Fluke 540B -- new LiPo battery system

[essele]

I thought I'd share a bit of progress on a project I've been working on to build a complete Fluke 540B power pack, but based on LiPo's.

It's made up from 9 low cost LiPo's, with 5 dual isolated 5V AC/DC converters along with 9 LiPo cell chargers.

Four of the LiPo's feed a low-noise voltage regulator for the +13.5v supply, there are another two regulators (one positive/one negative) creating a +/- 6.5v supply, and then a final regulator providing a +1.35v reference.

All of the supplies can be adjusted, and I have a provision on the board for (a) a feed for status led's, (b) ability to disconnect the batteries from the regulators, and (c) a feed into the 540B to allow for proper battery level display on the meter.

Originally I had planned to use a second PCB as the actual back-plate and interface the two, but I'm starting to rethink that now.

There have been quite a few schoolboy errors along the way ... wrong footprints, reversed footprints (that was a magic smoke moment), and finally a bit of stupidity with the ac/dc supplies ... I didn't read the data sheet properly and the second of the two 5v supplies is unregulated, so it generally drops quite a bit and then only recharges very slowly, but it does work. Please don't laugh (or be too worried) about my mains connection ... it's only for testing.

Anyway ... I've just removed the original battery pack from one of my 540B's and crimped on the connector so I can easily add or remove the new pack. Sadly it didn't quite work as planned, the search and galv side looked fine, but the reference wasn't registering. Opening up the 540B and I found a transistor floating around the case having falling out of a socket on the bottom pcb, not a good sign.

So I'm going to have to try this on my only working unit ... a job for the next few days.

I will post more once I've hopefully had some success.

I am already thinking about the next version with the following changes:

1. Rework the shape so that I can use the original backplate. I was originally worried about stacking too many batteries, but I think they will be fine. I will need to find a different set of AC/DC converters for that.
2. Look at using a voltage reference rather than a regulator for the reference .. I'm getting spurious 50uV jumps with the current one.
3. Ditch the status leds.

I'm sure people will wonder why I went for 9 independent chargers etc. I simply couldn't find a simple multi-cell solution that I would actually be able to hand solder.

Any comments/suggestions welcome!

[essele]

Quick update ... it's working!

The 540B I was using had the reference battery mod installed (not sure how I didn't see that) ... anyway once removed, and with some tweaking around with the protection circuit (where the transistor fell out!) ... all is working!

The 540B itself is a bit flakey ... there are obviously some loose connections or dodgy switches in there that need fixing as taps and knocks have a variety of effects, but I successfully managed to measure a 1V AC signal and accurately map it to a 1V DC reference, so all good.

There is a small amount of wandering that goes one which I think is the reference supply as I suggested above, if you warm it up you get quite a variation, so I definitely need to look at alternatives. I can't believe the standard battery mod is that good, but that may explain why you have to keep repeating the measurements.

So I'm going to see if I can redesign it to fix onto the normal backplate, this will probably mean using a single 5V 15W main ac/dc converter and then 8 isolated 5v dc/dc converters which I'm hoping will mean more flexibility in terms of space utilisation. I may then have to stack the batteries into two stacks rather than three. I'll look at something like the MAX6070 as the voltage reference, but I need to do a quick check of current consumption first.

[essele]

Ok, version 2 is complete.

So I've massively simplified it ... removed all the battery charge status stuff as well as the ability to turn it on or off. It's a bit of a compromise, but it's going last pretty much a year between charges, so I'm not too worried,

I've reworked the board so it fits the original backplate, and switched to a single AC/DC power supply followed by a number of isolated DC/DC converters.

I've also switched to a MAX6070 at 1.25v for the reference.

I now have two 540B's that work brilliantly with this ... everything seems nice and stable, and the two units seem to agree to within 0.002% when doing a quick check of a Datron 4000A against a 4200. Now I need to order a few more bits to make up the second supply so I can leave them permanently in the units.

[amspire]

I've also switched to a MAX6070 at 1.25v for the reference.

In the one I fixed years ago, I used a standard 1117 1.25V regulator running of 2 AAA batteries and it is easily stable enough. I mounted the tiny 1117 regulator board on the switch, so it was disconnected from the batteries when the unit is off. The battery test resistor was increased so it works for the two AAA batteries instead of the mercury battery. The 1117 is actually easily stable enough for 0.01% measurements, so I guess the MAX6070 would be superb.

I actually quite like your separate chargers for every battery. You cannot just charge them in series, and you want it to be trouble free. Well done! You certainly haven't skimped on the battery capacity. 1AH batteries probably would have done the job fine.

Richard.

[essele]

Thanks Richard.

I had originally looked at mounting it all so that it was triggered by the switches, but by the time you do it for all four supplies it's gets a bit messy, and I did a check on the quiescent current when connected and it was minimal (I don't have the numbers to hand, but I recall working out the lowest capacity supply would last at least a year using very crude calculations) ... so leaving it on didn't seem such a bad idea, even if my calcs were way off.

The LiPo capacity choice was mostly driven by really good value pricing from HobbyKing .. they seemed to be the best value for capacity in a form factor that would work, and with a connector rather than having to solder to a tab. And the charging model has definitely grown on me ... it felt naive to start with, but it seems to work really well, I think I'll use it again in my next project, but perhaps with the 2W version of the DC/DC modules, these 1W ones get a little warm and the 2W versions are only marginally wider, and not much more expensive.

Lee.

[amspire]

I had originally looked at mounting it all so that it was triggered by the switches, but by the time you do it for all four supplies it's gets a bit messy, and I did a check on the quiescent current when connected and it was minimal (I don't have the numbers to hand, but I recall working out the lowest capacity supply would last at least a year using very crude calculations) ... so leaving it on didn't seem such a bad idea, even if my calcs were way off.

I fixed my 540B a long time ago so I just went to NiCd batteries as a direct replacement. It will be good for another 20 years, and the charger is just a resistor.

Just had a look and you can still buy NiCd batteries on eBay.