Repairs of ADM-3A by Nick Papadonis in 2011

This page last updated Apr 3 2013.

Nick Papadonis repaired and restored his ADM3A terminal during the summer of 2011. He and I discussed his progress, here's some of his notes and repairs. He did a lot of work but by October 2011 his terminal was operational. To see my ADM-3A's and repairs, check this Web page link.- Herb Johnson

[ADM-3A CRT leakage]

[ADM-3A CRT leakage]

0) problems from leaking adhesive

"There was some sticky brown substance that leaked from the CRT and onto the keyboard assembly. This caused corrosion on the solder terminals between the keyboard and the PC board....I think it's from the polyvinyl adhesive that attaches the lens to the CRT. My CRT has "cataracts", so I need to heat gun the lens off."

A number of people have reported on this. They remove the CRT, support it in some careful fashion, and use a heat gun to loosen the bonding material and push the "lens" off the CRT. Nick looked around the Web for more information, and describes his version of that process.

Nick found this video, titled "How to repair "CRT cataracts"" It describes how a CRT lens is bonded to the front of a CRT; over decades, the bonding material breaks down and turns cloudy. This is often called "cataracts". The youtube video by "douglas" (drh4683) produced in 2008, discusses this situation and describes a removal procedure, like the procedure above, to remove the lens without damage to the CRT.

About Nick's descriptions below: "Legal disclaimer, I am only providing a review of my experiences and am in no way providing or suggesting this is the procedure to safely fix a CRT. Consult a professional. :) " - Nick

[ADM-3A CRT leakage]

[ADM-3A CRT leakage]

1) removing the cloudy "lens" from the CRT

"There is the CRT tube and then there is the lens (what you can touch from the front). They are glued together by about 1/2" of Poly-Vinyl adhesive. It sort feels like 5-min epoxy at minute 4. Sticky and mushy. When it ages, it turns a bit yellow and leaks that yellow oil on the motherboard. Whatever the chemical properties are the wetness takes a long time to dry out, not like water."

"I didn't touch the CRT for at least 2 weeks from receiving the shipment to allow it to discharge. I then placed the CRT in a small trash barrel with the visible portion facing toward the ceiling. Some people use one of those utility buckets from Home Depot. I did not remove the metal retaining ring around the CRT because a) I couldn't figure out how to release it without damaging it and b) it looked like it wasn't entirely necessary unless the CRT was being replaced." Note by Herb Johnson: That retaining ring is part of the implosion protection of the CRT. DO NOT REMOVE UNDER ANY CIRCUMSTANCES.

"I then put on eye protection, a long shirt and pants in case mainly the CRT tipped over and imploded. I then used a standard heat gun (used for shrinking heat shrink) to heat the CRT all over. I waved the heat gun about 6 to 12" back from the lens just to get some even moderate heat applied. Then I proceeded to bring the heat gun closer to the lens, heating around the edges and working in."

"In the Youtube video [I found], you could see the adhesive separate from the lens when air bubbles appear. I didn't really see this and just continue to heat for about 5 minutes. I then became inpatient and used a rag (protection from heat) to move the lens around. It started separating on it's own! I simply had to carefully slide the lens off the CRT tube and onto a soft surface where it could cool off. Then I used a paint stripper knife (the kind with the single long blade) to dig off the layer of adhesive from the tube. It pretty much slides off in two rubber like pieces."

"In the end, I finished cleaning the tube with brake cleaner because it was recommended. I don't think it was necessary. I then used glass cleaner to clean the tube and lens. Finally, when all the streaks and spots were gone, I applied 1/2" double sided tape in a few spots around the edge of the CRT (non-visable) and places the lens back over this. The final step was to use packaging tape (I used the clear stuff, but other people used the brown type) around the lens and edge of the CRT tube. This way no dirt gets between the lens and CRT (the polyvinyl is gone so there is an air gap). That's it."

