A canister made without a lathe

Eskil

Newly Enlightened
Joined
Feb 11, 2019
Messages
16
Location
Denmark
Hi guys,

Thanks for your help on various aspects.

I'd like to tell a little about my build, struggles and choices, some of it might help you not to run into the same problems.
image034.jpg

My goal was to build a canister for heating, based on available plumbing materials, as I have no access to a lathe.
The choice fell on 63mm GF industrial PVC 10bar pipe.
As I have no lathe, it's hard to make a location for an o-ring. So I accepted to glue it together and charge it through the e/o cable.
To me it seems like the O-ring where you open for every dive is the vulnerable spots in camera housings and lamps. So I'd prefer not to have to open the water tight barrier every time.
On the other hand, you miss the possibility to look inside, to notice if something odd is going on.

It can hold 7 18650 cells in one "layer" or you could call it a flower, a ring of 6 cells with one in the middle.
Unfortunately minimum buy is 5m, so I have some for new experiments...
image036.jpg

Selected Samsung 3400/3500mAH cells.
Going for a Santi-compatible heat vest, requiring "12V", which is 3 cells or nominal 10.8V. And 3 hours of use (two dives and surface interval) with 40-50W of power consumption prescribes 4 parallel cells.
This can be done in the selected tube with two "layers" of cells with 6 in each.


Next step was to point weld the batteries.
We have some serious lab powersupplies for automotive eqpt. at work, 15V 100A, so I thought that could do it - optionally two in parallel. It just heated the strips - no welding.
So I had to make a welder:
image003.jpg

I had this 500VA ring core trafo at home from an AMP building project nearly 30 years ago. On a ring core, you can simply add a new secondary winding by winding through the hole, you don't need to remove the old secondary to get space for it - just don't connect the old secondary windings.
It makes 0.5V/turn, I saw some other projects doing 1.5-3V, 2V measured during welding at a professional unit, so I went for 2.5V. 5 parallel wires of 4 square mm, 20square mm Cu in total.
Primary goes through a spring loaded lighting switch (those for kip relay lighting systems), used as a foot switch. A very short bump on the foot switch is enough:

image004.jpg

Now, as I found a nice protection board on ebay that would fit in the middle, and as the only place in these pipe end cups to place a gland or a piezo switch is in the middle, I went for the 6 cells in a ring setup - this should later prove to be a bad decision..

OK, next step is the gland. Wanted something with a long thread to be able to put a counter nut on the inside. And capable of handling Thor 9.6mm e/o cable
https://www.deepstop.de/en/light-accessories/945-eo-cords.html

Couldn't find Agro from reasonable sources. One of the few I could find was this:
https://se.lappgroup.com/fileadmin/documents/technische_doku/datenblaetter/skintop/DB53112005EN.pdf
After I bought them, I found that they seal using plastic against a brass cone, and I simply don't trust it.
I then heard that "blueglobe" glands have a good reputation among rebreather divers. So I bought some of those:
https://www.ibs-gruppe.de/shop/kabelverschraubungen/blueglobe/776/bg-816ms
They seem to work.
IMG_3682%5B1%5D.jpg

I grinded the typing away, drilled in the middle, and made a M16x1.5 thread. Got a nice 7-8mm deep thread there.

Now in to the electronics.
I bought these protection and balance boards:
image015.jpg

I got a bit suspicious when I saw that the power MOSFETs had the same name, but clearly were from different factories.
Currently it's hard to get MOSFETs due to allocations, and I've seen fake MOSFETs at work - re-branded, much weaker MOSFETs.
So I wanted to test functionality and power dissipation.
I didn't find any balancing current. So I made a schematic of the device, and realized that there is no balancing circuit on the PCB, even though the seller claims it.
Arrgh..

