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

[timginter]

It turns out that given the right (or really wrong) circumstances, a buck converter can also work as a boost converter in reverse, providing a high voltage on what should have been the input.
So far max measured current was ~17A (at 22V) and max voltage 40V (0A), which is a bit higher than the 18-35V rating.
It's probably not so good for the converter and has no real use anyways since this specific converter is not adjustable.
Oh well... too bad that bidirectional dc-dc converters aren't a thing on ebay and such.

Spoiler

Putting a high load on the 12V rail such that it drops to 12.3V causes 40V to appear at the battery V+ rail.

AC-DC ----------------------->|------------------------DC-ATX
(12.6V)        |                             |
               |                             | (out: 12.3V)
             BOOST                         BUCK
             CONVERTER                     CONVERTER
               | (out: 25.3V 1A)             | (in: 18-35V rated)
               |                             |
               -------------------------------
               |   40V when ac-dc at ~12.3V
               |
              BMS
               |
           6S BATTERY
           (18-25.3V)

Did all go OK, or were those the "really wrong" circumstances?

With stepping down, I just haven't found a converter which could handle high current for a more demanding rig. So far 3S 11.1V seems promising, just waiting for BMS to test the battery properly

 

[ruleh]

Those were the really wrong circumstances but everything seems somewhat ok, I think.
The converter might have taken a hit but it still works.
The battery also seems alive but I don't think that pushing >350W into it did it any good.
Then again it didn't heat up too much, reaching 45˚C, so who knows?
Interestingly the ac-dc psu didn't blow up despite supplying >500W. It might be rated for (fictional) 480W but still ran for some time without getting too hot or unstable.

For stepping down this one seems good if the data is correct.
https://www.ebay.de/itm/800W-30A-DC...Power-CV-CC-Adjustable-For-Solar/284043976423
It would only supply 375W at max (with 12.5V) though but that seems just about enough since it doesn't have to charge the battery like the ac-dc supply has to. That is assuming the gpu pulls ~200W.

For the 3S setup, you might be able to connect the battery directly to the 12V rail depending on the exact rail voltage and how much current would flow into an empty battery. It might even provide some filtering for the ac-dc psu.

 

[timginter]

Those were the really wrong circumstances but everything seems somewhat ok, I think.
The converter might have taken a hit but it still works.
The battery also seems alive but I don't think that pushing >350W into it did it any good.
Then again it didn't heat up too much, reaching 45˚C, so who knows?
Interestingly the ac-dc psu didn't blow up despite supplying >500W. It might be rated for (fictional) 480W but still ran for some time without getting too hot or unstable.

Whoah, sounds serious. Glad everything is OK, though.

For stepping down this one seems good if the data is correct.
https://www.ebay.de/itm/800W-30A-DC...Power-CV-CC-Adjustable-For-Solar/284043976423
It would only supply 375W at max (with 12.5V) though but that seems just about enough since it doesn't have to charge the battery like the ac-dc supply has to. That is assuming the gpu pulls ~200W.

True, for a "lower" draw rig and 5S or 6S battery that could work, it might be pushing that heatsink quite a bit, though.

For the 3S setup, you might be able to connect the battery directly to the 12V rail depending on the exact rail voltage and how much current would flow into an empty battery. It might even provide some filtering for the ac-dc psu.

My modded server PSU outputs 11.98-12.22V depending on load - it wouldn't charge the battery to full. I'd be really worried the battery would pull too much current at some point, too.

Can't wait for the BMS to come - with separate ports and the Constant Current Constant Voltage converter/charger the 3S battery setup may actually work

 

[timginter]

GxR-ONE comes with a female connector, but RC batteries and my modded server PSU also have a female connector.

I soldered a makeshift male-to-male XT90 connector, stretched the shrink wrap with snipe-nose pliers to fit around both connectors:

Spoiler: photos

This will come handy when I'm testing GxR-ONE at low voltage - I can take the shrink wrap off and have easy access eith a multimeter.

All works good, last few days I was running benchnarks and testing temperatures. Had to re-seat the CPU - I messed up the thermal compund and heat sink - I actually hit 90 degrees on benchmark

I also raised the motherboard 1.1cm from the aluminium plate - it took the bottom box to 6.6cm, but there's more circulation and temperatures are 31-40 degrees idle and max 79 on benchmark.

