Wow, Mad Catz TE2+ looks to have such a small input lag. Thanks for the new tests Teyah.
I also believe hori has upped there game with the hayabusa lineup, judging from what that youtube video has shown.
I wouldn’t mind someone putting us to shame with a creditable background.
Yeah you should not put 100% faith in whatever anyone says, even from your parents (seeds grow inside you).
Would be nice if we had a chart with one column with oscilloscope, another column with PS3 SF4, PS4 SF4 and etc with an absolute 0ms, but I do not see that happening soon, I just make use of what Teyah gives me, watch some videos on tests, then buy a few pcb’s/controllers and do my own test and pick my fav one.
Oh the vaping and torch community has the equipment we need, hopefully they are gamers too?
I would not mind donating, but trust is another issue.
are there enough of us interested in getting another test going? any estimate on cost? i know oscilloscopes can be pricey, but what else would we need? if its not insane maybe we can get some donations going. i certainly wouldn’t mind kicking some in.
Well need a list of boards to be tested and documented. A Procedure for testing including documentation of that boards version or revision.
Which we also should expect gaps in the database as it be hard to track down and find every single board revision.
Example for here
Mad Catz TE Round 1 Xbox 360 Rev A board
Someone photographs the board, take note of any version number printed, take notes of the board.
Like if there is a sloppy job of soldering from the factory, take note of that. the hertz of any clock crystals, condition of the caps on the board, ect.
If the Rev A and Rev B boards use different clock crystals take note of that as well, after testing we can compare our notes on the boards with test findings and maybe we can draw more conclusions or decide if more experimentation might be needed in the future.
There will also need a easy to understand explanation how to interpret the test data. Are the values absolute or relative, if known the margins of error, any estimates, and other impertinent information.
If an equation was used, lets show the actual equation so other people can plug in the values and see if they get the same results as us.
It be necessary to define any terminology used, I did not realize myself into recently and I had to do some fact checking that lag and latency are not at all the same thing.
That we all been using the terms incorrectly and the two terms are not the same.
some people argue for no real reason, they just like to always be arguing or posting nonsense.
no matter how you do these tests, teyahs test is accurate to a degree, it still shows which pcbs are slower than others… and thats the whole point. he could test it more accurately and it will still be the same results… you can shit post all you want… wont change anything.
the real argument is, does that little bit of extra latency really matter… thats up to the people using his results for making decisions on what sticks to buy.
I used the Kuro VLX and it was shit in USF4, thats my own conclusion and teyah’s test helped me feel a bit more assured about my conclusion.
I just wanted to say “Thank You” to Teyah and anyone that has helped or supply pcb for test. I understand that it is not 100% degree accuracy, but it is still 100% more testing than I have done, and give me an idea how one stick compares to another. I was on the fence between a VLX or TE2+ and his recent update was the deciding factor to just get a TE2.
Pretty much this. Other factors such as reliability should go into the pcb rating. Have a column for speed and have one for reliability or find a way to alter the rating based on an average of the two. There should be other thjngs as well like durability and ease of modifying, common ground and so on. Dock points for any of those being shit tier.
The point is its not just the values could be off that our point. This is not like were measuring with a ruler, the edge of that rule fell off so everything a millimeter off. We are basically claiming that their was never a ruler used in the first place. Every value is arbitrarily chosen from stick vs stick match ups were not every stick tested with every other stick. You get X number of test of Brand A stick with Brand B stick, but Brand A Stick never matched up with Brand C, D and so on. Brand B compared to Brand C and a guess is made based on that how Brand A would match up. Brand D and E was match up with each other but not with A and B only Brand C. An arbitrarily chosen board was picked based on incomplete data to test against other sticks. And no other factors was looked at or considered other than assuming the results are just Latency. Even though its been known that certain boards and converters will interfere with the operation of other boards. What those boards all are, we do not know as all boards for a console was not tested against other boards.
And not just because of all that, but numerous criticisms and objections are meet with stanch and obstinate denial instead of allowing for peer review makes it that the results now has to be rejected.
