Cromemco JS-1 Joystick Replica

nullvalue said:

Here are the readings taken from the original JS-1 @ 22V

Code:

AXIS            -           CENTER          +
Left/Right    14.51V        16.47V        18.50V
Up/Down       14.28V        16.07V        18.19V

Click to expand...

This basically confirms that the Cromemco document is correct and it is close to +/- 2V.

Therefore, since out Kraft units ,at least the ones I have (check yours) appear to have a tad more output voltage deflection, we should power them from +/- 10V.

To help confirm this can you repeat the the tests with the Kraft Joysticks, but powered from 20V not 22V.

Sure here are some more measurements for all in both 20V & 22V just to be complete:

JS-1

Kraft

nullvalue said:

Walt, the goal of this build was to stay as authentic as possible. For the buttons, we're going with the Cherry switches which are panel mount, therefore they won't need to be on a PCB. We may build out a PCB for the other components, not sure yet.

Click to expand...

Hi, Your goal is to build as authentic as possible, then the project still requires a DAZZLER board that doesn't need lifted IC PINs, added jumper wires and a joystick circuit board design that is accurate to the original too.

With the circuit boards correct, then anybody can build them up any way they want. That's my point, accurate to the original boards ... and make the Gerber files public and probably pass them onto Todd Goodman for future distribution.


.

wperko said:

Hi, Your goal is to build as authentic as possible, then the project still requires a DAZZLER board that doesn't need lifted IC PINs, added jumper wires

Click to expand...

I'm not sure what you're referring to here. The Dazzler board doesn't need any lifted IC pins (that was my RAM board that happened to be slightly incompatible with the Dazzler). The only jumper wires necessary on the Dazzler that I built were due to the PCB's basically being a prototype. But Gary & Ansgar have aleady made updates to their PCBs & Gerbers so any new runs of that board will not need any jumper wires.

wperko said:

a joystick circuit board design that is accurate to the original too.

Click to expand...

Not sure if you saw the inside of the JS-1, but there is no circuit board at all in the original. Just a tag strip and a few inline soldered components. I was just considering using one in this instead of the tag strip to make it easier and more straight-forward to assemble.

wperko said:

and make the Gerber files public

Click to expand...

Definitely plan to make everything here public. I'll start a GitHub when we're done to share all any/all files.

There is one thing about this unit that I'm not fond of, the cable hanging out of it. For some reason this bugs me, the dangling cable dilemma. It might just be an OCD I have, all of my lab equipment that had line power cords hanging out, I fitted IEC line panel connectors !

I think ideally it should have a connector.

Not a wonderful scenario as it is a 12 core cable, a D25 connector is a bit big and bulky and the cables are thick (often), a good choice would a D9 because existing cables can be quite thin and flexible, but what to do about the required number of conductors ?

Also I think it would be better to supply the Joystics pots with a precision regulated voltage, those zeners are "ok" but 10V regulators would be better.

I was going to add an extra 10 way lug strip to my unit anyway for the extra transistor required to run the 8 Ohm speaker. Looking at it I decided that the +5V and -5V feeds could be eliminated (getting rid of two cable conductors) and derived from the 10V supply, if IC regulators were used. So I re-designed the arrangement to eliminate the zeners too.

These are the lug strips I am going to use they generally mount on spacers, but many kinds would work:


If the +/-5V supplies are derived from the 10V supplies in the unit, in total 10 conductors are required.

Interestingly , this still works for a standard D9 cable because of the outer shell & cable sheath is another conductor and when they are screwed up with the thumb nuts a very solid connection is made. It is also good that this results in grounding of the control box, all metal objects attached to line powered gear should really be earthed. (The computer's supply common should be connected to line power earth, it is in my SOL-20 and in my PET).

I have attached the layout I am going to use.

It appears the +/- 10V supply is suited to the Kraft Joystick, however if a tad more output swing was required the 7810 regulator could just be swapped for a 7812. The 78L series 100mA regulators could also work as it is a low power affair, but the bigger ones are much easier to wire to the lug strip.

I get what you're saying, but DB9 are just so big & bulky, would you really rather have that hanging out the back of a joystick? Literally every other joystick I own (probably 30+) just has a dangling cable. It's just how joysticks are an always have been i guess. I have no problem with it.

nullvalue said:

I get what you're saying, but DB9 are just so big & bulky, would you really rather have that hanging out the back of a joystick? Literally every other joystick I own (probably 30+) just has a dangling cable. It's just how joysticks are an always have been i guess. I have no problem with it.

Click to expand...

DB9 bulky ? Not compared to the size of this box.