I had a very big 3S protection/balancing board in the drawer.
to use it, I had to re-arrange my 6 cells, and also cut some of the board away, and replace some power tracks with wires...
image018.jpg

image017.jpg

Continued in the next post
 
Last edited:

Eskil

Newly Enlightened
Joined
Feb 11, 2019
Messages
16
Location
Denmark
Now, I wasn't fully dicided on the method for power on/off.
I've bought some piezo switches. I could have one in the bottom cap of the canister.
But that would require a power LED to show if the cannister is on or OFF, especially for charging.
When making a power switch with a MOSFET, it will turn load current on or OFF, but current the opposite way (charging) would see a diode voltage drop. So if I set the battery to charge when it is OFF, it would charge a little and then appear charged - too early. While the charger sees 12.6V, there is only 11.9V on the battery.
On the other hand, if there is a power LED, there is a power consumption when the canister is accidentially turned on.
I didn't find an easy way to make a LED window. I considered clear PC plastic screws into a thread in the cap. But PC is brittle, and I didn't find a glue claiming it could glue PC and PVC together.
This made me go back to the good old reed tube - an external magnet will be a mechanical indicator of power on or OFF.
I wanted to use the MOSFETs on the protection board as the power switch also - no reason to add extra FETs, I thought.
I noticed that my protection board had a set of terminals marked T - and I found I could remove the gate voltage there with a reed tube, turning power OFF.
image022.jpg

image019.jpg

It worked.
I added an LED on the power out cable as indicator. And then it didn't work any more. I could only switch unloaded.
A protection latch turns OFF the MOSFETs of there is voltage over them - made to latch the protection in case of error, until the load is removed.
Now, added a little capacitor/transistor circuit to reset that latch in the moment the reed tube goes on:
image027.jpg

Now it worked.
Until I connected a heavy load. Now the over current protection detects too high voltage over the MOSFET when the MOSFET is in the process of turning on.
Yet another ARRRGh.
We had a suitable 3miliohm 25V MOSFET in the drawer at work, but, in a horribly small package: a DirectFET:
image028.jpg

IMG_3687.jpg

3miliohm will dissipate 300mW at 10A, but Directfets are not a pleasure to solder, and that source terminal was like 1x1mm, where I wanted 10A out.
Soldered it upside down to a small un-etched PCB.
I put a 100kohm resistor source-gate, and reed tube to pull up. That would dissipate an acceptable amount of power if accidentially ON in the locker. But it would make the MOSFET turn off rather slowly, and dissipate much more power when in the process of turning OFF.
Testing it proved that it didn't feel like it heated up with a 6A load, and also importantly: I didn't feel any heating of the FET when repeatedly turning on and OFF.
In other words, it might have been possible to use a larger resistor value.
100K will use 120uA, which will last 13 years from my 13.5Ah battery if accidentially left on. I'll better remember to turn it OFF every time ;-)

Now, as this canister will be glued together - can only look at the batteries if I saw it open - I wanted a safety valve.
As DIWdiver has pointed out, a first stage OPV doesn't have the area required by UL. Anyways, I bet it will in most critical occations be much better than nothing. And I have the correct thread tap. So I went with one.
Made a flat surface on the cap, drilled and made the correct threading. Counersunk a little to make a cone for the O-ring.
Pressure testing Tuesday. Diving ASAP.
 
Last edited:

Eskil

Newly Enlightened
Joined
Feb 11, 2019
Messages
16
Location
Denmark
Hey, I forgot to show my magnet ring, to hold the magnet for turning on and off through the reed tube:
image035.jpg

I made this ring, where I made a chamber for the magnet in the PVC, and made a PVC glued cap/"button to feel". I hope that as the PVC glue makes a kind of welding, it will be sufficiently watertight for the neodymium magnet to stay well.
I made a hidden grove in the ring, and put an A4 stainless steel pin in the bottom cap to make end stops for the ring movement.

I would also like to add a little comment on balancing - how do you see on the board that there is balancing on it, and how much balancing current is needed?

So the interesting thing here are those black square power resistors on the board. On my large board there are 5 of them, 2 marked R005 (that is the over current sense resistors). Then there is another group of 3 marked 101 (100ohm).
For balancing, there'll normally be the same number as cells in series.

