PiDP-10 production update

[oscarv]

Hi,

I have been a bit silent recently about communicating about the PiDP-10 'release date'. Apologies for that, I seem to have a bit of a foggy brain the last few months. But never mind, progress is being made.

Here is the injection mold for the case, now finished:

And some test cases have been produced, tweaks and improvements added. So now the process is on hold as I finish the (hopefully) final PCB, back and front panels for the kit. 

I made a major revision to the PCB, taking the cues from Angelo and Rich that an I2C chip is not too elegant in a classic design. So I took Angelo's PiDP-1 circuit, which uses just a 74HC238 and -138. Much nicer.

All that should be done in two weeks from now, and then it's a matter of production. Maybe October is too close, so now I make a little retreating move: November. 

Many thanks to Lars, and Richard C, and again - apologies for my poor communication. The foggy brain is partly due to the simultaneous project of setting up this micro-factory in Panama to make the kits. Doing this in Panama was a silly decision (but rational, I love the idea of this adding some activity in that area I started to care about). Just for fun, this is part of the preparation: delivering bubble wrap all the way from China. It is strange but yes - we will use up this much bubble wrap in about 2 or 3 years. Ridiculous but true.

Kind regards,

Oscar.

[oscarv]

Oh, the bubble wrap delivery photo... a container full of the stuff! Silly hobby, this.

[DR]

My but  you have a big workshop and tool set!  You have a better
injection molding device than I do.

Thx for the  update and teasers.

[johntk...@gmail.com]

Wow it’s really happening!!

[Randy E (OoMOR)]

I'll be sure to wave when I pass through Panama City in December!

Randy

[Ken Hansen]

We'll, shipping the bubble wrap from China seems excessive, but odds are if you found a local source in Panama for the bubble wrap, they likely had it shipped by container from China...

I bet it's about the lightest container ever shipped!

Thanks for the update, it was great to meet you in Dallas, TX earlier this year,

Ken

On Sep 15, 2023, at 06:53, oscarv <vermeul...@gmail.com> wrote:

Oh, the bubble wrap delivery photo... a container full of the stuff! Silly hobby, this.

<bubble-wrap-truck.jpg>

--
You received this message because you are subscribed to the Google Groups "PiDP-10" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pidp-10+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-10/f9eb0d5c-b1b2-4e3c-9df2-5529982b1216n%40googlegroups.com.

<bubble-wrap-truck.jpg>

[Eric Bruno]

Found this FYI:

How much does it cost to ship a 40ft container to Panama, the average price to ship a 40ft container to Panama is from $880 to $4550. The shipping timeline to ship a 40 foot container to Panama ranges depending on the route.

Sent from my iPhone

On Sep 15, 2023, at 13:43, Ken Hansen <n2...@w5fc.org> wrote:

We'll, shipping the bubble wrap from China seems excessive, but odds are if you found a local source in Panama for the bubble wrap, they likely had it shipped by container from China...

To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-10/A4E222FF-2D6F-45F7-8F02-E537B2EB03D8%40w5fc.org.

[Andy]

Fantastic Oscar!  This is a huge endeavour... really appreciate this labour of love

[tszo...@pacbell.net]

I am interested in  Angelo's PiDP-1 circuit.  Where can I find a schemcatic?

Thanks,

Tom

On Friday, September 15, 2023 at 4:50:34 AM UTC-7 oscarv wrote:

[Andy]

Hey Tom - open https://cdn.hackaday.io/files/1919858213886240/blincolnlights18.zip and there's a file blincolnlights18.kicad_pcb that should give you the schematic you need.

cheers!

[a...@papnet.eu]

Actually not! My panel is mostly identical to previous PiDPs. The important change for the PiDP-1 was to use decoders to drive the rows instead of using GPIO pins for those directly. If you do the latter you run out of pins for a panel with as many elements as the pdp-1 or -10.

A picture of the relevant change is attached (addr0-3 are GPIO pins).

[AB]

Hey Tom - thanks for writing back - Yep, we are pushing the limits on those Pis! Be sure to keep us posted on your progress 😀

cheers, andy

--
You received this message because you are subscribed to a topic in the Google Groups "PiDP-10" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/pidp-10/GTYjOBVCjdY/unsubscribe.
To unsubscribe from this group and all its topics, send an email to pidp-10+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-10/632be173-bee4-42dc-b357-5b35d00463f6n%40googlegroups.com.

