The Three Stooges, I mean Bezerks. Blowing life into them again....

[level42]

I knew you guys were reading...just had so many replies "to myself"

Did a bit of work on the two (spare) dead speech boards I have here.

Board A was not causing any errors on start-up, but it simply didn't put out any speech. This board was originally in my machine and worked for a couple of hours and then suddenly died.

Board B was causing errors (apparently on the data/adress bus). The game only beeps 3 times with this speech board connected and gives some error screen that normally would indicate bad RAMs on other boards.


Started with board A and swapped my working speech chip (S14001A) with the one on that board. Tested and....working fine !

Next I figured it could be worthwhile to try the S14001A speech chip from Board B on Board A and low and behold, it worked fine too.

So Board B is "double" dead now and Board A is 100% again.

I think Board B may have some bad data buffers...we'll see about that later...much later I guess.

The only speech board untested so far of our "batch" of machines and boards is the one on Ully's machine. If his is still working OK, we all got a spare working board too. That will be nice.

About the S14001A: It's pretty rare, but I did find one source for it. Chips costs 25 bucks though......
The chip is a pretty rare beast. I did some digging around and there's some interesting info on internet about it.

Here's a PDF with some interesting facts:
http://si.fileburst.com/s14001a_reveng.pdf

This doc was created to reverse engineer the speech for Berzerk in Mame. So contrary what I thought, Berzerk should not be playing samples in Mame but actually emulates the speech chip !

From this doc, I understand that the ROM's on the speech-board (there are 2) each hold different data.
One holds the syllables, one the phonemes. These are combined to create the speech.

The fun is that there's even a table with all the words and sounds and the corresponding hex-data. So, IN THEORY, it would be possible to change these words it's saying. If you put even more work in it, IN THEORY you should even be able to let Berzerk speak YOUR language......I kinda like that thought !
(Watje, vecht als een robot !)

Do you guys remember Ghostbusters, Kennedy Approach and Impossible mission on the Atari 8-bits and C64 ?
The speech on those games was done by the same guys who did this speech chip...

I've been reading some docs on the internet about speech synthesizers in those days and it's quite interesting, although I don't understand all of it I can understand the basics.

Anyway, one less dead board...

[petieken]

Haha, that's cool with the speech stuff. It's a lot of work though to find out all the sounds you need to make a single word. It does remind me of my early computer days in MS-DOS, editing game files so people actually swear instead of '!#?@' texts .

Now that I'm remembering this, I'm surprised of myself I could pull that off when I was about 8 years old .

I also did it a few times more recently as experiments with arcade roms (with success). Although this was done more for a self-educational purpose, to find out how to merge rom files with byte interleaving, byte swap & stuff like that.

[ckong]

......I kinda like that thought !
(Watje, vecht als een robot !)

that would be nice, like a kind of Berzerk clone.

[level42]

Well, my first idea was that it would be very cool if there now would be a software that could do the encoding.
The encoding originally was done by a quite extensive hardware set-up involving a couple of mini-computers, and from what I've read about the TI speech chips used by Atari (they tested a TI "PASS" system to encode the sounds at Atari, haven't read yet if they actually bought it) it was still required to make manual adjustments to the encoded sounds.  

Any way my guess is that a run-of-the-mill computer of today should very well be able to do it, BUT this doc actually makes it easier because we now have a table we "only" have to choose the already existing sounds and put them in the right order. Indeed, that could probably be done by a computer program as well. For testing, MAME would be ideal because it saves having to burn EPROMs and put them on the speech board for each try.

In theory, the sounds created by the chip should cover all languages because the synthesizer actually (tries to) reproduce the human "construction" to produce speech. However, I wouldn't be surprised if the sounds were tailored to the English language. F.I. I wonder if it will be able to produce the hard G we have in the Dutch language. But still it would be fun to be able to try it.

I've read on some sites that Berzerk actually had tailored speech for f.i. Spanish markets, but I've never seen the ROM files for that.

Sidenote:

Something that I find amazing about the sounds created on Star Wars is that the voices and sounds actually sound like the actors.  Even the "fly-by" sound, R2-D2's sounds etc. are created through the speech synth.  
This means that in some way they must have used the original sounds from the movie and "sampled" them, but, in theory, speech synths like this don't play back samples like we are used to now....interesting stuff, I plan on asking Jed Margolin some questions about it when I've read through all his docs.

The TI speech synth in SW is more advanced than the one on Berzerk though. The S14001A was developped in 1975, the TI was from early 80's.

[level42]

Another thing that I find pretty amazing is that I've never heard of a Berzerk bootleg.
It's strange because almost all the hit games have bootlegs. Maybe the set-up of the system (lots of PCB's) and the speech prevented it from bootlegs. The speech plays such a key role in the game (it's crazy how different it the experience is when speech is missing).

[level42]

Not a biggie to replace starters. Saint posted that my starter was the wrong kind for the tube, but this maybe due to the cab being for 230V ? I never gave starters any thought, I thought that a starter is just that and would work with all (most) tubes. The new one's at least seem to be able to.

Can you guys check what starter model you have in your Berzerks ?

Well I indeed had to replace the starter in the Berzerk. It was aimed for too high wattage tubes and it didn't always (more not) start the tube. Put in a brand new one and works fine now.

Ully's ZPU: Still going strong without a single problem !

Put some nice scores in the top 5 to keep Ully busy for some time once his machine will be working   (somewhere in the 11000, still not my highest score though...)

[Etienne MacGyver]

Not a biggie to replace starters. Saint posted that my starter was the wrong kind for the tube, but this maybe due to the cab being for 230V ? I never gave starters any thought, I thought that a starter is just that and would work with all (most) tubes. The new one's at least seem to be able to.

