No Dollar Wasted PC Build Guide

[mariushm]

Nice video, enjoyed it.

 

The case was a pain in the ass though... I wish you would have chosen a 5-10$ more expensive case ... considering you paid so much for a RTX 2070, it's kinda ridiculous to go cheapest case route anyway.

 

The video card was really close to the bottom of the case, so probably got hotter and noisier compared to other cases.

 

Some other places where you could saved money... just go with a single 1 TB SSD at 100$ a piece and you'd probably save 20-30$ which could be used on better case.

User can always add mechanical drive later for even more storage.

 

 

 

[Juventas]

I registered just to say that this is the best no-nonsense build guide I've seen, but there was one significant mistake that wasn't mentioned.

 

At 18:20 it appears the IO shield's ground tab for the HDMI port is inside the jack, instead of on top.  Even though that jack won't be used, it is worth noting, and if the tab is long enough to reach any of the conductors, it could cause problems.

 

[rulerruler]

Great video, I enjoyed both the content and the presentation!

It would have been nice to have a short overview of the components and the prices you bought them at in the beginning though.

[seon123]

I would be interested to see how much the higher temperatures from the low end case would affect the performance. We all know how sensitive GPU Boost 3.0 is to temperatures, and Ryzen too. Let's say the performance drops 3% compared to a case that's $20 more expensive. That would mean 3% reduced performance for a 2% cost saving (~$1000 build).

 

For the PSU. Going by warranty and 80+ rating is a terrible way to pick the PSU. How do you think it would it go, if you picked the CPU by the warranty and TDP? Instead, you could go to an in depth review, and see that it has proper protections, that it's primary is LLC resonant (instead of half bridge, which is more likely to have coil whine, and also performs worse), that the regulation and ripple is fine, that it uses a rifle bearing fan, and that all of its rails are independently regulated. That is how you choose a PSU. By reading reviews. 

 

With all of that said, the build is quite decent, and a lot better than pretty much all the rest of tech YouTube. Kudos to you, good job. 

 

On a side note: the 2070 actually makes sense over the 2060. From the current prices on PCPP, it's a $100 price difference between the two GPUs. From Gamersnexus' review of the 2060, the 2070 performs 11,9% better on average. So the 2070 is better value than the 2060 if the build (including the 2070) costs $844 or more. 

 

Another side note: doing the same for the Vega 64, ($70 price difference, 6,14% performance difference) the 2070 build has to cost $1139 or more. So for anyone buying the PC today, the Vega 64 is objectively the better value option.

[Emily Young]

19 minutes ago, seon123 said:

For the PSU. Going by warranty and 80+ rating is a terrible way to pick the PSU. How do you think it would it go, if you picked the CPU by the warranty and TDP? Instead, you could go to an in depth review, and see that it has proper protections, that it's primary is LLC resonant (instead of half bridge, which is more likely to have coil whine, and also performs worse), that the regulation and ripple is fine, that it uses a rifle bearing fan, and that all of its rails are independently regulated. That is how you choose a PSU. By reading reviews. 

Given that the ATX spec defines +/- 5%, ripple only comes into play when it's REALLY out of spec, which is unlikely for a unit within the build guide's guidelines. Obviously it's best to reduce this as much as possible, but most people won't care because it doesn't really do anything tangible other than potentially give them more overclocking stability.

 

At the end of the day, a reputable company, decent warranty, minimum rated efficiency, and enough power on the rails generally are going to be enough for most mid-tier builds, and rail power most likely won't even come into play these days anyway. I'd recommend reviews to mostly just confirm that the choice has a low chance of being a dud, but that's true of every product.

 

Things obviously change when you have more, higher draw components in your build, but at this level I would argue that it's really not a concern.

[Cohacq]

I have a DVD burner in my case! But it hasnt been plugged in in years.

