Time Sleuth Lag Tester

[orange808]

My Time Sleuth lag tester arrived today and it's really cool. After spending a couple hours with it, it's hands down the best latency tester I've had. Highly recommended.

I wanted to take a moment to share a positive review. The Time Sleuth is awesome and it does what it should. I think good stuff deserves to be recognized. Not all community products are good and I've bought a few disappointing community products. There's always pressure to cheer and say nice things about every community product, but that's bullshit. Only good stuff deserves recognition. The Time Sleuth is a good one.

First of all, the latency results appear to be accurate. I got the expected results with everything I tried.

Like the Leo Bodnar, it uses three flashing bars (top, middle, and bottom) to measure latency and displays the "total" display latency at each test point. However, it has many extra features that blow the competition away.

I only have issue. The Time Sleuth's weakness: it doesn't run on batteries and requires a separate power supply, so you'll need to keep a USB battery handy. No big deal, but worth noting.

It outputs test signals at 60Hz as 480i, 480p, 720p, 1080i, or 1080p. The test output display is more convenient than the Leo Bodnar; Time Sleuth displays the highest lag measured, lowest lag measured, and the average during testing. That's very handy when you don't have a locked frame rate.

The case shell doesn't come together perfectly at the seams like a consumer product, but that's not something I expect--or care about. The case gets it right where it counts. It feels more than solid enough in hand. It fits very well in my palm--and it feels more comfortable to use than the Leo Bodnar. I always found the Leo Bodnar to be a bit too wide for my palm and I hated holding down a button to test. The Time Sleuth is a better size and it's always working when it's plugged in; by comparison to the Time Sleuth, the Leo Bodnar feels awkward.

The cuts in the shell are properly sized and I had no problems with cords. Both an HDMI cord and the USB mini power fit into the device with the right amount of tightness. The shell doesn't prevent plugging cords in and the ports themselves are not too small or too large.

The knob to select testing signals turns and snaps into each resolution in a satisfying way. The knob feels sturdy and secure enough as well. It should last a long time.

The front light indicates power. It also has a satisfying flash during testing when the Time Sleuth is working; it blinks when the device detects an on screen bar that is flashing.

Really like this one. It raises the bar for testing devices and the Leo Bodnar is officially irrelevant.

[nmalinoski]

I don't have much use for one of these, but I want one now.

[BONKERS]

At that price I think i'll have to buy one of these myself now.

[Bahn Yuki]

https://youtu.be/tDLQrGKEnVI

Sent from my SAMSUNG-SM-G891A using Tapatalk

[citrus3000psi]

Thanks for all the kind words everyone! This project all started because I have a 1080p leo that wouldn't work on a my DVDO DUO processor and Seiki TV. My Leo uses an internal oscillator that created to much jitter video clock output. TV's that did work have this edge where you can actually see the clock problem: https://www.youtube.com/watch?v=pdWNqc8TYxc

I'm not sure if this is a new design, to try and save money during production.

Hats off to chriz2600 for this project too, couldn't have happened without him.

Just wanted to reminded everyone that you can easily build your own custom resolution profile here: https://time-sleuth.i74.de/

I will do a video on how to update the device (Change resolutions), as its very simple. You just need a USB-Blaster and quartus programmer installed. No need to take the case apart or solder soldering anything to update.

[Ikaruga11]

So all "lagless" 60Hz CRTs without any processing or delay still take 8.3ms to draw half a frame?

[Bahn Yuki]

Thanks for all the kind words everyone! This project all started because I have a 1080p leo that wouldn't work on a my DVDO DUO processor and Seiki TV. My Leo uses an internal oscillator that created to much jitter video clock output. TV's that did work have this edge where you can actually see the clock problem: https://www.youtube.com/watch?v=pdWNqc8TYxc

I'm not sure if this is a new design, to try and save money during production.

Hats off to chriz2600 for this project too, couldn't have happened without him.