I kinda agree with your first point, but your last point about the information being rejected, i think it should stay for now until someone or a group do a more elaborate test.
Hmm I do not know how the oscilloscope works but if we did had the equipment, couldn’t each product be tested separately rather than against one another?
I.e. wire up the oscilloscope to the button so it can register a continuity or an led light to show when the button was pressed (this could be our 0ms indicator), then another set of wires to another input on the oscilloscope to measure when and how long it took for a data packet to be sent to the USB port and at the same time have a camera video recording the whole thing until we see the first frame of action.
so basically Led light/first reading on oscilloscope means start the timer (0ms) second reading on oscilloscope means data packet sent to port (3ms) first frame of action (13ms) - need to adjust for display lag -9ms’ - so it takes roughly 4ms. Rough idea and they are made up figures.
That way you do not have to have Ryu constantly punching another Ryu, so no need to test against another product and then it means less work, no need to test product A against B,C,D.
Someone who used an oscilloscope can correct me, I am just guessing how it is used.
I think between all of us here this is pretty doable, in terms of getting good photos of the boards we have and their revisions. The harder part would be the tests themselves. But maybe we could plan to bring some of these to bigger tournaments? or locals or something. I’m just trying to think of how the logistics would go.
Pho real?
Yeah I would like the camera to determine how long the console took to translate that data packet into an action we see on the screen, unless the oscilloscope can do that too, I have no idea.
Ah, i see. The problem with that method is the impossibility to sync the camera @ 60 fps to the screen’s refresh rate, which means high variance in desync. Good for a quick and dirty, but not accurate for a legitimate scientific test.
Correct me if i am wrong but wouldn’t the first reading from oscilloscope or led light be an indicator of one to start the time and if recorded in 120fps camera you can get almost accurate data to the nearest frame? then we have to account for display lag and how long the pixel takes to change, it won’t be super accurate but close enough.
You combat high variance with a crazy amount of tests…
Problem is its near impossible to know when one frame actually starts and stops.
When during a frame if a input is detected would the input be accepted for the next frame, is it dropped or is it carried to the following frame after?
Taking into account animation delay and non-cancleable animations.
Do certain boards claim a priority over the other or infer with the other board’s operations?
How many times is enough?
As far as 360 and ps3 goes, can’t we make a program for the system to do the tests? Program waits til a button is pressed and a timer goes off until a button from the other controller is pressed.
This can give us a measurable relative latency difference down to the ms.
The testing procedure would be quicker and more accurate.
Would be better yes. Ideal would be to have a usb relay perform the buttonpress then measure untill input comes in imho. But that would be very hard to do on anything other then PC.
Thanks for the well wishes. Just to be clear, my new career is not EVO, it’s to be announced 
But hope to see you guys at EVO though! 
It’s going to be a fun year for tech talk fans.
So if i understand correctly as a layman reader…
The frame input window thing is already explained by teyah how it factors into his test, certainly the average result over time should account for that as the faster stick should win more often regardless of not knowing where the input sits. Is there more to this that can disprove teyahs method?
The exact Ms time issue: it seems teyah is guesstimating these based on the averages as a difference compared to a control labelled as 0ms. I can see the detractors point here, there isn’t really a need to add an exact label if it’s not %100 exact, so why not just remove the ms number and leave the table as a ranking instead. It’ll still serve the purpose of showing what we can test, namely which stick beats another stick right?
The pcb interference claim: if one stick interferes with another that’s certainly worth investigating though it seems doubtful. Surely this is not that hard to test? If we took a random selection of a vs b tests the winner should always fall in line with teyahs table. If not then there really may be an interference issue. Also any interference would be the way the ps4 and the game handle input devices, not the device itself, so it should apply to a lot of input devices or none at all to my mind…
PCB revisions: this doesn’t negate teyahs results, bit could skew a model if there was really a difference in a revisions delay. Perhaps teyah could just start posting stick serials or pcb version when possible, for people who want to test if there may be a difference between like pcbs…
What are the other counterpoints? And has anyone disproven even one ranking on the table yet? If anyone has an argument it should be to improve this great resource not detract from it, let’s be positive not negative.