Arcade: using an LCD instead of a CRT

Part 1 – debunking other peoples work, as I would like to be debunked

I saw this video from TNT amusements:

As you can see, they swapped the CRT for an LCD. They use a converter board to get the 15kHz to 31kHz and in that process scale the image in the converter and in the monitor(1280*1024). They did this for one of their customers. I am unsure as to the result of this operation, mostly due to these statements made in the video:

1) They will install a brand new LCD monitor
2) They claim the image is crisp and clear.
3) They offer it as a service

So, I went and found that exact LCD monitor, the same converter PCB and since I do not own a Moon Cresta PCB, I used my Frogger which uses the same resolution.


The Dell monitor is from 2006. This is somewhat old in LCD terms. I got the monitor for 10 USD. Like my SCI project below, it has a resolution of 1280*1024 and a ratio of 5:4. The converter board does not output in this resolution, so you have to choose one of the available resolutions in the menu and then let the monitor scale that to its native resolution. This can have drastic effects on the image quality, especially in games with thin 1 or 2 pixel lines. These can jerk around when they disappear and reappear as they move.

The monitor itself is extremely slow and is rated at 25ms. I know these values are highly subjective, but the number is from the manufacturer’s own specs. I tried running Terra Cresta on it, to see what a scrolling game would look like. Results are jerky movements and tearing.




Running this test I got pretty nice colors. The picture quality was overall okay, but a little soft. Nothing that I would have a great problem with. However, the excessive blurring of moving elements is completely unacceptable. The frog(or ship in Terra Cresta) is completely blurred out when in movement. When it comes to Moon Cresta I suspect that the black background and being a fixed shooter will combat tearing but then again, bright space ships against a dark backdrop must look horrible when moving.

Here is a short video, please note this is recorded of the screen. Keep your eyes on the ship.

Since there are two parts to this setup, the converter and the monitor, I have tried another LCD monitor with the converter. In this case, an LG 22″ TV. The results were far better, but still not great. Installing an LCD screen in your arcade demands a fast monitor. These days 16:9 screens are 99% of the market, but there are still a bunch of 5:4 1280*1024 monitors being made, mostly for business and writers. These monitors go for about 200USD. One can only wonder what the image quality would be like if one of these(5 ms or lower) has been installed in the Moon Cresta from the video.
I tried the setup one last time with my 27″ AOC LCD monitor which is stated as having a response time of 2ms, and it is leaps and bounds better than the two other monitors.

My final conclusion is: The setup seen in the TNT video is completely unacceptable. I would have a hard time finding anyone who would think this image quality is OK. You need to find a fast LCD monitor for this to work, and it probably will never look right with a game like Moon Cresta. The monitor used by TNT is unsuited for any kind of gaming.



Part 2 – scaling at its best…doubling (or the like)!

First of all: I know this will be frowned upon by many readers. But please try to follow my thoughts. Maybe it is possible to implement an LCD monitor in a classic arcade machine, in this case my Special Criminal Investigation(SCI from here on out). I got this machine years and years ago, with graphics problems. At that time I did not know anything about PCB repair and while SCI is an okay game, I like the prequel, Chase HQ, a lot better. So, I decided to remove the original PCB and use PC emulation. The inside of the cabinet never got to that point of tidiness that I like, so recently I decided to go back and give it an overhaul.

This is when I decided to give the LCD idea a try. I have always been a ‘CRT-guy’, not from arcades really, but from CRT projectors. I still use a Marquee CRT projector daily, and the choice of changing a tube monitor to a flat panel does not come easy. But these days I try to keep a more open mind than I used to.

Going into this project I had some concerns/demands/wishes. Let me put them in a bullet point list.

1) No stretching of the image in any way.
2) How will rescaling be done, and can it be acceptable?
3) Lower total-weight of the machine
4) Less heat
5) Less reliance on old parts, especially dangerous parts

The first point mentions stretching. I have a Cruis’n World (with the original CRT), and I see a lot of people put 16:9(1.78:1) TV’s in their driving cabinets. When these guys do this, they stretch the image from 4:3(1.33:1) to 16:9(1.78:1) which is 100% unacceptable to me. If they do not strecth the image, the image will probably be smaller than the original monitor due to the format and width of machine. I remember of friend of mine having a Sega Model 2 cabinet(I think), in which he put a vertical 16:9 LCD. He turned on Galaga while I was there, and it was horrible, just horrible. The problem is, you cannot get 4:3 LCD monitors in these sizes.

