• Please review our updated Terms and Rules here

T11 clock running Forth

Hi Mark, attached are schematics.
One shows the time base used on the clock. Datasheets are on the web.
The other is for the T11 wire-wrap board. It's quite a mess...

The stuff in the lower-left corner (HC595's) was never implemented. It was a plan to make an interface that would be driven by a modern MPU, to DMA code into the RAM for the T11. I was going to use an MSP430. Every time on power-up, the MSP430 would load the T11's RAM and let it run. So, the T11 wouldn't need any EEPROM. The HC595's were a serial-to-parallel interface, so that only a few wires would be needed coming from the MSP430.
I've had thoughts of something similar with a uVAX-II board pair in a Qbus backplane for power: On power-up, an MPU would write code into RAM via the console ODT, then run it. But, it's only practical for smaller programs, and I can't think of what I would do with it...

One of the HP LED displays is shown in the lower-right corner, connected to the LS138. Those particular displays are sorta expensive today. I've got a few extra, but I'm not sharing :-) .

The LS374 on the right side was not implemented. It was for adding output bits, but the clock doesn't need it.

The LS240 in the upper-left corner has two sections - one is for the T11 power-up config bits, the other is 4 bits of Input, selected by the "BTN" signal, which is addressed at the upper-most 1K page of the "B6" 8K page. Two bits are for the buttons, one is for the 3 Hz input (not shown on the drawing, I think it's pin 8 of the LS240), the 4th is unused.

The notations next to most chips pins, such as "3-24", are the row/column for the physical pin on the wire-wrap board I used.

Pete

tb_sch.jpg

t11_ww.jpg
 
griffk said:
Very cool project!
I have one ? Why do you put the UART(?-can't see the chip#) in a ZIF socket? Do you regularly change them out?
thanks, gwk
Click to expand...
Hi, That's not the UART, it's the EEPROM. The marking "UART / 64" is because the code is configured to set the UART clock to the "divide by 64" mode.
The UART is an MC68B50, in the lower-right corner.
Of course, we're talking about the PCB that was made in this thread, not the clock which started the thread. The clock doesn't have a UART.

Pete
 
Last edited:
Pete,
Thanks for the schematics and the explanation below. Are the HP LEDs you mentioned the ones that have 4 bits of data input and a Chip Select, then they display the Hex code until the next data nibble is passed to it? I have a couple of those somewhere in my chip drawers. I used them a lot in debugging 6502 embedded controller projects. I'll have to look for some more of them. They would be a great way to provide the clock display with minimal effort.

The time base work you did reminded me of a project I did many years ago. The IT guys in my company were trying to read data from an old grain exchange that was sending (6 bit) data at 66.67 baud. The data was intended for one of those glass domed ticker tape machines but they were attempting to read it with a DZ-11 on a PDP-11/44. The DZ-11 had a 50 baud setting and a 75 baud setting but neither worked with the 66.67 baud. Also, the code was "ticker tape code" with 6 bit codes that mapped to print symbols for 1/8, 1/4 etc. They asked me where you could buy a translation box. I said I had no idea, but I could make one. They pulled a line with the data feed (which only ran for about 30 minutes a day) to my lab and I used a 6502 with a UART and a 6522 counter. I was using controller board sets that Rockwell used to make so the hardware configuration was minimal. I set the counter to count the 1 mHz clock and generate pulses at 16x the 66 baud rate and wired that to the input UART's external clock. With the UART set for 6 bits and a 64 byte look up table to the 8 bit ASCII codes the project was complete. One UART bought in the ticker tape data and another one output it at 9600 and ~128 byte program was burned to EPROM. A week later I gave them their translator box and it ran for several years until the exchange modernized.

I did really like the way you used an EEPROM and a static RAM on the T11 board to allow program changes. That was quite an elegant way to allow program modifications. I did use Forth a bit on the 6502, but haven't touched it for years so that is an added treat with the T11 board project.

Thanks,
Mark

saipan59 said:
Hi Mark, attached are schematics.
One shows the time base used on the clock. Datasheets are on the web.
The other is for the T11 wire-wrap board. It's quite a mess...

The stuff in the lower-left corner (HC595's) was never implemented. It was a plan to make an interface that would be driven by a modern MPU, to DMA code into the RAM for the T11. I was going to use an MSP430. Every time on power-up, the MSP430 would load the T11's RAM and let it run. So, the T11 wouldn't need any EEPROM. The HC595's were a serial-to-parallel interface, so that only a few wires would be needed coming from the MSP430.
I've had thoughts of something similar with a uVAX-II board pair in a Qbus backplane for power: On power-up, an MPU would write code into RAM via the console ODT, then run it. But, it's only practical for smaller programs, and I can't think of what I would do with it...

