cut out the windings and add as many turns of 4 AWG as possible. Reassemble with a 1/8" air gap between the "E" and "I" sections. Connect this filter choke in series with the rectifier output. Above about 20A load, voltage dropped from about 9V O/C to a steady 6V (maybe down to 5V up at 50A). I used a hunk of steel wire to take up the remaining 2-3V between supply and cell. (Typical starting current was 20A, up to 50-60A once everything warmed up.)
Tim

What exactly is the function of the air gap.
Putting in an air gap will decrease the inductance value (given MOT core size and certain max. number of windings).
The air gap is to stop the core from getting saturated as there is a (large) dc component running through the inductor? Will try in SPICE.
If you leave out the air gap will the MOT core (in the inductor for smoothing) saturate? At what value of DC will it approach saturation.

Someone suggested putting the (MOT cored) inductor before the rectifier (full wave four diode rectifier) so that now there is no DC in the inductor. Will this work.

I am trying some filter combinations (one Capacitor and one inductor) in PSPICE. But I don't know how to model an inductor with a given core size and type of core material. It's been a long time since I dabbeled with SPICE.

TIA,
Dann2

12AX7 - 11-6-2009 at 19:11

I tested an MOT primary (intact transformer) and found something like 250At saturation (which was 2A or so through that winding). You can easily get 50A from one MOT, so you'd only get 5 turns worth of filter choke from another, ungapped. Increasing the gap increases saturation magnetization (to perhaps 500-1k At) while reducing inductance proportionally; since energy goes as current squared, energy storage goes up linearly with gap.

If the core saturates, the "choke" chokes and ripple goes way up.

Putting a choke before the rectifier will soften the AC supply, reducing RMS current, wrecking voltage regulation and doing nothing to output ripple.

Tim

dann2 - 12-6-2009 at 09:16

Hello,

I seen a reference to 'putting the inductor in front of the rectifier' here (third post down the page by TVI). Thought I would mention it. Does not make sence to me.

I need to take an angle grinder to an MOT so as to get an air gap into it. Doing some calculations I think you will get 0.08mH from an MOT with a 0.5mm gap and approx. 18 turns (50 amps allowed giving a flux density in core of 2.1 Tesla, ie. close/at saturation). I have little faith in my calculations!

Book 'Electric Machines' on Google books page 16. 19 , 31 an and 35 if anyone is so inclined.

Does the formula
L = Number of turns * Flux give me a figure for inductance? (page 33 of book)

I cannot get my head around the fact that if you take a core, and separating the E and I pieces (put in an air gap) you will obtain greater inductance by being 'allowed' to put in more turns of wire than you could place on the core before seperation as when there is no seperation you will get saturation. (FFS)

The circuit shown below refused to be analysed by 5spice. Can anyone see anything immediately wrong with it. If you place a bridge rectifier and a single voltage source (not a tapped transformer) it analysis OK. It is pissing me offf.

Thanks,
Dann2

merrlin - 12-6-2009 at 10:36

L = (number of turns)*(flux)/current. LTSpice runs your circuit with a 100 ohm resistor as the load.

dann2 - 13-6-2009 at 07:39

Thanks for that.

Does the circuit run OK with a 0.5 Ohm load resistor.

I need to get myself a copy of LTSpice. The 5Spice seems to be problematic.

Dann2

merrlin - 13-6-2009 at 11:26

Quote: Originally posted by dann2

Thanks for that.

Does the circuit run OK with a 0.5 Ohm load resistor.

I need to get myself a copy of LTSpice. The 5Spice seems to be problematic.

Dann2

Attached is the .5 ohm version and partial plot showing ripple. The free LTSpice download is at:

http://www.linear.com/designtools/software/#Spice

12AX7 - 13-6-2009 at 11:35

Like I said, gap allows more energy storage.

For example, a ferrite dual E core such as used in your computer might have A_L = 2uH/T^2 and 20At saturation, ungapped. Let's say you can fit 100 turns of 18AWG wire on the core. Then it will have 100^2 * 2 = 20mH inductance, and saturate at 20 / 100 = 0.2A (18AWG is hardly necessary here!), storing a maximum of about 0.5 * 0.02 * 0.2^2 = 400uJ.

Increasing the gap arbitrarily (how much can be determined based on the core's reluctance) reduces the B field obtained from a given H (magnetization); since EMF comes from the change in B, the self inductance (L == EMF / dI/dt) drops. The same reduction in B allows more magnetization before saturation (saturation, in B field, is intrinsic to the material).

