Thanks for your suggestions Mick! And interesting to read about how you tackle the lookups.
I also use a quality Meanwell 5 volt power supply that is connected to a 230v UPS, and the Raspberry Pi is equipped with a PiJuice Hat( https://uk.pi-supply.com/products/pijuice-standard), so there are two layers of uninterruptible power supplies in the system. To reduce wear on the Flash-card it's only used during booting, the rest of the filesystem is on an external USB-harddrive. All 1-Wire devices are powered, no parasite-mode here, but I'm using telephone cables to connect the devices and not Cat5-cables. I used to have a similar DS2482-800 card as you described (*AbioWire, *https://axiris.eu/en/index.php/1-wire/abiowire) on top of my Raspberry, all the channels help to keep the number of devices per channel down. However that's in the past, instead of having "all eggs in the same basket" and where you need SSH knowledge to restart the "basket" I now try to be more microservice with small autonomous 1-Wire controllers that can be restarted by my wife with a button pressed. More suggestions about my code and lowlevel pitfalls of 1-Wire devices are welcome! Are there any alternatives to these traditional mailing lists? Like Discord-group? // Henrik Den mån 14 sep. 2020 kl 12:45 skrev Mick Sulley <m...@sulley.info>: > Hi Henrik, > > I can't answer you questions I'm afraid but I would add a few comments. > > I have been using Pi's since they were first released and yes, I have had > a few problems as well. The most common problem I have found is power > supply, always worth using a decent one, I use Meanwell units in critical > situations. Second on the list is SD cards, I have had loads of failures > there and I generally use hard drives recovered from old laptops instead of > SD cards. > > The biggest problem I have had with 1-wire is that it occasionally locks > up. I use a Sheepwalk RPi3 adapter to give me 8 channels, that uses the > DS2482-800 chip. When a lock-up occurs it needs a power cycle of 1-wire, > reboot does not reset it. The solution I use for that is to use a second > Pi which drives a relay (driven via GPIO pins) and 1-wire power is routed > through a normally closed contact on the relay. Pi-monitor reads a > heartbeat file on pi-control and if the heartbeat is more than 30 seconds > old it energises the relay for 10 seconds. This is controlling a solar hot > water system, which can easily boil (and has several times) if the controls > do not function correctly. > > Best of luck with your system. > > Mick > On 12/09/2020 11:32, Henrik Östman wrote: > > Hi! > I have been following the OWFS project for 20+ years and I really think > it's a marvelous piece of software. It's been run on a Raspberry Pi for > many years now to control the heating in our house. > However lately my Raspberry Pi has been more and more unstable, > overheating problems, diskfailure, hanging, owserver unexpectedly going > down and soo forth. It really got me thinking if my setup is the best > approche, my wife and kids should not have to be Unix admins and be able to > SSH to my server and restart the Docker container with Owserver just to not > freeze when I'm out on a business trip! So in a true Unix spirit I'm > splitting my Home automation system appart into well defined pieces that > "do one thing and do it well". > My house is heated by water-based floor heating, and there are three floor > heating control centers, with relays opening and closing the water-valves. > Each control center has a DS2408 controlling the relayes of 3-5 valves. A > made a small circuit board with a commonly used Arduino compatible > microcontroller that reads the temperature in the rooms using DS1820 > sensors, then instructs the DS2408 to open/close the right valve. There is > one circuit board for each control center, so if one goes down only a few > rooms get affected. There is a reset-button on each circuit board that > could be used to restart it for whatever reason, and since there are no > filesystems mounted there is no problem with corruption in case of power > loss, and restarting only takes a second. The Raspberry Pi is now only used > for plotting graphs, MQTT and other home automation tasks. > > Since I no longer use Owserver in my setup, I understand that it may be > inappropriate to ask for help within this forum. But I know that there are > some 10+ years skilled developers here with deep knowledge of 1-Wire > systems so I'm still going to pop the questions, maybe you could direct me > to a more suitable forum where we could continue the discussion? > > The problem I'm having is with the DS2408 devices, I only seem to be able > to communicate with them using the