PDP8/M with Teletype 33 through Modem-line

[MauriceH]

Hello all,
Back again and playing with my PDP8/M with RX02-Emulation project v4 AK6DN, normally run with VT220 terminal.
As I have a great working Teletype ASR33,I now manage to connect through a modem-line connection.
That is the Teletype is downstairs, the PDP8 upstarts.

So have a small Telephone home-exchange, some telephones and some very old modem.
The ASR-33 is original equipped with a 200Bd modem, designed by University school of LEUVEN(Belgium).
Back in the 1970's.
Have spoken with old employ man of that firm who converted these machines ,mine fully original to that rebuild.
Inside all extra PCB's Transformers etc, all due to this modem conversion.
A motor 15min time out, and auto start up motor after some DATA code , a 20mA-to-RS232 conversion, 3 switches to pick-up telephone line and a local print stop switch.
X-on/X-OFF function to remote start-stop puncher
etc.

Teletype it self does not have issues as far as I tested it over period of time.
Punching, reading tapes, and running it directly with the VT220 terminal.
Machine is quit in a mint condition.


Question for help:
So this year I got a second exact same MODEM, what a great find.
Hook up the both through a home telephone exchange for 6 phones.

First try with just a terminal VT220 at110 Bd, modem 1, telephone exchange, Modem 2 in the Teletype 33.
That is working like a charme, no failures etc.

Then hook up all to the PDP8/M.

Sofar connection is made and I can give commands and receive the output from the PDP8.
Done with Serial card M8655 at 110Bd.
Full wired RS232 cable to Modem.

Running OS8 on the RX02-Emulation-project, normally that runs fine on the VT220.

But then after f.i. a DIR command, the ASR33 looses track in print out.
The timing point of that failure is at the same lines after several attempts of same DIR command at same disk.
(Also with a second serial card M8650 at 110Bd, same failure at same lines in the printout.
4th line and 8th line in de directory listing.
12th line is again good.

4th "CCL .SV * then rest of that line is printed at the end of carradge.
8th "SET .SV * then rest of that line is printed at the end of carradge.

Is there some special settings in OS8 for a Real Teletype?

Regards
Maurice
Netherlands.

 

[DougIngraham]

There is no special setting in OS/8 specifically for a real teletype.

Try changing the M8655 to send 2 stop bits. I believe this was the default setting for use with a Teletype. My guess is that the timing error creeps up and just happens to land on the last character of the 4th and 8th lines. There might be a Disk I/O operation somewhere after line 8 which would give the teletype a chance to resync.

That is my best guess.

If that fixes it you will probably want to change the VT220 to match the stop bits because the stop bits setting is a jumper that is either unsoldered or cut. And I think I remember that there are two places this could be jumpered to get 1 stop bit. I am not at home or I would go look at it.

Best wishes!

 

[Mike_Z]

Maurice, I use my ASR33 with my PDP8E regularly. I use two stop bits and don't think I ever tried just one. Since you mentioned that the Vt-200 works fine, and looking at your print out, I'm wondering if the problem is with the ASR33. It appears that the Line Feed at the end of the line is working every time, but occasionally, the carriage return is not. The errant line is directly below the last line, as if the LF worked, but the CR did not. You may just need to lube up the ASR33 carriage return linkage. If you set the TTY to local, and type on the keyboard, do you have the same problem? You may want to look at the mechanism that is being actuated and use some light machine oil on all the pivots, the ASR33 may just be hanging up. They do need oil now and then. My 2 cents or is it Guilder now. Mike

 

[MauriceH]

Both thanks for the tips.
My bad .. of course 7 bits 2 stops is the Teletype ASR33.
M8655-YA:
W4 now -OUT (Was in)
And put in TTY-Filter (Was out)
Only M8655 could be 7Bits/Char? Now still NB1,NB2 are both out (8 Bits/Char)

Is better to put at 7 Bits/Char? NB1,NB2; 0 1 ?

This setting is now working fine, see printout.

