EEVblog Electronics Community Forum

Computing => Vintage Computing => Topic started by: gardner on October 07, 2017, 04:00:52 pm

Title: Repairing a VT220
Post by: gardner on October 07, 2017, 04:00:52 pm
I have a project that involves a DEC VT220.  I have clean looking 1983 late 1986 one with an amber tube.  The tube lights up and I get a "keyboard error 4" diagnostic and the LEDs on the keyboard all remain lit.  I have the EK-VT220-TM-001_VT220_Technical_Manual_Nov84.pdf and what looks to be at least a partial schematic that I downloaded from somewhere.

So far I have confirmed the keyboard cable is alright and the voltages inside the keyboard check out.  I'll next sniff the comm lines to the keyboard and see if the two units are talking.

I'm curious if anyone's done work on a VT220 and has pointers what is most likely to be failing.  This one was last known working in about 1999 and could have physical damage internally, but most likely some of the elcos are failed.

EDIT: my unit turns out to be a later one, made in December 1986.
Title: Re: Repairing a VT220
Post by: fourtytwo42 on October 07, 2017, 04:53:51 pm
If it's working well enough to display an error code there isn't much wrong with it so I should hitch up it's serial port to something and see if its able to receive and transmit. I imagine the manual you have would have a list of error codes anyway!
Title: Re: Repairing a VT220
Post by: Gyro on October 07, 2017, 05:59:06 pm
Damn, last time I used a VT220 it looked so modern, what happened!  :(

Yes I know, I got old.

Keyboard error 4 does ring a very faint bell though.  :-\  It's a keyboard micro reset error - something to do with dip switches on the underside of the keyboard maybe??? (It's a very faint memory, maybe nothing at all)
Title: Re: Repairing a VT220
Post by: pfrcom on October 08, 2017, 02:14:53 am
Faint bell for me too - could it be a key stuck on the keyboard, or something holding a key down, while the terminal is powering on ?
Title: Re: Repairing a VT220
Post by: IanB on October 08, 2017, 02:21:38 am
Damn, last time I used a VT220 it looked so modern, what happened!  :(

Yes I know, I got old.

The 320 and later models did get slimmer and sleeker.

But if you think that looks clunky, imagine what the VT100 looked like!
Title: Re: Repairing a VT220
Post by: gardner on October 08, 2017, 03:35:32 am
I desoldered all the elcos and checked them out with my part tester.  They are not shorted and seem to have plausible values.  While probing around I managed to blow the fuse on the keyboard which is actually okay since I can now power it from a bench supply whilst leaving the main unit connected.

The keyboard is pulling a somewhat concerning 200mA and the 7805 gets pretty warm if I leave the keyboard powered too long.  There are no other signs of parts getting warm, so I am not sure whether this is a problem.  The 7805 is normally attached to a large steel plate and is amply heatsinked when the keyboard is assembled.

Probing the serial lines was educational.

could it be a key stuck on the keyboard

That is what the reset sequence from the keyboard reports.

Byte 3 keydown on powerup ( 3D hexadecimal)
Byte 4 the first keycode detected if there was a key pressed on power-up

E0 is the "7/&" key.  Nothing obvious, but I will have to have a closer look.

The initial 01, 00 are the expected hardware and software ID for an LK201 keyboard.  This one doesn't have DIP switches, only solder bridges that could not normally be changed.

Further investigation will have to wait 'till Monday.  Thanks for the ideas.  I'll follow up.
Title: Re: Repairing a VT220
Post by: SeanB on October 08, 2017, 09:02:20 am
200mA is kind of normal for most MSI TTL IC's in there, and yes a 7805 will get hot doing that without a heatsink. Check no switches have become stuck, and at lerast operate all of them a few times with power off in case there is just dust sticking them. What switches do they have there, many used a hall sensor on a ceramic carrier, or used a capacitive switch contact, which do mean a modest power draw when scanning the matrix.
Title: Re: Repairing a VT220
Post by: Gyro on October 08, 2017, 12:31:40 pm
Damn, last time I used a VT220 it looked so modern, what happened!  :(

Yes I know, I got old.