Let's say the gap was increased such that saturation = 40At and A_L = 1uH/T^2. Now the same 100T winding has only 100^2 * 1 = 10mH inductance. But it can handle 40 / 100 = 0.4A, and store 0.5 * 0.01 * 0.4^2 = 800uJ. Indeed, this core would benefit from even more gap, since we still aren't using that 18AWG wire at its ratings. Let's say we want 2A through it: we need 100 * 2 = 200At for saturation, so the gap has to be increased by that much more (for this core, I would guess about 1/16" or so). Inductance goes down by the same amount, to 0.2uH/T^2, so inductance is now 2mH. Stored energy is 0.5 * 0.002 * 2^2 = 4mJ, ten times the ungapped value.

Same thing works for laminated iron cores, of course; Bsat is just a little higher (figure 1T max) for a given permeability. I've measured far more ferrite and powdered iron cores than laminated iron; I should measure a bunch some day.

Notice that, if you can't reach a high enough current density to saturate the core, adding a gap only reduces the inductance; you don't need a higher saturation value if you aren't reaching it in the first place. This might've been your source of confusion.

Tim

dann2 - 13-6-2009 at 17:03

Thanks for that.
There is a discussion regarding gaps (no diagrams unfortunately)
here . It's a bit out of my depth but though you might be interested.

At what air gap size does diminishing returns set in?
Surely you cannot simply go on increasing air gap size and getting good returns (in this type of application)?

@Merrlin
I changed a setting in 5Spice from Fine to Course (dynamic step size) and it is now analysing OK. HNGAFC what dynamic step size is all about!!!!!!!!!
I sawed an MOT in half and can get 0.35mH per turn with an air gap (piece of plastic) of 0.4mm.
Now all I need is the ballast on the input as Sir Frank Rizzo suggested at the start.

Dann2

merrlin - 13-6-2009 at 20:31

Quote: Originally posted by dann2

HNGAFC what dynamic step size is all about!!!!!!!!!

Dann2

In numerical solutions there is usually a tradeoff between speed and accuracy. In solving, the timestep may be variable. In LTSpice a maximum timestep can be specified and at times I have found that a very small maximum timestep can be impractical due to the solution time required. Dynamic step size appears to be similar to maximum time step.

12AX7 - 15-6-2009 at 07:45

Quote: Originally posted by dann2

I sawed an MOT in half and can get 0.35mH per turn with an air gap (piece of plastic) of 0.4mm.

How did you determine this? Your units aren't correct: it's inductance per squared turn. That could be a big error.

Tim

dann2 - 15-6-2009 at 12:31

Got that totally wrong.
I would 25 turns on an MOT core and got 8mH (measured with a miltimeter, no DC present in the coil). Then I wrongly assumed I was getting 8/25 = 0.32 mH per turn. (I actually wrongly reported 0.35mH per turn (doubly wrong)).

Dann2

12AX7 - 16-6-2009 at 06:23

Ah, then that's going to be 8000 / 25^2 = 12.8 uH/T^2. But that's also initial permeability. Higher up the B-H curve, permeability rises (maybe about double), then falls as saturation is approached.

Tim

dann2 - 18-6-2009 at 19:10

Hello,

Got around to putting a 'Killawatt' meter on the input of a rewound MOT. The thing dissipates 160 Watts of power with no load connected to the output. The core gets too hot to touch. It draws 2.4 Amps at 234 Volts with a Cos Theta of 75 Degrees. When you put the Voltage down to 210 Volts the primary now only draws 1.0 Amps. I guess core is getting close to saturation as the number of windings on the primary is very skimpy.
This would be a problem with PVC covered wire as the PVC may soften enough to become cut by the sharp edges of the core. Keep it in mind. Better to use two transformers with the primaries in series if you have them, or a Variac on the input would be great.
The one in the picture can suppy 50 Amps (into a Graphite rod resistor, after a rectifier and 68000 Micro Farad capacitor) continous. I put 173 Volts into MOT using a Variac.

Doing some more testing.

Regarding the inductor I was messing around with (gaped MOT). When I wound on some turns onto the sawn-in-half MOT with a 0.4mm air gap, I can only get approx. 0.08mH which is a dissapointingly small inductance. It helps no doubt but I would like about 2mH minimum.
I have a large transformer from elsewhere. I might try that.
Setting up a 'raw DC' power supply to see just what sort of Current waveform goes into a Chlorate cell when powered with it.

See
http://www.qsl.net/i0jx/supply.html
for a description of the Resonant Choke Power Supply. You only need one tenth (approx.) the inductance for to get similar smoothing. Would this work for us in a Chlorate supply?