[Andrew Wasson]

Exciting stuff Oscar,

I've been following your progress on this project since early 2020. My KIM-UNO and original run PiDP-8/1 can hardly wait. I'm happy to be an enthusiastic enabler of this silly hobby. 

Cheers,

Andrew

On Friday, September 15, 2023 at 4:50:34 AM UTC-7 oscarv wrote:

[Mark Lawler]

I so need one of these new PiDP-10 kits when available.  When will you be accepting orders / pre-orders?  :D 

Best,

-Mark

[Richard C]

I have gotten KL and KS to display on PiDP10, there is some limits to what can and can't be done at panel switches. Such as being able to address all of memory, or boot machine.

Rich

[steve...@gmail.com]

Oscar and all,

Thanks for the status report. Looks like I will have to dust off my PDP-10 Site Preparation Guide and get my facility set up in time for delivery. It looks like quite a substantial undertaking:

http://bitsavers.trailing-edge.com/pdf/dec/pdp10/KA10/DEC_10-HAAB-D_SitePrep_May70.pdf

Hahahahaha.

-- steve

[Alan Abbott]

At the risk of sounding like a heretic, Has anyone used a non PI solution?

I have started to go off PI's as I am finding them not value for money and not as reliable as they were.

I have been looking as RISC V boards but no idea if a pos alternative .

The basic part (simh etc shouldn't be a problem but the GPIO bit? I am a hardware guy not software.

Alan

--

You received this message because you are subscribed to the Google Groups "PiDP-10" group.

To unsubscribe from this group and stop receiving emails from it, send an email to pidp-10+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-10/78419f63-6944-401b-86a6-8cf78adb3422n%40googlegroups.com.

-- 
"Those who do not learn from Dilbert are doomed to repeat it"

[Clem Cole]

below...

On Sun, Oct 8, 2023 at 2:26 PM Alan Abbott <alan....@gmail.com> wrote:

The basic part (simh etc shouldn't be a problem but the GPIO bit? I am a hardware guy not software.

That indeed is the issue.   OpenSIMH should work with any reasonable UNIX/Linux implementation, but ... it turns out RPI's SW GPIO interface is not the same as the "official" Linux interface -- although it's the most popular since most people use RPI's (since they had a lot of inexpensive Linux based boards with GPIO early on).  My analog is this is a little like in the 1970s when DEC marketing convinced people VMS Fortran was F77 ( it wasn't ).  However, most people wrote their code using the VMS extensions - so folks like Masscomp/Apollo/Sun et al. found themselves implementing all of the VMS extensions if they wanted to capture the existing FTN code.  Sadly the Debian folks have been unwilling to do this and are forcing people to rewrite any code that used the RPI GPIO pins heavily (like Oscar does).

Anyway, I have been looking at trying to recast Oscar's code to use the official Linux GPIO library interface, and sadly it's a good bit of work.  Oscar maps each PI's different Broadcom chips into user memory via the /proc interface.  His code then does bit manipulation directly.  It's simple and clean.  You have to map the chips properly - i.e., figure out which board you are using - and then code is common between the different RPs.

The Linux GPIO library does not support the same user-space memory-mapped interface style in /proc -- PINs are defined in a higher-level manner, and then different routines are offered to twiddle them.  More importantly, the naming of the PINS is quite a bit different.  The bottom line is that it will take some major surgery. The good news is that Raspian does support the Linux routines as well as their own private ones.   So ... if done correctly, it could make support of RPI alternatives for Oscar's code possible - so it seems like a good idea in the long run.

FWIW: The Libre Computer folks have been pretty nasty about it when I inquired about them offering an RPI-compatible library - as they seem to have built a couple of nice RPI4 alternatives, and many things just work on their board.  FYI: I have their board, and a couple of different Banana PIs have had mixed results trying to get any of it to work well.  Once I realized it would take a major structuring to make a portable implementation, and RPI4s started to become more available, I held off due to the surgery Rich and Oscar are currently performing for the PiDP-10 - it seemed like a bad time to fork the code base. Contact me offline if you want more details.

Clem

ᐧ

[Alan Abbott]

Thank you for the reply.

As a hardware guy, I thought the point of the GPIO was that it was a universal interface especially to make it popular.

Obvioulsy there is a lot more to this then it first seems.

It would (imho) be a good idea to try to make it universal if pos though.

