Gadgetory

there are a lot of challenges with accurately testing the thermal output of cases coolers GPS and CPUs and getting a measurement that from product to product is actually representative of the real world performance without some kind of variance within the results that may actually invalidate some of the differences shown in charts and I am now confident in saying that I think we have one of the most accurate and best testing methodologies for thermals on the media side of the industry and the reason I am able to say it with such excitement and confidence is because we just validated our own testing methodology in a real thermal chamber several thousand dollars this thing we got access to through a friend and we're able to use this to see if our results are actually accurate within a 100-percent controlled environment so this is something that was pretty unique to us it was a new experience with validating our thermal testing methodology and that's equal parts nerve-wracking and exciting because obviously on the one end you could find out that Oh everything we've been doing is questionable but we were very happy to see that the results from the thermal chamber and our everyday lab setup are actually almost directly identical with the cpu load to load tests and the idle to idle tests and GPU tests they are right on the dot in terms of accuracy between each environment so I want to explain how this all works because it was pretty cool and I think that the enthusiasts in the crowd might enjoy this type of thing so first of all if you are not familiar with thermal chambers a thermal chamber is they vary in size they vary in climate control abilities some of them have humidity controls that you can directly access some of them are just temperature controls and the whole point of a thermal chamber is to contain your test environment and create constants you know exactly where the intake is know exactly where the exhaust is in the system they're often very low powered fans so there's really no reasonable impact on test results if you have a good one and for setups like the one we use there's actually a lot of extra thermocouples built into the chamber so we can position those all over the place and figure out exactly what is the temperature of the intake in front of the intake fan what's the temperature behind the intake fan what's the sure of the exhaust what's the temperature of the CPU if you place one of these things between the cold plate and the IHS all this stuff were able to collect data for and validate our testing methodology now having years of experience testing cases I can vary directly speak to the level of variance that you can see an environment if you're not well trained and you don't have a solid bulletproof methodology for making sure that the tests are accurate from day to day even or from case to case or from season to season there's a lot of stuff that changes even in a controlled house and these are only loosely controlled any kind of office or house that's on an open-loop HVAC system or heating and air conditioning system is going to have variants in the test results and that needs to be accounted for by producing Delta values we use things like this which I'll explain in a moment to make sure our accurate readings are achievable by basically measuring ambient constantly and checking that against whatever the diode temperatures are and I'll explain that more momentarily so a couple of concerns and challenges with the test engineering of methodology in a non chamber environment like this where we do our testing is you've got to control where your cases or device it has to be the same freaking spot every time and I know it sounds sort of insane that you needed in the exact same spot from product to product but it's really it's true because if you do move the thin the ambient over there might be cooler or warmer than the ambient over there and that might change based on the season this is the wall that's external so it's near a window and if I put the case over there and I've tested this with our internal testing the result might be that there's actually a slight few degree even 12 to warmer temperature when it's the summer when the Sun is hitting the wall or when it's winter or it might just be different because there's a vent in the ceiling over there and if you don't position the case correctly with the intake fan in the exact same spot every time that intake or whatever it is heating or air coming from the ceiling will impact results that's not good is not a subjunctive thing it will impact the results and I've tested that several times at least in this environment so these are all very big concerns the way to deal with it that we've done is by taking something like this which is a very accurate basically digital thermometer that has thermocouples linked out of it can fit a few of them in here and we measure every second the ambient temperature in specific spots like in front of the intake fan of the case will occasionally do measurements of things like the exhaust the output of the case what does it measure immediately outside of the rear exhaust fan what does it measure within the case between the intake and the GPU and you can do all these different measurements and get pretty accurate or actually well really within point one Celsius resolution results of the temperature of wherever that particular thermocouple is and we use k-type thermocouple so they're they're rated at these the ones we use are rated for 800 degrees Celsius so that helps make sure that these environments remain accurate so what you do or what we do i should say is we take a reading of the ambient we take a reading of the diode using some software normally 80 64 and we subtract the ambient from the diode to create a delta that delta value shows you the difference that change from case to case instead of using an absolute temperature if it is 21 degrees celsius in here today and it's 20.5 celsius in here tomorrow for whatever reason maybe the AC turned on a few minutes into the test or something you need to account for that and the only way to account for that is with