Completed DIY "laptop" / portable PC (7.39 litres, 17'' screen, 5700x, RTX2070)

[timginter]

I'd like to see how my power solution ends up working, too I'm updating the first post as I find new things, 3S1P 11.1V RC battery + own UPS looks like the final version, just waiting for the BMS from the list. Hopefully I won't have more odd electronics laying around after trial-and-erroring.

I'm working on a HP 750W server PSU mod and I'm planning to use 2x Dell DA-2 until I finish it. Led me to a cool side project to create easy connectors for Dell DA-2 to any output: https://egpu.io/forums/psu-cables/dell-da-2-to-any-output-adapter

Works really great, just waiting for the GxR-ONE plugin-in PSU to assemble the prototype (no battery so far)

 

[Choidebu]

For 3S 11.1V LiPo, it's advised to stop discharge at 10.5V. I've spoken to someone with more electrical knowledge than mine (not a tall order there ) and some 12V appliances should be ok with 10V.

Agreed with @ruleh on this one.

1. A nominal 11.1V 3S lipo means 9.6V - 12.6V range, and that's what mfgs use to determine their capacity. Using 11.4V to 12.6V essentially only using 40% of its stated capacity. Actually not quite, because..

2. Voltage droop on load. Once you connect a load to the pack, a 12.6V (open circuit, fully charged) pack would go 12V straight away, and within seconds of considerable load go to 11.1 - 11.4V area.

This is normal as the pack would be working like 75% of their capacity going from 11 to 10V (depends on discharge rate)

This means I'd venture a guess if you're doing anything on the pc and mains failed, you have prob seconds before GxR shuts off. If it's idling, maybe 2 minutes, 5 mins tops.

 

[timginter]

Agreed with @ruleh on this one.

1. A nominal 11.1V 3S lipo means 9.6V - 12.6V range, and that's what mfgs use to determine their capacity. Using 11.4V to 12.6V essentially only using 40% of its stated capacity. Actually not quite, because..

2. Voltage droop on load. Once you connect a load to the pack, a 12.6V (open circuit, fully charged) pack would go 12V straight away, and within seconds of considerable load go to 11.1 - 11.4V area.

This is normal as the pack would be working like 75% of their capacity going from 11 to 10V (depends on discharge rate)

This means I'd venture a guess if you're doing anything on the pc and mains failed, you have prob seconds before GxR shuts off. If it's idling, maybe 2 minutes, 5 mins tops.

Hmm? Crossed messages? You quoted me "should be ok with 10V", then your calculations are based on basic battery knowledge and standard spec, but I wrote I won't be using it at 11.4-12.6V. I also wrote I'll test the GxR-ONE on low voltages to check if it's stable. If even 10.3V is still stable, the recommended 10.5V cutoff for 3S 11.1V battery will be perfect. It won't work for a standard "within spec" solution, but really not sure what you mean in my case.

10.5-12.6V should give me around 12 minutes at 450W and around an hour of light use with a 8000mAh battery - that's plenty for me. I had a great discussion about DIY UPSes and was recommended a relay instead of diodes - that was a gamechanger. I also found by accident that BMS's come in separate ports config - another gamechanger. The cables, BMS and the relay switch should create slight voltage drop to knock down max battery voltage from 12.6V, just to be safe:

AC-DC ----------- RELAY --- 25V 1000uF --- GxR-ONE
PSU      |          |       DECOUPLING     ATX PSU
        CCCV        |       CAPACITOR
       STEP-UP   VOLTAGE
       12.6V 6A   METER
         |          |
      SEPARATE ------
      PORTS BMS
         |
     3S1P 11.1V
    8000mAh PACK

The key parts here are:

• separate ports BMS - will not feed the current back into the CCCV converter
• relay switch - power from PSU or from the battery without battery feeding "around" and back to CCCV converter. no need for diodes

I already have the CCCV regulator, relay switch, voltage meter and the battery, just waiting for the separate ports BMS to test everything

 

[CC Ricers]

I'd like to see how my power solution ends up working, too I'm updating the first post as I find new things, 3S1P 11.1V RC battery + own UPS looks like the final version, just waiting for the BMS from the list. Hopefully I won't have more odd electronics laying around after trial-and-erroring.

