- Wed Mar 22, 2017 6:22 pm
#194031
So as some of you probably guessed the title is a bit misleading. I'm not building a desktop PC. I'm building a mobile PC. The trouble I'm having is converting the battery voltage to a usable DC voltage. It has been suggested that I just use a DC-AC inverter and a standard power supply. That would work but...it's hardly efficient. There are losses in the inverter and losses in the power supply. Also it adds some excessive bulk and complexity (relatively speaking) to the system. You might also be asking, "why not use a laptop?" Upgrade-ability mostly. Laptops tend to have a very set lifespan ideally I could strip the case of everything and use more efficient hardware or get a new GPU or CPU. You can do that for some laptops, I've got an alienware right here that has a discrete gpu but there's only one other GPU that fits and I have to get it refurbed from a suspicious dealer based in china. That laptop is maybe 6 years old.
I'm planning on using openBMS to handle the li-ion battery pack(s) and then I was planning to use off the shelf parts to do the DC-DC conversion for each rail. This is fine if I do a single series system. 15 cells in series yields a 38-63v pack which slots into a lot of off the shelf parts. No problem, except if the wire from the battery to the PC happens to fray or otherwise become exposed then you have a very high voltage exposed. I'm just not comfortable with that. My research into "safe voltages" lead me to believe that 24v is about as high as you want to go without needing to excessively protect against loose wires. Depending on how this thread goes I may decide to go ahead with adding the precautions because it's too hard to use the lower voltage. So with 24V as my max that gives me a 5s 3p system. 21vpk good. and 12.5vmin...oh dear. A lot of the off the shelf products require 3-4 volts between vbat and vout. Anything with a lower dropout also has a very low current. That's a problem, as the planned computer would need 250w at least. (a bigger buffer would be preferred but start at 250w and work from there as I've measured about 240w) Why so much? Well there is a big honkin GPU in there for doing math and video processing, a 1070 right now. We'll see as time moves on if something more efficient shows itself.
So that is the project outline. I'm looking for advice on a the DC-DC converters. I'll need 12v 5v and 3.3v outputs. I think I can ditch the -12, but there are off the shelf -12v regs that would work as only a few watts are required. I have looked at devices like the LM3489 buck controller but, I wasn't able to do the math to get good component values out of the equations in that datasheet for the power requirements. I was getting a very low switching frequency (like 12hz) so maybe was doing something wrong. Or maybe the vmin is too close to the vout for it to work properly. The other option I've recently found is the OpenUPS by minibox. I was hoping to tweak the hardware to improve the current output to the required level but haven't been able to find a schematic. Maybe it's possible to team two or more of those or split them into a dual 12v rail system. That's something I'd need to keep digging into. I've also seen transistor trees that use a linear regulator to control voltage and a stack of transistors to supply loads of current. Could work, but I wonder about efficiency and response to sudden current demands. I'm open to thoughts ideas and suggestions, not just on the DC-DC converters. If you've feel I've made a misstep in logic along they way let me know. I may not have explained myself properly (happens a lot) or I could just be too set in my thinking.
TL:DR Need high current low dropout DC-DC conversion solution for a minimum 250W battery powered PC.
I'm planning on using openBMS to handle the li-ion battery pack(s) and then I was planning to use off the shelf parts to do the DC-DC conversion for each rail. This is fine if I do a single series system. 15 cells in series yields a 38-63v pack which slots into a lot of off the shelf parts. No problem, except if the wire from the battery to the PC happens to fray or otherwise become exposed then you have a very high voltage exposed. I'm just not comfortable with that. My research into "safe voltages" lead me to believe that 24v is about as high as you want to go without needing to excessively protect against loose wires. Depending on how this thread goes I may decide to go ahead with adding the precautions because it's too hard to use the lower voltage. So with 24V as my max that gives me a 5s 3p system. 21vpk good. and 12.5vmin...oh dear. A lot of the off the shelf products require 3-4 volts between vbat and vout. Anything with a lower dropout also has a very low current. That's a problem, as the planned computer would need 250w at least. (a bigger buffer would be preferred but start at 250w and work from there as I've measured about 240w) Why so much? Well there is a big honkin GPU in there for doing math and video processing, a 1070 right now. We'll see as time moves on if something more efficient shows itself.
So that is the project outline. I'm looking for advice on a the DC-DC converters. I'll need 12v 5v and 3.3v outputs. I think I can ditch the -12, but there are off the shelf -12v regs that would work as only a few watts are required. I have looked at devices like the LM3489 buck controller but, I wasn't able to do the math to get good component values out of the equations in that datasheet for the power requirements. I was getting a very low switching frequency (like 12hz) so maybe was doing something wrong. Or maybe the vmin is too close to the vout for it to work properly. The other option I've recently found is the OpenUPS by minibox. I was hoping to tweak the hardware to improve the current output to the required level but haven't been able to find a schematic. Maybe it's possible to team two or more of those or split them into a dual 12v rail system. That's something I'd need to keep digging into. I've also seen transistor trees that use a linear regulator to control voltage and a stack of transistors to supply loads of current. Could work, but I wonder about efficiency and response to sudden current demands. I'm open to thoughts ideas and suggestions, not just on the DC-DC converters. If you've feel I've made a misstep in logic along they way let me know. I may not have explained myself properly (happens a lot) or I could just be too set in my thinking.
TL:DR Need high current low dropout DC-DC conversion solution for a minimum 250W battery powered PC.