• Please review our updated Terms and Rules here

T11 clock running Forth

Neat clock! It is inspiring to see new projects with vintage hardware. Thanks for sharing.
 
PG31 said:
Please count me in for a PCB, I have a spare RQDX3 that I can borrow the chip from.
Click to expand...

For those who don't happen to have a spare PCB from which to repurpose its T11, but would still like to get in on the fun, there is a reputable vendor for chips:

decstuff2015@arclightindustries.com

$20 each, plus shipping -- best would be Priority $5.25 for a small flat rate box.

Order more than one; same shipping :->.

[Reputable means that I've known him for several years now and *I* can highly recommend him :->.]

You could also hang-tight and snag an inexpensive RQDX1 or RQDX2 on eBay. They come up occasionally and don't get much buyer-action (at least compared to other DEC-stuff ...).
 
I started working on the schematic last night.
I found that I have several extra MC6850BJCS UART chips - ceramic, gold leads, gold lid, 1978 date code. I will "share" when the PCBs are ready, so nobody (probably) needs to buy any, unless we have a lot of customers.
I can't find (yet) a reference that says what the "BJCS" variant is. So I would assume that they are 1 Mhz. The faster parts are normally marked "68A50" or "68B50". The package matches the "L suffix" in the datasheet. I'll plug them into my "development system" to be sure that they work.

Any opinions on what the baud rate should be? Does it need to be settable? I'm hoping to keep things simple. The 6850 needs a clock that is either 1X, 16X, or 64X of the bit rate. One approach is to use the COUT signal from the T11, and divide it down with an LS393. The T11's crystal would need to be a baud-rate multiple frequency, such as 1.8432 Mhz.

Pete
 
saipan59 said:
I started working on the schematic last night.
I found that I have several extra MC6850BJCS UART chips - ceramic, gold leads, gold lid, 1978 date code. I will "share" when the PCBs are ready, so nobody (probably) needs to buy any, unless we have a lot of customers.
I can't find (yet) a reference that says what the "BJCS" variant is. So I would assume that they are 1 Mhz. The faster parts are normally marked "68A50" or "68B50". The package matches the "L suffix" in the datasheet. I'll plug them into my "development system" to be sure that they work.

Pete
Click to expand...

https://books.google.com/books?id=e...&ved=0CB8Q6AEwAA#v=onepage&q=MC6850BJ&f=false suggests that the "BJ" corresponds to the temperature range -55C to +125C.

More common sources indicate that "CS" corresponds to a Cerdip package with a temperature range of -40C to +85C.

Agree that it's a 1 MHz part.

I would personally prefer a separately settable UART clock, but I understand your drive for simplicity. Perhaps the PCB could support a separated clock, but users could simply not populate the UART-clock subcircuit and jumper to the CPU-clock with the applicable restriction on the frequency?
 
Did some checking on baud rate options.
- Configure the T11 for "constant clock" on the COUT pin. This output will be half of the xtal frequency.
- Connect COUT directly to the MC6850.
- Configure the 6850 to divide the incoming baud clock by 64 (it supports /64, /16, and /1 ).
- Using a 1.8432 Mhz xtal, the baud rate is 14400.
- Using a 2.4576 Mhz xtal (with 68A50 or 68B50), the rate is 19200.
- Using a 3.6864 Mhz xtal (with 68B50), the rate is 28800.
I like this, because no baud oscillator or divider is necessary. For other cases, the user can provide a separate baud clock.

Pete
 
Schematic so far

Schematic so far

View attachment T11_bd.zip

Here is what I have so far - getting close.
5th page definitely not done.
Need to check if the pullups on page 1 are really required - I've been running without them so far... [Edit: T11 manual says those pins have internal pullups, so I'll delete those.]

Pete
 
Last edited:
EEPROM hot-swap support

EEPROM hot-swap support

A feature that I will add, if space is not a big problem, is support for "hot-swapping" the EEPROM chip.
That is, allow you to power up, boot from your "Master" EEPROM, do some code changes out of RAM, swap the EE chip, then burn your image to the new EE.
This allows you to keep your Master EE safe, since there is no mass storage interface.
The method is this:
- Use a jumper/switch to assert DMR on the T11. This puts it in DMA mode, so it is no longer using the address/data buses.
- Use a jumper/switch to turn off VCC to the EEPROM chip. Now the chip can be swapped in its socket. RAM stays powered up.
- Restore power to the EE chip, then de-assert DMR, and the T11 is running again.

It's a "cheap trick", but it works. I've been doing it on my development system.

Pete
 
approximate board cost

approximate board cost

The approximate bare-board cost looks like it will be about $25 for one, $46 for two, $67 for three (shipped, from me).
That's for a quantity of 8 total.
Qty 10 would be $23, qty 6 would be $28.
Pete
 
Last edited:
saipan59 said:
The approximate bare-board cost looks like it will be about $25 for one, $46 for two, $67 for three (shipped, from me).
That's for a quantity of 8 total.
Qty 10 would be $23, qty 6 would be $28.
Pete
Click to expand...