The M8650 had a simple failure of bad solder on 1 of the Address-lugs, cold joint, so resistance.
And yes that one has a Cristal 14.418 MHz,110Bd,(As 19.6608MHz =150Bd) at G 7-8 that I had checked. Also Capacitor 47nF
and jumper J 2-3 for 2 Bits.
Typical it made same result.

Both cards now running perfect with the Teletype.
Thanks.

 

[Mike_Z]

I never argue with success. But these old TTY's need their oil. I can remember back in my early days, when having trouble with a n ASR35 and trying to find the problem (similar to yours), the telephone repair man was there and told me to remove the mech, sit it on some towels and and just pour oil over the entire thing. I used mineral oil and it was clear on entry, but black on exit. I let it drip dry for a couple days, re installed it and it worked great. This is kind of hammer mechanics, and now I'll use some needle oilers, the point is the TTY's need oil. Good luck.
I kind of like your remote telephone connection. I have a couple of old Hayes modems, maybe I can connect up a set up like yours. Thanks Mike

 

[DougIngraham]

I am glad that worked. I was not certain it was the issue. Strictly speaking the Teletype wants 8 data bits and needs at least 1.5 stop bits to be reliable. It needs 8 data bits if you are going to punch binary paper tapes and use all bits. The 8th channel is not decoded when printing but most software sends it set to a 1 which gives an effective additional stop bit.

Mike has a good point about lubrication. I remember the repair guy doing PM on the ASR35 that was originally with my machine. It was wet with oil inside after he got done.

If the carriage is not returning at full speed the first character of a line may not strike on the first column The left column may appear a little jagged while columns further to the right all line up. Usually the time it takes for the linefeed is enough to allow for the carriage to travel completely to the left. This was not always the case and the OS/8 Super TTY handler would allow for additional null characters to be sent after either the return or the linefeed, I don't remember which. There is probably an OS/8 command to set this but it would only work with programs that call the handler to talk to the console. Most programs did not do this although system utilities mostly do. I just looked back at the printout in your first note and I can clearly see the that the first column does not line up although in your new photo it looks pretty good. Was this due to lubrication or simply the additional delay imparted by the extra stop bits?

 

[Mike_Z]

The mis alignment of the 1st column could be from the dashpot not being adjusted properly. When the carriage returns, there is a lot of momentum. Near the end of travel a dash pot, should take up the jar and stop any bouncing. There is a rubber plunger that enters the dash pot cylinder. The air compresses and acts as a shock absorber. On the back of the dash pot there is a small adjustable hole that allows the compressed air to escape. Could be the mass of the carriage is bouncing slightly and occasionally printing the first character on a bounce. Again my 2 cents. Mike

 

[MauriceH]

Mike thanks,
I'll go look into that "dash pot" if that is working correctly to damp the carriage return.
Machine is kept in full oil ).
Machine I already have couple of years.

Have all the manual volumes, schematics and all schematics of that original conversion made back in 1970.
That firm "Koopmans" Sold those converted to 200Bd modem new.

As I also have the Anderson Jacobson A211 data-coupler ,some old Rotary dial phones,
A Wyse WY-85 Terminal I can hook up now a whole system of Modems.
Sadly all 3 modems does not auto-pick-up the teleph line ,so have to walk , haha.

So, upstairs, PDP8/M, VT220, 200Bd modem, old rotary dial-phone.
Data switch to switch serial data VT220-Modem.
Down-Stairs , Telephone exchange, 2 lines with Old Rotary dial phones.
1 line 200Bd modem and the Teletype
1 line 300Bd Data-coupler modem and the Wyse 85 terminal.

So I now can choose down stairs between a quite Terminal or the noisy Teletype to run the PDP8.
Or directly call down-stairs Wyse terminal to the Teletype and just work without the pdp8, not very useful.
But in case of some quick demonstration to some friends and family.