Just wanted to reminded everyone that you can easily build your own custom resolution profile here: https://time-sleuth.i74.de/

I will do a video on how to update the device (Change resolutions), as its very simple. You just need a USB-Blaster and quartus lite installed. No need to take the case apart or solder soldering anything to update.

That's exactly what I used to flash my firmware yesterday. Straightforward and easy to use. I went with 240p over 480i due to my use case.

Was impressed the tcls I own maintained low lag in 240p. Also the Sony hd crts are excellent displays whether component or HDmi as long as you are doing 1080i.

I have more results in my video description of anyone cares to see.

Sent from my SAMSUNG-SM-G891A using Tapatalk

[orange808]

So all "lagless" 60Hz CRTs without any processing or delay still take 8.3ms to draw half a frame?

Yes. On a CRT with no processing delay, the Time Sleuth will read something very close to that at the middle testing bar.

It naturally follows that you would also see about zero ms at the top and ~16 2/3 ms at the bottom.

[Bahn Yuki]

So all "lagless" 60Hz CRTs without any processing or delay still take 8.3ms to draw half a frame?

Yes. On a CRT with no processing delay, the Time Sleuth will read something very close to that at the middle testing bar.

It naturally follows that you would also see about zero ms at the top and ~16 2/3 ms at the bottom.

That seems to be correct. Are the readings not valid from the middle bar then? I thought that was the testing standard. I can redo from the top if that is the accepted standard.

Sent from my SAMSUNG-SM-G891A using Tapatalk

[orange808]

So all "lagless" 60Hz CRTs without any processing or delay still take 8.3ms to draw half a frame?

Yes. On a CRT with no processing delay, the Time Sleuth will read something very close to that at the middle testing bar.

It naturally follows that you would also see about zero ms at the top and ~16 2/3 ms at the bottom.

That seems to be correct. Are the readings not valid from the middle bar then? I thought that was the testing standard. I can redo from the top if that is the accepted standard.

Sent from my SAMSUNG-SM-G891A using Tapatalk

I'd like to see all three numbers if possible.

For instance, some displays refresh the screen very quickly and the bottom number is much closer to the top than you might expect.

[Bahn Yuki]

Yes. On a CRT with no processing delay, the Time Sleuth will read something very close to that at the middle testing bar.

It naturally follows that you would also see about zero ms at the top and ~16 2/3 ms at the bottom.

That seems to be correct. Are the readings not valid from the middle bar then? I thought that was the testing standard. I can redo from the top if that is the accepted standard.

Sent from my SAMSUNG-SM-G891A using Tapatalk

I'd like to see all three numbers if possible.

For instance, some displays refresh the screen very quickly and the bottom number is much closer to the top than you might expect.

I can do that. The time sleuth is very easy to use. Perhaps we can make a thread for lag results. I have 15 displays in my home that I can verify.

Sent from my SAMSUNG-SM-G891A using Tapatalk

[orange808]

Perhaps we can make a thread for lag results. I have 15 displays in my home that I can verify.

A database would definitely be useful. There isn't any really useful display latency information available anywhere right now.

Also, to be perfectly honest, the Time Sleuth is the first device that is properly suited to the task. It's easy to use and a standardized build from a manufacturer (a small manufacturer, but a manufacturer nonetheless).

I believe the most relevant signals are:
480i, 480p, 720p, 960p, and 1080p.

240p isn't supported on most displays and it generally looks awful (blurry) on displays that do support it. Serious gamers are going to use an upscaler. 240p lag times are an edge case on modern displays.

1080i measurements are useful for syncing audio to broadcast television material, but we're testing video games. 1080i is an edge case for most gamers--and I never ever use it for games. 1080i lag doesn't need to be recorded.

480i deinterlacing time should be in there. This will help people understand the real cost of using the display (or video processor) for deinterlacing--and highlight the time savings of line doubling. In many cases, I think it will also ignite resurgent interest in Framemeister when people see proof that the Framemeister is outperforming most displays in every way possible when deinterlacing games.