After a few hours under heavy load it looks like temps got a few degrees higher - hot air from the CPU and GPU is flowing back to the fan? I ordered an extra fan, I'll install it in the front to separate the hot GPU and CPU air, and create a bit more flow. Really happy with the setup, though

 

[ruleh]

Something that worked for me was reversing the gpu and cpu fan direction so they exhaust all the hot air out of the case.
It might be worth a try.

 

[timginter]

Something that worked for me was reversing the gpu and cpu fan direction so they exhaust all the hot air out of the case.
It might be worth a try.

Are your temperatures ok with the fans sucking air instead of blowing on the radiators? I've read cooling is a bit worse that way.

I tried reversing my Noctua CPU fan to test it myself, bit the screws only fit one way :/

 

[ruleh]

With open air I got worse idle temperatures but mostly same load temperatures. (Somehow I expected the exact opposite.)
With everything closed up:
cpu temp went down from 75 to 67
gpu temp went down from 90 to 70
case air temp went down from 60 to 45

The gpu exhausts to the bottom and the cpu to the top.
Tests were done on a carpet without any extra spacers or standoffs.
The gpu temperatures are skewed because I cut an extra vent between the swap. Also the gpu load wasn't 100% but I don't remember the exact load.

For the noctua I used some 16mm long screws I think.
The original screws are too short and the extra screws are too long. The gpu fans are held in place with cable ties.

 

[timginter]

With open air I got worse idle temperatures but mostly same load temperatures. (Somehow I expected the exact opposite.)
With everything closed up:
cpu temp went down from 75 to 67
gpu temp went down from 90 to 70
case air temp went down from 60 to 45

Wow, nice. That's a big difference, really interesting.

I noticed a similar thing with open air - initial idle temps are very low, load temps are similar, but idle after load are higher. I wonder if on higher fan speeds air gets circulated better, but when going to idle the fans don't push the old hot air away and keep recirculating it.

I have a 60mm and a 40mm fan coming soon - I'll try testing different setups

 

[timginter]

Extra fans came, they fit quite well to push the air around. Idle temps are pretty much the same, but 9 degrees cooler at load:

Spoiler: extra fans

GPU has great airflow, there are no higher parts to block the flow so extra fans are focused on mobo. I've read more on case designs and cooling. My last PC was a beige beast with a floppy drive... quite a bit changed since then I see how a closed case with properly directed air would work better than an open one with no air direction, really interesting.

A little surprise from ASRock - drove me crazy why sometimes WiFi wouldn't work - connected but no pages loaded, Windows would refresh the connection and no internet or "Default gateway is not available". Finally found the little shite - when the antenna in the bottom slot was folded, WiFi stopped working. When I unfolded it (any direction) - all worked perfectly. WTF!?

Spoiler: 1st - no connection, 2nd - all working ok

I tested both antennae in both slots together and separately - didn't matter which antenna was in the bottom slot and which direction it was turned - as soon as I folded the antenna in the bottom slot even a little bit, WiFi stopped working... seriously, WTF!?

 

[timginter]

With open air I got worse idle temperatures but mostly same load temperatures. (Somehow I expected the exact opposite.)
With everything closed up:
cpu temp went down from 75 to 67
gpu temp went down from 90 to 70
case air temp went down from 60 to 45

The gpu exhausts to the bottom and the cpu to the top.
Tests were done on a carpet without any extra spacers or standoffs.
The gpu temperatures are skewed because I cut an extra vent between the swap. Also the gpu load wasn't 100% but I don't remember the exact load.

For the noctua I used some 16mm long screws I think.
The original screws are too short and the extra screws are too long. The gpu fans are held in place with cable ties.

Can you post photos of the closed rig?

 

[Valantar]

I tested both antennae in both slots together and separately - didn't matter which antenna was in the bottom slot and which direction it was turned - as soon as I folded the antenna in the bottom slot even a little bit, WiFi stopped working... seriously, WTF!?

If you take off the antennas and look at the inside of the connector, does the central socket move when you bend the antenna? AFAIK these antennas are just a wire on the inside, so it stands to reason that bending it will stretch the wire, which in turn might pull on the connector. If that's enough to pull it clear of the connecting pin, then you'd lose signal. Similarly, take a look at the sockets on the WiFi adapter - do the central pins protrude all the way to the end of the threaded part, or is it recessed into the connector? It might be a case of either the antenna or the WiFi adapter socket (or both) having slightly shorter than average internal terminals and thus failing to connect when the antenna is bent.

