Plan would be to travel from camping site to camping site. Charge and rest there for the nights and ride during daytime. Need to carry some camping equipment with me though. Packing might prove to be a challenge.
I’ve been using the HLG-150H-48A and it ain’t bad. I just personally like having access to the voltage and current values graphically and being able to use the power supply for benchtop use.
I got the 0.5 HW boards this week, haven’t gotten around starting assembly on them yet. Probably gonna start tomorrow (Sunday).
Spent a couple more iteration rounds on the Benchtop power supply 3D-printed parts to optimize it’s airflow for cooling, it can now handle continuous 330 Watt pretty easily (PSU rated for 350W, so maybe actually a bit overloaded with 330W going to load through the buck converter), while staying moderately cool (warm, but not burning to hand). The problem was that the warm air moved by the PSU’s fan, kept circulating inside the 3D-printed enclosure, so I added a wall inside to stop it from being able circulate inside and it now pulls in fresh air from the front air ports and exhausts it from the back. I integrated the wall into the 3D-printed mount DPS-buck converter later (no pic).
On saturday I finally got back to building my DIY reflow oven that has been sitting on my kitchen table for at least 6 months. Planning on making a little production line… with a pick-n-place machine…
Couple test runs without any extra control electronics, just measuring temperatures from the thermocouples inside the oven (only upper TC in the first pic, both in the second one) with the heating elements wired straight into the wall plug. X-scale is time in seconds and Y-scale is temperature in Celsius. It has a pretty decent temperature curve, as these show the fastest possible heating time.
I have partially assembled one 0.5 board to test the new buck IC and it’s layout. So far it seems promising, as it’s outputting voltage, and that voltage is what it should be, but I will do a noise and thermal test with an external load to test/stress it a bit more.
I’m taking inventory of my current component wares as well as ordering the necessary parts for the upcoming hand-assembly batch.
I’m also designing my production testing jig, which is gonna be a bed-of-nails type of deal with a clamping connection for the battery and charger pads. Idea is to break out all the connections from the board by placing it in a jig and having all the testing equipment connected to the jig. Board doesn’t need to have any connectors on it to test it.
Point is to make sure that everything essential on the board has been tested and verified to work before shipping to them customer to reduce amount of DOA or defective items. This is also because battery equipment related malfunctions have the potential to be catastrophic. I’ll also probably ship the testing result certificate paper to show that the board has passed testing.
Here’s a list of the planned things that will be tested in order (questions/conversation welcome):
Short-circuit test
Flash production testing firmware
5V buck output voltage
3V3 linear output voltage
Status LED
TX/SCL and RX/SDA pins
CAN-comms
Cell voltage measuring
Cell voltage balancing
Battery and Charger voltage measurement + calibration
Charging current test + calibration
USB-connectivity
If all tests passed up to this point, write board HW version in one-time-programmable memory
Flash application/user firmware
Print test result and calibration verification
This test procedure should be able to detect the majority of defective boards while they are still in my hands. I can then set them aside, diagnose, fix and run them again through the procedure.
Most likely this will be a semi-manual test at the beginning that I will automate further with time, as the volumes increase.
The cell voltage measuring is done through via the LTC6803-3 battery stack monitor IC and referring to
it’s datasheet, the cell voltage measurement error should be within a couple ± milliVolts and the whole stack’s voltage error within ± ten milliVolts and I have seen and verified the cell readings to be within a couple milliVolts with external measuring equipment. So it is a pretty dang accurate chip (and better be for it’s price…). I personally run the balancing on my packs so that the cell voltages are within 5 mV between the bottom and top cells at top SoC.
Noise measurements
Yellow trace is battery input (my 10S pack, fully charged, ~41,5V). Blue trace is 5V output from the buck regulator. We’re measuring AC ripple on the voltages.
Summary:
10µF output cap, no load ~175 mV noise, full load ~100 mV noise
22µF output cap, no load ~125 mV noise, full load ~70 mV noise
47µF output cap, no load ~90 mV noise, full load ~40 mV noise
22µF output cap is what I have specified in the schematic and with it the output is pretty decent in terms of ripple. What it actually needs is a bit more capacitance on the input side, as can be seen with the relatively large ripple voltage, but currently all 63V high capacitance ceramic caps are absolutely out of stock everywhere, so you have to jump to 100V rated ones to find more stock and then you are much more limited in the capacitances available in the 1206 package size. And then the costs start to go increase… The next capacitance size from the 1µF is a 2.2µF and it’s almost double the price of the smaller one. I’m thinking of ordering a couple of the 2.2µF ones to test what the performance would be with them and maybe upgrading to them.
The ripple performance is very close to what is shown in the datasheet with just the single 22µF output cap, whereas the datasheet measurements have 2x 22µF ceramics on the output. Also we do have a much higher input voltage, but it doesn’t seem to matter that much with how this IC regulates the output.
Just a crazy idea: As we are talking 100V components - Have you thought about making a version for up to 20s? Or would it be possible to intelligently combine two for a 24s BMS? That would make these really versatile!
I’ve been thinking of a concept design for a 6S-18S BMS with that would support daisy chaining the BMS-modules for higher S-packs, but I need to focus on getting this project first out of the vaporware realm.
If you’re thinking of using CAN-bus for high voltage pack setups, then you need to design the CAN-bus system around galvanic isolation and isolated power, but if you’re going for dedicated isolated communication bus then you can essentially use any communication bus for that matter.
True isolation is of course better and will be able to support to hundreds of S-packs.
Communicating from the bottom pack to rest of control system and not the other series’d up packs? Or for communication between the series’d up packs.
Normal CAN-bus ICs have a maximum common-mode (voltage to GND) voltage that the ICs can handle. There are then the special ones that support isolation, but usually need a dedicated isolated power supply for them.
What current/voltage controller are you using for the PSU? I was looking at a similar setup to monitor charging but it didn’t support above 50v for 12s
When I started using the DPS5015 the 5020 wasn’t in the market, so that’s the reason why I have stuck with it, as it has proven to be a decent design & quality and I really haven’t reached that 15 Amp limit in my use. Now that you mentioned it, I’ll probably do a cost analysis between the 5015 and 5020 and see if the 5020 is more cost-efficient with the higher max load amperage.
No reasonably priced AC/DC-converters to supply even 10S at 15 Amps with high end being over 600 Watts.
The power/cost-ratio sweetspot is around 350W mark with roughly 10.3W/€ with the LRS-350.
7.90W/€ with SE-600, 6.79W/€ with SE-1000.
Granted, you can find better power/cost-ratio if you’re willing to source directly from China, but I want to ensure the AC/DC-converters I use are of guaranteed quality and certification with a name brand in case I want to start selling said PSUs combined with the DPS-buck regualtors for example. I don’t want to risk it with a possibly unknown certification of the PSU, especially as we’re dealing with mains power and voltages.