Jump to content
IGNORED

838-PEB


Shift838

Recommended Posts

4 minutes ago, Shift838 said:

 

I'm thinking for the card guide a simple 3D printed guide to bolt into the same holes as the standoffs just with longer screws.

 

for your second question maybe @InsaneMultitasker could answer if it is even possible.

If the question is related to maximum slots, I don't know of any restrictions though others may chime in here that have experience with extender boards, etc. 

Link to comment
Share on other sites

5 hours ago, atrax27407 said:

I ran two P-Boxes using one of the Y-cables from the console. One P-Box on each. The only restriction was that the Y-Cable length must be kept to a minimum. Jim's side-port splitter works great.

I thought I remembered Jim mentioning on the 4A, the Flex Cable card was limited to a total of 8 devices on a single PEBox bus due to an issue with transceivers(???) but that with the splitter, it was no issue to add an extra PEBox.

 

What I did not know was whether with a Geneve system if more slots were possible?  I just did not know if this was a 4A only issue, or was also a Geneve issue as well.


Beery

  • Thanks 1
Link to comment
Share on other sites

55 minutes ago, atrax27407 said:

As long as the cards have different CRU addresses, you can utilize 16 slots. There just need to be an interface card in each box. So one could have as many as 14 cards.

That's for the TI which I understand.  What my original question was, could a 838 PEBox be designed to have more than 8 slots to work with the Geneve if the case would handle the larger size?  The Geneve does not allow an interface card in the PEBox.

 

Beery

  • Like 2
Link to comment
Share on other sites

23 hours ago, 9640News said:

That's for the TI which I understand.  What my original question was, could a 838 PEBox be designed to have more than 8 slots to work with the Geneve if the case would handle the larger size?  The Geneve does not allow an interface card in the PEBox.

 

Beery

If the Geneve does not have a limitation of 8 slots, which I don't think it does.  Not sure if it has a limitation and if so what that is.

 

as far as the ATX case, given the internal drive cages the how I am mounting the connectors the max really is 8 slots.  I may be able to push it to 9 but thats top.  thats because regardless of the ATX case, (mid tower, full tower) i believe the depth is the same.  flipping the connectors to mount the cards is not feasible as many of the cards like the Geneve have the overhang that is large like the IDE, SCSI, etc..

 

another solution i thought of last night for this very issue would be to add a 60 position male header as a expansion slot.  you could then place the 838 PEB's next to each other and using a short cable with a female 60 pin IDC connector (the shorter the better the signal) you could chain them together.

 

I believe that would possible work.  I can test the theory once I get the boards in. (I have ordered the first prototype).  The prototype does not have that though.

 

or maybe even design a 60 pin breakout for the PEB slots (I already have that) and wire it up on a cable on each end and plug in slot 1 of each.  You would loose 2 slots but you would have a total of 14 to deal with then.

 

just a thought though.

 

 

 

pebbreakout.JPG

Edited by Shift838
  • Like 1
Link to comment
Share on other sites

1 hour ago, Shift838 said:

 

 

or maybe even design a 60 pin breakout for the PEB slots (I already have that) and wire it up on a cable on each end and plug in slot 1 of each.  You would loose 2 slots but you would have a total of 14 to deal with then.

 

 

I had an inquiry about this a month'ish ago wondering if two of your PEBox cards with a connecting cable would be able to accomplish this very thing.  I think at 14 cards, that would be enough <grin>.

 

Beery

  • Like 1
Link to comment
Share on other sites

The slot limitation is solely a matter of the driving power of the transciever chips (74LS244 and 74LS245 in most cases, although some boards also use 74LS373s). A chip can safely drive up to seven receivers, which would be 8 cards (1 driver and 7 receivers). The splitter gets past this by having the console drive two receivers (instead of one), feeding two PEBs, each driving 7 receivers from the Flex Cable Interface. Cable length is then limited by the tolerable timing delays of the system--shorter cables are somewhat better here.

