3 ms·

> As far as I know, Lead-Acid still is very popular in UPS, which is basically what this SolarPi is. No, SolarPi is a solar powered server. The application her

by shkkmo 3y ago

> As far as I know, Lead-Acid still is very popular in UPS, which is basically what this SolarPi is. No, SolarPi is a solar powered server. The application here is a "solar generator" type setup that has no grid connection and is very different from a UPS style setup.

dragontamer 3y ago
Please see my durability calculations: https://news.ycombinator.com/item?id=37224926 https://news.ycombinator.com/item?id=37224926 A 3W Rasp. Pi with 12V 20Ah lead-acid backup with a 500 cycle lifespan should last 1500 days at a minimum. Because it takes 3+ days for a discharge cycle to occur, so that's easily 4+ years of worst-case durability. And I'm willing to bet that if you design the solar panel CORRECTLY, you ain't gonna be doing full discharges every 3 days (IE: enough solar panels to have 99%+ uptime, meaning you've overbuilt solar panels to handle the worst case scenarios and try to stay at 100% power as much as possible). > No, SolarPi is a solar powered server. The application here is a "solar generator" type setup that has no grid connection and is very different from a UPS style setup. If you're hoping for 90% or 99% uptime, this solar application is going to look closer-and-closer like a UPS. 90% means you're down at most 36-days of the year (pretty crap reliability, all else considered). ------------ > Grid backup Also, the SolarPi guy is looking into Grid-backup systems. So yes, we're back to UPS
- shkkmo 3y ago
  Your durability calculations use made-up numbers and don't reflect a great understanding of lead acid and it's limitations. Assuming 12hrs of solar charge is ridiculous and a 12v 20ah lead acid will provide ~120wh (since you can only use 50%) which seems like about half the capacity you would need for your 72 hr "worst case". While you can get 500 "deep" (50% usage) cycles out of a lead acid battery that is specifically designed for that, you'll have lost almost half your capacity by the end. You also ignore the extra wear from partial recharges which will be pretty common since the top 20% capacity charges so slowly You have to buy 2x as much lead acid capacity, it lasts less than half as long and you get worse charging efficiency. The only time lead acid makes sense is when you need a lot of capacity but that capacity is only used infrequently and you have plenty of time to recharge between usage events. That is not the use case here. > If you're hoping for 90% or 99% uptime, this solar application is going to look closer-and-closer like a UPS. No, it is still a solar generator with a solar generator usage cycle. You still have the paired problems of reduced lead acid charging efficiency at the top 20% combined with extra battery wear for every day that you don't get the batteries topped up despite that efficiency. In order to reduce this, you have to have enough extra batteries that they can soak in your solar output even when in the slower charging of the top 20% while also having enough solar capability to get back up to 100% on partially sunny days. These issues significantly hamper the viability of lead acid for solar generators and are why LifePO4 has completely dominated Lead Acid in that market. > Also, the SolarPi guy is looking into Grid-backup systems. So yes, we're back to UPS The goal here is to switch to grid only when the battery runs out. That is the opposite of a UPS and is still a solar generator.
  - dragontamer 3y ago
    I think you're forgetting that this is a $30 Rasp. Pi battery solution that I'm talking about. When your battery wears out, recycle the old Lead Acid battery and buy a new one. Its not that big of a deal. The battery I spec'd out is a 15lb battery (~7 kg). Heavier than LiFePo4 for sure, but no one is going to have issues moving, replacing, or buying new batteries. -------- Anyway, I recognize that LiFePo4 is superior durability by a long shot. Thousands of cycles instead of hundreds, yes. My overall point however, is that "hundreds of cycles" is still years in this use case, and I don't think you've challenged the math too much from that perspective. Obviously, I'd prefer to keep on the cheaper-side of Lead-Acid batteries. But the more expensive Trojan AES batteries or "Gel" batteries have far better durability (rather than the cheapest AGMs that I've been talking about) I'm talking 1400 cycles at 100% DoD without any precharge or low-charge penalties for Trojan AES (though this is a very new battery only coming out this year). Trojan is working on making Lead Acid competitive vs Lithium Ion and have the cream-of-the-crop Lead Acid chemistries. But even "Gel" (more expensive than AGMs) type Lead Acids have specifications for 100% DoD cycles, 300 to 500 depending on brand. --------------- Anyway, this has gone on for many posts. But here's my arguments to finalize. 1. Float-charging "feature" for Lead Acid grossly simplifies circuit design. You possibly don't need a charge controller at all anymore (feasible design with just diodes), although if you want decent charge speeds you'll need to build your own charge controller. (though this seems doable with uCs and/or OpAmps to make CC-CV circuits). 2. Very low explosion risk, so this is appropriate for hobbyists (unlike Li-ion which even LiFePo4 have had some worrying fires). So DYI controllers for LiFePo4 is an issue of trusting the battery-controller and the brand you buy, details I don't have to worry about with Lead Acid. 3. Lead Acid has less durability, "hundreds" depending on brand (and yes, I'm talking about 100% DoD specs, not 50% specs like you wanna talk about). By my estimation, this provides years of service though this obviously depends on various assumptions. 4. Lead Acid's linear voltage is an advantage when determining DoD when building your own controllers (ex: 13V OCV means you're likely full, and 11.6V OCV means you're likely empty, and 12.3V means you're at 50%). LiFePo4 has "constant voltage", which is better for efficiency and power but it makes state-of-charge far more difficult to perform if you're building your own charge circuitry. ---------- LiFePo4 is a "superior" battery from a durability perspective. But no one seems to want to float-charge this thing for extended periods of time, and I certainly don't want to have an explosion risk while I test that out. And "downsides" (like Lead Acid's linear voltage drop, leading to lower efficiencies) are advantages with regards to circuit simplicity and circuit design. A *simpler* design is more suited for hobbyists, even if its got worse technical specs than a more sophisticated chemistry.