Cheers,

More later.

Dann2

dann2 - 25-6-2009 at 09:33

Hello,
Could get no satisfaction from an inductor made from a gapped MOT. The inductance that is obtained is just too small. Tried a large transformer but it is not worth the bother (IMO) unless you happen to have a large transformer core to hand.
See attached file for some observations regarding an MOT Chlorate cell supply.
Dann2

Attachment: MOT supply waveforms.mht (178kB)
This file has been downloaded 20518 times

dann2 - 28-6-2009 at 11:22

Hello,

During my travels on the interwebs I came accross the following (real mans) welder. It DID make me laugh. I actually though it was a joke but it is not. The transformer tied together with cord takes the biscuit. Then there's the welding goggles.................
http://www.jalopyjournal.com/forum/showthread.php?p=3385547

Can any helpful soul explain to me how a constant current transformer works. I am talking about one of the type that has two coils on the core. One coil is stationary and the other is allowed to move (aided with a counter weight) so that the current is regulated in a very exacting manner.
There is a doc. attached.
It does not make sense to me. I always thought that it did not matter too much where on the core you put the second winding (within reason). According to the constant current transformer stuff if you move the coils some inches towards and away from each other it makes a hugh difference and allows very precise current regulation????????
The core seems to look like an ordinary core but it is a bit longer to allow for movement of the coils.?

See here page 98

US 2807787 is another example.

Dann2

Attachment: CCR BROCHURE.pdf (653kB)
This file has been downloaded 895 times

watson.fawkes - 28-6-2009 at 14:33

Quote: Originally posted by dann2

I always thought that it did not matter too much where on the core you put the second winding (within reason).

Often yes, but really no, as you've found out from reading. The magic label in the diagram you referenced is the one labeled "leakage flux". This is magnetic field flux that is made by the primary but not picked up by the secondary. Controlling the leakage flux is the way that the device produces constant current.

dann2 - 29-6-2009 at 02:31

The core looks reasonable well dimensioned. It is not very thin or very long (it is somewhat longer that other transformers though), does not have any air gaps or shunts to incourage leakage flux. I simply cannot understand how simply spacing the coils a bit turns the thing from what we normally get (a 'normal' transformer) into something that gives out a constant current.
Anyhow, I guess it does. Some of those text books need rewriting when they show a diagram of a core with coils on it and state "It makes no difference where you place the coils on the Iron core as all (or almost all) the flux is contained within the Iron core" etc etc.
No doubt you would expect some differences to occur as the coils are moved closer together but not the hugh difference that these constant current transformers are giving, simple my moving the coils some inches.

Let me see: Two MOT's cut open, and two 'E's' placed together (hopefully no air gap)with one mains coil left on one of them. A few turns for the secondary and make the secondary bobbin winding in such a way that it can be slid up and down a bit for giving different currents. Can't see it working but I guess it will?????

Dann2

watson.fawkes - 29-6-2009 at 06:16

Quote: Originally posted by dann2

"It makes no difference where you place the coils on the Iron core as all (or almost all) the flux is contained within the Iron core"

Are you certain that the center post (the one that the secondary moves on) is made of iron?

Look up "magnetic permeability", and then look up its value for iron. The answer won't make sense unless you understand the concept and know the value for iron.

dann2 - 29-6-2009 at 08:00

Hello,

I am not certain that the core if made of Iron.
I have presumed (perhaps I am presuming wrong) that the core is made similar to other power transformer cores (the usual laminated Silicon steel). The only thing I can see different with cores for the constant current transformers it the fact that the core is a bit longer to allow (I think?) for movement of the secondary.
(This extra length will add very little reluctance to the core compared to a shorter ('normal' shorter length for more conventional power transformer) core of similar cross section if the core is made from Si Steel.)
____________
Edit:
Hipersil steel is used in the CC Transformer in the brochure above. It's a trade name for High Permeability Silicon Steel. Would presume the whole core is made from it.
___________
The whole concept beats me. I cannot find much any info. on what the cores are actually made from. There appears to be no magnetic shunts or air gaps in the core, just the moving secondary on Iron core. Will have another look for more info.

Dann2

[Edited on 29-6-2009 by dann2]

12AX7 - 29-6-2009 at 09:31

If the secondary remains in a constant position, there will be a certain amount of leakage flux around it, due to the gap between the primary and secondary. In current-limited transformers, this is often accentuated with a high permeability material in the gap (yielding a smaller package), i.e., magnetic shunts. Air works the same way, it just takes a lot more of it, hence the much longer travel necessary.