[Tchaikovskonovich]

Any particular reason to choose the 2 TB HDD over the 3 TB HHD? The price change isn't significant, and it still runs at 7200 RMP. I know 2.25 TB isn't going to be enough for me since I use almost 3 TB already. (Sorry if this question is kinda dumb, I'm still pretty new to this sort of thing)

[seon123]

20 minutes ago, GabenJr said:

Given that the ATX spec defines +/- 5%, ripple only comes into play when it's REALLY out of spec, which is unlikely for a unit within the build guide's guidelines. Obviously it's best to reduce this as much as possible, but most people won't care because it doesn't really do anything tangible other than potentially give them more overclocking stability.

 

At the end of the day, a reputable company, decent warranty, minimum rated efficiency, and enough power on the rails generally are going to be enough for most mid-tier builds, and rail power most likely won't even come into play these days anyway. I'd recommend reviews to mostly just confirm that the choice has a low chance of being a dud, but that's true of every product.

 

Things obviously change when you have more, higher draw components in your build, but at this level I would argue that it's really not a concern.

First of all, 5% is for the regulation. Except for on the -5V, where it's 10% (iirc). The ripple is max 120mV for the 12V rail and 50mV for the minor rails. While it does not necessarily matter for the performance, it can affect the longevity of the components. Granted, that's hard to measure the impact, unless it's really out of spec to the point where the whole PC dies within a year. 

 

That's why I mentioned all of the other points. E.g. the Seasonic M12 II would seem like a good option, with its 5 year warranty, reputable brand and 80+ Bronze rating. However, it completely lacks OCP on the minor rails, it doesn't have OTP or UVP, it's likely to have coil whine with higher end GPUs (including the 2070), it's loud compared to other options, and because it's group regulated, it will have issues with some of the new Intel low power standby states, and the voltages will go out of spec during unbalanced loads. 

 

With a 550W PSU, you can power quite high end components, and you can reuse the PSU for future builds. With a decent one, you won't have issues with Vega. I would have no issues with someone doing that with the CX550, but I would have issues with someone doing that with an M12 II 620W. 

[KEIN NEIN]

5 minutes ago, seon123 said:

First of all, 5% is for the regulation. Except for on the -5V, where it's 10% (iirc). The ripple is max 120mV for the 12V rail and 50mV for the minor rails. While it does not necessarily matter for the performance, it can affect the longevity of the components. Granted, that's hard to measure the impact, unless it's really out of spec to the point where the whole PC dies within a year. 

 

That's why I mentioned all of the other points. E.g. the Seasonic M12 II would seem like a good option, with its 5 year warranty, reputable brand and 80+ Bronze rating. However, it completely lacks OCP on the minor rails, it doesn't have OTP or UVP, it's likely to have coil whine with higher end GPUs (including the 2070), it's loud compared to other options, and because it's group regulated, it will have issues with some of the new Intel low power standby states, and the voltages will go out of spec during unbalanced loads. 

 

With a 550W PSU, you can power quite high end components, and you can reuse the PSU for future builds. With a decent one, you won't have issues with Vega. I would have no issues with someone doing that with the CX550, but I would have issues with someone doing that with an M12 II 620W. 

i know it's not related to the initial topic, but do you know any electronics forums?

[mariushm]

1 minute ago, seon123 said:

First of all, 5% is for the regulation. Except for on the -5V, where it's 10% (iirc). The ripple is max 120mV for the 12V rail and 50mV for the minor rails. While it does not necessarily matter for the performance, it can affect the longevity of the components. Granted, that's hard to measure the impact, unless it's really out of spec to the point where the whole PC dies within a year. 

 

That's why I mentioned all of the other points. E.g. the Seasonic M12 II would seem like a good option, with its 5 year warranty, reputable brand and 80+ Bronze rating. However, it completely lacks OCP on the minor rails, it doesn't have OTP or UVP, it's likely to have coil whine with higher end GPUs (including the 2070), it's loud compared to other options, and because it's group regulated, it will have issues with some of the new Intel low power standby states, and the voltages will go out of spec during unbalanced loads.