SKIP-ROM command, if I try to address > them individually then I only get garbage and errors back. The same code > works great when addressing DS2423 and DS18(B/S)20 devices, so I think the > code itself should work. Maybe I have missed something when communicating > or initializing the DS2408? I tried to read all the Maxim specs, and it > feels like I'm doing everything right. During startup I begin with > initializing the DS2408 and set the pins to output and to a known state. > > void ds2408_reset(DS2480B &ds, onewireNode &node) { > ESP_LOGD(TAG, "Reset DS2408, id: %s.", node.idStr.c_str()); > if (existTestMode(ds, node)) { > // Configure RSTZ as STRB output. > //ds.select(node.id); // reselect last selected device. > ds.write(SKIP_ROM); // HACK, this select all devices on the bus, but we > should select only this single device. Though I get CRC-errors using above > line. > ds.write(0xCC); // Issue Write Conditional Search Register command > ds.write(0x8D); // TA1, target address = 8Dh > ds.write(0x00); // TA2, target address = 008Dh > ds.write(0x04); // Write byte to Control/Status Register, RSTZ as STRB > output > // Verify configuration setting > if (!ds.reset()) { > ESP_LOGW(TAG, "Reset DS2408 failed after non-success configure RSTZ as > STRB."); > node.errors++; > return; > } > ds.write(RESUME); // reselect last selected device. > ds.write(0xF0); // Issue Read PIO Registers command > ds.write(0x8D); // TA1, target address = 8Dh > ds.write(0x00); // TA2, target address = 008Dh > auto status = ds.read(); // Read Control/Status Register and verify > ESP_LOGD(TAG, "DS2408 verify configuration setting: %s.", String(status, > HEX)); > // Set all relays off. > setState(ds, node, B11111111); > } else { > ESP_LOGW(TAG, "Reset DS2408 failed for id: %s.", node.idStr.c_str()); > } > } > > /** > * Exit test-mode. > * "The DS2408 is sensitive to the power-on slew rate and can inadvertently > power up with a test mode > * feature enabled. When this occurs, the P0 port does not respond to the > Channel Access Write command." > * @return 0=failed, 1=success > */ > bool existTestMode(DS2480B &ds, onewireNode &node) { > // RST PD 96h <64-bit DS2408 ROM Code> 3Ch RST PD > if (ds.reset()) { // onewire initialization sequence, to be followed by > other commands > ds.write(0x96); > for (uint8_t i = 0; i < 8; i++) { > ds.write(node.id[i]); > } > ds.write(0x3C); > if (ds.reset()) { > return 1; > } > } > return 0; > } > > int16_t setState(DS2480B &ds, onewireNode &node, uint8_t state) { > if (node.id[0] != DS2408) { > ESP_LOGW(TAG, "Device is not a DS2408!"); > return -1; > } > if (ds.reset()) { // onewire initialization sequence, to be followed by > other commands > ESP_LOGD(TAG, "Set DS2408 state to: %s.", String(state, BIN)); > uint8_t retries = MAX_CONSECUTIVE_RETRIES; > //ds.select(node.id); // issues onewire "MATCH ROM" address which selects > a SPECIFIC (only one) 1-Wire device > ds.write(SKIP_ROM); // HACK, this select all devices on the bus, but we > should select only this single device. Though I get CRC-errors using above > line. > do { > ds.write(0x5A); // Issue Channel-access Write command > ds.write(state); // Write byte to PIO > ds.write(~state); // Write inverted byte to PIO > auto status = ds.read(); // Read for verification (AAh = success) > auto newState = ds.read(); // DS2408 samples PIO pin status > if (status == 0xAA) { // AAh = success > ESP_LOGD(TAG, "DS2408 current state: %s.", String(newState, BIN)); > node.success++; > return newState; > } else { > node.errors++; > ESP_LOGW(TAG, "DS2408 setState failed, trying again..."); > if (ds.reset()) { > ds.write(RESUME); // reselect last selected device. > } else { > ESP_LOGW(TAG, "Reset DS2408 failed after non-success setState."); > node.errors++; > return -1; > } > } > } while (--retries); > return -1; > } else { > ESP_LOGW(TAG, "Reset DS2408 failed."); > node.errors++; > return -1; > } > } > > When using "ds.select(node.id);" it does not work, with > "ds.write(SKIP_ROM) it works better, the question is why? And if it does > matter, the circuit-board is using a DS2480 to communicate with the other > devices. > > Any help or suggestions would be more than welcome! > Thanks! > > // Henrik > > > > > _______________________________________________ > Owfs-developers mailing > listOwfs-developers@lists.sourceforge.nethttps://lists.sourceforge.net/lists/listinfo/owfs-developers > > _______________________________________________ > Owfs-developers mailing list > Owfs-developers@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/owfs-developers >
_______________________________________________ Owfs-developers mailing list Owfs-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/owfs-developers