Great to hear that your thermals have improved with some added airflow btw

 

[ruleh]

Can you post photos of the closed rig?

Here are some...

Spoiler

This thing is heavy.

 

[timginter]

If you take off the antennas and look at the inside of the connector, does the central socket move when you bend the antenna? AFAIK these antennas are just a wire on the inside, so it stands to reason that bending it will stretch the wire, which in turn might pull on the connector. If that's enough to pull it clear of the connecting pin, then you'd lose signal. Similarly, take a look at the sockets on the WiFi adapter - do the central pins protrude all the way to the end of the threaded part, or is it recessed into the connector? It might be a case of either the antenna or the WiFi adapter socket (or both) having slightly shorter than average internal terminals and thus failing to connect when the antenna is bent.

Thanks, I'll have to have a look, but it's an RMA anyway. I tested the top slot and it doesn't work at all - checked both antennae in it - no difference in signal to when there are none. Guess I got unlucky there.

Great to hear that your thermals have improved with some added airflow btw

Thanks, appreciate your earlier messages, too - I didn't expect lack of directed airflow would make such a difference. Cooling and airflow info/solutions really improved in the last 20 years

Here are some...

Spoiler

This thing is heavy.

Wow, thanks. It looks really good! Sounds like the hardware mounting and airflow work good, too.
Great keyboard! I didn't know they made it with USB! Absolutely loved the keys and trackpoint on my X230.

Weight will not be pretty, but it's a whole PC wherever you want it Mine is 4.2kg without the battery, but with the shape and folding screen it's actually easy to move. Great performance and freedom in a small package

 

[Valantar]

Thanks, I'll have to have a look, but it's an RMA anyway. I tested the top slot and it doesn't work at all - checked both antennae in it - no difference in signal to when there are none. Guess I got unlucky there.


Thanks, appreciate your earlier messages, too - I didn't expect lack of directed airflow would make such a difference. Cooling and airflow info/solutions really improved in the last 20 years


Wow, thanks. It looks really good! Sounds like the hardware mounting and airflow work good, too.
Great keyboard! I didn't know they made it with USB! Absolutely loved the keys and trackpoint on my X230.

Weight will not be pretty, but it's a whole PC wherever you want it Mine is 4.2kg without the battery, but with the shape and folding screen it's actually easy to move. Great performance and freedom in a small package

Heh, I have that same keyboard - it's fantastic. AFAIK it's the same keyboard as they used on the early-ish T4XX laptops - T420 and so on. Mine has lasted for quite a few years without visible wear and quite heavy use, though I have sadly damaged the USB socket (used it with my HTPC for a while and tripped over the wire). If you're interested, it's called the ThinkPad Compact Keyboard with Trackpoint (there's also a ThinkPad Compact Bluetooth Keyboard with Trackpoint and a newer, updated ThinkPad Trackpoint Keyboard II with BT and USB-C charging, but AFAIK neither of the BT variants support USB connectivity).

I would actually think this would be an excellent candidate for DIY laptop builds, as it disassembles relatively easily and has a simple and mostly self-contained construction - the front bezel clips off around the outer edge, allowing the glued-in keyboard+trackpoint combo to be prized out (carefully, apply some heat if necessary) alongside the tiny controller PCB with its micro USB socket. The double sided tape already on the back of the aluminium base plate should let you just stick it onto your deck, though you'd need to cut a hole to clear the back of the trackpoint plus some slots for the two ribbon cables to pass through.

That's too bad about the WiFi, though at least you spotted it early. Hopefully the RMA goes through smoothly!

 

[SiKiaTriK]

Being vastly overpassed by the electrical knowledge of the people here I'll just reffer to my RC experience so... regarding the DC feed, have you considered a 4S LiFePo source? the nominal voltage sits around 12.8v, being more confortable against the 11.1v of a 3S LiPo. They're a bit more flexible about charge/discharge tolerances and within similar price could be an interesting option.

