• Please review our updated Terms and Rules here

More PDP-8/X blank OMNIBUS PCBs

daver2

daver2

10k Member
Joined
Jun 19, 2012
Messages
15,340
Location
UK - Worcester
Continuing with what Tom has started (and with many thanks to Roland and Don for their hard work) I am planning to do a run of some more PCBs as follows:

RX8E (RX01/RX02 OMNIBUS interface).
32KW RAM card.
H851 top connectors (PCB only - no physical connectors).
RX01/RX02 emulator (requires the addition of an Arduino Mega 2560 card).

All from Roland's GitHub repository at https://github.com/Roland-Huisman?tab=repositories.

My plan is to do a run of 5 cards (minimum from JLCPCB) with an ENIG gold finish (no hard gold for the connectors unfortunately).

Keep one of each PCB for myself - and sell the remaining off for cost + postage and packaging.

Quite happy to increase the order quantities (in units of 5) if there is a high demand for one (or more) types.

Let me know your interest and I will get a price and we can take it from there. I will be posting from the UK (as that 'feature' has now disappeared from the forum).

For the avoidance of doubt - the items supplied will be blank PCBs with an ENIG gold finish with no furnished components or connectors.

Dave
 
Hi Dave,

I would be interested in the following:

1 x RX8E (RX01/RX02 OMNIBUS interface).
1 x 32KW RAM card.
1 x RX01/RX02 emulator (requires the addition of an Arduino Mega 2560 card).

Could you please work out prices including shipping similar to what I have done and I can pay by Paypal.

Best regards
Tom Hunter
 
I'll post this here from my blog entry on the RX01/2 emulator:

As of 15-Jun-2021 I still have a number (> 10) of the bare emulator PCB fabs available; they are US$10 each. SparkFun microSD breakout/adapter boards are available for US$5 each (and are also available direct from SparkFun). Also a kit of all discrete parts needed; cost is US$15. So US$30 for the whole set, just supply your own ArduinoMega2560, a MicroSD card for data storage (a small 4GB one is more that enough storage) and of course you need an interface board for your system (RX8E/28, RX11/211, RXV11/21). Send me a PM if interested in a purchase.

An additional comment: I have shipped boards/kits to both Europe and Australia before, and the cost for shipping even a small package is substantial, in many cases (ie, Australia) dwarfing the cost of the full kit at $30. So for any requests from areas outside the USA it would make sense (to me) to have a single person be a focal point for a group of orders to amortize the shipping, receive a bulk order, and then redistribute. Just a thought.

At the present time 6/17/2021 I have (9) full kits (pcb+parts+microSDadapter), (22) microSDadapter, and (19) bare PCB that are available.
 
Last edited:
Cheers,

That sounds a better solution than reinventing the wheel...

Can you put me down for a complete kit and I will send you a PM tomorrow regarding payment and shipping?

Cheers,

Dave
 
daver2 said:
Cheers,

That sounds a better solution than reinventing the wheel...

Can you put me down for a complete kit and I will send you a PM tomorrow regarding payment and shipping?

Cheers,

Dave
Click to expand...

Actually the shipping cost is a problem.

Dave if you supply all the boards at cost plus shipping, then that is substantially cheaper then if I get most boards from you and just the RX01/02 emulator from Don because I have to pay shipping twice which is the expensive bit. I am almost better off to just order all the PCBs myself from JLCPCB.

Also having the components already included may mean that for the remaining components I am under the US$50 free shipping treshold of Digikey so I will have to pay US$20 for the component shipment which would otherwise be free.

Best regards
Tom Hunter
 
Last edited:
Just on the 32KW RAM card - Roland confirmed that his board does not suffer from the glitches Vince sees on the combined RAM+M847 board when using the current 62256 SRAM parts made by Alliance (the no longer available original Cypress chips were fine).

Best regards
Tom Hunter
 
thunter0512 said:
Just on the 32KW RAM card - Roland confirmed that his board does not suffer from the glitches Vince sees on the combined RAM+M847 board when using the current 62256 SRAM parts made by Alliance (the no longer available original Cypress chips were fine).
Click to expand...

Hi Tom, just to be clear... I don't have experience with the Alliance chips. The only problem which I have is that Vince his board does not work with the memory chips which I have working flawless on my own RAM boards... I've contacted Vince a while ago about this issue but I just put my own RAM boards back into my machine. I have not spent any time on figuring out what the problem is... But I could try to order two Alliance chips to see how these work in my board.
 
Tom, I would be interested in 2x 32k boads and sharing shipping costs to Australia, if that tips the equation one way or another.
Leon
 
Roland Huisman said:
Hi Tom, just to be clear... I don't have experience with the Alliance chips. The only problem which I have is that Vince his board does not work with the memory chips which I have working flawless on my own RAM boards... I've contacted Vince a while ago about this issue but I just put my own RAM boards back into my machine. I have not spent any time on figuring out what the problem is... But I could try to order two Alliance chips to see how these work in my board.
Click to expand...

