DIYBMS v4

scttnlsn · 2020-12-23T15:25:57.260Z

stuart:

the reference voltage bounces around with the supply voltage - the exact thing you don’t want

Yeah, good point. Looking at the datasheet (the ATTiny85 in my case) the voltage reference does indeed drift quite a bit: https://ww1.microchip.com/downloads/en/DeviceDoc/Atmel-2586-AVR-8-bit-Microcontroller-ATtiny25-ATtiny45-ATtiny85_Datasheet.pdf (page 191). I suppose you could try and compensate for that curve in software a bit but that’s not ideal.

stuart:

The V4.21 design uses a 1.25V reference (0.5% tolerance)

Are you referring to the TL432 here (which I see in the jlcpcbassembly branch)? Looking at the datasheet (http://www.unisonic.com.tw/datasheet/TL432.pdf) I’m not totally clear on how much current it is constantly drawing from the cell. I would guess it’s “Minimum Cathode Current for Regulation” which is 0.6-1mA and that seems quite high to me.

Looking at the LM4040 (which I see in the master branch) - the operating current is bit lower at 45-60uA (https://www.ti.com/lit/ds/symlink/lm4040-n.pdf page 7) which seems more acceptable for smaller packs of cells.

What sort of acceptable idle current draw are you aiming for with v4? Thanks for walking through all this with me - appreciate your insights!

stuart · 2020-12-23T16:51:20.306Z

If you look at the circuit, the TL432 is driven from one of the ATTINY pins, so its only “on” for a fraction of a second whilst in use, then off for the remaining time.

The V4.21 boards sit idle most of the time consuming nano amps, and only a 5-10 milliamps when they process a packet of data for a brief moment. I spent a long time getting the power consumption as low as possible, based on the V3 DIYBMS design.

My test rig contains 8 single cell 18650 cells (less than 2000mAh each), and they take months to drain
down from the BMS module alone.

Chapulino · 2020-12-25T12:43:39.297Z

Hello guys .
After mounting the V4.21 modules, and of the 30 that I have prepared there are two that give me an error, the voltage reading is 0.5V higher than the real one and also does not give a temperature reading.
I have reviewed the soldering of the Attiny again and again and I have also changed the D1 for a new one (finally, as JLC did not have and sent the modules without the D1, as a solution I put AZ432ANTR-E1 LCSC code C84139 with satisfactory results). Where do you think the error may be, what components should I check or replace?

Captura de pantalla 2020-12-25 151108858×616 12.6 KB

Thanks to Stuart for sharing this exciting project.

stuart · 2020-12-26T08:38:03.419Z

Hi, if you have built 30 modules and 2 are not working as expected then there must be a bad soldering joint, do you have a magnifying glass?

If the voltage is wrong and no temperature reading it suggests that it’s the voltage reference chip.

Chapulino · 2020-12-26T10:41:56.059Z

Thanks Stuart, it didn’t look like it, but there was indeed an Attiny pin on each of the modules with bad contact.
Now I have 30 cell modules, which is good because being able to attend two 15S installations, mine and a friend’s.

IMG_14751024×768 84.1 KB

stuart · 2020-12-26T18:15:49.263Z

5 posts were split to a new topic: Module report zero volt and -40 temperature

Daemon256 · 2020-12-27T16:27:49.822Z

My system is working using DIYBMS

Thanks for helping Stuart

Maarten_van_der_Voor · 2020-12-28T08:35:41.709Z

So far I am pretty impressed, I moved my nissan leaf batteries to my shed, and although I am still waiting for a half decent solution to charge my test setup, things look pretty good.
I found my own mistakes in soldering and programming. Lost a virtual computer running grafana and influxdb for a day or 2. found things I can an cannot do with 3d printing mounts and I have to wait sometime to be able to get some dinrails.
Still I would like to share some graph. I graduatly drained to cells with 0.01 volt per night to get them pretty much balanced. I am impressed with how much the cell modules are able to bleed off, even when you set the temperature tresshold lower then the 65 degrees.

Thanks Stuart!!