When a cell reaches something like 4.2V, the resistor will load that cell, so those cells that haven't reached 4.2 yet can still be charged, while this cell is kept at 4.2V.
This only works if the charging current when the first cell reaches 4.2V is the same or less than what the resistor drains.
When I charged my pack, I found that the first cell reached 4.2V when charging current was a bit over 100mA. Now as this balancer uses 100 ohm, its balancing current will be 42mA. So with my pack, it's not enough...
I have to admit that even though I should have changed the resistors for something like 39ohm, I didn't have 1812 size resistors of 39 or 33ohm, and wanted to finish it.. so I just closed my eyes.
But next time I will sure want a balancer that can do more power.

I hope the description was entertaining.
I plan to be back when I've got my PCB for my power regulator home.
The principle is pure pulse width modulation - turning power on and off with a really low switching frequency. The reason for that is that if you do like that there is no coil power loss, an much less switching loss. And you can drive the big power MOSFET directly from simple logics/opamps/comparators. So I'll switch at close to a Hz. Then you can also see on the LED if you are modulating. (edited to add extra notes)
 
Last edited:

Eskil

Newly Enlightened
Joined
Feb 11, 2019
Messages
16
Location
Denmark
I've added more descriptions, including drawings on some of the pictures, to explain things better.
I must say I'm really overwhelmed by the number of comments;-) No, basically I hope it can help someone.
 

DIWdiver

Flashlight Enthusiast
Joined
Jan 27, 2010
Messages
2,725
Location
Connecticut, USA
Hey, nice build!

I think that's about the simplest battery welder build I've ever seen. Most of the build videos are long and overly complicated. Yours is just two stills and four sentences!

Don't feel too bad about the lack of replies. The Dive Lighting forum has been a ghost town for quite a while now.

I'm not seeing where you got 10A from. 40-50 W at 10ish volts is only half that. Plus, at 10A you'd drain that pack in less than an hour and a half, not the 3-4 hours you were aiming for. Oops, got it. 10A modulated down to around 50% duty cycle.

As I learned after building my first magnetic ring for a dive light, neodymium magnets are actually mostly iron. Iron plus salt water spells disaster. The nickel coating ought to protect it, but it is rarely perfect. I think mine worked for two years before they would no longer turn on the light. By then, they had expanded and could not be removed. Since they are ceramic, you'd need carbide or diamond to drill or grind them out. Since then I've been using samarium cobalt for anything that is exposed to the elements. More expensive, harder to find, and not quite as strong, but they don't corrode at all in habitable environments.

Sounds like your experience with the protection circuit is about like mine with blender firmware yesterday. Issued three firmware updates to the customer yesterday (the fact that it was April Fool's Day truly was a coincidence). This morning they told me they need another change. I told them it already did that.

Oh, and I imagine the T terminals are for a thermistor. Batteries get hot, thermistor resistance drops, current is cut off. The reed relay should have worked, as charge/discharge should be re-enabled when temperature drops.

I wish I could help you with the balancing issue. I did some research on it a couple of years ago, and came up pretty empty. I ended up designing something a bit more sophisticated, but similar to what you described. I'd share it, but the design belongs to my customer. I'm told it's going to UL next week...
 
Last edited:

Eskil

Newly Enlightened
Joined
Feb 11, 2019
Messages
16
Location
Denmark
Hi DIWdiver, thanks for your comment!

Congratulations on your design going to UL - I know the feeling! (just a pity they have to destroy so many nice units :)


I got it pressure tested yesterday, and it passed - so on to the water!
The 10A is just a safety margin, to (try to) make sure I don't realize tomorrow I should have tested at higher current. It is after all a rather "final" build as it is glued.
Good point with the Thermistor/PTC.

About balancing and protection, I'd prefer to build my own.. or maybe use the new TI chip bq77915. But I also wanted the project to succeed, be finalized, not go to sleep on some stupid hindrance.