The other factor is the active control loop produced by allowing the secondary to move. This turns the characteristic from ohmic to nonohmic (over the long run), controlling current much more precisely. (Notice that, in the short term, the heavy secondary will not have time to move and fairly large peak currents can be drawn.)

Tim

merrlin - 29-6-2009 at 11:51

Quote: Originally posted by dann2

The core looks reasonable well dimensioned. It is not very thin or very long (it is somewhat longer that other transformers though), does not have any air gaps or shunts to incourage leakage flux.
Dann2

The gap between the core and the secondary that allows the secondary to float will increase leakage inductance. Since you seem to be putting a fair amount of effort into your project, I suggest that you download the student version of Ansoft's Maxwell 2D E-M modeling software and review the examples before building your own models.

http://www.ansoft.com/maxwellsv/

dann2 - 29-6-2009 at 15:57

Hello,

Thanks for replys and I will download and see can I use that software.

I guess the following is where the design of the CC transformer differs from a transformer designed for good Voltage regulation.(deliberately put into the design of this type of transformer as suggested by yourselves to obtain leakage flux):
Large spacing between the inner side of the windings and the middle core leg +
The core is longer than a transformer with good Voltage regulation of similar VA +
The coils being spaced apart from each other on this longer middle leg

I have attached a picture from the brochure for the CC Transformer.
It would appear from looking at it that the coils look to have alot of space between their innner (rectangular) winding layer and the core. There appears to be a spacer (the dirty yellow coloured thing that looks like a big ice cream wafer!) along the core middle leg to fill up the space? Just guessing from looking at it.

I have tried to give the MOT I was working with a more constant current nature by adding more magnetic shunting. Also added an air gap in the core (along with the shunts accross the core between coils). They all help but you still do not get a 'real' constant currrent transformer, just a transformer with less and less Voltage regulation and more and more current regulation. I would like a transformer with a more regulated (fixed) current supply value. I do not think this is possible with simple shunts and/or air gaps unless you have it designed to give out quite a high Voltage (say 50 Volts) when there is no load attached. When this (50V with no load) transformer is then loaded so that output Voltage is dropped way down from 50 to say 8 or less Volts then you will get a fairly well defined programmed current. ie. a large difference between open circuit Voltage and (roughly 6 to 9 Volts) the working Voltage makes. There is no space in the MOT for the number of windings to give me an open circuit Voltage of (say) 50 Volts.

Will try to combine two MOT "E" pieces (of similar size) to give a long core. Primary will be one MOT primary (which unfortunately will be tight on the core which I do not want) + secondary on a moving bobbin. May work / may not, like a pipe dream.

12AX7 - 29-6-2009 at 17:22

Quote: Originally posted by dann2

They all help but you still do not get a 'real' constant currrent transformer, just a transformer with less and less Voltage regulation and more and more current regulation.

That is exactly how a current source can be defined. An ideal current source's Thevenin equivalent has Vs/Rs = Iout and Vs, Rs --> infinity. A real-life approximation is not only supposed to have a high O/C voltage, it is required to! Notice this CC transformer will likewise have a high OCV.

Tim

[Edited on 6-30-2009 by 12AX7]

watson.fawkes - 30-6-2009 at 16:47

The vertical member in the brochure looks like phenolic to me, not iron. The difference in permeability is the difference between "regulates" and "does not regulate".

dann2 - 30-6-2009 at 17:49

Another picture of CC transformer below from
http://www.tpub.com/celec/5.htm

The diagram suggests the middle leg is part of the 'core', ie. made of Iron.
The permeability of Phenolic would be far too low IMO. The input coil would look like a short circuit and bugger all coupling to secondary would occur.
Dann2

[Edited on 1-7-2009 by dann2]

watson.fawkes - 1-7-2009 at 05:17

Quote: Originally posted by dann2

The diagram suggests the middle leg is part of the 'core', ie. made of Iron.
The permeability of Phenolic would be far too low IMO. The input coil would look like a short circuit and bugger all coupling to secondary would occur.

Coils have inductance even in complete absence of a core, so-called "air core" coils. There are also air-core transformers, albeit specialty items. A core changes the inductance of a coil by a factor of its permeability, but this doesn't eliminate the inductance. Transformer coupling is about the ratio between mutual inductance and total inductance. And a core doesn't change the DC resistance at all.

This whole device hinges on an interplay between geometry of the device and permeability of the material. My intuition about the magnitude of permeability indicates that you're not going to use a maximally-permeable material, but I might be wrong about that. The reason you use a core is to minimize leakage inductance, which increases the linearity of the transformer. But a constant-current source is as non-linear an element as it gets!