Alan

Clem , Thanks I will contact you offline but in a day or two(just had joint covid and flu jabs so a bit under the weather)

-- 
"Those who do not learn from Dilbert are doomed to repeat it"

[steve...@gmail.com]

The software solution is to introduce an abstraction layer. Code in SIMH to an abstract interface that's defined in terms of the machine's front panel lights and switches. For example, the abstract interface could be defined in terms of functions like:

setRegister( nameOfRegister, bitPatternToWrite )

setRegisterBit( nameOfRegister, bitLocationToSet, newValue )

setStatusBit( nameOfStatusBit, newValue )

readIndividualSwitch( nameOfIndividualSwitch ): boolean

readSwitchSet( nameOfSwitchSet ): bitArray

That abstract interface can be implemented in many different ways, each of which maps the registers, bits, and switches to the given I/O technology, e.g., GPIO. Porting SIMH to an environment which has different I/O is simply a matter of mapping the abstract interface to the new I/O technology..

-- steve

[Richard C]

 I just have the tread read directly from the internal registers. The keeps the display update as realistic as possible.

Clem if you have suggestions, contact me at my normal email address.

Rich

[Clem Cole]

Indeed - that is the proper way in the long run and it is basically how the official linux interface works. But there are a number complications the least of which Oscar hw sharers bits, charlieplexs and does some low level twiddling (the official standard se interface does not support those modes easily).  

Moreover even with the RPIs themselves the GPIO ports have different names and are used differently.  Basically /proc exposes the different Broadcom chips and Oscar makes use of how the chips themselves work and how he wired against them ignoring some of the higher level RPI definitions.   

As I said the is nothing new in our business.  RPi was first and did something easy for them with /proc.  In fact early RPI programming examples in the wild of how to use RPI and GPIO do exactly what Oscar did as there was no standard.  

But it’s not quite that easy here.  You can sit on purity and try to force people to rewrite to a new an more portable interface which if Oscar wants to relax the requirement of using an RPi - seems to be needed.  Or the creator of a Linux for a board similar too but different from RPI offers an optional extension - which I would think folks that are trying to capture business from RPi might consider.   Or even consider changing the standard to include the de facto one — what people did and keep their code running. All of these schools of thought have value.   

I’ve lived and dealt with all of these over the years and there is value in each. 

Sent from a handheld expect more typos than usual

--

You received this message because you are subscribed to the Google Groups "PiDP-10" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pidp-10+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-10/b0bb6088-f077-4bed-979f-7a6f37def86bn%40googlegroups.com.

[Randy Mongenel]

One thing of note with the new RPi 5:  The GPIO header is no longer handled by the CPU, it is handled by the "southbridge" RP1 chip.

The Raspberry Pi Foundation says that the GPIO is "100% backward compatible" with the older Pi models, but with an additional chip handling the I/O, that's probably not 100% true unless you're using their tools and libraries.

They have made mention of the change in news articles and some videos, but the documentation hasn't been posted on their website yet.

To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-10/CAC20D2NcqvtGKJKU53R4d77XhaXVVDexvrVgUbiEMLN1J5jHpQ%40mail.gmail.com.

[Gavin Scott]

On Tuesday, October 10, 2023 at 12:57:49 PM UTC-5 mong...@gmail.com wrote:

One thing of note with the new RPi 5:  The GPIO header is no longer handled by the CPU, it is handled by the "southbridge" RP1 chip.

The Raspberry Pi Foundation says that the GPIO is "100% backward compatible" with the older Pi models, but with an additional chip handling the I/O, that's probably not 100% true unless you're using their tools and libraries.

They have made mention of the change in news articles and some videos, but the documentation hasn't been posted on their website yet.

RP-1 chip preliminary datasheet: https://datasheets.raspberrypi.com/rp1/rp1-peripherals.pdf

Techie interview with Eben and the RP-1 designers: https://www.youtube.com/watch?v=aioB40BGQYU

 

[Lars Brinkhoff]

Now that I have seen the real PiDP-10 case, I can attest "it's even yummier than my snack!"

[Ken Hansen]

I saw the case/prototype in Texas, and while I really liked it, "yummy" is a bit too far...

Ken

On Oct 20, 2023, at 14:00, Lars Brinkhoff <lars.br...@gmail.com> wrote:

Now that I have seen the real PiDP-10 case, I can attest "it's even yummier than my snack!"