something like this now for testers and sites or youtube channels that just check their thermostats you the thermostat says that's not good enough and that believe me that is nowhere near good enough first of all the thermostats probe might be in the hallway where the thermostat is that's not the temperature of this room that is actually several degrees different from this room and our other lab or main lab where all the production and video rendering happens that's got hot computers in it's got other electronics in it that room measures that I don't know the Celsius reading is measures at about 84 degrees Fahrenheit depending on how much work is going on here it'll measure it may be 68 max 70 Fahrenheit and in the hallway it might be 67 so this is all stuff that you can't just take a reading from a thermostat and think it's going to be accurate for a delta value in a case test and that's why we introduce these different thermal couples and stuff we've worked on with silverstone we worked on with course share with NZXT all these companies we've talked to them for years now to say hey what are the big problems that we should look for in case testing what do you guys have trouble with internally what do you have trouble with externally with media outlets that you think maybe aren't doing it 100% correctly or are doing it correctly and how do we improve our testing so we did all this we talked a lot sent back and forth ideas and came up with the methodologies today which I'm not going to reveal Foley on camera but i'll give you the overview of things to sort of share insight as to why our tests are accurate so first of all what were the actual results within the thermal chamber that we tested in which again I think they're ten thousand dollars when we were using which is really not bad for a thermal chamber but certainly not something that I can just throw down and put up in the corner of a room somewhere so when we used again was accessed through a contact of mine several thousand dollars very closely controlled and the results for that I'm showing you a chart right now it should be a chart on the screen the results for that are almost identical so for the stock test on the cob like halo 5 silverstone case we saw a difference of about point zero seven degrees Celsius from the lab test here and the test in the thermal chamber that's a very good thing that means that after we subtracted the ambient temperature for this lab and for the thermal chamber we saw and a readout that had a resolution of a point one Celsius reading margin of error with the devices so I read out I had a point zero seven Celsius difference between the values so I basically the same for all intents and purposes now with our next test with the bottom intake we saw one I think it was point 0 46 Celsius again so close it's just it's it's really exciting how close the results were and then the biggest difference we saw was 0 point 2 25 that's a bit larger than the other two but still pretty damn small difference between the two and I did talk with the reason why that happened we had some access to other tools we use thermal couples in different cases or a situation spaces within the thermal chamber to figure out why that happened just slightly different from the other two and it was because the I had installed a rear exhaust fan didn't have one stock and install the rear exhaust fan that was spitting heat into the wall which was a couple inches away three inches away so spitting heat into the thermal chamber wall and then that heat was recirculating back into the bottom getting sucked in below the GPU and that caused a very very slight increase in temperature from one test to the next so that's sort of how we validated everything at a top level and there's a lot more concerns too so when designing these test cases we look at everything with how close is the wall to the case is the exhaust facing the wall or is it facing away where are the vents in the ceiling on the floor where is the case positioned exactly in the room what is the surface of that the case is sitting on what's the material does the table run cooler or warmer than ambient on average depend on what kind of material it is should almost always be cooler if you're in the winter anyway that's what we've experienced and in the summer you may experience warm it just depends on sort of where the windows are basically there's a million things that can change within a non-controlled thermal chamber environment and the way we account for those so like I said this sits there measuring ambient every second and then we have the diodes measuring am you at every second we take this data remove it from that and then we have a delta value which gives us accurate results based on this thermal chamber test so the whole point of this we use a couple different tools we use one of these as well this is a laser you pointed at different surfaces you get a thermal readout I can see a table in front of me right now is 17.1 degrees Celsius and using this device if I had a thermocouple plugged in which i'll go ahead and do right now we will see that the ambient temperature of this room is 19.3 degrees Celsius where I'm standing right here and that's 17.1 on the table so you can see just how much the temperature can change from object object from location in the room one to the other it's right now we're at let's see here we are at 18.7 now that's how moment to calibrate I put it in front of this light here and now we are at nineteen point two and that's an LED light so that's pretty cool but you can just see how much things can change and if you're testing cases these are products that often maybe have a degree of difference between them if that so obviously you've got some margin of error built in and that will help you figure out is there a significant difference between two results one case the next if they're zero point maybe zero five zero point zero seven degrees apart then we know that it's actually almost within margin of error where they could effectively be the same and in terms of practical use they are the same because your CPU and your GPU don't care