[ADM-3A CRT leakage]

2) Cleaning the goo from the PC board

"I poured alcohol down the board while it was on it's ends. Then wiped with paper towels. I let the alcohol sit for a few hours. That seemed to get most of it. I suspect that Acetone would also work as long as it doesn't touch any plastic."

The "goo" leaked under the keyboard. Nick removed the keyboard and had to do some repairs of corroded keyswitch terminals, and replaced whole keyswitches.

"It was all goo under the keyboard. I used alcohol to remove it, however some brown still remains on the PC board. It disintegrated the keyboard terminals so badly green came out when I scraped the terminal with a dental pick..."

"The substance got between some of the ICs, but I cleaned it up good with alcohol. For the ICs that are soldered to the board I didn't see any corrosion, so that was good. There was nothing for the substance to "seep" into either. Just to be sure I let alcohol run down under the necessary ICs so any substence may be removed. I removed the DIP sockets that were contaminated and rusting at parts. "

After this and other repairs, Nick said: "I melted off the polyvinyl adhesive from the monitor lens and the tube is visible again. Now I'm replacing a brightness pot and will power it up."

3) Removing the keyboard assembly, replacing or repairing key switches

How did you get the keyboard assembly off the PC circuit board?

"I flipped over the circuit board so I was facing the underside. I then used a desoldering tool to desolder all terminals from the bottom. I made sure to do a good job of sucking out the solder. If a hole still had some solder filled, I added solder to it, then removed it again."

"I then placed a screwdriver between the keyboard and pc board for leverage. I simply moved the screwdriver and the plastic keyboard assembly popped off from the PC board. It release pretty easily and I didn't force it."

Nick and I had a lot of discussion about keyswitch repair, finding other keyswitches, and related issues. It was a lot of work on his part! He bought an Atari 800 keyboard to obtain some keyswitches but they had to be cut out: "I found a 3/4" hole saw fit just around the key. My plan is to cut out the keys and transport them as necessary to the original ADM3A keyboard."

"I'm going to get perhaps a "lino tool" used to scour groves in linoleum tile to dig into the plastic and search for the remaining portion of the contact. I tried an Xacto last time and that was scary. I then tried a dremel with a drill bit to drill down a little. I went too far and broke a spring! So I'm open to any other suggestions for digging into plastic."

Later he reported success, with some qualifications, as follows.

"Atari 800 keyswitch drilled out with [3/4-inch?] hole saw (longer version). This keyswap wouldn't really be that necessary if I were just more careful restoring the keyboard in the first place. I was able to use a dental pick and scrape some of the corrosion away inside the broken terminals. I then soaked the keyboard in alcohol and used a cup to pour it through the keyswitches. This ensured most of the PVA oil/goo was cleaned out. I then used a long fine tip (SMD) solder tip to get at the broken terminals. I placed a little solder on the little portion available to me. Then I used a tweezer to hold a component lead clipping and soldered it to the end of the broken terminal. New terminal!

"[As for the swapped keyswitch assemblies,} I just need to "weld" the plastic around the holed out keyswitch together. [MEK or Ambroid ProWeld (Tenax7R) did not work.]"

"I transported [the key assembly] to ADM-3A keyboard below and glued in place with epoxy. I made sure to cover moving parts with electrical tape before filling in epoxy. In the end, the yellow part of the Atari keyswitch was about 1/4 cm to high (key stuck up above others). I had to carefully shave the yellow box and bottom of the associated *key* with an Xacto. It's pretty close now."

4) Reconnecting all the keyswitches on the PC board.

"I already tested continuity of the [printed circuit] through holes on both sides. That was my first concern. They are in good shape. Of course, I won't solder the keyboard back in place until I can furter test that. Also, fortunately the tru holes under they keyboard are not used. The routing is underneath the board which will be accessible. So is the solder points. Not sure why they had them there unless it was just standard practice."

"Most of the original [keyswitch PC board] pins are toast. Given that, I think it will be easier to solder all the bare wire directly to the terminals. Then pull that through for everything. Then alignment issues are reduced. It's a matter of picking the correct wire for the hole."