-5v is deprecated (no longer required on power supplies, wire is missing in connector)

You probably meant -12v, which in modern computers is only used for serial ports so most likely computer won't complain if voltage is missing. It's also not that critical, since majority of devices working with serial ports will consider anything above around -8v (or even -5v) as a digital "1" and any positive voltage as a digital "0" (or the other way around, doesn't really matter and I can't be bothered to look it up to be sure it's correct).

Basically, it's not uncommon to measure -10v or something like that on -12v output on power supplies because the circuit is poorly regulated especially when unloaded (nothing consuming power from that -12v)

 

As for the rest... ripple is not really a big concern longevity wise, it's more or less a problem when it comes to overclocking... and it may cause coil whine generated in inductors (you hear whining sounds from the power supply or the video card or motherboard vrm)

 

As for that Seasonic M12 II ... over current protection is not really such a big deal, most people exaggerate its importance. You have power supplies drawing 250w ( 20A+) from two 8 pin pci-e connectors on same strand... where would you put the over current protection, at 400 watts? If something suddenly breaks and consumes 400w, you already have a crater in that circuit board and smoke and possibly even fire in the computer.

 

Problem with cheaper power supplies is that some are group regulated (the voltage on smaller rails like 5v could depend on the voltage on 12v or the other way around) so if the computer pulls a lot of power from one voltage, the other voltages could go slightly outside of specs.

 

For example, if you install 5-6 mechanical drives and some 5v led strips on a pc with an Athlon 200ge processor , you basically have around 40-60 watts consumed from 5v , but only around 20w on 12v ... that could make the power supply output something like 5.1v and 11.7v ... because the power supply was designed with the assumption that 12v is more used these days.

 

Older power supply designs could also have issues with computers going in deep sleep modes, where modern processors can drop to less than 1w of power consumptions ...some power supplies are designed with assumption that pc would always consume a few watts and the voltages could become unstable at such low loads.

 

@KEIN NEIN  check eevblog.com , badcaps.net forum and jonnyguru.com forums

 

[ChrisGaming]

I have a question. I saw him plugging in the strap into the outlet. Isn't that dangerous or something? I live in the U.S. and was wondering if I could too to upgrade my pc

[mariushm]

13 minutes ago, ChrisGaming said:

I have a question. I saw him plugging in the strap into the outlet. Isn't that dangerous or something? I live in the U.S. and was wondering if I could too to upgrade my pc

The US sockets have the earth/ground terminal as a hole .... he plugged into the earthing, at bottom of image.

The round bit of the strap would not go inside the narrow holes (live and neutral).

 

[KEIN NEIN]

1 hour ago, mariushm said:

-5v is deprecated (no longer required on power supplies, wire is missing in connector)

You probably meant -12v, which in modern computers is only used for serial ports so most likely computer won't complain if voltage is missing. It's also not that critical, since majority of devices working with serial ports will consider anything above around -8v (or even -5v) as a digital "1" and any positive voltage as a digital "0" (or the other way around, doesn't really matter and I can't be bothered to look it up to be sure it's correct).

Basically, it's not uncommon to measure -10v or something like that on -12v output on power supplies because the circuit is poorly regulated especially when unloaded (nothing consuming power from that -12v)

 

As for the rest... ripple is not really a big concern longevity wise, it's more or less a problem when it comes to overclocking... and it may cause coil whine generated in inductors (you hear whining sounds from the power supply or the video card or motherboard vrm)

 

As for that Seasonic M12 II ... over current protection is not really such a big deal, most people exaggerate its importance. You have power supplies drawing 250w ( 20A+) from two 8 pin pci-e connectors on same strand... where would you put the over current protection, at 400 watts? If something suddenly breaks and consumes 400w, you already have a crater in that circuit board and smoke and possibly even fire in the computer.