that much about one degree but within the world of cool and it's also a unique challenge for these manufacturers because cooling is the type of thing with computers anyway we're ASA texts and cool it's and all these a politics all these companies of the world they are trying to compete with one another for maybe one degree at a time it's a big deal if you can drop your temperatures by one degree Celsius so that's why this accurate testing is useful because even though it's practically not that different we can still see what is the absolute best product in each category within a slight margin of error built in there what is the best product we can rank them in a hierarchy and then we can talk about it in the article and in the videos hey how does this impact you practically and sometimes there's a pretty damn big difference so if you have one case it might be we test the revolt to leave all to was what was it I think it was almost 30 degrees Celsius I want to say warmer on the 980 TI hybrid GPU than when we tested it on an open bench so that is a giant difference and that just shows why the accurate testing is important for finding those types of things now other tools we use we do use multimeters as well this is a bit more involved then probably a lot of sites think is necessary but I do it just as a precaution so this is especially something we do for testing fan controllers and that's because fan controllers if they're cheap they don't output a constant voltage they might fluctuate even from something like 11 12 volts the high speed setting so that's a pretty fair fluctuation i will drop your rpm a good bit so we use multimeters to sit there and read out with probes during testing or after testing what I ever read out what is the voltage throughput what's the the current and all that stuff and that's something that we use to help validate as well we use this to check the temperature in a couple different spots do spot checks and then we use this for ambient put it on a spreadsheet do a bit of work there crunch the numbers and count with the charts that you all have become familiar with so that was the adventure of testing all this stuff in a thermal chamber some of the photos should have gone through during this time if you have any questions let me know in the comments below I'm pretty pretty happy to see that the testing methodology is accurate that its sound and next we are working on DB testing some noise testing I've said this many times actually in every article about cases lately and it's for good reason so we don't currently test noise and that's because it's really hard to test knows you need to know some math about how the decibel levels work need to know some math about the deltas and it's logarithmic it's not just a simple subtraction formula if you add more fans it's a logarithmic change in the noise output so that's my next goal is to start adding DB tests and we're going to buy some expensive equipment after we talked to a couple of manufacturers figure out what's what's good for our purposes I'm not going to go build a half-a-million-dollar anechoic chamber that's just absurd frankly I just want to do the best test we possibly can within a realistic real world everyday environment and hopefully over the next few months figure out a methodology that is so accurate and reproducible that we know definitively from case to case GPGPU which one's quieter and even if that's in a somewhat relative setting so I'm not building any anechoic chambers I do have access to them and we're going to be validating our methodology and those as well when the time comes so i will be able to use those zero DB noise floor anechoic chambers and figure out if we set up a case in here how loud is it and no absolutely how loud it is and then set it up in a lab environment in our lab this one and figure out okay what's our noise floor how loud is the case here and how do we compute that elta to figure out one to the next what's louder especially because things like noise can change on a day-to-day basis see there's all kamillion more concerns there does your air conditioning turn on that's a noise increase is there a train going by is there a plane there's a car on the street driving by all kinds of small things can impact noise testing it is a nightmare to get into and do accurately but now that we've conquered thermal at least the Psalms agree i'm not going to say we're the best at thermal dynamics there is because it's not really what we do we're not thermal engineers just test engineers so I am pretty confident that we've basically conquered that problem in terms of measuring thermal i will say that but DB is next and that's going to be a very big challenge but i'm excited to tackle it and for anyone who wants to help us all this stuff it is expensive to do so I you know there's sponsors that definitely helps and then we've got patreon link to the postural video or in the description below if you want to help us out directly but other than that just watch the content if you like it then great please comment stuff like that all the usual youtube BS people spew you know like comment subscribe what it actually does help but thank you for watching the discussion on how this stuff works and behind the scenes of what we're doing to make sure our tests are accurate and not just I frankly excuse but not just that people post up with some kind of half-assed you know check the thermostat and subtract that or even worse just post absolute values where you might have a temperature fluctuation I mean we've with this thing let's see what it is now it is now it's now 19.1 here and here a minute ago it was I think it went from 18.3 at 1.2 19.1 over there it's like 19.2 and that has increased as well so just standing here has increased the heat of the room the lights and you know I the point is that there's got to be some kind of science behind how you account for that and that's what we do and that's what we do pretty well so thank you for watching I will hopefully see you in the next video where we talk about some more product stuff and news and yeah I'll see you all next time thanks