If you're trying to regulate over a large range, using an iron core may not be what you want. But at this point I'll simply second the recommendation to run a simulation.

dann2 - 1-7-2009 at 12:39

Hello,
I put together the device in the picture. It is two 'E' MOT pieces on top of one another (laminations not interlayered, there are three gaps). It does not have alot of range. As I start to add more air gap, the range goes up a bit. Maximum air gap size I have used in 0.8mm. When I say air gap I actually mean three air gaps, one in each leg (middle core and two outside legs). It would be best if I only gapped the middle core (correct?). I will have another go at the thing with the middle core (damm near) removed. I can obtain the two 'I' pieces and put them into the two outside legs. I have a choice of three thicknesses depending on what way I put the 'I' pieces between the top and bottom 'E's. This will make the overall core much longer with a hugh gap in the middle (core) leg. A picture will explain better next time.

The set up pictured gives a noticable force between the primary (bottom) and home wound secondary at 15 amps and above. The secondary is lifted to the top at 25 five amps or even less. The vibration (noisy) core helps to stop the (not very precisionly built) secondary bobbin from jaming. The 'control' is yours truly moving the coil at this stage.
I cannot see this thing being something anyone would want to build as you will need quite a sweetly built sliding bobben/weights/pivots etc, system.

There is another picture of a CC Transformer here page 352.
Hard to figure out what the picture is saying about the core type/shape, or if it is trying to say anything at all.

Thanks for replys.

Dann2

watson.fawkes - 2-7-2009 at 06:36

If you'd like to modify your transformer for greater range, take out some of the iron in the center section. If the layers are stuck together with some resin (as typical), you should be able to notch the sheet metal, get a tool between layers, and peel off the layer. There's no need for a reduced core in the center of the fixed primary; the rest of the center section is what to reduce.

12AX7 - 2-7-2009 at 08:29

Mind that adding air gap will increase magnetizing current even further -- and it's usually over the edge already.

Tim

Vegemeister - 3-7-2009 at 06:35

I think y'all are greatly overcomplicating matters. The cell couldn't care less about regulation. As long as it gets DC, it's happy. Heck, if you're using MMO on Ti for both the anode and the cathode you can probably give the cell AC and it wont notice.

I have much more MMO plated expanded Ti than I have bare Ti or welding equipment, so I use an MMO cathode and make mechanical connnections above the surface of the electrolyte and protect them with hot melt glue. I have not yet implemented this AC plan, but I hope it might do something about the carbonate collecting on the cathode. Try it at your own risk.

You really don't need to worry about constant current. If it concerns you, put two splices 1 foot apart into a lengh of 10 AWG copper (preferably solid core) and put it in series with the cell. the potential between the splices in millivolts is a reasonable approximation of the current in amperes. Do a trapezoid approximation of the current*time integral if you want to calculate your current efficiency.

Your main concern should probably be reducing losses in the transformer and resistive losses from unnecessary overvoltage. Microwave oven transformers are, by design, very non-ideal transformers. The plate current in domestic oven magnetrons is regulated in a manner which depends on the transformer being operated well into saturation. Also, copper and silicon steel are at the expense of the oven manufacturer while electricity is at the expense of the consumer. These transformers, while accessible, are quite under-engineered. My recommendations are as follows:

1. Knock out the shunts.

2. Wind some extra turns on the primary (make sure you put them in phase with the rest of the primary). Or, put two transformers in series, both windings.

3. If you want to go DC, use a center-tap full wave rectifier, not a bridge, and use schottky diodes. This means one small schottky drop in between the transformer and the cell instead of two big pn drops. With the voltages we're running, those diode drops contribute significantly to inefficiency. A center tap is easily made by using a bifilar winding.

If you use two transformers in series, DO NOT use the point in between the two cores as the center tap. If you do this, each transformer core will only see DC. Instead, make a bifilar winding that includes both cores.

4. You should probably not use a filter capacitor. This forces the transformer to deliver large pulses of current when the output voltage exceeds the bus capacitor voltage. Transformers do not like this, and the problem gets worse the bigger you make the capacitor. Without a cap, the tranny still sees pulses as the output voltage exceeds the cell voltage, but a capacitor is worse. Furthermore, getting ripple under control at tens of amps and 60 Hz requires a very expensive capacitor. The cell really doesn't mind.