 

Problem with cheaper power supplies is that some are group regulated (the voltage on smaller rails like 5v could depend on the voltage on 12v or the other way around) so if the computer pulls a lot of power from one voltage, the other voltages could go slightly outside of specs.

 

For example, if you install 5-6 mechanical drives and some 5v led strips on a pc with an Athlon 200ge processor , you basically have around 40-60 watts consumed from 5v , but only around 20w on 12v ... that could make the power supply output something like 5.1v and 11.7v ... because the power supply was designed with the assumption that 12v is more used these days.

 

Older power supply designs could also have issues with computers going in deep sleep modes, where modern processors can drop to less than 1w of power consumptions ...some power supplies are designed with assumption that pc would always consume a few watts and the voltages could become unstable at such low loads.

 

@KEIN NEIN  check eevblog.com , badcaps.net forum and jonnyguru.com forums

 

Ok thanks

[emanresu2]

Do not buy any of those Gigabyte Gaming cards. Their fans literally can't spin at speeds up to around 30% (varies slightly from card to card), which is totally fine according to Gigabyte. So what they'll do in the 1-30% range is constantly starting and stopping to spin, with a clearly audible clicking noise (again, normal according to Gigabyte) and the expected sound of sudden revving.

 

I don't consider a custom fan curve a solution. You either have to run the fans at ~30% at all times (which is just noisy and throws their advertised semi passive cooling out of the window) or you set it to either run at 0% or at least 30%, which just makes it happen a bit less often.

 

I have no idea why this isn't a bigger deal and never mentioned in any tests. It has apparently been going on since the 10 series cards.

 

 

[DataStorm]

The ESD / Antistatic is useless as shown there. You as a person being directly "grounded" means that the anti-static mat below on the desk isn't and that the motherboard on top of the box for sure is NOT as well. The ESD wristband should lead to a connector on the anti-static mat and (often) from the other side of the mat to the ground THROUGH a 1 megaohm resistor. Why the resistor? Because if some is put on it with a high charge, the 1 megaohm resistor will slowly dissipate the charge to protect the equipment. Also, any charge coming from ground is also attenuated, because there is other equipment on the ground connected, you'd gain charge from it if not through a large resistor. The idea is that it's all at the SAME charge (the difference in charge is what the "static" electricity is making it flow).

 

Basically you have in this video a different charge all the time because you're wrongly grounded.

 

So in short:

[your wristband] ----> [Anti static mat] ----> [1 megaohm resistor] ----> [ground connector at socket]

 

Sometimes a floormat is also being applied for higher sensitivity.Those are connected commonly together before the resistor. If multiple workplaces have to be setup, those should all be connected together so no one can charge someone else due having different charge.

 

Also, it does happen that the ground in a socket is not connected, I've had issues in buildings where they put 20 cm cable of "ground" in the pipe, but didn't connect it through.... so that needs to be checked.

[mariushm]

@DataStorm You're not correct.

 

The ESD(static dissipative) mat is completely optional. If you just pulled the board out of the bag (which is anti-static... eh.. sort of) and you put it on the desk... it's not gonna be charged with static electricity, so it doesn't necessarily have to be on a mat, for the mat to dissipate static electricity.

The computer is a fixture, in one place, on a surface that's insulator and can not charge with static electricity randomly, like wood or cardboard or glass (or can extremely slowly, practically doesn't matter for this conversation). It will not charge itself with static electricity just by sitting in one place.

 

You're using a strap to give static electricity an easy and safe path to discharge instead of going into components. The reason you'd use one is because you may accidentally charge yourself with static electricity - could be due to dry air around you, could be due to friction between your body and clothes you wear (wool or synthetics are big culprits), could be the sole of your shoes rubbing on carpet or whatever material is on the ground.

