Ex-Goose Battery conversion

[ex-Gooserider]

Batteries have arrived - more details later... Haven't checked all, but every one I've looked at has had a 2017-4-21 date code, so it looks like all are from the same lot, and about as 'fresh' as one can expect...

Only problem was one holder had a couple of it's legs broken off, fortunately I ordered a couple of extras.

Now I need to start on locating all the software for the Hyperion and get it going on charging them.

ex-Gooserider

[Burgerman]

Software

[Burgerman]

firmware

Non of this will run on a betamax. And probably XP as its ancient! Nearly 2 decades out of date. Works fine on 7 or 10 64bit though.

[ex-Gooserider]

Thanks for the software links - I think I may already have found the right software on the Hyperion website, but will check to see if what you posted is more recent...

I have the software running on an ancient Dell 'palm-top' mini-notebook, running XP - not at all sure it could run anything more recent as it is using an Intel "ATOM" CPU... (I use the same machine to run the Pilot+ programming software...)

I think I have the charger set up right - I'm currently bulk charging the cells 4 at a time in parallel, using a jig I made (more on that later)
I'm charging to 3.6V nominally, the display keeps flipping between 3.599 and 3.601V... I set the memory to 60,000mAh (4 x 15000) and the maximum charge rate to 20A - but the display seems to show a max rate of about 3.5A which falls off rapidly. So far the charge reached '100%' in about 35 minutes, but I let the batteries soak for a little over 2 hours, when the charge current display is around 0.050mA... Is this a good stopping point?

I noticed the Ri value seems to start around 50 mOhms, but seems to go up to about 70-75 as the charge goes up... Is this normal?

As promised a bit more details (with pictures) on the batteries....

I received the order in three boxes - UPS tracking had them listed as 12kg each, but this doesn't seem to match the weights I felt picking them up, or what I found for contents...

The UPS tracking was a bit strange in that for several days it said that the shipping info was generated in China, but no shipment was actually in their hands - and then suddenly it was about to be delivered... I just checked the tracking record in UPS, and the shipment seemed to come from California... - the record shows:

Code: Select all

NORTH BILLERICA, MA, US    05/16/2017    5:42 P.M.    Delivered
Chelmsford, MA, United States    05/16/2017    2:32 A.M.    Out For Delivery
Chelmsford, MA, United States    05/15/2017    11:34 P.M.    Destination Scan
   05/15/2017    9:40 P.M.    Arrival Scan
Baldwin Park, CA, United States    05/11/2017    12:30 A.M.    Departure Scan
United States    05/03/2017    7:00 A.M. (ET)    Order Processed: Ready for UPS


Note the California, USA return address....

The packaging was also strange - it didn't look like the usual yellow cardboard from China, and wasn't in a typical 'buisness' uniform size box - it looked like home-brew recycled packaging, including resizing boxes to fit....

Interesting - ABSOLUTELY NO Labelling to indicate Lithium Batteries as content....

When I opened the first box, I found 3 large inner boxes:

[ex-Gooserider]

top view

Each of the inner boxes contained 4 small boxes that held 2 cells each in foam packing - All the cardboard was the brown shade I am used to seeing with US origin packages, not the sort I'm used to seeing with stuff from China.

All the cells that I've looked at so far have had the same date code, and the serial number labels on the back have been in a relatively narrow range (probably a good thing in terms of getting a well matched pack)

Hopefully the label shows - it looked OK in the camera...

[ex-Gooserider]

All three of the boxes were the same basic idea, although the number of inner boxes varied... They had one of the inner boxes that was used to hold all the building blocks and bus bars, but I didn't take pictures of the rest of the packages.... One building block had a couple of it's legs broken, but otherwise no damage... I'm not going to sweat the broken block as it doesn't seem worth the hassle over such a small amount...

Once I had the cells in hand, I decided to build a jig to make it easy to charge them - All the wood came out of the scrap bin, and it's a bit ugly, but it works....

Empty Jig fixture

The cell connections on each end are made with some heavy tinned copper braid - on the negative end, the screw in the end of the cell (BTW, the screws are all button head Allens, with a flat and lock washer - no need to buy replacement hardware...) makes contact with the braid directly

Note braid in bottom of hole... Hole is sized to fit end of battery.

[ex-Gooserider]

The plus ends each have a spring (they came out of my odds and end collection of parts - I think it was an HF 'assortment collection' that I picked up a while back on the theory that using one out of the package would justify the purchase...) that is sized to just fit over the screw head, and sit on the flat washer.... The two end blocks are spaced so that I have to jam the + end into the spring and compress it enough to barely slide the negative end into the hole - the spring then makes a compression contact...

