Making a Bluetooth adapter for a Car Phone from the 90's

[UselessPickles]

I need a way to test my battery module prototype. After some searching and reading/watching reviews, I ordered a Kunkin KP184 DC electric load. The lowest price I found was on Banggood.

It seems to have a lot of good features for the price (optional remote voltage sensing, constant current/voltage/power/resistance modes, battery discharge mode, over-current protection test mode, remote/automated control features, etc.) and good accuracy. One "negative" I've seen reported in a review actually seems like a good thing to me: while the load is "on", adjustments are immediately effective. As long as you are aware of this and don't do stupid things, it seems useful to be able to manually sweep the current and monitor voltage readings.

I hooked it up to my power supply and multimeter to play around with it and get familiar with its features, and also learned that my power supply under-reports current draw quite a bit more at lower amperage compared to higher amperage (-2.3% @ 200mA, -0.4% @ 500mA, -0.2% @ 2+A). But voltage output/readings all seem to be pretty close, and the electric load seems to have pretty accurate control of amperage.

I also made some custom cables for testing the battery module:

Cables for connecting the power supply and electric load to the input/output of the battery module, potentiometers that I can connect to the board to simulate temperature readings from thermistors, and cables for eventually connecting the battery module to the car phone Bluetooth adapter motherboard for a full integration test.

The potentiometers include a resistor to offset the lowest possible value (giving me a more useful range of adjustment), and intentionally bare wire for connecting a multimeter to monitor the resistance as I adjust the knob.

The assembled boards have shipped already, so I should be able to start testing by the end of this week.

 

[UselessPickles]

My battery module unfortunately does not work

Here it is all assembled and hooked up for testing with the LEDs lighting up as expected and giving false hope.

First, the protection IC wouldn't enable charging/discharging of the battery. I spent a long time trying to figure this out, double-checking everything I could think of, and eventually discovered some strange voltage readings at the per-cell connections for cell balancing. This led me to discover I had a mistake in my schematic where I mixed up two connections because I simply placed two "hierarchical sheet pins" out of order, but connected them as if they were in order:

So I corrected the mistake on the PCB:

Once again. I had an initial false sense of hope. The protection IC was now enabling charging and discharging. It generally appeared that external power was able to charge the battery, and I could draw power from the output both with and without external power connected.

But I found that I was still able to draw power even when I simulated a temperature fault to force the protection IC to disable charging/discharging, although with substantial voltage drop. Then looked at voltages more closely when discharging was enabled and found that there was a substantial (but less) voltage drop in this situation too. The charge-enable MOSFET was never fully opening or closing and was the source of the voltage drop.
After some more research, I discovered that I did not understand MOSFET specs when I chose a MOSFET. I only looked at specs like "max continuous current", etc., but had not looked at the "safe operating area" chart:

Despite having a "max continuous current" rating of 3A (which I thought was plenty for my target of supporting a max of 1~2A, but realistically only using a max of about 300mA in the car phone project), it only handles about 100mA at the battery voltage I'm working with.

I definitely had cranked up my load to 2A at some point, so this seems like a reasonable explanation: I must have fried my MOSFET with too much current.

So then I assemble my second PCB and correct the mixed up connection before I install batteries or apply input power. My plan was to test with low loads (about 50-100 mA) to stay within the limits of the MOSFET so I can at least verify that everything else is working properly with the charging/discharging and battery protection for everything other than over-current protections.

It initially looked good. I installed the batteries, and the protection IC initialized into a low voltage fault as it is designed to do: charging enabled, but discharging disabled. I was unable to draw any current at all from the output, and the voltage reading of the output was very low: the MOSFET appears to be working properly when "closed" this time.

I connected external power, which should clear the under-voltage fault and allow discharging, but now I encounter a completely different problem with the second board. The "input OK" LED flickered crazily for a second before remaining solid, then the charging LED started flashing (indicating charging). But my power supply was showing that it was not supplying any current at all.

The protection IC did, however, enable discharging, and without any substantial voltage drop, so it now seemed that the discharge-enable MOSFET was working correctly.