One of the HP LED displays is shown in the lower-right corner, connected to the LS138. Those particular displays are sorta expensive today. I've got a few extra, but I'm not sharing :-) .

The LS374 on the right side was not implemented. It was for adding output bits, but the clock doesn't need it.

The LS240 in the upper-left corner has two sections - one is for the T11 power-up config bits, the other is 4 bits of Input, selected by the "BTN" signal, which is addressed at the upper-most 1K page of the "B6" 8K page. Two bits are for the buttons, one is for the 3 Hz input (not shown on the drawing, I think it's pin 8 of the LS240), the 4th is unused.

The notations next to most chips pins, such as "3-24", are the row/column for the physical pin on the wire-wrap board I used.

Pete

View attachment 25642

View attachment 25643
Click to expand...
 
Yes, the displays are HP 5082-7340. There are a couple of other 73xx parts that are very similar. One negative on these is that they use a lot of power, and get really hot. In the old HP test equipment, they have to be heat-sinked, because they (of course) right behind the front panel where they don't get any air flow.
Very good on the 6502 stuff - I also did a bunch of 6502 stuff starting in 1979.
Pete
 
I was wondering if anyone (other than Pete of course) has gotten their board going. I spent some quality time at the bench tonight and got the board 80% of the way together. I socketed everything but still have to stuff the sockets. I accumulated HCT logic for all the 74XX and a zif socket for the eeprom. I also bought two more eeproms. I am nearly ready to start playing. I'll use one of those cheap little boards with max3232 and 9-pin d-sub connector to get to RS232 to connect the console terminal. I was thinking of providing power from one of those wall-plug-in 5V switching power supplies with the USB connector. I think they are regulated, but have never looked at one on the scope. I'll check it out under various loads to see if it looks safe enough. We could be up and running by mid week.

A number of years ago (maybe eight?) I went to my first VCF East. There was a fellow demonstrating analog computers (and yes, I am old enough to have used one in college.) For some reason, this fellow also had copies of this article at his table : http://www.unz.org/Pub/ProgrammersJournal-1988nov-00056. I picked one up along with the stuff on analog computers. I recently came across the hardcopy again and finally gave it a read since I will be learning Forth for the first time for this project. It is an amusing article and provides a good historical perspective and introduction to the language.

Lou
 
Pete,

Well, I am on track to power up tomorrow night. I finished the build this evening:

http://www.vintage-computer.com/vcforum/album.php?albumid=300&attachmentid=26215
http://www.vintage-computer.com/vcforum/album.php?albumid=300&attachmentid=26214
http://www.vintage-computer.com/vcforum/album.php?albumid=300&attachmentid=26213

I did solder in the crystal, committing to 14400 baud, which is more than fast enough for me. I was sure to keep a gap between the bottom of the can and the board so as not to short out the traces under the can. I had some classy old tall TI sockets for the wide dips. The zif socket is plugged into a tall dip socket. This allows access to the screw hole and to install a right angle double row header at J6/7 (I will add that later when I get some right angle headers)

I will add a pin to the +5V bus trace by J4 to supply power to the little MAX3232 board. That's the last bit of soldering left.

The R1 that I used has a gold tolerance band, but I did hand sort my 1K resistors to select the one closest to 1K. The one I installed was closer than 1%. I didn't have any 1/8W resistors, so I used the 1/4W I had, installing them at an angle to take up the increased body length.

Looking forward to tomorrow night.....

Lou
 
Looks great, Lou.
You will still be able to do a different baud rate, by removing the jumper and providing your own baud clock signal.
I did the same thing for choosing R1...

I'm still trying to decide on an "application" for one of the PCBs. I already have the clock running on the original wire-wrapped version.
Clocks are good because they have a reason to be left running 24/7.
I'm open to suggestion.

Pete
 
saipan59 said:
Len definitely got his running, about a month ago...
Pete
Click to expand...

Yes, I'm up and running. I have a few other projects to get completed first before I jump back to the clock display and time base. I did buy the hp displays like Pete has. Pete also pointed me to a glass enclosed crystal that I plan to use for the time base. I found a schematic of a hallicrafters 100kc tube based frequency standard that should be adaptable to my project, and then it's just a 100,000 divider. Oh, and a 150v ps.