Of course, the problem is getting Time Sleuth devices into enough hands to make a dent in the massive pool of displays out there... and getting people to share results.

-----------------------------------------

Also, I worry that a manufacturer will push out a firmware update and we'll get flamed for accidentally recommending a laggy display--when it's not our fault the manufacturer changed things and our recommendation was good at the time we posted it.

[nmalinoski]

I believe the most relevant signals are:
480i, 480p, 720p, 960p, and 1080p.

240p isn't supported on most displays and it generally looks awful (blurry) on displays that do support it. Serious gamers are going to use an upscaler. 240p lag times are an edge case on modern displays.

While I agree that those five are the most relevant, I would like to see 576i@50 and 576p@50 included. I don't necessarily care about the lag with these input modes, but I live in the US, where PAL support is either not advertised or non-existent; and, if I want to play PAL games at 50Hz, I might not be able to. My current display doesn't support PAL modes, which forces me to use a video processor that can do framerate conversion.

So, if a bunch of people are going to be testing displays for a database that ostensibly would be used to aid in the buying process, I think it would be trivial to include these additional video modes in the testing procedure, so that the proposed database could, as a secondary function, reflect PAL support. (The same could probably be said for 240p support, but I would say that's less important.)

Also, I worry that a manufacturer will push out a firmware update and we'll get flamed for accidentally recommending a laggy display--when it's not our fault the manufacturer changed things and our recommendation was good at the time we posted it.

A legitimate concern. Perhaps including the firmware version in the report, when possible, can help indicate those issues.

[Fudoh]

240p isn't supported on most displays and it generally looks awful (blurry) on displays that do support it. Serious gamers are going to use an upscaler. 240p lag times are an edge case on modern displays.

but it's a great to have a 240p lag tester to run the signal through all the external upscaling solutions. This let's me bechmark the whole processing chain. And when it comes to 15khz inputs, 480i isn't really an alternative, is it? And right - 1080i on the other hand is pretty useless ? Vita TV maybe, but anything else ?

[maxtherabbit]

240p isn't supported on most displays and it generally looks awful (blurry) on displays that do support it. Serious gamers are going to use an upscaler. 240p lag times are an edge case on modern displays.

but it's a great to have a 240p lag tester to run the signal through all the external upscaling solutions. This let's me bechmark the whole processing chain. And when it comes to 15khz inputs, 480i isn't really an alternative, is it? And right - 1080i on the other hand is pretty useless ? Vita TV maybe, but anything else ?

like 2 xbox games and gran turismo 4? lol

[nmalinoski]

240p isn't supported on most displays and it generally looks awful (blurry) on displays that do support it. Serious gamers are going to use an upscaler. 240p lag times are an edge case on modern displays.

but it's a great to have a 240p lag tester to run the signal through all the external upscaling solutions. This let's me bechmark the whole processing chain. And when it comes to 15khz inputs, 480i isn't really an alternative, is it? And right - 1080i on the other hand is pretty useless ? Vita TV maybe, but anything else ?

like 2 xbox games and gran turismo 4? lol

Looks like the count is at least 7 for the Xbox and at least 5 for the PS2. GameCube can apparently do 1080i with the Game Boy Interface software, and I imagine some other homebrew, but I'm not aware of any retail games that will do better than 480p60.

[orange808]

Good points, but there are limitations to consider.

The Time Sleuth has a limited amount of available test signals. There are five slots.

It's unreasonable to ask potential volunteer testers to reflash the firmware multiple times for each test display.

So, we need to agree on a firmware with all the base signals required.

Adding the need for an OSSC to properly test will complicate matters. Disconnecting the OSSC and taking it out of the house is a pain is the arse. I would prefer to test five signals stored in the Time Sleuth.

I understand that my suggestion is imperfect, but I think it's the lesser evil.

[nmalinoski]

I was unaware of the 5-slot limitation. Can that be expanded, or is it fundamental?

[orange808]

I was unaware of the 5-slot limitation. Can that be expanded, or is it fundamental?