by shkkmo 3y ago

dragontamer 3y ago

Please see my durability calculations: https://news.ycombinator.com/item?id=37224926 https://news.ycombinator.com/item?id=37224926 A 3W Rasp. Pi with 12V 20Ah lead-acid backup with a 500 cycle lifespan should last 1500 days at a minimum. Because it takes 3+ days for a discharge cycle to occur, so that's easily 4+ years of worst-case durability. And I'm willing to bet that if you design the solar panel CORRECTLY, you ain't gonna be doing full discharges every 3 days (IE: enough solar panels to have 99%+ uptime, meaning you've overbuilt solar panels to handle the worst case scenarios and try to stay at 100% power as much as possible). > No, SolarPi is a solar powered server. The application here is a "solar generator" type setup that has no grid connection and is very different from a UPS style setup. If you're hoping for 90% or 99% uptime, this solar application is going to look closer-and-closer like a UPS. 90% means you're down at most 36-days of the year (pretty crap reliability, all else considered). ------------ > Grid backup Also, the SolarPi guy is looking into Grid-backup systems. So yes, we're back to UPS

shkkmo 3y ago
Your durability calculations use made-up numbers and don't reflect a great understanding of lead acid and it's limitations. Assuming 12hrs of solar charge is ridiculous and a 12v 20ah lead acid will provide ~120wh (since you can only use 50%) which seems like about half the capacity you would need for your 72 hr "worst case". While you can get 500 "deep" (50% usage) cycles out of a lead acid battery that is specifically designed for that, you'll have lost almost half your capacity by the end. You also ignore the extra wear from partial recharges which will be pretty common since the top 20% capacity charges so slowly You have to buy 2x as much lead acid capacity, it lasts less than half as long and you get worse charging efficiency. The only time lead acid makes sense is when you need a lot of capacity but that capacity is only used infrequently and you have plenty of time to recharge between usage events. That is not the use case here. > If you're hoping for 90% or 99% uptime, this solar application is going to look closer-and-closer like a UPS. No, it is still a solar generator with a solar generator usage cycle. You still have the paired problems of reduced lead acid charging efficiency at the top 20% combined with extra battery wear for every day that you don't get the batteries topped up despite that efficiency. In order to reduce this, you have to have enough extra batteries that they can soak in your solar output even when in the slower charging of the top 20% while also having enough solar capability to get back up to 100% on partially sunny days. These issues significantly hamper the viability of lead acid for solar generators and are why LifePO4 has completely dominated Lead Acid in that market. > Also, the SolarPi guy is looking into Grid-backup systems. So yes, we're back to UPS The goal here is to switch to grid only when the battery runs out. That is the opposite of a UPS and is still a solar generator.
- dragontamer 3y ago
  I think you're forgetting that this is a $30 Rasp. Pi battery solution that I'm talking about. When your battery wears out, recycle the old Lead Acid battery and buy a new one. Its not that big of a deal. The battery I spec'd out is a 15lb battery (~7 kg). Heavier than LiFePo4 for sure, but no one is going to have issues moving, replacing, or buying new batteries. -------- Anyway, I recognize that LiFePo4 is superior durability by a long shot. Thousands of cycles instead of hundreds, yes. My overall point however, is that "hundreds of cycles" is still years in this use case, and I don't think you've challenged the math too much from that perspective. Obviously, I'd prefer to keep on the cheaper-side of Lead-Acid batteries. But the more expensive Trojan AES batteries or "Gel" batteries have far better durability (rather than the cheapest AGMs that I've been talking about) I'm talking 1400 cycles at 100% DoD without any precharge or low-charge penalties for Trojan AES (though this is a very new battery only coming out this year). Trojan is working on making Lead Acid competitive vs Lithium Ion and have the cream-of-the-crop Lead Acid chemistries. But even "Gel" (more expensive than AGMs) type Lead Acids have specifications for 100% DoD cycles, 300 to 500 depending on brand. --------------- Anyway, this has gone on for many posts. But here's my arguments to finalize. 1. Float-charging "feature" for Lead Acid grossly simplifies circuit design. You possibly don't need a charge controller at all anymore (feasible design with just diodes), although if you want decent charge speeds you'll need to build your own charge controller. (though this seems doable with uCs and/or OpAmps to make CC-CV circuits). 2. Very low explosion risk, so this is appropriate for hobbyists (unlike Li-ion which even LiFePo4 have had some worrying fires). So DYI controllers for LiFePo4 is an issue of trusting the battery-controller and the brand you buy, details I don't have to worry about with Lead Acid. 3. Lead Acid has less durability, "hundreds" depending on brand (and yes, I'm talking about 100% DoD specs, not 50% specs like you wanna talk about). By my estimation, this provides years of service though this obviously depends on various assumptions. 4. Lead Acid's linear voltage is an advantage when determining DoD when building your own controllers (ex: 13V OCV means you're likely full, and 11.6V OCV means you're likely empty, and 12.3V means you're at 50%). LiFePo4 has "constant voltage", which is better for efficiency and power but it makes state-of-charge far more difficult to perform if you're building your own charge circuitry. ---------- LiFePo4 is a "superior" battery from a durability perspective. But no one seems to want to float-charge this thing for extended periods of time, and I certainly don't want to have an explosion risk while I test that out. And "downsides" (like Lead Acid's linear voltage drop, leading to lower efficiencies) are advantages with regards to circuit simplicity and circuit design. A *simpler* design is more suited for hobbyists, even if its got worse technical specs than a more sophisticated chemistry.