The spring is woven through the braid to ensure contact on that end.

Our electronics lab has a mOhm meter and I checked the jig with it - the test leads measured 27mOhms by themselves, going from the banana plug to the contact for each battery was about 50mOhms, which I think is low enough not to be an issue...

Will keep people posted as things progress further....

ex-Gooserider

[Burgerman]

I think I have the charger set up right - I'm currently bulk charging the cells 4 at a time in parallel, using a jig I made (more on that later)
I'm charging to 3.6V nominally, the display keeps flipping between 3.599 and 3.601V... I set the memory to 60,000mAh (4 x 15000) and the maximum charge rate to 20A - but the display seems to show a max rate of about 3.5A which falls off rapidly.

Cos they are basically charged so you are on CV stage.

So far the charge reached '100%' in about 35 minutes, but I let the batteries soak for a little over 2 hours, when the charge current display is around 0.050mA... Is this a good stopping point?

As long as you do the same with all. The PL8 allows a specific setting to be set. 2 hours should be long enough for cells to reach saturation.

I noticed the Ri value seems to start around 50 mOhms, but seems to go up to about 70-75 as the charge goes up... Is this normal?

Its very hard to measure low resistance readings. Esp with your springs, and strange contacts and thin cables. And accuracy falls with the charge current. Plus voltage measuring accuracy drops without balance connector. So yes probably.

[Burgerman]

As for parallel charging, I just screwed a dozen cells or so together with the bus connectors at a time. And connected with a single 4mm banana/ring cable.

[ex-Gooserider]

I notice that the per-channel display in the software is blank, I'm assuming because there is no balance wires hooked up... I didn't think they were needed since I'm only doing 1P.... Other than possibly getting a few more numbers to display, would there be any benefit to adding a balance wire setup?

ex-Gooserider

[Burgerman]

Not really on 1S.

[ex-Gooserider]

Not really on 1S.

OK...

All batteries are now charged to full (took about 4 days at 3 batches of 4 / day) and sitting....

I've started a spreadsheet with the number I wrote on the cell, plus the MFG serial number - the range of numbers was relatively small, with many 'clumps', and as I noted previously all had the same production date on them.

I also noted which slot on my charger jig a given cell was charged in. Hopefully it won't make any difference, but I figure that if there were any problems with the connections in a given slot, there might be a pattern when I go to the next step in a month or so. If there is I will have to decide how to handle it then...

I didn't do any sort of partial discharge on the cells after charging, from the different threads on how to do the pack set up, it seemed like an optional step, and just extra complication and time...

ex-Gooserider

[sin85]

The core of Rachele's speech output computer is a nearly 10,000 line set of PerfectScript macros and what she sees on screen is several hundred tables in a WordPerfect document. I challenge any Word expert (or OpenOffice expert for that matter) to reproduce even 10% of what we've been able to do with WordPerfect - and I'll even give you a copy of the source code to help you get started. My guess is that just trying to create two different-sized tables on successive pages, then edit the first one without screwing up the second, will prove to be a disaster.

i agree WordPerfect is by far superior and more logical to use. in my time as an interpreter i used it not just for word processing but for publishing as well

[ex-Gooserider]

Bringing this back up, I did a lot of searching and the only connectors I found that seemed likely to work well as 'Blind Mating" item to go between the battery boxes and the frame for balance wiring, were two series of Molex connectors, either the "Mini-fit" or the "Micro-Fit" series.

The Mini-fit has slightly wider pin spacing, and takes up to AWG 18 wiring / 9A per ckt, while the Mini-fit was a smaller spacing, that only went up to AWG 20, and 5A per ckt. Either offers more current capacity than needed for balancing. Interestingly both require about the same height panel cutouts. (length depends on the number of circuits, and wasn't a big concern to me) Both series had panel mount housings for both sides, and I am sure that I could have made either series work. It looks like I will be having to make mounting plates out of some 1" x 1/16" aluminum angle extrusion, and bolting them to the battery box lids and the battery frame.

However the Mini-fit documentation called for 1/2" between the panel mounting surfaces when mated, and the Micro-fit only calls for 0.24" - which is much closer to the fit of the existing chair parts, so it should give me a nicer installation. In addition, since the Norcomp connectors that Seem to be the best option for the connection to the charger only use AWG 20 maximum for the signal wires, I wouldn't gain any benefit from the Mini-fit's larger wire size capacity.

Therefore I've decided to go with the Micro-fit series, and since the Norcomp connector had 10 signal positions, I went with the 10 circuit size on the Molex connectors (IIRC, they come in 2-ckt increments from 4-28 ckts....)