SCI and Chase HQ run 320×240 resolution. The 5:4(1.25:1) LCD monitor I ordered have a resolution of 1280×1024. This works out awesome, since 1280/320=4 which gives perfect scaling. No guessing, just multiply by 4. On the ‘line’-side(to stay in old-land), the game has 240 lines. 1024/240 = 4.27 which is not very even scaling. But if we instead do this: 240*4 = 960 then we get perfect scaling again, but leaving a small black bar in the top and bottom of the screen, which also gives you the 4:3 picture! And that is it! It cannot be better than this. Imagine off-scaling like this: You have two lines, one is black the other is white. Scale that to three lines; you have to flip a coin to decide if the third line should be black or white. Doing even scaling, no guessing is required.

I weighed the items. The CRT + chassis came in at 15.5kg, the LCD at 3.5kg. This makes the already insanely heavy cabinet 12kg lighter. I have not accounted for the Hantarex US250 power supply in the bottom yet, which is, of course, also gone.

I know that nobody has been injuried greatly from the anode cup or otherwise, from working on CRT chassis. However, these CRT pcbs are getting older and older, maybe pushing 25+ years. Also, how many people have worked on the chassis before it reaches you? Should this be cause for concern, or at least cause for concern for the next guy when you have worked on it?

All these points lined up, and I took the plunge. I am not the type of guy that forces himself to like something just because he has it, so I will be completely honest.

Image quality:
Very, very nice. Nice strong colors and good blacks(CRT-guy, remember?). Does the image actually warrant an ‘upgrade’ from CRT to LCD? No, they are on par. Colors and precision goes to the LCD, slightly. Everybody will love the image from the LCD.

I get a bit worried with myself about my connection with curved CRT’s in arcades. I just expect to see that curved screen when I stand near an arcade machine. It is not bad on the SCI, but on more classic machines, it really pops out at me. Not in a horrible way, but just a gut feeling that something is wrong. I once saw a picture of an arcade machine with an LCD running Pac-man, and my mind could not handle seeing the maze un-curved. It looks absolutely crazy, but it is, of course, right. The maze should have perfect geometry. Of course a picture should be flat, and CRT TV’s in homes has been flat from the mid-90’s. That is 20 years! However, not all cabinets will look right with an LCD. I think that older machines with the screen almost lying down, and the graphics bezel ‘try’ to surround the picture, but fail because of the curve. Would such a game, like Galaga, be so bad if the LCD was all the way up to glass and going to the sides of the bezel? This is more difficult on games where you have the glass close to the players face.

Picture of SCI with LCD.


I also need to address the problem of changing a CRT for another CRT. You can get into a lot problems with a new tube. I think about that plastic bezel that often surrounds the tube, and follows the curve of that exact tube. I have tried several times where I needed to discard the plastic bezel after a tube swap.

So, can my SCI project be translated to other games? Well, that depends. I saw a Youtube video where a couple of guys put an LCD in a Moon Cresta. Now, being a PC monitor, it starts at 31kHz but the game board is 15kHz. The resolution of Moon Cresta is 256×224 and the monitor is 1280×1024. 1280/256 = 5 and 1024/224 = 4.57. Pretty uneven scaling. So, let us do as before and use 4. 256*4 =1024(of the 1280) and 224*4 = 896. This will leave black bars all around the image, and it will not be doable in their setup. What they do is: They use a converter that takes the 15kHz RGBS from the original PCB and outputs 31kHz VGA(RGBHV). At the same time, it scales the image, but the device does not have the resolution of the LCD. This means that they scale the image in the converter and then again in the LCD. I would love to see how that image looks in real life. *I have one of these converter units on order to test it*

The main question must be: one of these years I am going to get a Galaga. My favorite game of all time. Could I put an LCD in that? No, I could not. I think Galaga would suffer a lot from off-scaling. If Galaga was 320×240, like SCI, then maybe! But it is not. These games with very fine, straight lines do worse compared to games like Out Run and maybe Neo Geo games. But for games that has resolution that you can multiply up, think about it! Consider it.

As a final note, I also feel that some games deserve to be kept original more than others. Donkey Kong is more a piece of arcade history than SCI. Same goes for Galaga compared to Cruis’n World. I love Cruis’n, but it is really just a racing game. And what about the future? Maybe when OLED TV’s comes about, we can have those curved since they use no backlighting, like LCD uses fluorescent or LED to light up the panel.


Picture of free space where the CRT used to live.

Sonic: Rise of the Hedgehog finished!

After about eleven months, today is the day I finish my arcade game, Sonic: Rise of the Hedgehog. I started the project to make the first danish arcade machine. The scope of the project went way above what I had in mind. I have spent countless hours sitting in front the computer making the game. The cabinet(or hardware) side of things was a lot easier, but both put together was, in hindsight, way too much for a one-man project. But I did it.