I'm working on a HP 750W server PSU mod and I'm planning to use 2x Dell DA-2 until I finish it. Led me to a cool side project to create easy connectors for Dell DA-2 to any output: https://egpu.io/forums/psu-cables/dell-da-2-to-any-output-adapter

Works really great, just waiting for the GxR-ONE plugin-in PSU to assemble the prototype (no battery so far)

There are loads of RC lipo batteries though not one I found that would be in the form factor that would fit my tiny case. I'm no electronics wizard and don't feel up to the task of ripping apart the pack and using it for a custom fit yet.

Will you be using a RC lipo charger for those batteries? One thing I've considered but feels janky using yet another power charger and unload the battery whenever you want to charge it... unless yours is like a custom UPS too.

 

[ruleh]

I think you might need a buck-boost converter now that the battery voltage can go lower than the converter input voltage. Though I guess that would depend on how the boost converter does its current limiting.

 

[timginter]

There are loads of RC lipo batteries though not one I found that would be in the form factor that would fit my tiny case. I'm no electronics wizard and don't feel up to the task of ripping apart the pack and using it for a custom fit yet.

I wanted to avoid DIY battery, too. Had the same problem with dimensions when I wanted to keep the battery at max 20mm, RC plane batteries looked good, they are usually quite short and flat, but a bit more expensive.
You don't have to cut open an RC pack, you can buy cells separately and wire them up yourself - electric skateboard videos on youtube have good instructions. It will probably end up more expensive than an RC pack, though, I'd probably also use battery holders not (spot) solder straight on the battery

Will you be using a RC lipo charger for those batteries? One thing I've considered but feels janky using yet another power charger and unload the battery whenever you want to charge it... unless yours is like a custom UPS too.

If you look at the rough schematics above, the CCCV (constant current constant voltage) will charge the battery (provide 12.6V at 6A) and the BMS (battery management system) will balance each cell (charge/discharge every cell separately) and protect the battery from overcharge/overdischarge. Schematics above are pretty much a DIY UPS. It will be clunky but it should support up to 60A and I couldn't find any "official" 12V UPS boards for such high current.

I think you might need a buck-boost converter now that the battery voltage can go lower than the converter input voltage. Though I guess that would depend on how the boost converter does its current limiting.

Not sure I understand. The only converter is the CCCV to charge the battery and it's used only one way - from ATX PSU to BMS, to charge the battery. BMS has separate ports for input and output, so current from the battery will not go back to CCCV converter. Relay switch controls the flow direction - if ATX PSU is unplugged, relay will only allow the path from battery to GxR-ONE. With the ATX-PSU plugged in, relay will cut off battery and open the path from ATX-PSU to GxR-ONE

 

[ruleh]

What I'm saying is that the CCCV converter might not actually limit the current when its input voltage is greater than its output voltage. At least both of my (step up) converters behave like that. I don't know about yours but you haven't put the exact converter in the parts list.

 

[timginter]

What I'm saying is that the CCCV converter might not actually limit the current when its input voltage is greater than its output voltage. At least both of my (step up) converters behave like that. I don't know about yours but you haven't put the exact converter in the parts list.

Thanks, totally misunderstood. I clarified the parts list and added a link to the one I bought (might look for a shorter one in the future, though). Because of wide range voltage input and output I couldn't give a search link. As long as you have separate ports BMS, any [around 12V input] to [around 12V, set to 12.6V output] @ 4-8A CCCV step-up will work.

Input from ATX PSU is around 12.33V (minus around 2% from cabling) and I'm stepping up to 12.6V. In my case input current will never be greater than output, there will also be no current flowing back into the CCCV step-up

 

[ruleh]

So the bms also does current limiting? Otherwise the battery+bms will pull the converter output voltage down to the battery voltage, at which point the current limiting from the converter won't work anymore.