 

[timginter]

Heh, I have that same keyboard - it's fantastic. AFAIK it's the same keyboard as they used on the early-ish T4XX laptops - T420 and so on. Mine has lasted for quite a few years without visible wear and quite heavy use, though I have sadly damaged the USB socket (used it with my HTPC for a while and tripped over the wire). If you're interested, it's called the ThinkPad Compact Keyboard with Trackpoint (there's also a ThinkPad Compact Bluetooth Keyboard with Trackpoint and a newer, updated ThinkPad Trackpoint Keyboard II with BT and USB-C charging, but AFAIK neither of the BT variants support USB connectivity).

I would actually think this would be an excellent candidate for DIY laptop builds, as it disassembles relatively easily and has a simple and mostly self-contained construction - the front bezel clips off around the outer edge, allowing the glued-in keyboard+trackpoint combo to be prized out (carefully, apply some heat if necessary) alongside the tiny controller PCB with its micro USB socket. The double sided tape already on the back of the aluminium base plate should let you just stick it onto your deck, though you'd need to cut a hole to clear the back of the trackpoint plus some slots for the two ribbon cables to pass through.

Thanks for the info! I found the wired version earlier but didn't have a clue they made a bluetooth one, those keyboards were absolutely great. They're a bit pricey - I'll stick to the Microsoft wireless keyboard for the prototype, but the ThinkPad bluetooth one would be great for the official build.

That's too bad about the WiFi, though at least you spotted it early. Hopefully the RMA goes through smoothly!

Thanks. ASRock's RMA sent me to the retailer, luckily it was from Amazon - a week after return ended, but Amazon's customer service is great - sorted the return quickly over the online chat.

Being vastly overpassed by the electrical knowledge of the people here I'll just reffer to my RC experience so... regarding the DC feed, have you considered a 4S LiFePo source? the nominal voltage sits around 12.8v, being more confortable against the 11.1v of a 3S LiPo. They're a bit more flexible about charge/discharge tolerances and within similar price could be an interesting option.

Thanks. I had a look at LiFePo4 earlier when I was planning 5/6S - I've read they are safer, but the choice was very limited.
Found a sensibly priced 4S 8400mAh, but the voltage range is too high - I'd have to regulate voltage to always stay below 12.6V stepping up/down at high currents is the biggest problem.
Funny you mentioned it, though - BMS finally came... and they sent a 4S 40A LiFePo4 instead of 3S 60A LiPo I started a return and hopefully they'll send the right one soon.

If 3S LiPo without voltage regulation will work relatively well, it will be a perfect solution - no conversion and the UPS will be much simpler. I'll probably need a MOSFET diode between the relay and decoupling capacitor, though, to prevent stored battery current flowing back into the power supply.

 

[timginter]

I came across Noctua NH-L12S - that cooler design should work great with open cases, blowing hot air away from the mobo, but it's a bit too high. After some digging I found ID Cooling IS-47K and Alpenföhn Black Ridge - apparently the B550M-ITX/ac mobo is fully compatible with the Black Ridge! They may need LP or even VLP memory, but that's ticked off in my build, too.

Black Ridge is available in UK at a sensible price - I'll see how it works, reviews are really good. I'm hoping I'll be able to rest the top of the radiator against the aluminium plate and fit a noctua fan "underneath" blowing the air out.

 

[SiKiaTriK]

I came across Noctua NH-L12S - that cooler design should work great with open cases, blowing hot air away from the mobo, but it's a bit too high. After some digging I found ID Cooling IS-47K and Alpenföhn Black Ridge - apparently the B550M-ITX/ac mobo is fully compatible with the Black Ridge! They may need LP or even VLP memory, but that's ticked off in my build, too.

Black Ridge is available in UK at a sensible price - I'll see how it works, reviews are really good. I'm hoping I'll be able to rest the top of the radiator against the aluminium plate and fit a noctua fan "underneath" blowing the air out.

With the BR you'll need just LP ram sticks, like the Corsair Vengeance LPX. Then you can use a 120x15mm noctua on top, ending with a total height of ~62mm (vs the 70mm of the noctua). On my setup, being inside the DAN-A4 I had to stay with the 92mm fan under the fins, but a 120mm fan on top would perform like a charm, efectively cooling the ram and other PCB areas, wich is always advisable. Same goes with the IS-47K, wich is an almost identic copy to the BR (with a better compatibility towards big heatsinks on the PWM area)

 

[timginter]