Yes, there is typically a 1 megaohm or some other high value resistor involved, but it's typically CLOSE to you, usually INSIDE the strap, right by the metal bit that touches your skin.  The actual purpose of that resistor is two-fold... to limit the current... one, like you said to make that "zap" from you, flow at a slower pace in the mat or ground over a few milliseconds and give the mat time to "dissipate" the charge, with minimal transfer of the energy into the product sitting on the mat and two... it's also there to minimize the chance of you getting zapped by something else connected to earth (imagine some faulty device pushing 230v into the earth wire for a few ms and you are the closest to that faulty circuit and you get 230v going up into your hand, flow over your heart and go down into earth through your feet)

Read this if you want to learn more : http://www.circuitnet.com/experts/87211.html

 

And you can see in the video below at around 9:00 , that a strap is not even necessary, you can just discharge yourself periodically (or before every time you plan to touch components) by touching an earthed surface.

 

See video below which shows inside some of these devices, including those fake "wireless" esd straps.  One of them has the resistor in the cable, one has in the bit with the metal touching your hand and so on ...

 

 

 

 

[YogiBehar]

Really I understand you guys are making fun of the verge but come on. You know the antistatic band does not need to be used. I build pc's all day and every video and now tech geek ever uses one. In the old days yes but not today hell I some times I build a pc on my carpeted floor.

[DataStorm]

Ehm, @mariushm , you missed some... static discharge works both ways.

 

Touching ground can charge or discharge you with an amount. The idea of using a wristband is to be connected at all times to keep the charge DIFFERENCE as close as can be at all times.

 

You touching a grounded surface can discharge you, or charge you, doesn't matter, it's bout the difference of charge between you and the electronic component you're working on, so you charged or discharged, but the electric component will be a different charge anyhow. This is why the wristband with wire is connected to the antistatic mat. The antistatic bag does give off the charge, but it basically insulates the content from external influences. Once you grab the component.... it's charge will be equalized with you, Every time after that that you touch ground and the component again, it will have to re-equalize with your charge.

 

As for the 1 megaohm resistor between you and the mat, that will invite charges to go through your hands instead of via the wristband off to the mat. I've set up a lot of workplaces for this, generally you better use a low value resistor between you and the mat if absolutely needed, couple kiloohm's at best. Only exception is when you work with live high voltage equipment... you don't want to become the pathway to ground then...

[LienusLateTips]

6 hours ago, GabenJr said:

Not sure where Vega 64 was $420-440, everywhere I looked at the time had it for $500 (yeah, including PCPartPicker with no retailer filters). Anyway, it wasn't really coincidence that I landed on the Corsair CX, those guidelines mostly funnel you into it given its usual price point. The EVGA 500B would have also been in that range though, and isn't great. I would have commented on it had it been my choice; It won't really matter much at this performance level either way, but yes, something to be considerate of.

The 500B would do a lot worse in terms of longevity and stability... it has an old, outdated group regulated design. The grey CX is a DC-DC design, where the voltages won't go out of spec...

 

There's also a few Thermaltake units that are pretty rubbish with 5-7 year warranties and 80+ Bronze...

[Stefan Payne]

6 hours ago, GabenJr said:

Given that the ATX spec defines +/- 5%, ripple only comes into play when it's REALLY out of spec,

Nope, according to one of my contacts, who talks to Mining people, he said that everything over 60mV on 12V causes increased failures...

The spec is also pretty old and lose. What you'd expect for early-mid 1990s technology/specification..

 

 

The main problem is that even today many PSU are still violating Intel Haswell spec, wich is tested at 0A on Minor Rails and 100% on 12V and vice versa. 
Because there are still PSU that regulate multiple voltages together -> Group Regulation.

 

The Corsair CX550 you went for, does not do that. It has so called "DC-DC" Technology (Marketing Term, Technical its dual buck converter wich regulate 3,3V and 5V out of the 12V)

 

 

Quote

which is unlikely for a unit within the build guide's guidelines. Obviously it's best to reduce this as much as possible, but most people won't care because it doesn't really do anything tangible other than potentially give them more overclocking stability.

Naa, I disagree.