 

[Choidebu]

I wrote I won't be using it at 11.4-12.6V. I also wrote I'll test the GxR-ONE on low voltages to check if it's stable. If even 10.3V is still stable, the recommended 10.5V cutoff for 3S 11.1V battery will be perfect.

So you're betting on the GxR to not shutdown below 11.4V?

My recommendation is to choose an AC-DC PSU with 11.8VDC~12.3VDC output but you can run it safely with 12.6VDC maximum and 11.4VDC minimum.

Cause the maker here says that's the minimum.

I'm not talking out of napkin math; just my personal experience. Woukd be great if GxR does still run on 10.5V.

Atx spec iirc allows for +/- 0.5V. Most 12V plugin board I know does no regulation on input, only filtering.

 

[timginter]

So the bms also does current limiting? Otherwise the battery+bms will pull the converter output voltage down to the battery voltage, at which point the current limiting from the converter won't work anymore.

BMS doesn't limit the current. I'm not sure I understand - why would the BMS limit/change CCCV's output? I don't think BMS regulates voltage "down to battery voltage", as far as I know it takes whatever voltage comes in and applies it to the battery (plus protections/balancing). Even if BMS regulated the voltage it would still take 12.6V but dissipate any difference as heat? CCCV would always output 12.6V? If I'm being really daft here, please explain step-by-step like for a 5 year old

So you're betting on the GxR to not shutdown below 11.4V?

I'm not betting on it - I already wrote a few times that I'll test it first Battery solution is a work-in-progress, if lower voltage doesn't work well it means back to the drawing board and I can return CCCV, BMS, relay and the battery. I wouldn't go over the spec'd voltage, but I'm not too worried about testing the other way - at lower voltages GxR-ONE should just pull slightly more amps to compensate, but it will still be well within 50 amps spec. If the voltage is too low, GxR-ONE should just shut down.

According to spec, Daly's BMS have customisable overdischarge protection so if all works I'm hoping to set the discharge protection to 10.6V

 

[Choidebu]

Just trying to warn you of pitfall(s) I've seen firsthand, on my own battery project.

I agree though, that discussing battery at this point of your project seems silly. I'll come back later when you got some chance to finish up other parts and familiarized yourself with batteries' stuff.

at lower voltages GxR-ONE should just pull slightly more amps to compensat

@REVOCCASES , does GxR-ONE 'compensate'? No need to answer here, just tag me somewhere else.

 

[Choidebu]

So the bms also does current limiting? Otherwise the battery+bms will pull the converter output voltage down to the battery voltage, at which point the current limiting from the converter won't work anymore.

Any worthwhile bms should have ocp, scp, at least on output. On bmses with separate charging port I never really notice ocp on input - seems like a very case-by-case protection (depends on how many P group and C rating you want to charge at). At the very least you got ovp for sure, it's going to trigger at the end of cv cycle.

In any case I don't think worrying abt that now is needed.

 

[REVOCCASES]

does GxR-ONE 'compensate'? No need to answer here, just tag me somewhere else.

not for the 12V rail (CPU and GPU power), only for the 24P plugin part

It might still work though. Motherboards and GPUs can actually accept much lower "12V" voltages. I do not recommend doing it but I have tried running my PC at something around 10V and did not experience any issues. At some point if the voltage is too low, your PC will just freeze/crash. But never ever go over the 12.6V, this might end bad. And definitely watch out for the current on the cables/connectors when running it at lower voltages...

Again: GxR-ONE official input voltage spec is: 11.8VDC to 12.3VDC

but @timginter told me, he is aware of potential risks involved running it out of specs - so no guarantee from my side if/how it actually works for this application - guess I would try it with some old hardware first

I mean this project is all about "no risk - no fun" and "learning by doing" and I really like that...