I will give an overview of the process below. It will be with the oldest lowest and the newest at the top. Looking back, everything is a blur. A blue blur(!).


Video of the attract mode:

– August 22 2014

Made a small digital flyer for the game:


– Late August 2014

Sent off the hi-rez front art for printing. Turned out great!


– Late July 2014

The poster on the right went up when the project began. The sun is a harsh mistress.


– Late July 2014

Control Panel all done. The colours might change later. I also almost finish the game’s RUSH MODE.


– Early July 2014

I finish the ULTRA-UPDATE video:
– Late June 2014

Testing the game on a normal TV with a XBOX controller.


– Late May 2014

Finishing the artwork for the marquee, getting it printed. Also started on Zone 2, which is space-themed.

– Mid May 2014

First draft of the indicator for the control panel:

– Early May 2014 mention the progress with the game again.


– Late April 2014

I finish the I/O interface. This will take inputs from the control panel to the PC and it will drive the lamp on the control panel. This is for when Sonic is close to a secret in the game.

Also, testing gfx that end up not being used.


– Late April 2014

This is the event horizon point. This is where there is progress in strides. Now I have invested so much time into this project that I have to finish it.

I set the title to Sonic: Rise of the Hedgehog. I release the GIGA-UPDATE, a video that covers progress on the game:


– Late March 2014

After fixing the monitor in the cabinet I was ready for the first test, running the early draft of the game on the cabinet. At this point the game is running at 480p 60fps.

– Early March 2014

I get the donor cabinet. Working with wood is another of my shortcomings.
You can see the classic Sonic poster on the wall. This is what I call method-coding

– Early February 2014 runs the news of ROTH being developed:


– Late December 2013

Finished the first intro screen with controllers for coin inputs and start.

– Late November 2013

Working title is set to Sonic: Ring Master/Untitled. Also, I make the game to run with several layers of parallax scrolling.

– Mid November 2013

I decide that my game has to be 60fps at a resolution of 640x480p 31kHz. Since I wanted to the game to be FAST and PRECISE I needed peak performance for the twitch gameplay.

I also make my first test joystick, using a NES Advantage:


– Early November 2013

Working on platforms. You cannot have a platforming game without moving platforms, right? So, I start making platforms that move up and down. Making games really amazes me with the amount of different ways to do stuff.


– Late October 2013

I begin coding the game. Since sprite drawing and artwork is not within my capabilities, I ‘borrowed’ Sega’s graphics, hook, line & sinker. My original plan was to make a high-res game where the camera is placed far from the action. This can be seen on the below screenshot. I have always loved a pulled-back view, but both my wife and a couple of early followers of the project frowned at this. Also, opting to use an arcade monitor forced me closer to the action. This is the first compromise.


Pinball lamp measurements v4

I am working on a project, that gives me the chance to document some data regarding different lamps used for pinball machines. The lamps used are incandescent(regular) 44s and Cointaker frosted cool white LEDs. LEDs equivalent to the Cointaker should give similar results. I chose Cointaker since it is what I use myself.

Version 4 final. Updated January the 6th 2015.

1. Power consumption

Standard 555/44 incandescent bulb. Used for inserts and GI:

Running as CPU controlled(+5V) these draw 0.2 A.
The wattage is 5*0.2 = 1W

Running as GI(+6.3V) they draw 0.24 A
The wattage is 6.3*0.24 = 1.5W

Cointaker 555/44 LED cool white. Used for inserts and GI:

Running as CPU controlled(+5V) they draw 0.02 A
The wattage is 5*0.02 = 0.1W

Running as GI(+6.3V) they draw 0.05 A
The wattage is 6.3*0.05 = 0.3W

2. Temperature measurements

Using a standard 44 incandescent bulb and a Cointaker frosted LED 44 for measurements. The thermometer was placed at the very tip of both subjects. I will make measurements regarding the heat build-up and the heat decay when turned off. Possible error sources are the thermometer itself which needs to be heated and cooled. However, this is the same for both subjects. Seen in relation to each other, it should cancel out.

First, here’s a graph with four lines, two tests for the incandescent and two tests for the LED. This is to see how each behave as both GI and feature(6.3V and 5V)

The graph shows a radical difference between the incandescent and the LED. You can see that all protrude from the same point on the Y-axis. This is the room temperature(23.4 C) when measuring on the desk. The feature(5V) LED almost doesn’t deviate from this value. At the other end, the incandescent(6.3V) is almost at 44 C. Looking at the graph one can tell that there is huge difference between the subjects.

Going to the heat decay time graph(below). This is to show how long it takes for the bulb to reach the room temperature again, when turned off. As seen in the graph above, this is 23.4 C.