However, at some point while trying to investigate what was going on with the charging IC (disconnecting/reconnecting external power, removing/reinstalling batteries, etc.), the MOSFET fried again. Same symptoms as the first board (MOSFET never fully opened or closed, allows current draw when gate signal is low, etc.)

I still have no idea what's wrong with the charging IC on my second board, but it has a very unstable "startup" for a second or two every time external power is connected, and it never draws any current from the power supply to charge the battery or provide output power. The batteries are always supplying output power, even when external power is connected (external power is supposed to take precedence over batteries for supplying output power).

• Is there a manufacturing defect from JLCPCB that is shorting out some pins on the IC or something?
• Did some inrush current from the battery take out both the MOSFET and the charging IC?
  • This doesn't seem likely because the protection IC did not enable discharging until AFTER the charging ID finished it initial glitchy power-on, so I'm pretty sure something was damaged/faulty before there was any chance for battery inrush current.
• Was I too careless about ESD risk while handling the PCB to solder my connectors, and damaged the charging IC myself with static?
• Did I just receive a faulty charging IC?

And I'm also less confident about the cause of my MOSFET failing, because I thought I was staying within its limits this time, but it still failed. Maybe the unstable glitchy voltage from the charging IC contributed this time? Maybe there was still some current spike at some point when disconnecting/re-connecting things that briefly exceeded limits enough to fry it?

The best I can think of doing next is to choose a MOSFET that handles much higher continuous DC current at 12-14V, and order a new batch of PCBs (with my mixed up connections fixed too, of course), and hope that those were my only significant problems, and that the charging IC issues on my second board was just bad luck with manufacturing.

I'm currently leaning toward this MOSFET: Vishay SUM70060E

• Over 10A continuous current at my battery voltage levels.
• It specifically lists "Battery management" as one of the intended applications.

 

[UselessPickles]

I'm currently leaning toward this MOSFET: Vishay SUM70060E

I just had a brilliant idea. I looked at the Dayton Audio battery module I'm currently using in my prototype to see what MOSFETs it's using. There's a pair of UMW 50N06 next to each other, arranged like they are probably used for charge/discharge enable/disable. If that MOSFET is sufficient for 3 larger 18650 batteries in series, then it should be more than enough for my smaller 14500 batteries with less capacity and discharge current.

 

[rjenkinsgb]

Be careful with the FET polarities - they are often used "backwards" in applications like this, to avoid the internal diode passing current when the FET is supposed to be shutting off the power.

This is a typical connection, with separate charge & discharge enables & the two FET drains linked.

 

[UselessPickles]

Be careful with the FET polarities

Thanks for the tip. Fortunately, this is one detail I double/triple/quadruple checked because I am aware that there are a variety of pin arrangements for MOSFETS. I was careful to select the correct symbol in KiCad that matched the pin order of the MOSFET I had selected, and double-checked the pin arrangement of the PCB footprint.

It seems very reasonable that my problem is simply an under-sized MOSFET. I also realized that even when I thought I was testing within the limits of the MOSFET (<= 100mA), I probably still exceeded the limits as the MOSFET warmed up, because that 100mA limit is at 25C. This makes sense now that I think about it: the MOSFET seemed to work fine until after it had been switched on for about about 30 seconds. It was probably initially within limits, then warmed up and its current capacity decreased below the 100mA I was drawing.

I also have come to the conclusion that the additional diodes I placed across the MOSFETS are unnecessary. I initially did this because the current rating of the body diode of the MOSFET was lower than the amount of current I wanted to support. But the protection IC datasheet specifically explains that it protects the body diodes by opening each MOSFET in various conditions based on the direction of detected current flow.