 

[ruleh]

BMS doesn't limit the current. I'm not sure I understand - why would the BMS limit/change CCCV's output? I don't think BMS regulates voltage "down to battery voltage", as far as I know it takes whatever voltage comes in and applies it to the battery (plus protections/balancing). Even if BMS regulated the voltage it would still take 12.6V but dissipate any difference as heat? CCCV would always output 12.6V? If I'm being really daft here, please explain step-by-step like for a 5 year old

I don't really know how to explain this... I'm not sure if I'm even right or not...
Well anyways here is an attempt:

A current limited voltage supply will not usually supply its max current and max voltage at the same time.
If the load is low (high resistance), it will supply the max voltage and let it draw whatever current it wants as long as its below the set max current.
If the load is high (low resistance) and tries to draw more than the set max current, the voltage regulator will lower its voltage so that the updated current draw will be less than max current.

As an example let's take Vmax=12.6V and Imax=6A.
A 5Ohm load would get 12.6V and draw 2.52A
A 1Ohm load would only get 6V and draw 6A.
If the load got 12.6V instead, it would be draw 12.6A.

However with a step-up converter taking a 12V input, the load will always get at least 12V because a step-up converter can't supply an output voltage lower than its input voltage.

Any worthwhile bms should have ocp, scp, at least on output. On bmses with separate charging port I never really notice ocp on input - seems like a very case-by-case protection (depends on how many P group and C rating you want to charge at). At the very least you got ovp for sure, it's going to trigger at the end of cv cycle.

In any case I don't think worrying abt that now is needed.

The problem is that in this case we have an unregulated 12V charger for battery voltages <12V, a CC cycle only from 12V-12.6V and a CV cycle finally at 12.6V.
I'm not sure if depending on ocp as a regulator for battery charging at the lower voltages is a good idea.
(On the other hand it might allow for pulse charging )


One minor problem with having a low 12V rail is that fans won't spin at max rpm. (And maybe some harddrives will not like it?)

Spoiler: random battery runtime test

A battery test with a "good" 6S 5600mAh battery.
total runtime: 130min
5min at 200W (had to stop because gpu overheated)
80min 1080p video playback (without hw decoding because I'm a dum-dum and deleted the driver)
45min idle at 18-30W

During the video playback there was also some misc loads here and there (compiling, compressing, decompressing), maybe for a total of 10min or so.

 

[REVOCCASES]

One minor problem with having a low 12V rail is that fans won't spin at max rpm. (And maybe some harddrives will not like it?)

That's correct. Doesn't work with traditional HDDs and fans would spin a bit slower.

 

[Choidebu]

The problem is that in this case we have an unregulated 12V charger for battery voltages <12V, a CC cycle only from 12V-12.6V and a CV cycle finally at 12.6V.

See his updated diagram; he's not using step-up, he's using CC-CV charger.

I see you understand how CC supply works, but it baffled me that you seem to think a CC supply needs a variable CV supply to start with.

The simplest CC supply you can make with an LM317, a capacitor and a resistor. It takes a (constant or not) voltage supply. It outputs voltages anywhere up to input voltage - 4.25 V as longs as it meets the set current (which is calculated as voltage drop of 1.25V across the 'sense' resistor)

See modern CCCV doesn't need variable supply - it can be buck boost, buck only. Doesn't matter.

 

[Choidebu]

Just realised I'm not as adventurous as I thought I was. In fact I'm quite the strickler for specs compared to you guys in this thread.

 

[ruleh]

The parts lists mentions this boost converter:

So I assumed that it is being used to charge the battery directly.
If it is being used directly, it needs to have buck-boost topology.

I'm not exactly sure which the updated diagram is, but the latest one I can find says "CCCV STEP-UP".

To clarify a bit, I don't think that a CC supply needs a variable voltage supply but think that it becomes the variable voltage supply.

(I thought that the CV meant constant voltage which makes the "variable CV supply" hard to understand. Then again, programming has "static constant variables".)

 

[Choidebu]

Variable CV supply means a CV supply with output voltage you can adjust. Constant just means it'll try to maintain that set voltage.

A CCCV module would always need a buck. If it says step up then it's probably buck boost.

Speaking of stepping... Lemme just steps.. away and let people learn their lesson... Again... In the name of fun..