5. Adjust your tuns ratio such that the current (once the cell temperature has equilibrated) is reasonable for your electrode surface area.

dann2 - 19-7-2009 at 17:07

Greetings Cabbage head,

One step in the MOT conversion that you left out is getting rid of the secondary winding. I find it is inclined to take up too much room if you leave it there

How do you bifilar wind the two transformers to get the center tap in such a way that you will have no 'transformer seeing DC only, problems'.

I set up the two MOT cores as the last picture above shows and tried it with no air gap, 0.4mm gaps and 0.8mm gaps. I only put half the normal input Voltage into the MOT using a Variac.
I used a large Graphite Anode in a Chlorate cell and moved the Anode in and out to simulate the Voltage variation accross a Chlorate cell. I only varied the Voltage by (approx.) 0.85 Volts.
To make a long story longer the best results were obtained when there was a 0.8mm gap in the transformer (in all three legs).
With the moving coil furthest away from the primary:
The current when Anode was in the cell fully was at 15 amps, 3.25 Volts accross cell.
Then the Anode was moved out, so that Voltage accross cell was now 4.1 Volts and current into cell had dropped to 11.1 amps (simulate a rising Voltage accross cell).
The coil was now moved untill it was at it's closest to the primary and the current went up to 17.5 amps with 4.63 Volts accross cell.
So I suppose there is scope to use a contraption like this to controll the current into a Chlorate cell and keep it constant if you added a smooth mechanism to allow the output coil to move but it would simply not be worth the bother. The Voltage variation simulated was very small anyways and it would not be representative of the wider Voltage variation you will get with a Chlorate cell when temperature, concentrations etc change.

Will try larger gaps in middle leg with no gaps in outer legs if I get the inclination.

Cheers,
Dann2

[Edited on 20-7-2009 by dann2]

WizardX - 19-7-2009 at 18:12

This power supply is great for powering all kinds of electronic projects. It produces a well filtered, variable 1.2-30 volts at 5 amps.
http://www.aaroncake.net/Circuits/supply.asp

dann2 - 19-7-2009 at 19:28

Hello,

I hate being a party pooper BUT that supply will not supply five amps and avoid meltdown of the LM338.

It will supply 5 amps (guaranteed) when the Voltage between output and input is 10 Volts or less with the case at 25&degC (large heatsink). If the Voltage between input and output is 30 Volts it will only supply one amp. It's all about power dissipation in the device and how much heat is generated. See the data sheet page three forth row of the table up from the bottom. Or the graph just below table on left (page three).
Its a mistake made quite a lot to assum that when a device is declared to have a (say) 5 amp capability that it can output 5 amps under all input/output Voltage conditions. It will go on fire if you put 30 volts accross the device from input to output and draw 5 amps out of it. (This device wont burn up because its internal circuitary will limit the current (it will stop working as you would like).
30 (Volts) * 5 amps = 150 Watts = device on fire!!!!!!

http://www.datasheetcatalog.org/datasheet/stmicroelectronics...

Dann2

CC Add-on

dann2 - 6-10-2009 at 16:58

Hello,

Got around to putting together a constant current add-on for a computer power supply. Works OK. The maximum amount of current the thing can put out will depend on what Voltage you will ask the IFRZ44N to drop but using the 5 Volt output of the PC supply and connecting to a Chlorate cell it will be OK for a modest 10 Amps. The IFRZ44N is heatsinked using a microprocessor heatsink with a small fan on the back.

R1 consists of 50 cm of 1mm Copper wire. A 741 op-amp will not do this job as it will latch up to the power rails. I used a CA3140E.

Current regulation is good when tested with an 0.3 Ohm resistor as load. When 10 AMps out was set it also give 10 Amps when the resistor was short circuited.

The whole thinn should fit inside the PC Power supply case.

Dann2

Attachment: cc.zip (83kB)
This file has been downloaded 583 times

12AX7 - 7-10-2009 at 07:44

The LM358 includes the ground rail, and even this lowly op-amp has better performance than a damn 741. CA3140 looks like a good choice, although like all MOS amps it's quite noisy.

Copper is a very poor choice for shunt because it has a massive tempco. Nichrome is much better, and better still is nichrome bonded to a base, especially a base you can heatsink. Aluminum body resistors are fairly expensive, but do a good job (and they're usually 1% too).

If you'd like more current handling, you can of course connect any number of these circuits in parallel. You might consider using larger transistors, as the TO-220 package is really only good for 50W. TO-247 is good for 100W. A couple hundred watts is a lot to dissipate and you might also consider a circulating water supply for them -- saves on expensive heatsinks, and a bucket of water dissipates heat surprisingly well.

