Replacing the battery to Li-ion

[StuartL]

I know I'm dragging up an old thread but it has become suddenly relevant.

I've had my Bioloids for just under a year and achieved a lot. However they were sat in a box for about 8 months (it's a crime, I tell you) and it's just transpired that both batteries are now useless.

They'll charge fine but they'll not last more than a few seconds of operation now.

Has anyone else had the batteries in storage for any length of time and found the same thing?

Either way it's now time to change batteries. Li-po seems a good choice. Is there any increase in the percevied wisdom already dispensed in this thread?

Being an RC geek getting hold of Li-Po batteries is not a problem, I just want to check whether there's anything I need to know beyond what's already in this thread before purchasing. e.g. Are there any particular arrangements/sizes/manufacturers of Li-Po batteries which fit nicely inside the Bioloid? Has anyone worked on deliberate shut-down code for low voltage? I see from Jon's documentation that the CM-5 has an ADC input with the battery voltage monitor on it. Is there an appropriate output to either alarm (beep) or turn itself off? If not I figure that turning on all the LEDs and deliberately halting (I presume the AVR has a halt instruction??) is a fair way to alarm.

Thoughts?

[BillB]

Stuart,

Once your convert to Lipos there is no going back. The AX12 servos actually perform much better (more power and faster) under LiPos.

Don't rely on a low voltage detection program running on the CM5. Instead use a dedicated hardware Lipo voltage detector. That way you can ensure that you are warned of a low voltage irrespective of whether you are in Play, Program, Manage or Bootloader mode.

Also - make sure you connect the Lipo up so that it uses the internal CM5 fuse. I can't remember off had but only one of either the battery or DC socket is fused.

I also unsoldered the DC input socket at the front of the CM5. That way there is no danger of someone inadvertently plugging a mains DC into the CM5 when the Lipo is connected (in which case the CM5 may try and charge the LiPo - which would be catastrophic!!!!!). If you do solder the DC socket then do remember to solder together the GND and Negative. The DC socket does this automatically for you when the socket is empty, but it is not immediately obvious.

Hope this helps.

[StuartL]

Firstly thanks for such a comprehensive reply.

Once your convert to Lipos there is no going back. The AX12 servos actually perform much better (more power and faster) under LiPos.

Given the comments below what do you do about charging them? The circuit in the CM-5 has no temperature sensor so it's virtually impossible (as far as I know) to do safe Li-Po charging. I can't bear the thought of unscrewing the CM-5 every time I need to charge the batteries, although I could mount an external connector to reach the battery.

Don't rely on a low voltage detection program running on the CM5. Instead use a dedicated hardware Lipo voltage detector. That way you can ensure that you are warned of a low voltage irrespective of whether you are in Play, Program, Manage or Bootloader mode.

This is a fair point but I've written my own code from scratch anyway. The low voltage monitor is entirely interrupt driven and can turn off the servos when the voltage has dropped below a preset level.

Also - make sure you connect the Lipo up so that it uses the internal CM5 fuse. I can't remember off had but only one of either the battery or DC socket is fused.

It's the battery

I also unsoldered the DC input socket at the front of the CM5. That way there is no danger of someone inadvertently plugging a mains DC into the CM5 when the Lipo is connected (in which case the CM5 may try and charge the LiPo - which would be catastrophic!!!!!).

As it's my own charging code I can guarantee the charging circuit won't be switched on, purely because if I switch to LiPo the charging pin enable (bringing portB5 active low) code will be commented out

If you do solder the DC socket then do remember to solder together the GND and Negative. The DC socket does this automatically for you when the socket is empty, but it is not immediately obvious.

If (given all of the above) I did remove the DC socket it seems like a good hole in the case to put a charging connector for the LiPo batteries. But your comments about shorting ground and negative are well noted

Hope this helps.

Definitely.

I've also been spending the last 24 hours or so investigating Li-Po charging techniques, cell and charger costs etc...

I've concluded that I can't be bothered.

I know the AX12s perform better at the higher voltage (and at a higher temperature no doubt, which is already a problem for my snake :S) and I know that LiPos are a better battery technology but the absence of in-robot charging and the costs of the cells are severely off-putting. If I had some cells kicking around from the RC stuff (I'm entirely nitro powered atm) it'd be different.

So a big thank you for your post. For now I'm off to figure out how to find someone selling a 2x4 AA pack with decent capacity and a decent cable for high current draw. Then I can use the built in charger and leave each CM5 permanently in the snake and biped and avoid those damn JIS screws

[billyzelsnack]

Lipos can be had pretty cheap if you know where to look..

www.hobbycity.com

Hmm. This season I am doing A123's for my bigger RC heli. Maybe I should look into that for my Bioloid too.

A123's are MUCH safer than lipo. They are also much more tolerant of over discharge and you can fast charge them incredibly quickly. They also have very high continuous discharge rating and last a large number of cycles ( 1000+ ). Charged cell voltage is 3.2V with 2300mah of capacity and weigh 70g. Off the top of my head I think 4 just might fit into the bioloid battery box. I got mine out of a Dewault drill pack on Ebay for about $100 shipped a few months back( 10 cells in a pack ).