I was expecting (OK, hoping ...) that these PCBs wouldn't end up being so pricey given that they aren't all that large and there are no gold fingers involved. Oh well. I'm still "in", but only for one. Apologies that that doesn't do anything good to the per-PCB cost :-<.

Thanks for the continued progress!
 
Lou - N2MIY said:
This is a huge timesaver over wire wrapping.
Click to expand...

Yes indeed. I'd really like to do a DCT-11EM-clone at some point given that (thank goodness!) the ROMs are available. Think that I'll use the "front-side" wire-wrap approach using SIP-headers parallel to the socket-sides. Both neater and a bit easier to work with than dealing with standard WW sockets, although necessarily less dense overall.
 
Thanks guys. Just for "full disclosure", the price estimate doesn't include any profit for me. I took the delivered price of ExpressPCB, and added $4 per package to ship to you.
My motivation is just to reduce the cost to myself for 2 or 3 boards. I'm not sure what I'll actually do with the boards...
Based on the current board size, the cost delivered to me is:
6 boards for $143.
8 boards for $165.
The layout routing is proving to be a bit challenging at the current size, so the size may have to increase just a bit.

Or, eliminating the headers for address/data/control bus signals would help - let me know if we want to back off of that requirement.
A space-saving alternative for the CPU bus signals is to not include a header, but instead simply include a via on each signal that is accessible from the bottom side. Then it would be "possible" to solder individual wires to each signal (but certainly inconvenient).

Another (better) idea: Instead of the 2-row IDC-type headers on the edge of the board that I have now, how about a single row of twenty 0.1" pins on each side of the T11, and similar next to other chips that need signal access? The routing should be much easier I think. It would still be possible to buy or make a 20-pin ribbon-type cable if necessary, or mount a hand-wired perf-board above or below (if the SIP rows are populated with long WW posts).
Would that be OK with everyone?

Pete
 
saipan59 said:
Thanks guys. Just for "full disclosure", the price estimate doesn't include any profit for me. I took the delivered price of ExpressPCB, and added $4 per package to ship to you.
My motivation is just to reduce the cost to myself for 2 or 3 boards. I'm not sure what I'll actually do with the boards...
Based on the current board size, the cost delivered to me is:
6 boards for $143.
8 boards for $165.
The layout routing is proving to be a bit challenging at the current size, so the size may have to increase just a bit.

Or, eliminating the headers for address/data/control bus signals would help - let me know if we want to back off of that requirement.
A space-saving alternative for the CPU bus signals is to not include a header, but instead simply include a via on each signal that is accessible from the bottom side. Then it would be "possible" to solder individual wires to each signal (but certainly inconvenient).

Another (better) idea: Instead of the 2-row IDC-type headers on the edge of the board that I have now, how about a single row of twenty 0.1" pins on each side of the T11, and similar next to other chips that need signal access? The routing should be much easier I think. It would still be possible to buy or make a 20-pin ribbon-type cable if necessary, or mount a hand-wired perf-board above or below (if the SIP rows are populated with long WW posts).
Would that be OK with everyone?

Pete
Click to expand...

Unless we're suddenly crossing a major size/cost threshold I'd prefer to keep the headers, however I'm not opposed to the alternatives. Agree that one could obtain "Arduino-style" long-pin SIP-headers and then possibly mount as a daughterboard.

Anyway, if I ever do a DCT-11EM-clone I'll want to start from scratch and not try to piggyback this board. I like it as an extensible Forth-machine ... as intended by its designer :->!

And if anything that I said led anyone to think that I thought ill of anyone, well you heard/thought wrong. I'm just congenitally frugal; it's in my blood-n-bone.

paul
 
pbirkel@gmail.com said:
Unless we're suddenly crossing a major size/cost threshold I'd prefer to keep the headers, however I'm not opposed to the alternatives.
Click to expand...
I checked on the pricing formula: The size-related part of the pricing is simply 70 cents per square inch per board. The other costs are roughly fixed (setup fee of $61, shipping, a small charge for having lots of holes, etc.). Current size is 4.7 by 3.1 inches, but I expect that to increase somewhat.
So, I'll keep the 2-row headers in place for the moment. Maybe I'll experiment with the single-row option, just to see if there is a major reduction in board size (I doubt it).
pbirkel@gmail.com said:
And if anything that I said led anyone to think that I thought ill of anyone, well you heard/thought wrong. I'm just congenitally frugal; it's in my blood-n-bone.
paul
Click to expand...
No worries!

Pete
 
You must log in or register to reply here.
Back
Top