Tim

Swede - 7-10-2009 at 14:12

I'm a bit late into this thread, but for rewinding transformers and cabling cells, there are two outstanding sources of excellent heavy Cu wire that is both flexible, and in one case, has a nice, thin PTFE insulation. The first is welding cable, traditionally used to ground the work. Most good shops have big spools of #6 through #0 stranded copper that is limp as a noodle, very easy to wind, although the insulation jacket is fairly thick. Still, it is a MUCH better bet than household #2 Cu cable, which damned near requires he-man muscles to manipulate.

The other alternative is scrapped or surplus aircraft cable. They are almost always insulated with a PFA or sometime kapton sheathe that is thin and allows for a higher number of turns. One last option is to strip the welding cable of it's heavy rubber jacket, and insulate with a spiral wrap of Kapton tape.

I agree that a decent power supply is probably the most challenging thing a new experimenter must obtain. 20 amps or less is easy. 50 to 100 amps into 25+ liters, much harder. Having constant current is a joy if you are into calculating efficiencies, and experimenting with cell parameters to see how they affect CE.

One last thought - 500W to 1KW 5V fixed supplies are a dime a dozen on eBay, and can be found for $25 to $50. The vast majority have a trimmer pot that allows voltage adjustment through a narrow range. Most that I've seen allow 4V to 6V, a nice range for a chlorate cell. I've been tempted to export that trimmer into an external 10 turn pot, or at least a better potentiometer for everyday use, although I'd probably derate the supply by 20% or more when wanking with the voltage like that.

dann2 - 7-10-2009 at 15:41

Hello,

What do you mean by 'The LM358 includes the ground rail' ?

I actually did not choose the CA3140, it was what I had to hand (after messing with the 741 for some hours). Lucky guess I guess.

I thought the TempCo would matter damm all with this circuit as we want a relatively crude current out (not precision anyways). Setting the current at 10 Amps and holding a lit match under the Copper coil (0.01 resistor) the current droped to 7 Amps with just some heating. I then heated the resistor with a hairdryer, which would be a more realistic simulation of operating conditioins and the current dropped to 9 Amps. Will have to get out the piggy bank and purchase some Nichrome. I have 28 SWG Nichrome but it would take too many strands or/and the resistor would be too short and difficult to connect when constructed.
There is a wire resistance table here:
http://www.kayelaby.npl.co.uk/general_physics/2_6/2_6_2.html

I am tempted to try a small Graphite rod. TempCo is low (and negative), metals are +).
http://library.thinkquest.org/10784/tempcoeffs_resistivity.h...
Whats ya's think?

If you have a physically small meter (say 1mA FSD) and you want it to read 10 Volts exactly when your system is OK, it can be difficult to read the small meter. Is there any way to connect the same meter in such a way that it would read from 8 Volts (zero on meter) to 12 Volts (FSD).
This would greatly increase the resolution of small meters in the region where you want it.
A Zener in series with a smaller resistor (than you would use if there was no Zener) in series with the meter sounds like it would do the job. A small meter in the PC power supply case would be great.

About those hugh 4 to 6 Volt power supplies. I often though that you could turn them into a constant current supply by connecting a stepper motor to the adjustment pot and sensing the current and moving the stepper accordingly. This would require a microcontroller. I am messing aroung with PIC's at the moment. It would be a good project to take on and more importantly a GREAT CAUSE :cool:

Don't hold you breath for the final circuit draft..........

Cheers,
Dann2

[Edited on 7-10-2009 by dann2]

12AX7 - 7-10-2009 at 17:49

Quote: Originally posted by dann2

Hello,

What do you mean by 'The LM358 includes the ground rail' ?

The input common mode range. 741 and most FET amps (LF411, TL072, etc.) go from 2V above the negative rail to 2V below the positive rail, both input and output. Some amps are specified for operation including one or the other rail. The LM358 is good for -0.3V input. Most op-amps, when taken out of their common mode range, do weird things, like invert the output, or slam to +V.

Quote:

I am tempted to try a small Graphite rod. TempCo is low (and negative), metals are +).
http://library.thinkquest.org/10784/tempcoeffs_resistivity.h...
Whats ya's think?

Might work, but resistance varies with pressure (maybe unless you have a very dense piece of graphite). And you need a fairly thin piece for the current you're at. And you need to make contacts -- which isn't the hardest part, as you can plate copper onto graphite very nicely.

Quote:

If you have a physically small meter (say 1mA FSD) and you want it to read 10 Volts exactly when your system is OK, it can be difficult to read the small meter. Is there any way to connect the same meter in such a way that it would read from 8 Volts (zero on meter) to 12 Volts (FSD).