[billyzelsnack]

Ok. I just checked and 4 A123's won't fit in the CM5. Drats. I always forget how large these A124 cells are. 3 just might fit though. The only screwdriver I have that'll open up the bioloid battery case is at work, so I can't tell for sure until I can open it. If 3 works it should give better performance all around vs NiMH.

Higher discharge rate.
Less weight.
Faster charging.
Less self-discharge.
More cycles.

[BillB]

Stuart,

I've concluded that I can't be bothered.

It is not as bad as it seems, the very easiest and simplest way to add a LiPo is to cut off the battery connector of the standard Bioloid Battery and solder it onto the LIPO pack leads. You will need a dedicated LiPo charger to charge the Lipo battery, but it will work and the upgrade will only take about half an hour... but it is NOT a safe solution!!!

The recommendations in my post will help to prevent destroying your house by accidently mishandling the potentially explosive Lipos. You don't need to do all the recommendations in my post - but personally I like to be overcautious with Lipos.

Bill
[/quote]

[billyzelsnack]

Nope. 3 won't even fit.

http://www.flickr.com/photos/85327336@N00/2404735388/

[StuartL]

Seems that the NiMHs have finally given up. One of the packs just turned into a furnace and seriously melted the CM-5 case. The CM-5 PCB and top part of the case appears to be unaffected but the bottom part is completely distorted and the battery is, of course, a write off.

So LiPo it is. I think I'm going to get two of these per CM-5:

http://www3.towerhobbies.com/cgi-bin/wt ... XSUX2&P=ML

Battery charging then becomes an issue but I'll figure my way through that somehow

Gonna be an expensive month for the robots, what with the investment in the IMUs too. Watch for an independent review of the HUV Robotics IMU soon.

[billyzelsnack]

Watch out with those micro-deans. I shorted the exposed pin against my charger once (charger connects via the balance connector) and blew out a chip on the charger.

[tom_chang79]

Seems that the NiMHs have finally given up. One of the packs just turned into a furnace and seriously melted the CM-5 case. The CM-5 PCB and top part of the case appears to be unaffected but the bottom part is completely distorted and the battery is, of course, a write off.

StuartL, was there any indication of a failure before it turned into a furnace? I'm a little alarmed from your experience... Also, how long have you had the Bioloid and what is your rough estimate of how often you charged the stock NiMH battery pack?

[StuartL]

Seems that the NiMHs have finally given up. One of the packs just turned into a furnace and seriously melted the CM-5 case. The CM-5 PCB and top part of the case appears to be unaffected but the bottom part is completely distorted and the battery is, of course, a write off.

StuartL, was there any indication of a failure before it turned into a furnace?

Well, there was a smell and a fizzing...

I'm sure that there was a drastic temperature increase but at the time I was out and Matt was making tea, so he heard the fizzing and came running.

I'm a little alarmed from your experience...

So am I. Matt is still shocked nearly a week later. He's got visions of his house burning down. While that's an excessive reaction it does illustrate the point.

Also, how long have you had the Bioloid and what is your rough estimate of how often you charged the stock NiMH battery pack?

Ok, I'm a slack geek. I bought the Bioloids for RoboGames 2007 but missed the actual games with them (they were on a plane somewhere). Around that time I probably spent a while using the Bioloid charging mechanism (i.e. Bioloid firmware and manually starting the charging) and then, because I started writing the library, I started running off the PSU.

In hindsight I think that there's a fair chance that one of the bugs in example.c that I duplicated in my library was to turn the charging circuit on permanently on startup. Over the few weeks I developed last year this probably caused a degree of over-charging.

The Bioloids were then relegated to project back-burner while other projects took my attention (anyone say ADD? Oooh look! Shiny!). This year they've been developed in anger and Matt has taken on one of the kits which I converted into a snake. He's been mucking around with it for a few months and certainly for a part of that the charging bug still existed.

As per this and the other thread I spotted this bug and worked on a proper charging algorithm. I believe the algorithm is largely fine but undoubtedly has the same problem as the Robotis one, which is that the resolution of the ADC doesn't allow accurate end-of-charge detection.

I do not know how Robotis have done their end-of-charge detection (timer, delta-peak, current monitoring) but I'm guessing they undercharge the pack sacrificing battery life for safety. I was unaware of such concerns so I concentrated on getting the pack fully charged.

I think between the development last year, the bug in example.c and the constant PSU running this year for snake development the pack was slowly destroyed by overcharging, and eventually had nothing to burn off except itself.

I suspect from circumstantial evidence that the Robotis charging algorithm doesn't have this problem because it just under-charges the pack instead, hence the 8-15 minute run times people are getting. With charged 2.3Ah NiMH cells a Humanoid should easily run for an hour, maybe even more.