Source: ANTIC: The Atari 8-Bit Podcast
Interviewer: Kevin Savetz of ANTIC: The Atari 8-Bit Podcast (www.AtariPodcast.com)
Michael Phillips worked as a bench technician at Atari from February 1981 to June 1984, doing component level repair of Atari video game systems, personal computers, and peripherals.
Because Michael is a lifelong stutterer, he didn’t want to do a voice interview — but he was willing to be interviewed by email. Because this is an audio podcast, I’ve enlisted Randy Kindig to read Michael’s responses — you can hear that audio version here: http://ataripodcast.libsyn.com/antic-interview-35-michael-phillips-atari-bench-tech
The interview was conducted via email, February 2015.
“Beating the device in question...was part of being a good tech. The key is knowing how hard and where to hit.”
“Misspellings, bad English and odd terminology were par for the day. One guy once referred to the I/O cable as a ‘hose’.”
“One I vividly remember was an 810 [disk drive] that came back 3 times. The guy claimed it would randomly erase disks, but we could never find a culprit...”
How did you get hired by Atari? Looks like it was your first (real?) job after school. As I mentioned in our pre-interview dialog, I've had a stuttering issue all of my life. So, getting through school was stressful to the point that I had no interest in dealing with college. Plus, a neighbor had gotten me a part-time job at a print shop, which I worked at in my senior year. I kind of liked it, so I decided to go full-time after graduating in ’79. But it quickly became clear that what was fun part-time was not so enjoyable full-time. It was a very physical, dirty, on-your-feet job. I think I lasted 3 months before saying adios. So, now I had no career lined up and needed a plan. I liked to work on cars and had a strong audio hobby, so something hands-on sounded good --but what? Being I was from a barely middle-class family, whatever training I was to get had to be on the cheap. A friend turned me on to a state-based 6-month vo-tech program that would pay me $3.25/hr… just to attend. I was like, "sign me up"! I wound up choosing electronics for 2 reasons: One, it was the only trade listed that I didn't understand that well, and two, I already had the audio hobby so it seemed like a natural fit. We were taught using the Heathkit course, which was very basic but good for getting an entry-level job, and I received my Certification in late 1980. I started mailing resumes and landed an interview at Atari. They were growing in leaps and bounds at the time, and I was hired in February of ‘81 to repair the 2600 systems for $5.42/hr. It wasn’t that bad for the day, plus they offered good benefits. We had a stock ownership plan (which I still own remnants of today) and a “sabbatical” – every seven years you would get an additional seven weeks vacation on top of the three you would have earned. We all fantasized about that one! Of course, none of us got even close, and I think they dumped the program for newer hires after a year or two.
Tell me about the repair hub. Where was it? It was a warehouse on Belmont Dr in Somerset (NJ), but as warehouses go, this one was nice. There was a small reception area where people could drop off their systems for repair, although most were mailed in. It was kind of surreal; this big, open warehouse with a large expanse of carpet for the repair and office areas, the latter being walled off of sorts by shelves and filing cabinets. I personally liked the open environment.
I joined a team of about 7 techs doing only 2600s at that time, being the computers weren’t widely in use yet. Our workbenches, while being nice in that they were a thick, maple butcher block, weren’t exactly ideal for electronic repair work. The most obvious blunder was a total lack of ESD (static) protection, but in hindsight, it really wasn’t a tremendous problem.
Anyway, you’d have 1 or 2 people opening the returns and putting everything into a “carrier”, which was essentially the original box’s insert. I remember doing this the first couple of days as a way to get acclimated to the product. We would make the carriers using new cardboard and hot glue, which I had never seen or used before – in fact, the smell of that glue will always take me back to Atari without fail! As an aside, it was also used in many repair operations, one being to reinforce the early 9-pin connectors. The pins could easily be pushed backwards, so to avoid replacing the entire connector (which was a pain), we’d push any culprits back out and douse the rear of the connector with hot glue. (They eventually designed a newer version that prevented the pins for pushing back, but the glue worked great.)
Continuing on, we would fill up rolling racks with these trays for movement to the various ports-of-call. First was accessory test, where a young lady would test joysticks and paddles. If they failed, new ones were supplied at this stage. Then the racks were queued up for the techs. You were always expected to take the “next available” box, since it was regarded as bad form to pick and choose your next victim. It was fairly easy to tell quick repairs from the pigs, so we were on our honor here. Plus, the layout of the area made it easy to watch for cheating, so everyone usually cooperated.