Hi Roland,

Sorry I misunderstood what you wrote.
What exact SRAM chips are you using which work reliably (Manufacturer, model and date stamp)?
It would make sense for us to buy the exact same chips rather than experiment.

Thanks and best regards
Tom Hunter
 
Roland Huisman said:
I don't have experience with the Alliance chips. The only problem which I have is that Vince his board does not work with the memory chips which I have working flawless on my own RAM boards... I've contacted Vince a while ago about this issue but I just put my own RAM boards back into my machine. I have not spent any time on figuring out what the problem is... But I could try to order two Alliance chips to see how these work in my board.
Click to expand...

Roland, I just looked at your board schematic for the first time and I don't see any changes from Vince's schematic except for layout. Did I miss something? I spent a lot of time looking at Vince's design and at the original Lafferty design and was never able to come up with a definitive reason for the issue. There are several of DEC's omnibus design guidelines that are violated but none of them by themselves should cause the problem seen with the Alliance parts and not with the Cypress parts.

On Vince's board one of the rules that is violated is the signal distance from the edge connector. The guideline is to keep them less than 6 inches (15 cm). This is really the only thing I can see that is different between the two layouts. On all the designs the address lines should have been buffered because they are driving more than one load (the two memory chips). This is not a problem for the memory board if it works but could cause instability somewhere else on the bus. My belief is that the real problem stems from the logic levels being slightly different between the different vendor's parts. The levels on the omnibus open collector drive are adequate for either part but we are seeing quite severe ground bounce at the memory chips at a critical time and I think the logic levels the memories are seeing is not good enough at that moment. It is almost good enough because the board will boot and run, often for several minutes before the memory test finds a problem or crashes completely.

The chips that are driving the omnibus should have been open collector. With the way it is now the 74LS245's could be attempting to drive the bus to a logic high while the CPU is trying to pull the bus to the logic low state. If this happens you can see high transient currents in the 74LS245's and the open collector drivers on the CPU or other bus master. If all drivers are open collector then even if several devices are trying to talk at the same time you never see these transient currents because only the passive pull up resistors on the terminator card are pulling towards a logic high and this is the norm for them. I don't think this is the issue causing the difference we are seeing in the memory chips. The bus master should be listening for a long time and not driving the bus.

The last place I think could be a problem is during read when the memory chips start to drive their local outputs but the 74LS245 is also driving those same lines as if it is trying to write memory. They are driven by the same signal which would be mostly fine if they were the same logic types. This is the memory read signal. It feeds the direction pin on the 74LS245 and the Output Enable pin of the memory chips. Since they are not the same logic family they have different timings and there is a brief interval where they can be driving pins in opposite directions the result being high currents inside these parts. Keep in mind that we are talking about only a few nanoseconds at most where these high shoot thru currents exist. I looked at DEC's semiconductor board layouts to see how they addressed this issue but they didn't have to since the memory chips they used have separate pins for input and output bits. If this is part of the issue it could probably be addressed by adding in a small additional delay in driving the output enable line on the memory chips. You want it to turn off immediately but be delayed during turn on. This is probably the source of the ground bounce we are seeing. Probably is the key word here as we were not able to isolate this. The norm is that tri-state drivers take longer to enable than disable to avoid this problem. The expectation is that all devices used have similar delays.

I am looking forward to hearing the answer to Tom's question about what parts you found that work with your layout but not with Vince's. Another data point!
 
Hi Doug,

The RAM schematic is identical. Vince was out of stock back then and I just wanted to have a cheap board. So I reduced it to 1/3 of the original board size. That is why I made the small version. The layout is completely different as you said. I kept the chips as close as possible to the bus and I added a ground plane. The RAM chips are Cypress versions for sure. I've sent one of my 32kw boards to m_thompson so maybe you can borrow that board to do some tests?

Regards, Roland
 
Hi Roland. If you get a chance could you copy down or just take a photo of the exact part number of the chip that doesn't work on Vince's board but does on yours. I expect that the ground plane and shorter runs might be what makes your board work. We tried adding additional grounding but that is far from the same as a ground plane layer in the board. Vince's board almost works with the currently produced chips. Even a tiny improvement would probably push it over the edge to functionality.
 
I have still a few blank boards left if anyone wants them:

1 x M869R US$ 17.00
3 x M8357R US$ 15.00 each

I sell these at cost.

Shipping is extra for up to 500 gr (2 boards):

USA: US$ 18.00
UK: US$ 22.50

Best regards
Tom Hunter
 
You must log in or register to reply here.
Back
Top