Schematic and layout in Eagle PCB for a quad bm1387 standalone miner
// skip if deemed off topic!
// Other versions made
// A single BM1391 version, obsolete as information on the asic is too scarce
// aSiNine Pharaonis: dual BM1387 version, compatible with the quad version (this)
// Unnamed project single BM1397 (in pre-alpha stage, pcb's are on the slow boat from China)
Motivation:
I'm running an ESP32C3 with a RGB LED to monitor traffic (flickers on inbound TX's and glows on memory pool size) to my node.
Why not let it do something useful while idle?
Properties (concept):
- Standalone miner to be connected to a mining pool or a private Bitcoin node that will serve the work (over websocket)
- Size: 80x60mm
- Programmable Vcore for the bm1387 asics
- ws2812 led for special effects ;)
- max. power usage targeted at ~10-15W
- Cooling: M.2 SSD cooler, tie-wrapped to the board with 5V fan.
- Low(est) cost possible!
Major parts used:
- 4x BM1387xx asics (tested bm1387bf & bm1387b)
- ESP32-C3-mini-1
- SY8208A or B step down regulator (8A)
(currently controlled by PWM, considering a cheap ($0.10) DAC)controlled by a dac. - XC6206P-18 lin. regulators for 1.8 & 0.8V
- SY8009B step down regulator (overkill) for the esp32c3, limiting the power supply to 5V only..
- usb-c connector for power (5V) and debugging
Revisions
* Don't build anything alpha, components and schematic/layout will change in future revisions. *
v0.2-alpha (see tag) 7-12-2022 upto today
Not actually build/tested yet.
All bugs from v0.1 corrected.
Included dac to control Vcore (in 19.2uV steps, overkill).
Added temperature sensor (not yet calibrated or tested).
Added fan (5V only) with pwm control.
Added feedback of Vcore to the ESP32 to calibrate the power supply.
No more (big) hardware changes anticipated.
No more (big) layout changes aticipated. Ok, now though.
v0.1 29-10-2022
Debugging stage
pre alpha version
(For the unpublished 2 asic version (Pharaonis): Added a pt8211 DAC to control Vcore instead of a PWM from the esp32c3, the pwm caused some problems during wifi transfers and reconnects. I'm currently working on this version, some results found during testing: power usage ~2.8W @ 44GH/s ~40GH/s, so it can be passively cooled.)
!BM1387 apparently only seaches through roughly 4000 mega nonces out of the 4.295 mega nonces available...!
Function tested ok but unstable (related to pwm)
todo: final speccing of decoupling capacitors, redo some of the power&ground planes, more stability testing, add dac to schematic and layout, change mounting holes for heatsink.