So just a few minut5es ago, I put in an order to Digikey for 10 eqch of the Molex housings, and 100ea of the pins for them, and a bunch of the Norcomp connectors, enough to do at least a couple of chairs...

I did cheap out a bit on the housings for the Norcomp connectors - I'm not convinced that die-cast zinc (AKA "pot metal") is any more durable than plastic, and since we are basically looking at DC for the charging circuit, there is no need for the shielding offered by a metal housing - so I got plastic housings that are about half the price... Looking at where I'm planning to mount the Norcomp plug on the chair, I opted to get both the usual 90* exit housings and some that have a 45* exit which I think will be a better fit for running back to the Molex connector on the battery box frame....

I still need to order wire that I can use to make up the balance harness, and the cable to go between the charger and chair...

I'd really like to use AWG 20 for the balance wires, and Ideally would like to use a 'ribbon cable' just because those are easier to wire and keep track of which wire goes where (I'd rather not have to buy several spools of hook-up wire in order to get color coding...)- but most of the ribbon cable I've found is AWG 22....

Also, what size should I look at for the power wires? The Norcomp connectors can take up to AWG 8, but I seem to recall that the Hyperion and PL/8 chargers can only put out enough to need AWG 10 or 12, especially if using the super-flexible silicone wire....

I have had the Headway cells doing the self-discharge for over a month now, so I hope to start the next steps and actually build the pack as soon as I get the rest of the parts in.

ex-Gooserider

[ex-Gooserider]

Minor additional note - I'd requested samples of the Micro and Mini-fit connectors from Molex, and they sent me one side of each, and said the other side was on backorder.... Today I got a package with the other half of the Mini-fit, and confirmed that it does have a very wide spacing requirement between the two panels...

Glad I decided to go with the Micro-fit parts...

ex-Gooserider

[snaker]

@ex-Gooserider: Can you post the digikey links and some pics of Micro and Mini-fit connectors? Are they different sizes of D-Sub?

[ex-Gooserider]

They aren't "D" connectors at all... They are rectangular connectors made by Molex, with two rows of contacts and a variable number of columns, so 2-4-6, etc pins up to about 22 circuits... I ended up ordering the 10 circuit size, so 2 pins by 5

They have several styles of housings in each family that allow various types of arrangements, like putting them on circuit boards, etc... What I wanted and why I chose these was the ability to "Blind Mate" which is when you have the connector in a place where you don't have either end in your hand to plug it in... The classic example is a "hot swap" power supply on a server, where the plug is on the end of a box that you slide into a hole, with the mate at the bottom of the hole - you don't touch the connectors, they plug in by the box being guided into position.... The battery boxes on the Invacare H-frames do the same sort of thing, using guides to force the Anderson SB-50's to line up and connect as you slide the boxes into the frame....

The Micro-fit have a 3mm "pitch" or spacing between pin centers, and the Mini-fit have a 4mm pitch. This spacing difference allows the Mini-fit to take bigger wires (AWG 18 vs 20 for the Micro-fit) and handle more current per circuit... However because of the way the connectors are designed, the Mini-fit requires more space between the connectors. The Micro-fit has more than enough power capacity, and is a better fit for the way the battery boxes go together...

I tried to upload a Zip with the technical docs on them, but it was to big to include in the post...

Send me a PM with an e-mail that I can use to send about a 4MB file...

I don't have the Digikey links, but the Digikey part numbers off my order are -

WM3715-ND Conn Recept 10 Pos 3mm Dual Row
WM1839-ND Conn Term Female 20-24AWG Gold
WM3611-ND Conn Plug 10 Pos 3mm Dual Row
WM1843-MD Conn Term Male 20-24AWG Gold

ex-Gooserider

[ex-Gooserider]

Minor progress - I've done LibreCAD versions of the panel cutouts needed to mount the Micro-fit connectors, including getting all the sizes I need to program our CNC mill to cut them out....

I had a length of 1" x 1/16" aluminum angle in my material stash, and I've cut it up into 8 pieces of a little over 2.5" each, which is two sets worth of brackets...

I have milled all of them to 2.5" length.... Not absolutely necessary that I do this, but it makes for a bit neater installation later....