Len
 
8008guy said:
Pete also pointed me to a glass enclosed crystal that I plan to use for the time base. I found a schematic of a hallicrafters 100kc tube based frequency standard that should be adaptable to my project, and then it's just a 100,000 divider. Oh, and a 150v ps.
Len
Click to expand...
Hey Len, I'll post a schematic for the 100KC tube-based oscillator I use on another clock. I borrowed the design from a 1950's vintage HP frequency counter. Note that many tube xtal oscillators will work just fine on a relatively low voltage - mine works down to 40-50 volts, maybe less. Worst case, you might have to lower the resistance of the plate load resistor. The old designs will usually call out 150V or more because that voltage would already be available in an old shortwave radio.

Pete
 
tube time-base oscillator

tube time-base oscillator

Here is the tube time base that I used (just the oscillator, tube V1).
time_base.jpg
In the HP521 counter, it runs from 200V, but I run mine from about 85V, I think. The tube filament can be run from 5VDC - it would technically want 6.3V, but 5V is fine. The filament on that type of tube is 12.6 V center-tapped, so you can wire it for 6.3V, but just give it 5V from your logic supply. The current will be less than 300 mA.
I have lots of 5963 tubes - I'd be glad to send you a couple. It's very similar to a 12AU7, which is used in a lot of tube audio gear. In this circuit, they should be interchangeable.
If you want to go crazy, you can build the whole circuit shown in the schematic, which includes freq dividers to get down to 100 Hz.
See how those dividers are called "Phantastrons" - it's really a type of non-retriggerable monostable circuit, or "one-shot". You can make the same thing with a 74121 chip. Or you can do the smart thing and just use CMOS decade counter chips... ;-)
Pete
 
8008guy said:
Hi Pete,
The "Phantastrons" are really interesting. How accurately does your time base end up being?
len
Click to expand...
It can be tuned to be within 5-10 seconds per month. The most important things for accuracy are keeping a constant temperature, and using NP0 capacitors in the circuit near the crystal, and keeping the voltage fairly constant.

Pete
 
I still have a bit of troubleshooting after last night's work. I am unsure if the use of the MAX3232 was a good idea. Although powered from +5V it's supposed to have 5V TTL thresholds, I think they might be too low. I still get a lot of garbage coming through, send and receive. I may also need to work on shielding. If I hold the grounded metal spacers I have on the board as legs, things settle down a bit. A lot is working though since I do see the FIG-FORTH V1.3 banner through the garbage occasionally. I also see "OK" a lot. It appears that the output is 7 data bits, one stop bit. Is it even parity?

I may try a MAX232 tonight to ensure proper TTL voltage levels. I also need to look at all signals on the scope for voltage level and cleanliness.

Also, 14400 baud is not supported by any terminal I have, from VT100 through VT420 (and not even the Briel pocketterm). I had to use a PC as the console terminal :( .

Lou
 
The UART is set up for 8-N-1.
Here is the code that sets it up:

934 010364 112737 000003 160000 MOVB #3,@#160000 ; RESET MC6850 UART
935 010372 112737 000024 160000 MOVB #24,@#160000 ; INIT UART, RTS LOW

Pete
 
OK we got it! Finally today I got a little "hello world" program running! Indeed the store-bought DE-9 MAX3232 boards were not happy. I made my own DE-9 MAX232 board and it works fine!

I need to find some 1.2288 MHz crystals so that I get 9600 baud, but they don't look common. At the moment I am stuck with PuTTY on the PC as a console. However, I am excited that it all works!

I am thinking about my clock now. I have some RED60 power line clock dividers (they output one pulse per second) (datasheet : http://www.lsicsi.com/pdfs/Data_Sheets/RED Series.pdf) I was thinking I would like to display hours, minutes, and seconds. I was considering outputting each in binary from the six general purpose output lines when requested with two of the four general purpose input lines. One of the input lines would be the 1 Hz pulse from the RED60. The binary hours, minutes, or seconds would be decoded to two decimal digits and then transformed into 7-segment display segments, latched into latches and sent to pairs of 7-segment displays. I would do all that with 74XX TTL. I have everything to do all that. Does this sound reasonable?

I would have one input left. I might be able to figure out a clever way to set the clock with it, or else set the clock over the console port with a terminal connected.

Lou
 
Good job Lou! Glad to hear you got it running.
Baud clock: If you're going to add logic for a clock display anyway, you could add a 74LS393 and a DIP-can oscillator at 2.4576 Mhz, and get 9600 or 4800 easily.
http://www.ebay.com/itm/QTY-5-2-457...EE-SHIPPING-/111154373954?hash=item19e1512542
Clock logic: You could simplify your logic by using 3 of the decoded LS138 outputs as strobes for your hours/minutes/seconds latches. And it would free up the input bits for time-setting buttons.

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