 

[UselessPickles]

Now for a bit of a side quest. Once I have the new battery/power management circuitry worked out, I plan to position the 3 batteries and all the battery charging/protection circuitry in the area currently occupied by the red battery management circuit board in my current prototype:

This will free up a lot of space in the lower section of the PCB near all of the connectors/ports, including the space where the original car phone could include an optional RJ11 "data" port for connecting to modems and fax machines:

This port is a small PCB that mounts via 2 pin headers/sockets:

I'm creating a replica of the overall shape of that data port PCB so that I can add an RJ12 port in the same position (same body/connector as RJ11, but with all 6 positions populated with contacts) and use it as a programming/debugging port:

I need 5 pins for programming/debugging of the MCU, and I can use the 6th pin for my I/O UART TXD that I use for logging to a terminal on my computer. I'll just have to make a custom adapter from an RJ12 cable to a pair of connectors that connect to the PICKIT4 programmer and my UART/USB adapter. This will allow me to fully debug and program the MCU without opening up the car phone at all, while still keeping the car phone 100% original in appearance. Plus, the completionist in my likes the idea of having all possible ports populated on the car phone.

The existing programming and I/O pin headers on my PCB are on the opposite side of the board from where this programming port will be, so I'll run some ribbon cable between those pin headers and the the right-angle socket on the programming port PCB. The pair of pin headers/sockets used to mount the programming port to the main board will be purely structural (no electrical connections).

In the future, I may consider a major reorganization of my PCB layout that could move the MCU closer to the programming port so I can make use of the mounting pins for the electrical connections, and possibly move power-related circuitry closer to the external power supply socket. But for now, I'm taking things incrementally.

Conveniently, the exact RJ11 port originally used by the car phone is still in production, and there is a 6-contact variation of it available: Hirose TM5RE1-66(20)

The pin headers will be a bit trickier, because the exact JAE parts are no longer available, and I need to get the total height of the mated connectors right. The dimensions of the male pin headers on the main board seem to be pretty standard/common, but the female part will require some creativity:

That is about 8.4mm tall with a 1.1mm spacer for a total height of about 9.5mm. Typical height of female pin headers seems to be 8.5mm. I haven't yet found any options that are either 9.5mm, or have options for spacers that make it 9.5mm. But I can buy sheets of 1mm thick ABS plastic and make my own spacers.

 

[UselessPickles]

I updated my battery module design to use the UMW 50N06 MOSFET. Conveniently, this is a part that is in stock at JLCPCB. Unfortunately, I still need to pre-order a couple more of the battery charging ICs before I can place an order for the updated boards, so it will probably be about 3 weeks before I have these new boards in hand.

These MOSFETS are substantially bigger than the ones I used the first design. Hopefully I have enough copper area for heat dissipation (it's on both sides of the board with many vias). But considering I'll be running much less continuous current through these than they are rated for, I expect heat to be a non-issue.

 

[UselessPickles]

Today I started learning how to create 3D models, using "onshape" (free web-based 3D CAD design software). I made this model of the antenna connector on my car phone so I can better visualize my PCB layout and ensure I don't place any components or traces too close to it.

In my custom Bluetooth conversion, this antenna socket isn't electrically connected to anything, but it is still physically installed (with the wire cut off) so I can still attach the original antenna for the sake of complete appearances.

 

[UselessPickles]

And here's a sneak peak at what my board will look like with the programming port added on, and the battery holders relocated. I still need to fine-tune the positioning of the pin headers for the programming port a bit more.

I really need a better 3D model for the RJ45 port and the special connector that sticks up out of the case. Sounds like some more fun 3D modelling practice

When I place an order for a new batch of battery module prototypes, I plan to also order some bare boards of the programming port and the motherboard so I can partially assemble them with ports, pin headers, and battery holders to test overall fitment/alignment of everything.

 

[UselessPickles]

Here's my second attempt at creating a 3D model of a part: the RJ45 port that the handset plugs into. It's much better than the basic cube that came as the model for this part from SymacSys.

Here's the real part:

It's slightly different from the RJ45 port on the original car phone, because the exact part is no longer available (Molex MXJ 52018-8816). But I found a part from a different manufacturer that is compatible (Kycon GLX-N-88M).

 

[UselessPickles]

I created a 3D model for the special obsolete connector (JAE DRA-8SC-F0) on my car phone's PCB that is originally used to connect the portable battery circuitry, but I am reusing in my Bluetooth conversion to connect to the Bluetooth module.

The 3D render of my PCB is finally pretty accurate.