The 320 and later models did get slimmer and sleeker.

But if you think that looks clunky, imagine what the VT100 looked like!

Yes, I know. When I worked for DEC I had a VT125 (Regis graphics) as a home terminal. We used to have them so that we could dial into the office and work from home when we wanted to. Looking back, I guess there weren't that many able do that for technical jobs at the time (pre-PC era), at least in the UK. I had a VT52 for a while too (the one with a relay mounted underneath the keyboard chassis to simulate that familiar teletype mechanical feedback feel).

Sorry, I digress!
Title: Re: Repairing a VT220
Post by: Gyro on October 08, 2017, 01:40:32 pm
To make up for my digression, here's the LK201 keyboard interface spec. It indicates max consumption of 350mA max. @ 11.8V (4.2W), so it's not unreasonable that the 7805 is getting quite warm... (
Title: Re: Repairing a VT220
Post by: gardner on October 10, 2017, 12:26:55 am
Well simply operating the key's didn't fix anything, but after disconnecting the flex connections to the key matrix, the keyboard boots okay and the terminal boots and self tests successfully.  I've had a poke at the key matrix.  There is nothing visible under the keycaps.  The switch mechanism is a spring leaf that runs parallel to the keyboard plane and the key presses down on it to make contact somewhere underneath.  The keyboard is made up of a number of modules all heat-staked into the steel frame of the keyboard by about 10,000,000 heat stake posts.  It looks like someone's already had a go at one of the modules.

Looking over the mechanical matrix assembly I can't see any parts that would be harmed by a soaking, so I think I may try flushing the whole thing out in water before attempting a disassembly.  The docs do show the pinout of the flex and I may have a try at discovering what sort of impedance of crap is in the phantom switch closure.
Title: Re: Repairing a VT220
Post by: gardner on October 11, 2017, 03:00:43 pm
A more careful look at the manual and some poking at the key matrix shows that the actual switch matrix is formed by the large sheet of flat flex folded back on itself.  There is no simple way to flush out something that is between the layers.  I have tried to shove the tail of a zip tie down between the membrane layers but it is a bit too thick to go in very far.  It's looking like I will have to attack the 10,000,000 heat stakes.
Title: Re: Repairing a VT220
Post by: gardner on October 14, 2017, 03:33:23 am
A bit of key matrix surgery and I tracked down all of the problem keys -- in the end there were at least three. They keyboard now works but I will still have to do some surgery to replace the fuse.  I don't have an appropriate 500mA or 700mA one handy.

The flex connectors on the circuit board are an absolute bear.  There is a bunch of spring fingers that press the flex directly onto the corresponding traces on the board.  Getting the flex located and the retaining clips positioned then the tabs bent over without anything moving is a challenge.

Title: Re: Repairing a VT220
Post by: gardner on October 18, 2017, 06:58:05 pm
In the end the only solder-in fuse I could easily lay my hands on was 2A -- 8 times the current rating of the correct (400mA) fuse.  But it's all back together now and seems to be working.

My next task is to figure out how to inject TTL level serial into it.  I've had a look at the schematics, but the only version I can find online is from the Internet Archive:

This is where I found the technical manual and the scan of that is quite good, but this one of the schematic is quite difficult to read in some areas, particularly around the EIA and 20mA loop input multiplexer.  The JP2s are a bit better than the PDF, but still pretty crappy.  If there is a better copy of this schematic available, that would be quite helpful.

My vague plan is to disconnect the 20mA current loop inputs and take these over for TTL level inputs, then I can still use the EIA-RS232 inputs if, for some reason, I need to do that.  The terminal setup screen can be used to choose either.  I may have a go at the ROMs to see if I can change the label in the UI for the current loop selection.  Anyone know offhand if the ROMs are socketed in this?
Title: Re: Repairing a VT220
Post by: gardner on October 19, 2017, 03:00:18 am
So now that I'm back at the bench and can actually get the board out, it turns out that all the pouring over the schematics was a total waste of time.  The board in this unit is totally different from what the one in the internet archive shows.  This one is a good deal simpler and looks to be using a Signetics version of the 8051 and a Motorola USART.  I can see what looks like the opto-isolators and the transistors that form the current loop current source, but it is a much more blind game.