Can you guys check what starter model you have in your Berzerks ?

Well I indeed had to replace the starter in the Berzerk. It was aimed for too high wattage tubes and it didn't always (more not) start the tube. Put in a brand new one at works fine now.

talking to yourself again ??   

just kidding my friend   

[level42]

Since the little guy had fever today I was forced to stay home from work so I did some work on the Berzerk Joystick.

I still had problems to hit the diagonals and I really wanted it to improve.

First a little explanation about the construction of the joystick:


The red arrow points to the PCB. This carries the opto's (2 are draw, incicated by the yellow arrows). These point DOWN in a right angle. You can see that when the joystick is centered, the light will travel side-ways of the reflective disc (indicated with the green arrow). If the stick is moved, the disc will come over the opto and will reflect the infra-red light back onto the opto.


If you look at the design of the construction it's easy to understand why the diagonals are a problem:
The opto's are only on the primary directions (i.e: up, down, left, right):


The opto's are U2,3,4,5. The red arrows show where they should have added the other 4 for the diagonals....

The joystick works so, that the shiny disc that is connected to the joysticks tilts over to a certain direction when you move it. If you move it (f.i.) up, it will come close enough for the opto to "close" (start seeing the reflection of it's own infrared beam).
However if you move it diagonally, it will have to cover two opto's. Since the disc is circular shaped and even everywhere, the distance will be less then when you hit a primary direction.
This causes a very narrow angle to give f.i. both the up and left signals.

To be honest, Stern screwed up. They should have put opto's on every possible direction IMHO. But it was (as almost always) probably a cost thing. Anyway, that would be the best possible solution.

Since these opto's are long out of production, and I don't want to seriously hack the PCB (it would also require some extra TTL logic) I thought about a simple solution: how about bringing the PCB close to the disc.
This was very easy to do by adding some extra rings on the supports.

I first tested it with two added rings on each support. It worked great on the diagonals, but now I started having problems going in the primary directions. Apparently, if the disc gets TOO close, it's not working either.

The little graph in the datasheet proved this "theory":


You can see that the further away the reflective surface is, the less current it will conduct, which is logical. But also when the distance is very close, it will stop conducting (in fact, it drops much faster there).

So, I needed to find the right "balance". Thus I removed one of the 2 extra rings on each support and tested again. This time it worked perfectly !
I hit diagonals as easy as the primary directions !!! It's really like a different controller now !

I also even further loosened the spring-tension. I don't really know why they included it in the first place. The stick does self-center by itself, but it does springs a little back and forth if you let it go. I played it a while without the spring attached and I was almost perfectly happy with it. It totally got rid of the required force to push the joystick. IMHO a joystick is not a means of muscle-exercise but this one sure felt like one (esp. before I loosened the spring). THe only problem I had without the spring connected was moving over VERY short distances (I mean one pixel here). This is often what you want to do with Berzerk. Stand around the corner of a wall and shoot EXACTLY 1 pixel over the edge of the wall. It was almost impossible to do without the spring.

So, I re-attached it and loosened it even further than before. I will need to seek for the ultimate compromise....

There's another thing I don't like about this stick:
The end-hit....(or how do you call it ?) When you hit the end of the travel it hits a very hard rubber like stuff. I don't know if this has become harder over the years or if it was intended this way, but I don't like it. Is there any stuff to weaken the rubber around ? Or should I try to get some new and cut a new block (if possible) ?

Anyway, I'm already MUCH happier than I was before. Almost all my games after I  moved the PCB up a bit I got through to 10000 points with 3 extra men. I lost them quickly every time though so I still didn't brake my 12xxx record but I'm sure I will soon....

Did I mention this game ROCKS ?

To be honest I was a bit hesitant about the re-playability of Berzerk but man does it draw you back and back and back ! I love it !

[Superully]

you really dug deep into the topic, andre! R-E-S-P-E-C-T. curious how my joystick is going to feel / work in comparison to yours. perhaps we also need to make some adjustments ...

[level42]

There may be a second way to adjust / correct this issue.

See the schematic:


The signal of the photo-transistor in the opto is compared to a reference voltage created by R2 and R3 (Schematic only shows 1 opto, but of course they're all the same). R2 and R3 could be replaced with a 10K pot. This way it's possible to vary the voltage that the opto's signal is compared to by the LM339. In theory this should allow to change the "range" of the opto. However, this will also require careful testing and finding the right balance.
As the extra rings on the supports are much simpler and seem to work great I settle for that, but in theory this should work.I think

[level42]

Wahoo, a box from the US arrived today !

And you gotta love Bob, he even separately packed the stuff for all of us ! 


Of course I can't wait to use some of the stuff. The old buttons:




Old and new. As you can see, red colored stuff ALWAYS fades....


"Small" difference !


Now, all we need to wait for is the repro CP's from Cmoore !

[Etienne MacGyver]

wow what a difference !

[Superully]

bob really is a genius - look at that packaging. i'm happy for you, andré, that you didn't have to sort the stuff on your own.

and then those new buttons. man, this game is going to look kickass with the repro control panel! 

[level42]

Yes. Now it looks like a "flag on a shit-boat" as we say in Holland.......

Indeed, I think a good number of suppliers would be annoyed by the way I ordered because some of us had the same parts (i.e. the buttons). I sent him a list per person and instead of Bob complaining about it he did this. Big plus for Bob.

Oh, and no "shipping troubles" at all.

[Muerto]

wow! - thats the way to do business! maby some can learn somthing?