I entered the program for the receptacle side into the Sharp CNC mill, and cut the 4 receptacle brackets.... The smallest bit that I have is 3mm, so I need to do a bit of hand filing work to clean up the inside corners of the cutouts..... A smaller bit wouldn't have helped that much unless I went so small that it would have been difficult to cut without breaking bits at any reasonable feed rate... I still need to do the plugs, which require a different program and two different part setups because one of the brackets is flipped over because of the way the mounts work...

ex-Gooserider

[expresso]

Sounds like your putting alot of work into this - looking forward to seeing the end result when done - i am sure will be nice - knowing you did it - knowing you build it better - with wiring and details important to you -

didnt think you were going this far - thats great you can do all these little details - makes the end result so much nicer

[ex-Gooserider]

Got the plug brackets milled... Turned out I was able to use the same program for both versions by simply rotating the angle so that the leg I wasn't cutting was against either the fixed or moving jaw of the vice... I also was able to pick up a 1/16" diameter bit ~1.5mm, which was a bit of a mixed blessing...

Since it was half the diameter of the 3mm bit I'd used on the receptacles, I got a lot more precision on the cuts, and will need to do less filing on the inside corners...

OTOH, I was able to run the 3mm bit at a 5 IPM feed rate, and full depth cut, so it was only a couple of minutes run-time per part - I probably spent more time on swapping out the parts as I cut them.... I also had spares, so I could push a bit knowing that it wasn't a big deal if I broke a bit) However I only had one of the 1/16" bits, so I had to be real conservative - ONE IPM, and 0.025 cut depth, so 3 passes per part - can you say BORING....

Still need to do filing on all the parts, but my first test fits seem to say they came out just as I planned...

ex-Gooserider

[ex-Gooserider]

Question on balance wiring...

As mentioned, I have two battery boxes wired up in the usual series sequence for lead bricks. The way the chair works with the bricks, is the boxes slide into tracks on the frame, with the front box holding the 12-24V brick, and plugging into a connector on the frame that goes to the controller and other stuff. The rear box has the 0-12V brick and plugs into a connector on the front box.

I'm trying to keep this setup, and am simply adding some Molex Blind-Mate connectors to the boxes so they will do the balance wires the same way the main power wires work.

I'm going to be using a 6P, 8S setup with 15Ah cells, so cell groups 1-4 are in the rear box, and 5-8 are in the front box... The main power wiring has the - lead going straight through from the frame connector to the - terminal on the brick, which will be the negative of Cell group 1. The positive on the rear box goes to the negative on the front brick, which will be the positive of Cell 4 going to the negative of Cell 5...

Most of the balance wiring is obvious, but I'm wondering what the best way to handle the balance wire that goes between cells 4 and 5... The two cells will already be connected by the AWG 6 main power wiring, but what about the balance wiring?

Should I connect it to the positive of cell 4, or the negative of cell 5, or do a jumper with the balance wiring? (Does it matter?) No real difference in the work needed to do any of these, just a question of where to put the connections...

ex-Gooserider

[Burgerman]

I do both 4 and 5 so you can charge each pack as a 12v pack if you want. And it keeps resistance low.

[ex-Gooserider]

OK, that makes sense... Easy enough to do given the connectors I'm using.

ex-Gooserider

[ex-Gooserider]

I'm making up cable sets to go between the chair and the charger. I'm going for a 5' length, which I know is more than really ideal, but I want to be able to plug the chair in without needing to worry about where I have the charger and power supply...

The PL-8 cable was pretty easy, albeit annoying as the JST-PA balance connector is fiddly to make up, and only supports AWG 22 wiring...

I am now trying to make a similar set for the Hyperion 1420 - but I can't find a good description of how to wire the balance connectors on the charger end (the chair end is the same....)

I know I use a JST-XH-8 connector, but how do you hook it up for an 8s pack? I've done a lot of searching and can't find any good pictures / diagrams...

ex-Gooserider

[Burgerman]

Pins 1 to 8 on the A connection are for the first 7 cells. Wire as a 7S pack.

Pin 1 on the B connector goes to the same place as pin 8 on the A connector. (Thats the connection on the pack that is both the pos on cell 7, and the neg on cell 8. Pin 2 on the B connector goes to the pos on cell 8...

In other words the A socket has 7 cells, 8 wires as you would expect.
B socket has the two wires from cell 8 only.

[ex-Gooserider]

OK, many thanks...

ex-Gooserider

[ex-Gooserider]

Actually, I'm finding it even more confusing to a certain extent...

I dug through some of the stuff in the garage and found the box I got the Hyperion in a few years back, which contained some of the factory official connectors that went between the charger and the adapter boards.... The Hyperion connector is NOT a JST-XH, though it appears that an XH connector would fit the socket on the charger. The Hyperion connector has a locking tab on it, while the JST-XH does not...