"I was able to transport all keyswitches to the ADM keyboard. If you look at the attached pictures component leads were soldered to many keyswitches where the leads corroded due to PVA leak. I tested continuity of each keyswitch on and off everything works. Now I'm trying to solder the keyboard leads back and realized without the original terminal ends (they were aligned at the factory) it's almost impossible to get the leads to line up into the pc board holes."

"Yikes! Now I'm thinking about using wire wrap either short lengths to the top (facing up) side of the board or perhaps running it all the way around and under. I'm thinking under would be less stress and provide more clearance. I'm also considering using thin gauge bare wire (28 AWG or 26 AWG). Solder to terminal, use 14"+ length, pull through PC board hole, when all wires pulled through tighten slack and solder."

"In regards to picking which wires go where. I'm thinking I'll have to hoist or hold up the keyboard with two bricks about 10". I can then go underneath with the 14" wires and put them in the correct holes. When done, drop the keyboard slowly while pulling the wires through until it meets the board."

Nick followed this procedure - I think he used bare wire - and later wrote: "Thought you would be [pleased] to know I was able to test continuity of all keys after wires were soldered. AOk so far."

5) Repairs to the logic board.

"Herb, I powered up the system. I hear a constant speaker tone and the following output on the screen. [random "!" characters between blanks]. Pressing keys has no effect on the screen. The screen remains static like this. I guess it looks like character generation is working, however the orientation of the scan (too close to bottom and twisted), random characters and diagonal scan lines are symptoms something is wrong. I checked all DC voltage levels from the VRMs and at the input to the TV section. Everything looks ok."

We discussed settings for all the switches. After testing, study and examination, Nick said: "Herb, I tested out the terminal today. I had an IC inserted into the socket at the wrong pin offset. I set the terminal up for 8-Odd-1 and connected it to my laptop via [the terminal's] Port B modem connector."

"When typing characters on the laptop they appeared on the terminal CRT, however they did not always match. For instance, if I pressed 1 on the laptop, it would show up as 0 on the terminal. If I held down 1 on the laptop, it would show up mostly as 0's on the terminal and then a couple 1's here and there. Some keys mapped 1-1. I noticed "W" mapped to "V" and would switch correctly to "W" after holding the key in for a full line of characters."

"The terminal is equipped with the 24 line U-L case option. I set the terminal to full duplex and saw the same behavior. Other then the character generation issue I just described, the terminal also appears to "wipe out" a line every few seconds. So I would type maybe 70 characters, they would be displayed on the screen, then disappear and the cursor end up at position 0."

"This is almost like a hard reset on the motherboard. The screen flickers and it's like everything is wiped out. I would think clear screen would be more graceful. There was no other device connected to the ADM when this occurred."

I suggested running the terminal at "half duplex" so it echos whatever characters you type. And I suggested the errors in the characters were likely due to bad bits in the RAM. The ADM-3A has one or two rows of eight 2102's (1Kbits by 1). Study of the ASCII codes of the characters will suggest which bits are bad. The other problems are likely logic problems and need some study of the schematics and design notes (available in ADM-3A documentation) to narrow down where they may be in the logic. I suggested looking at the DC power with an oscilloscope to make sure it was "clean".

About a month later, Nick reported as follows:

"I found the source of the screen cursor moving around to different positions and screen flicker occurring every 10 seconds. I used the oscilloscope and saw that 5V was dropping momentarily during that time frame. I disconnected the CRT section from the motherboard and the problem went away. I then inspected the CRT and found that one of the green ground wires attached to the spring [which is wrapped around the back face of the CRT] was broken where it attaches to the CRT PC board. I fixed the connection and the problem was resolved. I suspect this involved high voltage buildup due to improper grounding."