Well as least the changes only requires a 10 conductor cable, likely thinner than a 12.

Hugo Holden said:

Interestingly , this still works for a standard D9 cable because of the outer shell & cable sheath is another conductor...

Click to expand...

Is it? I thought it was fairly usual, or not uncommon at least, to ground the shield at only one end of a cable.

Hugo Holden said:

DB9 bulky ? Not compared to the size of this box.

Click to expand...

Yes, if a DB-25 is considered "bulky" a DB-9 would be equally bulky. But I've never even seen a DB-9; most people needing just 9 pins use a DE-9 instead.

cjs said:

Is it? I thought it was fairly usual, or not uncommon at least, to ground the shield at only one end of a cable.


Yes, if a DB-25 is considered "bulky" a DB-9 would be equally bulky. But I've never even seen a DB-9; most people needing just 9 pins use a DE-9 instead.

Click to expand...

"Equally Bulky" ?? a 25 pin plug is about 2 1/4" wide !

All of the 9 pin cables I have , the metal shell is connected to the metal shell at each and the 9 pins are isolated from that,

Do you have commercial 9 pin cables that are different from that ? Can you check ?

The 9 pin connector is much less bulky than the 25 pin and a lot of them have thin flexible cables only about 6mm diameter or less (just less than 1/4").

I must say I'm baffled about the remarks that a DB-9 is "Bulky". Many of my cables that have this connector have only 4mm soft flexible cables and the connector body is only 1.25" long (about 32mm ) and 1/2 wide. That looks pretty small on the back of an enclosure that is 6.5" x 2.5" in size.

I have attached a photo of another suitable DB9 cable. The cable is only 5.6mm in diameter , soft and flexible, ideally suited to the job.

Hugo Holden said:

"Equally Bulky" ?? a 25 pin plug is about 2 1/4" wide !

Click to expand...

I was being a little tongue-in-cheek there, and I think you misread a single letter in one or two places in what I wrote.

Yes, a DB-25 connector is about that wide. A DB-9 connector (if such a thing even exists) would be the same width, because it's also using a "DB" shell. If you want a smaller connector, you need to switch to, e.g., a "DE" shell, which is a bit over a centimetre wide.

For reference, here is a comparison of the standard shell sizes.

Note that there can be different connectors with different numbers of pins using the same shell size. For example, the connector commonly called "VGA" is a DE-15 connector. It's quite different in size from a DA-15 connector, also used for video. And I have friends who have ended up with the wrong part because they weren't paying attention to what they were ordering.

Hugo Holden said:

All of the 9 pin cables I have , the metal shell is connected to the metal shell at each and the 9 pins are isolated from that,

Click to expand...

I am not surprised, since this seems to be the most frequent arrangement, but I am not sure that the standard specifies it. Perhaps it does, since RS-232 doesn't use differential signals, so there's the potential for a ground loop over the ground pin anyway. I don't really know enough about this to be sure.

But what I do know is that shields are not designed to carry signals, and thus are probably even worse to use as a ground reference. A moving ground might not be such a bad thing on an RS-232 connection, since the signals are digital and swing quite far past the ground in either direction (minimum ±3 V, but usually about ±12 V). But you're running analogue signals over this, which I think could be affected by even small changes in the ground level at the joystick vs. the computer.

cjs said:

I was being a little tongue-in-cheek there, and I think you misread a single letter in one or two places in what I wrote.

Yes, a DB-25 connector is about that wide. A DB-9 connector (if such a thing even exists) would be the same width, because it's also using a "DB" shell. If you want a smaller connector, you need to switch to, e.g., a "DE" shell, which is a bit over a centimetre wide.

For reference, here is a comparison of the standard shell sizes.
View attachment 1278360
Note that there can be different connectors with different numbers of pins using the same shell size. For example, the connector commonly called "VGA" is a DE-15 connector. It's quite different in size from a DA-15 connector, also used for video. And I have friends who have ended up with the wrong part because they weren't paying attention to what they were ordering.


I am not surprised, since this seems to be the most frequent arrangement, but I am not sure that the standard specifies it. Perhaps it does, since RS-232 doesn't use differential signals, so there's the potential for a ground loop over the ground pin anyway. I don't really know enough about this to be sure.

But what I do know is that shields are not designed to carry signals, and thus are probably even worse to use as a ground reference. A moving ground might not be such a bad thing on an RS-232 connection, since the signals are digital and swing quite far past the ground in either direction (minimum ±3 V, but usually about ±12 V). But you're running analogue signals over this, which I think could be affected by even small changes in the ground level at the joystick vs. the computer.