Naamloos1767×850 166 KB

mspencer · 2020-12-28T14:16:29.953Z

I am powering the ESP from a dc-dc powered by my main 7s pack. The ESP with the standard program is rather power hungry (70-80ma @ 3.3V). I have several other ESPs running with an average of ~30ma. They are taking advantage of the modem sleep modes and delays in the code allowing periods of ~15ma consumption during idle followed by periods of ~70-80ma during mqtt transmission.

If I force the modem to sleep (wifi.forcesleepbegin()) the current drops down to a steady 20-27ma. Adding in any amount of delay into the main loop seems to cause a crash of the program…not sure why, it just appears to be resetting. There should be a lot of dead time between the tx/rx packets and other updates which could take advantage of sleep.

stuart · 2020-12-28T17:31:36.577Z

mspencer:

If I force the modem to sleep (wifi.forcesleepbegin()) the current drops down to a steady 20-27ma.

Does the controller web interface still respond to requests when in sleep? Does it automatically wake up when a web page is requested?

Ross · 2020-12-28T21:58:37.850Z

20201229_0855111908×4032 1.79 MB

Just commissioning a 16S with heatsinks

38120 commissioning diyBMS1232×825 193 KB

Emonpi logging data

noddle · 2020-12-29T03:52:09.203Z

Hi Stuart
Where is the Min, Max voltage range in the web front end hiding now,
I found this in the thread but it seems the file is not been used anymore,

Capture774×301 9.39 KB

stuart · 2020-12-29T08:44:17.097Z

That’s been added back into the next release

mspencer · 2020-12-29T13:12:40.235Z

No, the wifi.forcesleepbegin basically turn off the wifi, so no webpage is available. It does seem to still work as the lights flash and look correct…just no webpage. It was just a test to see the power if the wifi is turned off.

Any idea why a delay() added into the main loop causes so much problems? I am okay having a slower update of maybe 10-20seconds between cell updates as needed.

kaeferfreund · 2020-12-29T19:15:49.863Z

wow! That looks really good. I am planning to make a tree 6p13s packs. If that works well, I’ll make another 3x 6p13s config. As the cabling is a lot of unnecessary work and the stacking looks really good, I’m wondering if we can make a stackable version that automatically connects Data between the stacked modules. Even more helpful would be if one of the battery contacts could also be passed between the stacked modules, so that we’d only need one contact/plug for each cell in series

stuart · 2020-12-29T22:26:03.506Z

Just to repeat what’s been mentioned before, the modules are designed to be as near to the cells as possible, using short power wires to keep resistance low.

Sid1 · 2020-12-30T18:11:31.462Z

Is there a difference between the software version v2020.11.02 and v2020.12.16?

stuart · 2020-12-31T11:26:28.832Z

Sid1:

v2020.11.02 and v2020.12.16?

No, I’ve been experimenting with github actions to automatically build the code and release it - this was just an example/test. There are major code changes coming soon though.

volkergi1970 · 2020-12-31T14:58:02.074Z

Hello Stuart, at first thanks a lot for your work. I use your BMS since 3 months with an 48V 200Ah Bank without Problems. Since i have changed the Power inverter i have problems with the communications of the controller with the modules. During the measering with an oscilloscope, i have killed the controller. Now i load the last version from github to the new controller and the controller starts up, i can connect with him, but it doesnt show the menu to set the SSID Name and password. I only see a white screen under the adress 192.168.4.1. He outofill /softap.htm behind and nothing is visible on the page. I`ve checked it with different controllers.

stuart · 2020-12-31T16:22:34.027Z

I bet you didn’t upload the file system image?

Platform IO V5 "Upload file system image" option moved diyBMS

192.168.4.1 “Can’t handle this request right now” ? Has anyone seen this before? [edit] It’s not much fun when you download the ‘newest’ zip file for the controller code, open the workspace included in the zip file, then click on platformio.ini–click on the alien head–scroll down to envesp8266_d1minipro–(even though it is not the pro, very confusing)–once loaded, click on upload-- then look for the ‘upload file system image’ and it is NOT there. NOT fun. So if it’s been a while and I forget, o…