After repair, you’d hit one of two automated test jigs to validate your work. There were no exceptions; the printout from the machine was taped inside the game with the tech’s initials and date. Each jig had a noisy little dot-matrix printer that spit out a pass or fail code. Pass was a single line of that distinct dot-matrix zipping sound, while failures were a fail code plus the fault(s) detected, meaning a double or sometimes triple sound. The printer was clearly audible to everyone, so it was often a source of giggles when someone was having a problem with a tough unit. You’d hear them cycling through the various tests, then everyone would be listening for that fatal second chirp, which was often followed by an expletive from the tech. Finally, it was on to the burn-in racks for overnight cooking. We took turns checking them each morning, making sure the Combat test game was still on and able to start. Failures went back to that tech and didn’t count on the new day’s tally.
After about a year or so, Atari rented a bigger building down the road on Worlds Fair Dr to house the coin-op guys and newly formed computer group, with the idea to eventually merge the 2600 group in after they were settled.
How many people were there?
The 2600 shop on Belmont was the largest, about 20 people if memory serves. Coin-op was small (less than 10), and the computer group started with about 5 or 6. In the peak times, I’d say there were close to 50 people combined.
What was the work environment like?
Worlds Fair Drive was far more suited for this stuff than Belmont. It was more professionally equipped, as expected since the products being serviced were more complex and expensive. There were much nicer workbenches with equipment selves, drawers and lots of outlets, and a proper tile floor to keep the static down.
I recall being a bit miffed when new techs were hired for the computer systems instead of training up the 2600 folk. (To say one got sick of fixing the 2600 after a couple months is an understatement of grand proportions.) We were told we would ultimately be cross-trained (which we were), but they wanted more experienced /better educated techs to start things off. Funny, once we merged, the resentment reversed: WE were the intruders, and a couple of the computer techs were a bit condescending at times. But with the dealer-repair network building up to provide 2600 service and computer sales on the rise, they needed help and we were all they were getting. After that rough patch, things got better quickly. Most of us were young and we ultimately got along pretty well, and lots of good friendships were formed throughout the company.
How many things were you expected to fix daily? Quotas?
At Belmont, there were never really quotas, but with the volume of consoles coming in, high production was certainly welcomed (and noticed). Before I got there, the techs were routinely doing about 10-15 systems per day. The given primary tools were Soldapults (hand powered desolderers) and manual screwdrivers, which don’t exactly inspire productivity. Then this energetic 20 year-old punk comes along (uhh, me) and starts fixing out 16-20 per day, and it was being noticed by all in a good way (and bad). Now from my perspective, compared to my print shop job, this was a piece of cake. I sitting on my ass in a clean environment, doing something I was enjoying very much. Being I played around with cars, spinning a screwdriver was right in my wheelhouse. And let’s not forget there were TONS of these to fix. It’s not like I was hogging all the work! While the management loved it and started pushing the others as a result, I quietly became persona non grata, especially when it was noticed I didn’t take breaks, either. Remember, in my mind there was no need…I was already sitting down.
To top it off, I was good. I moved fast, but made sure I was thorough. too Besides, there was no sense in spending too much time looking for hidden problems when every repaired unit was burned in and rechecked in the morning. I played the odds; a calculated risk, so to speak. If something did fail overnight, it was almost always one of the socketed ICs, which was a quick repair.
I honestly didn’t fully realize how mad some of the techs were, but after being educated by one of my friends there, I adjusted by taking my breaks and not going full-bore every day. Like any young kid, I was just trying to impress management and possibly score some sort of promotion. It finally hit home when someone was almost fired for being slow. I remember my dad scolding me when I told him that my performance was partly responsible. As someone who grew up through the Great Depression, he knew the value of a job to someone. Live and learn, I guess, and learn I did.
For the computers, after the big merge, the focus was more on quality of repair, so there was no real pressure to put up numbers, at least in the beginning. Towards the end of the run (late ‘83 to mid ‘84), we were pushed. But by then, most of us knew our time was short and didn’t really give a rat’s ass.
Tell me about the failure modes of the products, back then.
I’ll spare you the 2600 stuff, except for this little snippet. You may be aware that the early 2600, known as the “heavy sixer”, is highly coveted now and is lauded for its “quality of build and performance”. In reality, those early production units needed loads of mods to make them reliable. If one lands a modified one, then yes, the RF performance was a tad better, for what that’s worth with the inherent crappy graphics. But otherwise, they're not even remotely special. Just heavy!
On the computer side, the common failures that I recall are as follows. The early 800s’ motherboard sockets and/or plug-in card edges were tin plated, which caused crashes galore once some oxidation set in. The 400 keyboard, which sucked when working, was prone to failure. The socketed LSI ICs (the CPU, POKEY, GTIA) were regular culprits. The polystyrene capacitors (or poly caps as we called them) would frequently break and take out the audio. Also, there were always the I/O connectors that people would butcher, along with the 9-pin joystick ports. The problem descriptions were always a potential source of amusement. Misspellings, bad English and odd terminology were par for the day. One guy once referred to the I/O cable as a “hose”. Of course, that became the inside name for it from then on!