  • Like 4
Link to comment
Share on other sites

The limitation on how many cards can be driven: the driving card with LS244 and LS245 can comfortably drive 8 other such chips. These would be the driver chips in the Geneve, or the flex cable. 

 

The trace length in the backplane is also a factor but I don’t think you see it with 3 MHz. 
 

If your peripheral cards had their driver chips replaced by the HCT or ACT versions, you could have more peripheral cards. The CMOS family (the C letter) does not suck in current from the driver, like the standard LS parts do.

 

 

You need the TTL compatible family (the T). The HC versions aren’t the same voltage levels. 
 

You see a lot of HCT chips replacing LS. The switching speeds of replacements are sometimes slower, sometimes faster—but lower power and less cost. 
 

The calculations are given in detail in the E-Bus systems design book. 
 

The fan out of a LS244 can be 8 other LS chips, or 30-100 chips of the HCT family. (Compare data sheets, lookup LS244 I_OL and I_IL then HCT244 I_IL.)
 

 

  • Like 1
Link to comment
Share on other sites

Oh, and I think I’ve heard to solder another LS244 on top. Double the drive power. Not sure where I might have got that. That could be the way to go, if you need 16 slots. 

In extending buses: I’ve seen an S-100 card cage where the backplane has jumpers to another backplane. It’s for a 1MHz Z80A, so no high speeds. 
 

Another issue with a longer bus is that the signals can arrive more out of phase. Eventually you randomly lose a bit. Probably not an issue at 3MHz, but it starts to be at 10MHz.

Worst case, you get a driver  bogged down and having a slow rise time (transition from low to high) and “metastability”. You’re screwed. 

The TM990 backplane manual also gives equations for driver loading. Beyond 4 cards, you need some terminating resistor packs. Another technique is “active termination”. 
 


Out there, you see a lot of equipment with no more than 8 slots (parallel.) The 12 and 16 get real expensive. PCI, PXI, and NuBus for instance can cover 16 slots, but in practice you see just 4-8 in a chassis. PCIe is a whole different world. 

Geneve2020 conceptually has 16 slot IDs (CPU is one) but I’m only building an 8 long. 
 

I’m still learning about this stuff. 
 

  • Like 4
Link to comment
Share on other sites

13 hours ago, FarmerPotato said:

Oh, and I think I’ve heard to solder another LS244 on top. Double the drive power. Not sure where I might have got that. That could be the way to go, if you need 16 slots. 

In extending buses: I’ve seen an S-100 card cage where the backplane has jumpers to another backplane. It’s for a 1MHz Z80A, so no high speeds. 
 

Another issue with a longer bus is that the signals can arrive more out of phase. Eventually you randomly lose a bit. Probably not an issue at 3MHz, but it starts to be at 10MHz.

Worst case, you get a driver  bogged down and having a slow rise time (transition from low to high) and “metastability”. You’re screwed. 

The TM990 backplane manual also gives equations for driver loading. Beyond 4 cards, you need some terminating resistor packs. Another technique is “active termination”. 
 


Out there, you see a lot of equipment with no more than 8 slots (parallel.) The 12 and 16 get real expensive. PCI, PXI, and NuBus for instance can cover 16 slots, but in practice you see just 4-8 in a chassis. PCIe is a whole different world. 

Geneve2020 conceptually has 16 slot IDs (CPU is one) but I’m only building an 8 long. 
 

I’m still learning about this stuff. 
 

Can you replace LS244 with HC244 ?

(I am reaching way back but I think HCmos can drive more current than LS TTL)

 

Here is quote from the web:

The  MC54/74HC244A  is  identical  in  pinout  to  the  LS244.  The  device inputs  are  compatible

 with  standard  CMOS  outputs;  with  pullup  resistors,they are compatible with LSTTL outputs.

This  octal  noninverting  buffer/line  driver/line  receiver  is  designed  to  beused  with  3–state 

memory  address  drivers,  clock  drivers,  and  otherbus–oriented  systems.  The  device  has 

noninverting  outputs  and  two active–low output enables.The HC244A is similar in function to

the HC240A and HC241A.