Click to expand...

I cannot agree:

The DC signals from the Joysticks are derived from a Thevenin resistance of 2.5K and not prone to interference, nor are those signals from the switches.

From what I can tell the connection between the shells of the typical DE(or DB) or D9 connector, is via the shield of the cable around the 9 internal conductors. It has a lower resistance per meter than any of the internal conductors, at least if the cable is a shielded type. And, if it is not a shielded type, no matter. Then the shells are just linked by one of the cable conductors, it would be no worse than the single common conductor in Cromemco's 12 core cable, which carries all of the ground currents, which in the application are very low. It also helps that the joysticks are powered from a differential +/- voltage supply.

The highest earth/common current is in the speaker return and at max continuous audio volume is only in the order of 60 to 70mA. Most of the time it would be in the range of 20mA. I think if Cromemco were worried about a high resistance in their ground conductor in their 12 core cable, they would have used more than one conductor, but already they were up to 11 or 12 conductors.

But, like most of these things, it requires an experiment.

I set up my current limited lab power supply and applied current to the metal shells of a standard RS-232 style 9 pin cable. This one is 2 meters long (about 6 foot). I wound up the current to 300mA, Testing with the DVM the voltage drop across the cable was 0.055v, suggesting the resistance is 0.18 Ohms. It is perfectly satisfactory. The speaker currents at worst would cause a 10mV offset.

Maybe your concern would be valid if the cable was 60 foot long.

Just to be clear, these are the sort of panel mount connectors I am referring to, they come in male & female versions and fit the standard 9 pin cables.

www.jaycar.com.au/db9-male-connector-solder/p/PP0800?pos=2&queryId=2f1e6fa41349a509ce6f4b000d6e7d70&sort=relevance&searchText=DB9%20connector

The straight through cables are quite cheap, even in the 3m lengths too. They come in M:F of F:F commonly

I understand your desire to add a connector, but I don't mind the cable coming directly out of the the enclosure. In fact, I think the unruliness of two JS-1's being directly connected to the chonky edge connector for the D+7A just adds to the "charm". Let's just keep the design flexible enough to support adding a connector if you want to.

Hugo Holden said:

The DC signals from the Joysticks are derived from a Thevenin resistance of 2.5K and not prone to interference....

Click to expand...

So if the level of the ground changes, this won't change the values you read? Ok.

(Though, come to think of it, this particular device, unlike a modem or similar, doesn't have any other connection to ground, so a ground loop is unlikely.)

Hugo Holden said:

Just to be clear, these are the sort of panel mount connectors I am referring to, they come in male & female versions and fit the standard 9 pin cables.

Click to expand...

Right. Those are mislabeled; they are clearly not DB-9 connectors, or DB-anything connectors. (See diagram above, and note how different that is in size from the DB connector in the diagram.)

nullvalue said:

I understand your desire to add a connector, but I don't mind the cable coming directly out of the the enclosure. In fact, I think the unruliness of two JS-1's being directly connected to the chonky edge connector for the D+7A just adds to the "charm". Let's just keep the design flexible enough to support adding a connector if you want to.

Click to expand...

Yes it is kind of charming.

What I will do for mine is get one 5m long or similar F:F or M;M RS232 cable (and cut it in half) since they are cheap enough and the cable good quality and wire the cut ends directly into the edge connector for the D+7A board, this way the two control units can be separated if they have the 9 pin panel connectors on them, and the cable kept separate too.

It can be difficult to find short lengths of multi-core cable (meaning less than 5 meters or so) that is relatively thin and flexible, especially in a 12 core format, Ideally the cable is in the range of 4 to 6mm OD. Otherwise it can be quite stiff and awkward. At least adding the regulators inside the box cuts the required cable back to 10 core, which essentially is what is used on most RS232 interconnect cables which gives more cable options because there are so many pre made RS232 cables out there, and it can actually be cheaper to buy that and cut off plugs to get the cable than buying actual cable itself per meter from a supplier.

On the naming of the panel connectors and plugs, since they are D connectors, I just call them by their pin numbers like D9 , D25 etc, some companies call the panel connector shells DBx etc but still call the plugs Dx, why I am not sure. If you search ebay for "DB-9" connectors & DB-9 plugs there are thousands of listings. I think DB9 must be an abbreviation for D Sub 9.

I checked all of the cables in my collection, they all have the shells connected , they are all 10 conductor cables or cables with a conductive sheath (have to cut them to find out). None have brands, some are much more compact at the plug end , like the one in the attached photo, most have cables in the vicinity of 5.6mm OD some around 4mm. One problem on ebay, most of these products are unbranded. Searching RS232 on ebay cable yields 10,000+ hits. Some are things like adapters & null modem cables.