I have omitted the two LED tests. Their heat build up was too low to include here or even worry about. Looking at the graph you can see that it takes 12 minutes for the incandescent(5V and 6.3V) to return to room temperature. Being a little more fair, it takes about 3-4 minutes for most of the heat to disperse.

3. Light output, relative

Now let us look at the light output differences between the standard 44 incandescent bulb and the cointaker frosted LED. I have measured using the lux from each subject at both voltages(6.3V and 5.0V). Lux is the SI-standard for measuring illuminance. Since these sources(the incandescent and the LED) are at different places in the x, y, z – plane of a pinball machine, with some being hidden, some being under inserts and all of them being at different distances from the player it would be futile trying to measure the real scalar value of one of these. On top of this, pinball machines are being played in rooms that are lit. This means that the light output from the pinball machine would add almost no light to the room. Therefore, I have chosen to do the measurements of each bulb/LED and voltage in a dark room using a luxmeter at a set distance from the bulb/LED mounted in a 44 socket. Doing this gives me values that are unusable except when looked relative to each other. Here is a graph of the measurements:

As can be seen on the left half of the graph, the luxmeter read 250(no units) for the cointaker frosted LED, and 120(no units) for the 44 incandescent. This shows that the LED has double the output of the incandescent. The same go for the right side of the graph. Again, the LED doubles up. This must mean something when LEDs are used as inserts and you play in a well-lit room.
Possible errors/neglects in the above graph and statements can be that LED doesn’t generally spread the light around as well as incandescents. The lux measurements are made from the front, where the LED is stronger than it is from the sides. However, since it has double the output from the front and the LED is frosted, I will make the bet that the LED is better(higher output) from the sides as well. Keep in mind that colored LEDs take a dive in light ouput! (the same goes for incandescents)

4. Light output comparison between different colors, relative

I have measured the standard frosted LED against its colored counterparts. I wanted to see how much light output is lost to get color. The result are in the graph below. I have included an incandescent as well.

As earlier the numbers means nothing. They are just the numbers from the light measurement at a set distance. Only the comparison is valid. If we look at the white frosted compared to the red frosted, the light output is cut 60%. The other colors are in the same league, with orange being cut by 68%. Still, compared to the incandescent, they are still brighter. But this is very uncertain. In this test, the incandescent are 73% dimmer than the white frosted, but in the earlier test the difference was 50%. This is due to incandescent pinball lamps varying a lot in output. So, whether a colored LED is dimmer or brighter than an incandescent is up to chance or testing each for output.


The first power measurement tests means that a standard incandescent bulb uses 5 times as many amps as the Cointaker LED. They produce way more heat and still only half the light output.

Going LED is sure to decrease heat in the cabinet and go softer on the low-grade connectors found in almost all pinball machines. This will also make the boards run cooler. Not to mention mylar lifting above the insert due to the heat creating gasses from the plastic – which in turn will decrease the size of the insert, which will cause it to sink into the playfield.

I have been entertained watching the slow change in pinball-fans. It started out with them never wanting LEDs in their machines. Then it went to “inserts are OK, but not GI”. Guess where it is going next. Yes, the GI as well. Move faster, pinball-fans! You of all people should push for better tech!

There are no valid reasons not to use LEDs!


Developing for Android

I have just installed the ADK and countless other files. Now I will be able to develop software and games for Android. First of all, I will begin working on a game especially for that platform. I think it will be using the standard touchscreen controls, even though I am very opposed to a control scheme this lacking. I grew up on arcade games and thus I have great trouble with controls that are not razor sharp. Some people have gamepads for their Android devices and others have units like the JXD S7800 or Yinlips which has d-pad, normal buttons and even analog sticks. Still, I wonder how extremely small the total percentage is, compared to the total number of Android users.

A picture of the table setup. Please note the Ducky Shine 4 keyboard. Everybody should have one.

…and a picture of a test-run; small white blocks that disappear when you touch them. Not very interesting, but at least it works.


BaCo v1.4 Bacterial Count Calculator

BaCo v1.4 is available for download. Fixed some problems, and now the result is returned in scientific notation.

BaCo v1.3 is available for download. There are very minor fixes from v1.2, and they will probably go unnoticed by most.

Click here: to download the software.

About BaCo:

This is a project I started with the goal of making bacterial counts as fast as possible. The program is slim, has only one file and only needs the minus-key, number-keys and enter to be used.. Choose your plates, insert values, and you get the bacterial count. You’ll also get the count uncertainty expressed as percentage and +/- # CFU. Project time was 8 hours(v1) + 2 hours(v1.1) + 1 hour(v1.2) + 0 hours(v1.3) + 1 hour(v1.4) = 12 hours