With the BR you'l need just LP ram sticks, like the Corsair Vengeance LPX. Then you can use a 120x15mm noctua on top, ending with a total height of ~62mm (vs the 70mm of the noctua). On my setup, being inside the DAN-A4 I had to stay with the 92mm fam under the fins, but a 120mm fan on top would perform like a charm, efectively cooling the ram and other PCB areas, wich is always advisable. Same goes with the IS-47K, wich is an almost identic copy to the BR (with a better compatibility towards big heatsinks on the PWM area)

thanks, good to read about first-hand experience. the IS-47K is difficult to get in UK, found them only on aliexpress. The BR looks great with an option for a big fan on top!

it's been a rough ride with the B550M-ITX/ac... it's a great little mobo, but looks like ASRock is still driving it out of beta. Looks like the BIOS had some problems with temperature sensors - very erratic, often readings jumping really high and suddenly change by 10 degrees even though I had fans on full. Thought it was my cooling or lack of air circulation in an open case, but after updating BIOS to 1.6 temperature reading stabilised and show way lower with fans on silent...

Also, the issue I had with the WiFi (only 1 antenna port working and WiFi cut off when I folded the antenna in the working port) seems to have been caused by drivers - just finished a few day marathon of debugging.
I updated all drivers and BIOS around the same time so wasn't sure which one was causing the folding antenna problem. Even if I uninstalled all WiFi drivers and installed stock/older ones (even ones from ASRock website) the issue was still there. Just did a clean Windows install with BIOS v1.6 and issues went away - FINALLY stuff works as it should.

I'm still ordering Black Ridge, though - with the Noctua blowing air towards the mobo, everything on the backplate is getting a bit warm after heavy use (to the point that USB dongles are warm to the touch when I unplug them). With the Black Ridge I'm hoping the aluminium plate touching the heatsink will give extra cooling and I'll still be able to have a fan blowing away from the mobo. If I can return the Noctua cooler I'll just have to order an extra A9x14 fan - BR doesn't come with one.
GPU is screwed directly on the top plate getting extra cooling from there, CPU will get extra cooling from the bottom plate - fingers crossed.
That's assuming the aluminium plate won't get too hot - otherwise I'll have to separate them since I often use the new rig on my lap on the couch.

Can't believe some rubbish drivers were such a pain...

 

[Valantar]

thanks, good to read about first-hand experience. the IS-47K is difficult to get in UK, found them only on aliexpress. The BR looks great with an option for a big fan on top!

it's been a rough ride with the B550M-ITX/ac... it's a great little mobo, but looks like ASRock is still driving it out of beta. Looks like the BIOS had some problems with temperature sensors - very erratic, often readings jumping really high and suddenly change by 10 degrees even though I had fans on full. Thought it was my cooling or lack of air circulation in an open case, but after updating BIOS to 1.6 temperature reading stabilised and show way lower with fans on silent...

Also, the issue I had with the WiFi (only 1 antenna port working and WiFi cut off when I folded the antenna in the working port) seems to have been caused by drivers - just finished a few day marathon of debugging.
I updated all drivers and BIOS around the same time so wasn't sure which one was causing the folding antenna problem. Even if I uninstalled all WiFi drivers and installed stock/older ones (even ones from ASRock website) the issue was still there. Just did a clean Windows install with BIOS v1.6 and issues went away - FINALLY stuff works as it should.

I'm still ordering Black Ridge, though - with the Noctua blowing air towards the mobo, everything on the backplate is getting a bit warm after heavy use (to the point that USB dongles are warm to the touch when I unplug them). With the Black Ridge I'm hoping the aluminium plate touching the heatsink will give extra cooling and I'll still be able to have a fan blowing away from the mobo. If I can return the Noctua cooler I'll just have to order an extra A9x14 fan - BR doesn't come with one.
GPU is screwed directly on the top plate getting extra cooling from there, CPU will get extra cooling from the bottom plate - fingers crossed.
That's assuming the aluminium plate won't get too hot - otherwise I'll have to separate them since I often use the new rig on my lap on the couch.

Can't believe some rubbish drivers were such a pain...

Weird that you've had so many issues - I've got that board in my HTPC, and it works perfectly. No weird thermal readings (I've got the system's only fan set to switch off below 55°C for fully passive cooling, and it tends to hover below that pretty steadily with desktop and video usage - spikes would set the fan going, so I'd notice), and while I'm using Ethernet on it, I've tested the WiFi and found it ... okay. Nothing spectacular, but then the spot it's in is a worst-case scenario for router coverage.