I read your comment on neodymium magnets somwhere -thanks for sharing! I've also replaced magnets in china lights.
I made the precautions here that:
1: the hole for the magnet is way larger than the magnet
2: hoping that the PVC gluing a chamber (as opposed to epoxying the magnet into a hole) would be more waterproof.
3: As I can't replace the cells, I guess the canister will have a limited life - hopefully equal to normal cell life. But I can rather fast make a new one now.
4: Time kills DIY projects.. I had a useable magnet in my hand :) Might try to get a samarium cobalt magnet next time!
 
Last edited:
Joined
Feb 1, 2008
Messages
889
Location
US
As I learned after building my first magnetic ring for a dive light, neodymium magnets are actually mostly iron. Iron plus salt water spells disaster. The nickel coating ought to protect it, but it is rarely perfect. I think mine worked for two years before they would no longer turn on the light. By then, they had expanded and could not be removed. Since they are ceramic, you'd need carbide or diamond to drill or grind them out. Since then I've been using samarium cobalt for anything that is exposed to the elements. More expensive, harder to find, and not quite as strong, but they don't corrode at all in habitable environments.

look for ''plastomagnets'' http://www.arelec.com/en/plasto_detec (3).htm

my company also produce them but not in some usable shape ; they could be very strong
 

KrameE

Newly Enlightened
Joined
Jun 22, 2013
Messages
37
I just logged in for the first time in almost 5 years get another comment on this post. This is an awesome build. Hope it's working out for you!
 

Eskil

Newly Enlightened
Joined
Feb 11, 2019
Messages
16
Location
Denmark
Hi KrameE,
Yes, thanks, so far it works excellently. I've moved to the classic canister mounting method, but aside from that, all is good.
I find the heating level fine, so I put my controller project on hold.
 

arek98

Enlightened
Joined
Dec 21, 2006
Messages
424
Location
New Jersey, USA
Hi Eskil.
Nice build. I just saw you thread first time, it is a ghost town :)

Do I understand correctly that for welding batteries you just used this trafo core and DC input and it works?
That's quiet ingenious. It means that applying DC to primary creates pulse in secondary that is strong and long enough to weld.
I'm I right or I missed something?
It would also mean that switch is not important, just touch welding spots with electrodes and turn input DC. Secondary will get only an impulse, any current in secondary is gone after DC input stabilizes.
 

DIWdiver

Flashlight Enthusiast
Joined
Jan 27, 2010
Messages
2,725
Location
Connecticut, USA
I think he tried DC power supplies, found they didn't work, then switched to the AC powered xfmr. At least that's how I read it.
 

Eskil

Newly Enlightened
Joined
Feb 11, 2019
Messages
16
Location
Denmark
I think he tried DC power supplies, found they didn't work, then switched to the AC powered xfmr. At least that's how I read it.
Exactly.
Later, I've received one of those Chinese triac based welder controllers, and they make it a bit easier to dose the welding time than the manual foot pedal. But it also worked fine manually.
 

Eskil

Newly Enlightened
Joined
Feb 11, 2019
Messages
16
Location
Denmark
I just made some interesting observations, when replacing cells in a canister for a freind for a yellowdiving heating vest.
1: Both the original Sartek and Hollis canister had no protection (or it might be cells with built-in protection in each). I'm a little bit in chock..
2: The Yellowdiving vest has attenuation electronics that you control by turn off-on of the canister. It works just the way I planned to make my regulator, by low frequency hard switching (square wave). I wanted to go that way, since this will give the lowest loss=heating. And will make switching the FET very simple. And as that should be just fine for a resistive load vest. They switch at 2.3Hz. Now, the interesting/scary thing here is that at 100%, it only runs at 40% dutycycle, meaning that the current is very high when it is on, I measured 10.8A on the top of the square. So when you expect to deliver something like 4A 12V for 50W in the vest, you actually need to be able to deliver 11A. Just where I would typically place a current limit.
IMG_5675.jpg
 
Top