Get a better meter. Or subtract 8V from it so it reads 0 to 4V. (To cover the negative range, a series protection diode would be wise.)

Quote:

About those hugh 4 to 6 Volt power supplies. I often though that you could turn them into a constant current supply by connecting a stepper motor to the adjustment pot and sensing the current and moving the stepper accordingly.

Wow, that's a terrible waste. I bet you an entire supply that they are voltage controlled. The pot changes either the feedback or reference voltage. Simply replace it with a control voltage input, or an error amp in the former case, and there you go.

Tim

dann2 - 12-10-2009 at 02:14

Hello,

I did not think of using the -12V from the computer power supply. When the + and - 12V are used to power the op-amp then all the problems the 741 was having disappeared.

Also changed the values of the voltage divider at the input to the op-amp. I cannot obtain a 50 Ohm pot with a shaft for a knob anywhere. I can easily get 1k sliders and pots.

Job completed. I used a second cps box to house the bits and pieces + a meter.
The whole shaboo is a very salubrious, austintatious, plush piece of engineering! (as you can imagine).

Dann2

12AX7 - 12-10-2009 at 09:06

Yes, a 741 will be happy with +/-12V.

Tim

dann2 - 12-10-2009 at 17:55

Hello,

Final circuit attached. It varies current out from 0 to 15 Amps when the 5 Volts output from the CPS is used.
It cannot do that if you supply the 12 Volts from the CPS as the 12 Volts output on my supply will only supply 8 amps.

I put back the 741 as it works OK.

Just wondering could this circuit be used to 'smooth' raw DC coming from a rectifier?

Cheers,

Dann2

[Edited on 13-10-2009 by dann2]

12AX7 - 12-10-2009 at 20:20

It can be used for smoothing only as long as the voltage is high enough. You can't make voltage with it, it stores no energy. It can allow you to use less storage, so ripple can be a larger part of the supply voltage without modulating load current. Simple way to test: put an AC meter across the current shunt resistor (be sure it's an AC meter, not reading the DC component as well!), and connect more capacitors until AC goes to zero.

Notice if the circuit is oscillating, it will have quite a bit more AC on it, though it could also be past the meter's bandwidth. A 741 is pretty slow, but just in case, a series resistor between -in and Rshunt should be added, 10k is enough, and a compensation capacitor (100pF is enough) from 741 output (in front of the gate resistor) to -in, to basically make an integrator for high frequencies.

Tim

dann2 - 26-10-2009 at 10:48

Hello,

Great big lump of a power supply at link (in UK).
http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=200...

Constant current mode, 7.5 Volts and lots of Amps for the Chlorate and Perchlorate making guru.

Dann2

Swede - 15-12-2009 at 07:49

Quote: Originally posted by dann2

Hello,

Great big lump of a power supply at link (in UK).
http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=200...

Constant current mode, 7.5 Volts and lots of Amps for the Chlorate and Perchlorate making guru.

Dann2

I'm very late with this reply, and it no longer matters, but whoever stole that Lambda supply for 42 pounds sterling is going to be one happy camper. Lambda made a series of those switching supplies in various voltages, and they all have CC/CV operation. I have one, 10V 100A, that is my favorite supply.

There are a lot of them out there, available at really good prices with a bit of searching. Not trying to slam PC power supplies, but when you add it all up, something like this Lambda is tough to beat for electrochemistry experiments.

Edit: This came off sounding bad. I am a huge DIY fan, and the advanced projects here are excellent, but I point this out more for a beginner who doesn't have the skills, will, or time to put something together. A powerful commercial supply doesn't have to cost hundreds and hundreds of dollars.

[Edited on 15-12-2009 by Swede]

12AX7 - 15-12-2009 at 10:20

Indeed, switching power supplies are not easily designed by novices.

Tim

dann2 - 15-12-2009 at 18:42

Hello,

There is still an opportunity to shove one under your coat and skedaddle!!!!!!!

That guy has been selling them for quite some time. He must have a number of them.

Dann2

[Edited on 16-12-2009 by dann2]

Swede - 16-12-2009 at 07:18

Quote: Originally posted by dann2

Hello,

There is still an opportunity to shove one under your coat and skedaddle!!!!!!!

That guy has been selling them for quite some time. He must have a number of them.

If I lived in the UK, I'd be all over that supply. Two in series would give you anything (voltage-wise) you'd probably ever need. One alone, at 7.5V, should work fine for a well-engineered perchlorate cell. For chlorate, it's more than adequate.