 

  • Like 2
Link to comment
Share on other sites

@TheBF 

 

the HC series will output high and low to the other families, but it’s input voltage threshold is higher than LSTTL provides. That’s why the article says to use pull-up resistors. IMHO this is a kludge. 

Here’s a quick guide from TI. See chart on page 4.  I like to use the compatible LVC parts especially LVC245, after looking at the data sheet parameters.
sdyu001ab.pdf?ts=1651782206324&ref_url=h

 

In LSTTL it’s the low input state that sucks power. The output driver has to “pull” those inputs down. Overload the driver, and it might get there too slow (it’s draining charge held on the transmission line/trace) or not at all. 

HCT is cheap. You only need HC drivers if you’re interfacing to other CMOS chips.  


 


 

 

  • Like 3
  • Thanks 1
Link to comment
Share on other sites

  • 1 month later...

Since I successfully finished the 838-IO Plus run, i pulled out the prototype boards for the 838-PEB to start working on it.  I'll be trying to post updates weekly if I can.

 

Prototype Board:

IMG_0464.thumb.JPG.275f69509f60b20428b8869d5503e0fb.JPG

 

It will fit in a standard ATX case (i'm using a mid tower) with all mounting holes lined up.  I'll probably be fully populating it this weekend and getting it into the ATX case for testing.  Until I order a new set of boards I will just run the +5v Standby from the bottom of the board.

 

As you can see I have a DB37 connector on the back of the board that will come out of the slot where all the connectors for a PC mother are.  This connector then connects over to a IDE 40 pin that will connect over to the 838-IO Plus card via a small DB37 to IDC40 cable.

 

I also have plans for new blanks for the PC board slots on the case to be able to hook up VGA for the F18A, DB25, etc.  Working on something for the Geneve too for the slots.

 

I have tested all continuity, all power is provided when the power supply is engaged for the correct pins.

 

Of course, I did find an issue with keeping the power supply on the ATX engaged.  I completely forget it use a momentary switch instead of a latching.  a PC has a circuit already to act as a latching to keep the ATX PSU engaged.  So I have designed a circuit that does the same thing.  I will need to add a couple of headers for +5V Standby and another Ground to break it out on the 838-PEB board as I will just be creating a small board to use a few jumper wires to hook it up so that if I ever work on any improvements on the circuit and since it will not be built into the 838-PEB PCB, then it can be swapped out easily. 

 

This also allows for users that want to swap their momentary switch in a ATX case with a new latching or even just add a new latching switch to the case somewhere and just hook directly up to the 838-PEB board without needing this  additional board.  It will be up to the user to decide the route they want to go..

 

I decided to make my own test PCB for the circuit.  Now I have not had much luck following various guides to do a homemade PCB.  But I found one that works for me.

 

Since the etching solution only removes exposed copper, I figured I would use my wife's Cricut cutter. 

 

I designed the circuit with large enough traces to be easy to weed out.  Some users have actually cut the copper off of the board from around the traces using a 'Posterboard' setting on the Cricut cutter or a DIY milling machine.  I may try that next time on the Cricut cutter. 

 

This time I printed the PCB traces to a PDF in a mirror image as I was only using the bottom for the traces.  I then converted it to a PNG image file and made it transparent.  Loaded the PNG into the Cricut Design Space software and then set the height and the width to the correct size measuring the dimensions for the furthest outside traces from outside edge to outside edge (this ensures you get the correct size for the component footprints - hope to find a easier solution next time).

 

Here are my results:

 

PCB Traces printed and weeded out on vinyl:

IMG_0456.thumb.JPG.3e02b3cb2d780bd853b29bc29d474ca9.JPG

 

Transferred a PCB (did not want to cut it down as I have quite a few of these and just wanted to test the result)

IMG_0457.thumb.JPG.e0fb42741f6a29c833c2ad42373703ae.JPG

 

After etching and before removing the vinyl.