"I got a good price on 14 new 2102 RAMs which were slightly faster. I replaced all the RAMs with new DIP sockets and then installed the new RAM. That part of the terminal is working. I need to go through my old RAMs and find the bad ones. Probably some method of replacing a single RAM and testing out the terminal."

"Now I just have a constant BELL sound. I traced it back to the key code portion of the logic. I'll dig deeper another day." A few weeks later Nick added: "I was able to fix the constant beeping noise. I traced the problem to a faulted 74123N. I replaced the cap associated with that one shot also."

6)Cursor offset

"Memory bits look good but now I appear to have a new problem. When I type the cursor is in the correct position however the characters are showing up shifted about 32 columns to the right. I don't think my rework would have impacted this but anything is possible. I get an audible bell warning when the characters exceed the 78th column. Of course the cursor is still far to the left of the screen.

"Does this ring a bell? Looking at the schematics I was thinking it was the memory addressing section. Now that the bell is correctly rung I'm thinking the character position counter must be wrong." - NIck

Herb said: See if the cursor is EXACTLY 32 positions off. "32" sounds like 2**5 to me, maybe a bad output level on a counter chip at the 5th pin? Do you get cursor all the way past 64? Does it wrap around after 80 to 32? See if the appropriate IC counter has a pin stuck high or low.

Nick said: "I checked all of the clocks and they look ok. I also checked most of the counters with a logic analyzer and those looked ok. Cursor only until 32. There is no wrap around after 80. I can press enter and the cursor returns to 0. The characters are still output starting shifted 32 columns to the right. Up down works with the cursor and characters (characters still shifted)."

After some discussion: "I hooked up the logic analyzer correctly to MA0 - MA9 and read the service manual with regards to memory addresses carefully. Cursor position 0 (offset count = 0) and Line 0 should equal memory address 1. I watched the memory address change for what appeared to be the WRITE operation of the where the cursor was positioned. With the cursor positioned at column 0 (offset 0), I saw the memory address on the MA lines == 32. I hit the space bar, so the cursor was at column 1. Now the MA lines reflected 33. Again, space, the MA lines reflected 34 (decimal). "

"I believe you were correct, this is a Bit 5 stuck issue. I traced the stuck high bit between the 1602A UART and 74LS175 higher order. I lifted that pin on the 1602A and the bit remained high. I checked continuity around the areas and didn't find any shorts. I unsoldered the 74LS175 and probed the trace. It was now low. It appears the 74LS175 was driving the input pin internally? Anyway, I ordered a new IC and will see if replacing it fixes the issue."

7> Horizontal overscan

"Another issue I encountered today while probing around, was that the display horizontal width greatly increased! Now about 10 characters on each side are being output off the CRT visible space. Not sure what happened because everything was relatively centered yesterday. Perhaps I just need to adjust the horizontal width. I think it takes some sort of allen driver (different then the var pots)." [I cautioned him about using metal tools to adjust the horizontal coil with power on - DON'T DO THAT!, it will overheat due to the increased magnetic flux.]

"Maybe the horizontal width coil was damaged when I tried placing a screwdriver in it a few days ago. I'm guessing I would have noticed the problem immediately. The problem seemed to have occurred when I accidentally bumped mini-grabber against another IC pin on one of the counters. I don't think it damaged anything, but the horizontal width had a problem sometime after that."

"I followed up trying to isolate the Horizontal over scan problem. Visually the letters appear to run off each horizontal side of the CRT. The run off causes about 5 characters on each side to be non-visible (over-scanned off edge of CRT). The focus is out on the characters near the edge. I tried turning off the monitor and re-adjusting the horizontal width coil in both directions. No change in horizontal width."

"I'm wondering if I did damage it. If so, do you know what the specifications are for a replacement? I don't have much experience in the analog section and the schematic simply says "Horizontal Width Transistor". Mfg Pn: ? BBRC Pn: 1-016-0303. I'm assuming there is a common width coil for these monitors."

Herb Johnson
New Jersey, USA
here is how to email @ me

Copyright © 2013 Herb Johnson