The Hyperion cable has 7 black wires and one red one, and when it plugs into the charger, the red wire is on the left. The charger end HOUSING has a molded in "1" next to the red wire, and a "8" on the other end... The Adapter board housing has no numbering I can see. The board itself has silkscreened numbers on each connector, with number 1 going to the black wire that connects to # 8 on the Hyperion end, and #8 going to the red wire (which also appears to only connect to the 7S plug on the board...)

When I dig up an actual JST-XH connector, and the JST-XH datasheet, the notch that marks the #1 socket, and the molded numbering on the housing matches the marking on the Charger end connector - and is opposite the board silkscreening....

(It is hard to see the molded numbers, I'm using a magnifier lamp, and also coloring them with a Sharpie to improve the contrast...)

So to clarify, when I am looking at the balance connectors on the charger, is the most negative pin on the right?

And the most positive pin on the left?

ex-Gooserider

[Burgerman]

So to clarify, when I am looking at the balance connectors on the charger, is the most negative pin on the right?

And the most positive pin on the left?

Yes... They are backwards to common sense.

And when you get to pin 8 which is the last positive on a 7 cell battery, it goes to the very same place on the pack as pin 9 (pin 1 on connector B)
And the pos from the last cell, cell 8 in your battery goes to pin2 on connector B.

So:
Connector A the one on the right -- Starting from the pins on the RIGHT: Cell 1neg, 1pos, 2pos, 3pos, 4pos, 5pos, 6pos, and the last pin is pin 8 and is the positive from the 7th cell.
Connector B the one on the left -- Cell 14 pos, 13p, 12pos, 11pos, 10pos, 9pos, 8pos, 8neg

And. When using the hyperion, I found out what kills them. Always make sure its power up, before connecting the battery. Preferably connect the balance wires first. Doing it in reverse killed 4 of mine over 5 years. And do the BT trick, as your power supply can make an earth loop and put current through the USB. And use the latest firmware!

[ex-Gooserider]

OK... That's what I thought, but wanted to be sure...

Other added annoyances -

The Hyperion doesn't use the "Safety banana" plugs like the PL-8, I had to cut the guards off in order to plug the output cables in....

The connector shell for the Hyperion balance leads is not QUITE compatible with the sockets for the JST-XH connectors I was hoping to re-use those as I messed up when I got a big load of JST-XH parts a while back, and didn't get any XH-8's..... Guess I do now...

(aside from my general objection to splicing, the Hyperion connectors were made up w/ AWG 24, I want to use AWG 22, the maximum allowed in a JST-XH)

ex-Gooserider

[ex-Gooserider]

cell-grph-6-20.zip

normal .zip file, open graphs w/ PL8 S/W

(332.56 KiB) Downloaded 189 times

inv-pack-1-cells.ods.zip

remove the .zip, opens w/ Libre-Office (not sure if the MS product can deal with an ISO standard format or not)

(26.56 KiB) Downloaded 200 times

cell-grph-21-32.zip

normal .zip file, open graphs w/ PL8 S/W

(302.77 KiB) Downloaded 187 times

Bringing this thread back, as I hope I'm ready for the next step, though I'm not positive.

After much headache (discussed elsewhere) I have gotten all the cells charged back up, except for four that seem to have vanished on me... I suspect they are in the garage someplace, but I couldn't find them. Fortunately, I had purchased 56 cells, even though I only needed 48 for the pack, so I can build the pack at least...

These cells were charged back in late June, using the Hyperion. I did them in groups of 4, using a fixture I built for the job. I then put them in the garage and left them to sit and self-discharge, per BM's instructions.

Starting late December, I charged them using a PL-8 to 3.6V, 20mA termination preset from BM. I used the same fixture, but only position #1, to do each cell individually. I am attaching an ODS (Libre Office) spreadsheet with my results, and a few real .zip files with all the cell graphs that I was able to get (due to hardware / software issues I didn't get graphs on the first few cells)

What concerns me is that while most of the cells charged in less than 30 minutes, and gave a fairly narrow range of mAh return values, I had several that didn't stop until the "safety time-out' hit at the 4 hour mark. These had much higher values when they terminated... How do I treat these when figuring out the cells for each parallel string?

I didn't look at every graph, but spot checking the 'pack capacity vs time' screen, which seems to me like the best equivalent of mAh returned over time, it looked like the 4-hour cells all had similar curves where the mAh went up really fast for the first 20-30 minutes, and then switched to a slow but steady increase at around 1mAh every 10 minutes or so...

I'm wondering if it would be better if instead of using the mAh returned when the cells cut off at 4 hours, I used a number from just after that knee in the capacity curve - say around 25 minutes????

ex-Gooserider