Next step is to call out-side to an other Teletype 33 near by.
My brother just got a Teletype 33 and I got that running last week.
Luckily we both still have those phone-lines, ok they work on glassfiber-modems, but still.
Would be fun if it runs over some distance phone lines.
But then, he must have a modem also..
Have to see if I have some old Hayens modems, like 14k4, must have a 2400Bd also.
Lets see what that will bring.

Just having some fun.
Maurice.

 

[ElDuaneo]

I am glad that worked. I was not certain it was the issue. Strictly speaking the Teletype wants 8 data bits and needs at least 1.5 stop bits to be reliable. It needs 8 data bits if you are going to punch binary paper tapes and use all bits. The 8th channel is not decoded when printing but most software sends it set to a 1 which gives an effective additional stop bit.

Mike has a good point about lubrication. I remember the repair guy doing PM on the ASR35 that was originally with my machine. It was wet with oil inside after he got done.

If the carriage is not returning at full speed the first character of a line may not strike on the first column The left column may appear a little jagged while columns further to the right all line up. Usually the time it takes for the linefeed is enough to allow for the carriage to travel completely to the left. This was not always the case and the OS/8 Super TTY handler would allow for additional null characters to be sent after either the return or the linefeed, I don't remember which. There is probably an OS/8 command to set this but it would only work with programs that call the handler to talk to the console. Most programs did not do this although system utilities mostly do. I just looked back at the printout in your first note and I can clearly see the that the first column does not line up although in your new photo it looks pretty good. Was this due to lubrication or simply the additional delay imparted by the extra stop bits?

That is why we always send the carrage return before the linefeed. it gives the carrage extra time to get back.

 

[DougIngraham]

I looked at the Super TTY handler also called the KL8E to see about the additional delay and it turns out a 16ms delay can be specified after a specified character. They mention linefeed specifically. Of course this is a timing loop so it varies a bit based on the processor. It would fail badly on a simh based implementation.

I was thinking there were commands to configure these options on the fly but I haven't found any mention yet. These would be specific to a particular build of the handler and I am not at all certain how they would be applied in order to be sticky between boots. The KL8E is the handler that is currently included on the CSD boot disk. I should be able to put this inside the CSD server and save 2 pages of memory in OS/8. But that would require a completely different way of setting the options from OS/8 if there is such a way currently.

Anyone know about those commands? They don't seem to be part of the CCL, or at least not documented in the OS/8 handbook.

 

[vrs42]

Anyone know about those commands? They don't seem to be part of the CCL, or at least not documented in the OS/8 handbook.

There's a bunch of SET commands where you say
.SET <driver> <option>
but I don't know how they are implemented or what drivers have to do to support them.

 

[MauriceH]

Those Teletypes were sold to many PDP8 users, the guy of that firm told me.
He had at least a 1000 maintain contracts for these Teletypes and many more sold.

Translate this site with his fun service TTY story's Back then and lot of remarks he remembered on the changes
they made in those conversions ASR33's.
RetroComputer/Teletype-33

Machine shown on that website with the 3-colored buttons and towels under it to easy turn around, is mine.

Could be something to be find in the "MAINDEC-08-DHKLD-A-PB" date: 4/6/72
PDP8/E TELETYPE AND KL8 ASYNCHRONOUS DATA CONTROL TEST
REPLACES: MAINDEC-8E-D2AC-PB 1971,72

Or later replacement
MAINDEC-08-DHKLD-A-PB date: 6/1/76
PDP8/E TELETYPE AND KL8 ASYNCHRONOUS DATA CONTROL TEST
REPLACES: MAINDEC-8E-D2AC-PB

Think in those tests must be included right timings for the Teletypes.

 

[vrs42]

There's a bunch of SET commands where you say
.SET <driver> <option>
but I don't know how they are implemented or what drivers have to do to support them.

Made a little progress. CCL has code to recognize SET commands, and then decide whether to call a helper program or not. Apparently, SET.SV gets called to help out. SET.PA is apparently a MACREL program with a bunch of arcane knowledge about where various settings are stored.