The ROM is socketed in this version, but it is just one chip.  I don't recognise the part number -- another Signetics part CN83815N it looks like.

Title: Re: Repairing a VT220
Post by: rhb on October 29, 2017, 12:08:10 am
This is a fine mess you've got us in.  Now I'm wondering if my Falco still works ;)
Title: Re: Repairing a VT220
Post by: gardner on November 15, 2017, 08:33:41 pm
After a bit of probing and experimentation I found suitable points to connect TTL level TXD/RXD into the points that formerly handled the 20mA current loop connections.  I pulled out the opto isolators and just left the current loop circuitry hanging.  The RXD signal seems to be an open collector and sources barely a few uA at high logic level, and whatever the TXD is, it accepts a 3.3V high alright, so I have just hooked them straight onto a 3.3V Raspberry Pi for testing.

The long range project is to embed the Raspberry Pi with it's software stack right into the terminal.  Although there is a lot of room inside the case, most of it is open to the high voltage fields of the CRT.  The logic circuit is squeezed in next to the outer case fairly closely, but I believe I have room to stick a Pi-zero in there with some double-sided tape on top of the EIA-RS232 transceivers.   The SDCard storage on the Pi is pretty slow, but even so I have SIMH running a Microvax and VMS 7.1 and the overall performance feels like a VAX750 I used to work on.  The Pi has more memory, storage and a faster CPU, but the storage speed is really awful, even for an old VAX.

I wanted to customize the strings in the VT220 setup screen to make the setup screen reflect the repurposed 20mA current loop network connection.  I was able to read the ROM in my TL866 and have been poring over the 8051 code.  I can find the strings alright, but they are broken up and rearranged strangely like the string package has been compressed somehow.  I do not believe I will be able to accomplish my mission here without discovering the exact encoding algorithm that's been used and that is likely lost in history at this point.

My remaining tasks are to tinker with U-Boot a bit more and the FreeBSD init process to get the startup of the Pi software to work right.  Once this is done I should have a somewhat amusing nostalgia piece packed into the stand-alone VT220.
Title: Re: Repairing a VT220
Post by: landspite on August 12, 2018, 06:03:59 am
Hi, I have a bunch of VK201's which I have been trying to repair (get at least one working). I'm curious did you remove the heat stakes to repair your keyboard, if so, what did you replace them with?

Also did you end up soaking it, I have heard some people run them through a dish washer, sans heat/dry cycle...
Title: Re: Repairing a VT220
Post by: james_s on August 12, 2018, 06:41:25 am
I fixed the keyboard for my C-Itoh CIT-220 which is a VT220 clone, it had a failed 7805 in it.

I have a VT-220 too, seems like there was something odd about it too but I don't remember what it was.
Title: Re: Repairing a VT220
Post by: gardner on August 23, 2018, 03:00:40 am
I cut the melted-over part of the heat stakes off from the back using a sharp knife.  To re-attach them, I held the modules in place while squirting some hot-melt glue onto the back.  They're not all that strong now, but the key modules wind up sandwiched into the keyboard and the heat stakes don't get much stress.

I never tried freeing the connections using liquids.  I doubt it would have worked, and it could have left some inaccessible residue between the membrane layers that would require even more extensive teardown than already needed.
Title: Re: Repairing a VT220
Post by: markce on September 14, 2018, 05:00:39 pm
The VT220 was very sensitive to spoiled coffee between the keys. I took a damaged one home and showered it the same day. Still did not work reliable.
One other point. The series VT220's we used had a design error with the mains filter. When you pulled the mains cord with the terminals RS232 connected, it blew the interface.
DEC replaced 100+ filters at our site.