I would be interested, if anybody with a commercial RS232/DB9 cable like these, has one where the metal shells are not connected together ?

Further to this, testing more cables of the same length of 2m, with a 300mA current, the link between the shells drops around 50 to 60mV, but for the conductors on the pins, they drop about 110mV at that current, so the connector shells are most certainly connected via the cable shield.

nullvalue said:

I'm going to pick up some clear spray today to see how well it holds up

Click to expand...

FYI I got some clear gloss spray, and I'm really glad I did this test first. I really don't like the shininess that that gloss adds. I think I will go with a matte clear finish instead. Waiting for it to dry to test the durability, but the spray did not affect the transfer (no color bleed or anything) which is good at least.

Hugo Holden said:

I think DB9 must be an abbreviation for D Sub 9.

Click to expand...

Sigh. No. "DB-9" means "D-shell," "shell size B," 9 pins. It's just that random people (including parts vendors!) who didn't understand that "B" means a specific shell size started applying the "B" designation to "E" shell sizes. I already posted an image showing shell sizes A through E.

Nobody can stop you from using explicitly wrong technical terms, just as nobody can stop me from calling a PC a "hard disk" if I wish, but both cases are likely to lead to unnecessary confusion.

cjs said:

Sigh. No. "DB-9" means "D-shell," "shell size B," 9 pins. It's just that random people (including parts vendors!) who didn't understand that "B" means a specific shell size started applying the "B" designation to "E" shell sizes. I already posted an image showing shell sizes A through E.

Nobody can stop you from using explicitly wrong technical terms, just as nobody can stop me from calling a PC a "hard disk" if I wish, but both cases are likely to lead to unnecessary confusion.

Click to expand...

Yes it is all over the place on ebay, have a look at this German cable, the label in the photos, they called it a D sub 9:


In any case, there is no confusion about the connectors & cables I am referring to now, even if the terminology is all over the place.

@nullvalue wanted a pcb for this project.

After looking at this for a while, I tend to agree, it is the way to go. It is all about keeping everything as orderly and tidy as possible.

The original design was a little messy and it had the pull up resistors for the switches wired in a messy manner.

I decided to design a pcb as compact as possible. Also to reduce the conductors in the required cable down to 10. This means if one was so inclined an RS232 connector plug and cable could be used.(my preference but others might not want it)

The audio amplifier, which can run an 8 Ohm speaker, is fine running from +12v supply & gnd. All it means is that a coupling capacitor is required for the speaker without a split power supply. The +5V supply for the switches is easily created with shunt zener regulator from the +12V regulated supply, so there is no need to feed this box with the + and - 5 volt supplies.

I altered the design to allow for this. And also added some resistors so that the supply feeding the Joysticks could be altered in the range of 20 to 22V.

I made the pcb design "old school" (none of the thin tracks and auto router antics).

It is a single sided pcb, the attached photo shows the view from the top, it has one link required on the component side. I put in an extra pad for the transistor bases in case somebody wants to use TO-5 cased transistors like the 2N3053 and 2N4036.

I also used some 10uF axial capacitors (like the ones used on the Dazzler boards, on mine I used 10uF axial Tant caps and have some left over).

This arrangement only requires the one 7812 voltage regulator IC that sits flat on the pcb and its tab goes under one of the mounting screws. The board will simply sit on two mounting 4-40 screws & nuts with some 1/4" tall spacers. The board is so simple it lends itself to be home made, I'm going to make a couple of pcb's up for testing. Most of the pads sit on a 2.54mm grid.

(A number of speakers could be used with this arrangement, 45 Ohm, 32 Ohm, 16 Ohm or 8 Ohm. going below 8 Ohms is not helpful, above 45 the power output drops off)

I received the Hammond enclosures today, they have rectangular vent holes punched in the back and lower surfaces. Some of these will help to ventilate the speaker. I'm not sure yet if a few of these will require extending in length where the speaker mounts.

I'm checking the mechanical Joystick positioning/mounting, we may have to move the Kraft joystick a little off center toward the rear of the enclosure to accommodate it.

This is great! Will this need to be done in a Kicad project to generate the Gerbers?

nullvalue said:

This is great! Will this need to be done in a Kicad project to generate the Gerbers?

Click to expand...

I'll post just the track image and artwork. Probably this could be converted into a Gerber in little time ( by Gary maybe) in Kicad. I'll make a couple of boards by hand as well. I cannot see any issues, never hurts to double check everything.