The knob on the device has five positions. Even if the device was theoretically capable of more, you wouldn't have a way to select them.

[nmalinoski]

I was unaware of the 5-slot limitation. Can that be expanded, or is it fundamental?

The knob on the device has five positions. Even if the device was theoretically capable of more, you wouldn't have a way to select them.

Considering the positional switch on the existing hardware, I would think the firmware could be modified to treat any change from the knob as an increase, so every click, even back and forth, would increment the index through a list of modes. Maybe have it reset to a safe resolution, like 480p60, when selecting 1.

Otherwise, a hardware revision with either a momentary switch or some kind of incremental rotary switch would be needed to go beyond the five-position limitation.

[citrus3000psi]

I was unaware of the 5-slot limitation. Can that be expanded, or is it fundamental?

The knob on the device has five positions. Even if the device was theoretically capable of more, you wouldn't have a way to select them.

Considering the positional switch on the existing hardware, I would think the firmware could be modified to treat any change from the knob as an increase, so every click, even back and forth, would increment the index through a list of modes. Maybe have it reset to a safe resolution, like 480p60, when selecting 1.

Otherwise, a hardware revision with either a momentary switch or some kind of incremental rotary switch would be needed to go beyond the five-position limitation.

The FPGA only has enough resources to hold 5 resolutions at once.

[RGB0b]

A database would definitely be useful. There isn't any really useful display latency information available anywhere right now.

I started one here that will be moved to the wiki once its live: https://www.retrorgb.com/lagtest.html

It's just a Google Spreadsheet for now, but it's a great start. Email me if you'd like me to make any changes or if the spreadsheet isn't working for you. I'm only on here about once a week now, so I won't be seeing these replies in realtime.

[BONKERS]

So I just got mine in the mail. And i'm having issues with it working on two specific displays. They are older 1080p LCDs but the only other thing they have in common is that they both have a CCFL backlight. Not LED.

The only way I can get the blue indicator light to go off is to completely cover the diode with two fingers. Or cover it enough at just the right angle for it to flash in sync as it should. But like this it still can't pick up a proper reading.

It works fine with 3 other displays i've tested. All LED based backlights. (Two VA Tvs of different resolutions. And an IPS monitor)

For one of the VA displays it will not work unless it any room lighting is turned off.


Anyone run into this kind of thing? Using just the bare circuit board without the case.

[BuckoA51]

Yes I've come across one display that I couldn't get a reading from, an old Sony KDL-W4000. Leo's tester couldn't get a reading either.

On one display I had to hold Time Sleuth at 90 degrees to get a reading too.

[BONKERS]

When I emailed the site he suggested turning down the brightness/backlight. That gets the blue diode flashing but no reading still.

Very strange. Haven't tried that on the other display yet.

[strayan]

Hope this works with all my pre 2007 plasmas! Will be sure to update the spreadsheet with my results.

[strayan]

What’s with plasma having the same latency at the top middle and bottom?

[chriz2600]

What’s with plasma having the same latency at the top middle and bottom?

That's a result of how a plasma screen is drawing the image. See here: https://www.youtube.com/watch?v=tGpdah32n3c

[strayan]

Cool. I wonder whether this gives plasma any unique advantages over other display types. Maybe if you could find a plasma (doubt they exist) with less than 16ms of lag?

Pretty cool to find out that my plasma has 24ms though.

[Xyga]

Cool. I wonder whether this gives plasma any unique advantages over other display types.

Only for motion perception (smoother and closer to crt-like comfort) but that doesn't make'em ineherently better input lag-wise.
That use to be a widespread myth, probably cause by the better motion's nocebo effect in that aspect.
Anyway, plasmas can have as much lag as any other kind of display, input lag doesn't come from the panel technology.
There is a relationship to response time, to a certain degree but only accounting for a minuscule percentage of the total measured lag like 5% or less in many cases.

PS: question to the general, do the TS readings vary a few ms with the display's brightness ? (like the LB for instance)