IMG_0458.thumb.JPG.ff4250891882d5c71f661c775c8d6f10.JPG

 

vinyl removed:

IMG_0459.thumb.JPG.e96e5456f29e55f2ea77fd8996659fb5.JPG

 

Carefully marked where i needed to drill using a spring loaded center punch:

IMG_0461.thumb.JPG.765e188e7867206548cd2b06e4a61447.JPG

 

Components populated and soldered up with my hen-scratch.  Board tested and it all works.

IMG_0463.thumb.JPG.9d77494e1f274ccae4a9b2bfb021f611.JPG

Edited by Shift838
  • Like 8
Link to comment
Share on other sites

i went ahead and just replaced my momentary switch in my PC case with a latching for now. 

 

Like I have said you have options:

 

  • replace the switch with a latching as I have done.  Most of them use an 8mm x 8mm momentary like the one in the link below but measure yours to be sure they also come in other sizes like 7mmx7mm.  I may even offer to purchase them prewired with long header wires.

uxcell 30 Pcs 6Pins Square 8mmx8mm Latching Type DPDT Mini Push Button Switch - - Amazon.com

 

  • Use a completely different latching switch, drill a hole and mount and wire it up.
  • Use the circuit I designed to keep the signal engaged for the momentary until it is pressed again.  Acts as an On/Off switch for each press of the button.

 

Link to comment
Share on other sites

Tonight i decided to go ahead and partially populate the 838-PEB Prototype board with a couple of 60 pin edge connectors, the IDC40 and DB37 female to be able to at least test with the 838-IO Plus card and a TI 32k memory card to make sure it was functioning.

 

after populating partially and verifying that the correct pins were getting the correct voltages i plugged in an 838-IO Plus card and TI 32k card.  Turned the power switch on to the ATX case and then the 99/4A.  F18A screen error!

 

So how the 838-IO Plus card is hooked up.  It plugs into slot #1, then a small cable (DB37 male to IDC40) connects from the Card to the 838-PEB board.  Removed the cable that hooks to the board and hooked the standard DB37 Male to DB37 Male from the card to the TI and it works.  

 

So is it my cable... No and yes.

 

what I did wrong, is how I wired up IDC40 on the board.

 

So the DB37 counts normally of course.  but.. the IDC connectors.  All odd pins are on one row and even pins on the other.   That's where I messed up.  as using ribbon cables the pins will not be correct.  I rewired the DB37 connector manually and it works.  So something I have to correct on the final board.

 

But it is working.

 

Tomorrow I will populate the rest of the board and get a FDC, 838-Speech and TiPi installed to do my standard testing.

 

Look for final pics soon..

  • Like 4
  • Thanks 1
Link to comment
Share on other sites

so I had some time to work on populating the board.  I also worked on printing the PEB card rail system that @wierd_w designed.  I did have to change a couple of things on his base design but he did a great job making it modular and being able to mount it to existing ATX motherboard footprint mounts.

 

components populated and mounted in ATX mid-tower:

IMG_0465.thumb.JPG.81d4fcd6dd52501f657879aa529fe46f.JPG

 

Front and Back rail system installed:

IMG_0469.thumb.JPG.72eb369f4cc588b8300a45ee500477f5.JPG

 

838-IO Plus Red Version, 838-Speech Adapter, TiPi, TI 32k and Myarc FDC installed

 

  • TiPi is in slot #8 so it will not take up 2 slots and does not interfere with drives mounted.
  • 838-IO Plus in slot #1.

IMG_0470.thumb.JPG.d2eb6d2c75c1bcc5740874297de8ca13.JPG

 

Notice the rail system fits nicely for new cards without the shell like my 838-IO and 838-Speech card.  It also fits the new HRD4000B perfectly.  Lines up with enclosed cards as well.

IMG_0471.thumb.JPG.3c4d1d99e947744eacc2923a6b52c8a2.JPG

 

IMG_0472.thumb.JPG.2e93232099c3b47b20bfc58caa0bfb4a.JPG

 

I do have 1 or 2 surprises for the board coming out as well that I thought of while I was putting it together.  I even got one of them proved out and working already.