A not great PSU can increase the aging of the components in general. 

Its not just for OC that a good PSU makes sense...

 

Quote

At the end of the day, a reputable company, decent warranty, minimum rated efficiency, and enough power on the rails generally are going to be enough for most mid-tier builds, and rail power most likely won't even come into play these days anyway. I'd recommend reviews to mostly just confirm that the choice has a low chance of being a dud, but that's true of every product.

That's how it should be, yes.

Sadly that is not how it is...

 

There are a couple of subpar PSU still on the market, from "reputable Companies" because that's what people still buy and testing PSU isn't easy.

 

 

It starts with lack of protection. If you have a slight short, some PSU from "Reputable Companys" only have OPP for overload protection and nothing else (OK, UVP on minor rails and OVP on all). If you slightly mess that up, Huston we got a problem.

 

There is a review that shows it, sadly only in German but that shouldn't be a Problem for google translate or others.

https://www.hardwareluxx.de/index.php/artikel/hardware/netzteile/29916-enermax-triathlor-eco-650w-und-lepa-maxbron-700w-im-test.html?start=3

 

Here the last part, under the pic of the insides of the LEPA unit.

TL;DR: they brought the Voltage on 12V down to 6V at 100A. And OPP didn't switch off for whatever reason. UVP is just not there.

 

Quote

Things obviously change when you have more, higher draw components in your build, but at this level I would argue that it's really not a concern.

Yes, but the PSU is still really important as it can shorten the life of all components, especially if you don't have too much money, you should get a really good PSU because that can increase the life of your components.

 

I know that PSU are a black box that seems a bit abstract and are all too often ignored...

 

 

In the end the only mostly trustworthy companys can be counted on two fingers:

a) be quiet

b) Corsair (yeah, I have to say that).

 

Other "Reputable Brands" are not as good...

Especially som "highly regarded" brands...

[Tchaikovskonovich]

@Stefan Payne What power supply would you recommend for this build then? I was planning on building this one to replace my now 6 or 7 year old pc, and want to make sure it will last pretty long since I'm going into college and won't have a ton of money to spend on it.

[Stefan Payne]

6 minutes ago, Tchaikovskonovich said:

@Stefan Payne What power supply would you recommend for this build then?

Depends where you're from, what's available.

The one Gabe used for this Build was fine.

6 minutes ago, Tchaikovskonovich said:

I was planning on building this one to replace my now 6 or 7 year old pc, and want to make sure it will last pretty long since I'm going into college and won't have a ton of money to spend on it.

a) look into VEGA; if that's available at a good price and if that makes sense

b) get a better case. The cheapest of the Cheap Cases are pretty bad. Rather look at 10 year old used Cases than getting a new one. Cases rust, they don't age (the only thing you can use for decades in theory).

c) depending on availability and price: be quiet Pure Power 11, 400W or more, Bitfenix Formula, Corsair CX.

 

 

[LienusLateTips]

31 minutes ago, Tchaikovskonovich said:

@Stefan Payne What power supply would you recommend for this build then? I was planning on building this one to replace my now 6 or 7 year old pc, and want to make sure it will last pretty long since I'm going into college and won't have a ton of money to spend on it.

The one that he used for the build, the CX grey is fine. Pretty much anything in the PSU tier List linked in my bio better than Tier B will be fine, Tier A or better would be recommended for a higher end PC.

[Princess Luna]

10 hours ago, GabenJr said:

best bang for the buck!

We all know it can always be better someway or another but this is far the true best not just throw random parts together video of the sort out in a great while across YouTube.

 

I also believe that the V64 has it's place specially if NVENC won't be used for streaming, it represents a bit of a better value than 2070's and pricing is very competitive.

 

It was good work, gratz!

[aisle9]

@GabenJr Your voice does this soothing, rhythmic Kermit the Frog thing when you stop paying attention to the teleprompter and talk about what you're doing. Have you ever considered being a Muppet for a living?