<yummy-pidp10.png>

--
You received this message because you are subscribed to the Google Groups "PiDP-10" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pidp-10+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-10/8651cdfb-d894-498f-95ad-6e48397e6762n%40googlegroups.com.

<yummy-pidp10.png>

[Randy E (OoMOR)]

Ken,

Lars said it it was "yummy" compared to his snack, but failed to state how good or bad his snack was. For all we know, he was eating a week old gas station hot dog.

Just a thought.

Randy

[Andrew Wasson]

Good point but... I'm pretty sure that case is going to be next level awesome!

[Cynde Moya]

I can't wait to get one shipped to me in Australia!

Cynde

[Lars Brinkhoff]

Cynde Moya wrote:

I can't wait to get one shipped to me in Australia!

Very good, upholding a  long tradition of 36 bits down under.  The first PDP-6 sale went to to University of Western Australia, Perth.

[johntk...@gmail.com]

One is needed here in my office in Redmond. As we work on AI/ML systems, it’s imperative we have SHRDLU running at all times. 

[oscarv]

Hi,

I realise I am bad at communicating. And in planning, to some extent. 

But: the container with 1,000 PiDP-10 cases will sail from China on November 28th. So the race is on to get all the other parts in stock before the boat arrives in Panama. Which is where we will box up the kits before sending them to the fulfillment center in Florida. Strange story, I know, if you want to know, look at the PiDP-11 order page for pictures. But don't actually order on that page, it is still the wrong PiDP there, for now:-)

This is a bit scary. I had to order 1,000 of the cases to get a decent per-unit price on the part. But I'm not so sure the world market for faux PDP-10s is anywhere near 1,000 units. A wise man at IBM once estimated the world market for computers at five,which is cause for concern. But then again, that was the underestimate of the century. We'll see. It doesn't matter, the PiDP-10 is happening, finally, that is all I really wanted!

Kind regards,

Oscar.

[Andrew Wasson]

Fantastic Oscar! Thanks for the update and looking forward to more info.

Cheers,

Andrew

[Manuel Maseda]

Glad I live in Florida!!

Manuel

[paul.j.s...@gmail.com]

Thanks for the update! Can't wait for the order page to go "live".

Paul

[Malcolm Ray]

Does this mean I should be constantly hitting refresh on ceds.dev rather than tindie.com?

Regards,

Malcolm

--
You received this message because you are subscribed to the Google Groups "PiDP-10" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pidp-10+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-10/7cb957b9-3a8e-46ad-8a4a-5ce8f91b7d73n%40googlegroups.com.

[Andy]

Fantastic news Oscar - fingers crossed for no global shenanigans that will interrupt the shipment.... :-)

Manuel and I will be leading the Florida welcoming committee!

On Tuesday, November 14, 2023 at 8:29:56 PM UTC-5 oscarv wrote:

[Christine Munro]

On Wednesday, 15 November 2023 at 01:29:56 UTC oscarv wrote:

This is a bit scary. I had to order 1,000 of the cases to get a decent per-unit price on the part. But I'm not so sure the world market for faux PDP-10s is anywhere near 1,000 units. A wise man at IBM once estimated the world market for computers at five,which is cause for concern. But then again, that was the underestimate of the century. We'll see. It doesn't matter, the PiDP-10 is happening, finally, that is all I really wanted!

I think most of us are PDP-10 evangelists so should be able to get a few more people interested!  And IMHO it would do the world of modern computing a lot of good for more people to see how things were done on the 10, and what could be accomplished with comparatively meagre resources.  And the correct number of bits.  And lights.

[Lars Brinkhoff]

Christine Munro wrote:
> I think most of us are PDP-10 evangelists

Where do y'all do your evangelizing?

[johntk...@gmail.com]

I am planning on displaying mine at work. At least to start with. 

[Andy]

i evangelize everywhere to everyone.  Get totally ignored, of course :-)

[Christine Munro]

Anywhere there's someone who might listen!  I know various people in IT and try to go on about "the good ole days", as well as some people who were actually there and just "need" reminding; I also figure that some people I know who are into retro-gaming might be interested in retro-computing.  I mean if someone is happy to buy a retro Commodore 64 and/or Sinclair Spectrum, why not a PDP-10 too...?  And play ADVENT as it was originally intended!

[Mark Lawler]

Eagerly pounding refresh as I await the release and how to order mine.  :D 

Best,

-Mark