 

You can see it partially in a couple of the photos showing just the rail system.  I have designed a circuit to allow to hook up a Gotek to for the Rotary and OLED modification.  My system will not be close to me so I plan on using a RJ45 and a standard CAT5 network cable to be able to control them.  This way I can use a bigger display and just have it on my desk.  My plan is to integrate 2 RJ45 jacks on the board with headers for each jack for Rotary and Display based on Gotek 1 or Gotek 2 drives.

 

I have already ran through an entire test on this system with the current cards that I mentioned above.  

 

  • Copied over 200mb from one directory to another on the TiPi
  • Formatted, copied, renamed, deleted on the FDC
  • Tested the ram
  • Tested the speech

My 838-IO Plus has the 5 ft cable and a 15" cable that goes from the 838-IO Plus to the 838-PEB IO Plus connector on the board.  I have tested small cables to make the link.  a standard ribbon cable will not cut it.  It has to be a twisted paired ribbon cable like the one that is used for the 99/4A Side-port IO extension cable from IEC, just different connectors.

 

I'll work more on t his tomorrow.  I still need to route some cables, mount a couple of fans and some temperature sensors for my fan controller/temp monitor.

 

 

 

 

 

 

Edited by Shift838
  • Like 3
  • Thanks 2
Link to comment
Share on other sites

Right. This is good, This is how I roll too.

Did you have internal power for the pi? I didn't see it mentioned, but you might have mentioned it before and I just forgot that it was there. 

BTW, this looks great. I'll have to buy one around next tax season when all that govt money gets pushed back my way. Yeah ?. If everything is up to snuff and you have 1 Left over. I'm loving it. Really nice!!! Especially with power meter and temps in fahrenheit with glorious RGB fans. Bwhhaha 

Screenshot_20220613-054013-236.png

Edited by GDMike
  • Like 1
  • Haha 1
Link to comment
Share on other sites

48 minutes ago, SkyPilot said:

Looks nice, for the wealthy.  The new IO board alone is about $250.00, how much would a new P-Box cost?

I will only be selling the actual PCB, this way the user can decide on what ATX case and power supply they want to use.  it will keep cost lower.  Not sure what the price is yet.  The footprint is fairly large.  Also the best price so far I am finding before shipping just on the card edge connectors a bit over $6 each.

 

i'll work up a cost for the PCB and all the parts associated with it over the next few days.

Link to comment
Share on other sites

3 hours ago, GDMike said:

Right. This is good, This is how I roll too.

Did you have internal power for the pi? I didn't see it mentioned, but you might have mentioned it before and I just forgot that it was there. 

BTW, this looks great. I'll have to buy one around next tax season when all that govt money gets pushed back my way. Yeah ?. If everything is up to snuff and you have 1 Left over. I'm loving it. Really nice!!! Especially with power meter and temps in fahrenheit with glorious RGB fans. Bwhhaha 

 

I have not done anything for powering the Rpi as most of us that have the TiPi just leave the Rpi powered up all the time. The Raspberry pi requires a 2.5amp 5v power supply to be able to be powered up correctly and to be stable.  I thought about using the +5V Standby line, but after some research that line only gives a max of 1.5 amps (depending on which regulator was used) many ATX power supplies will only do up to 1 amp for that line.  So powering it up via that line is a no go and I would think that we just want to leave the rpi running.

 

I may be able to put in a USB port close to slot 8 to be able to plug a small microusb cable into and have it pop out the back to another port (maybe i said) then plug the actual rpi power supply into that port to provide the power. Just so we don't have to run a cable.  I will look into that.

Edited by Shift838
  • Like 1
Link to comment
Share on other sites

I will be putting 2 USB type A connectors on the board.  One near the back of the board and the other next to the 838-IOP DB37 on the board.  A user can use a USB cable to plug from a power supply to the USB A port that is mounted to the back of the board and then another USB cable to plug into their RPI from the USB A connector nearest slot #8.  This way the power is always on and there is no need to snake a cable inside the case.  It will only be used for power as only pin #1 and #4 hare connected.

  • Thanks 1
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...