On the peripheral side, there were two variants of the 810 disk drive; one with an MPI-brand drive mechanism (which contained the head and stepper motor), the other was made by Tandon. As you probably know, MPI had the pop-up, spring loaded door while the Tandon had the less exciting open slot and lever/latch system. The Tandon, being the later variant, needed fewer mods than the MPI. Now I know head alignment problems were common on the MPI, and I think the Tandon was prone as well, but can't remember exactly. Anyway, once the head loosened and moved enough from tolerance, the read and write errors began until the drive became unusable. I can’t recall exactly the root cause or what we did after realigning to make it more robust, but I don’t recall them coming back repeatedly for that one. The early MPI drives had two voltage regulators on a heat sink that connected to the rear board via little 3-pin sockets. The socket connections would oxidize over time, causing the drive motor speed to vary substantially from the spec. So, we had to pull the sockets and manually solder them to the rear board. The side boards always had issues, but they were tough to troubleshoot in a timely manner. So, if it wasn’t one of the socketed ICs that failed, we usually just swapped out the whole board.
Let’s see, what else. Oh, the 410 cassette deck. These were almost always head alignment, known as azimuth in tape speak. The little screw that pivots the head wasn’t sealed on the early units, so it would gradually go out of whack. We used trusty nail polish after realigning and sent them on their merry, slow-loading ways. The follow-up 1010 version used to snap the control keys with regularity. We had beefier versions for replacements. The 820 thermal printer had little microfuses that would blow, the result manifested by whites stripes within the printed text.
Another big problem was physical robustness, or shock resistance to say it another way. What your listeners might not realize is how high a tendency there was for this failure mode to occur, especially when compared to modern electronic stuff. With all the card and IC sockets and less-evolved connectors of that era, a good nudge or bump on an affected unit will promptly lock or reset it, which would REALLY piss people off in the day – remember, loading software took forever, and with the game consoles, losing a good run on a game was cause for heart failure. So, beating the device in question before (to find fault) and after (to ensure proper repair) was part of being a good tech. The key is knowing how hard and where to hit. That’s called experience!
What item, when it came in for repair, were you just like "&^#%!!! One of these, again?"
Oh jeez, where to start. The easy answer is anything old that needed a full accoutrement of mods – massively time consuming. With the aforementioned 800 tin-plating problem, the original solution before swapping out for gold was a liberal application of Lubriplate (a light grease) to both surfaces after cleaning off the oxidation. Unfortunately, it didn’t work for long, so the mandate soon became “all gold”. Being the sockets were usually all gold anyway, they became filled with grease from the tin cards and you had the fun job of extricating all that muck.
A close second was ANY item that some genius decided to open up to attempt repair. They were easy to spot: Screws mixed up or missing, covers and shields not secured properly, anything adjustable was tweaked. Why people do this, we could never understand. Even by today’s standards, these are highly complex devices and shouldn’t be touched unless one has the service manual, some soldering ability and electrical knowledge. 810s were the best. People would bend the head’s pressure arm in an attempt to clean it, change the drive speed, etc. One thing was for sure: It was definitely coming in for service after that, and it nearly always voided the warranty.
Finally, there were the returns that had bugs (the real variety) in them, or units that were just disgustingly dirty. The policy was to clean everything after repair, every time. It wasn’t always that way, though. Going back to the 2600 again (sorry), it was common for people to send in units without telling us what was wrong (the frequent “dead” or “doesn’t work” really wasn’t helpful) And more often than not, they would neglect to send the power adapter or TV switchbox, both of which were prone to failing. So of course, you’d get a unit that worked fine, even after a thorough test and burn-in, and we’d ship it back. They would get it, see the same problem as before, and because it was still dirty, think we didn’t do anything and call up screaming. After that, we were all given cleaning solution, a rag and a small scrub brush. So at minimum, at least they knew it was looked at.
Were you allowed to declare a thing "dead" and have it replaced (with a refurb?)
With the single board stuff (like the 850 interface, for example), that may have happened on rare occasion. For other stuff, you could almost always drop in a new sub-assembly if something was giving you a hard time. We had plenty of junk parts, being people regularly sent us stuff that looked like it fell off a building. For those, we’d send them a suitable replacement (for a cost, of course) and keep theirs for parts. However, there was the occasional exception. One I vividly remember was an 810 that came back 3 times. The guy claimed it would randomly erase disks, but we could never find a culprit. We had replaced the most suspect boards on visit #2, but when it came in that 3rd time, we just flipped him a new one, labeled him "nuts" and kept his for shop use. It always worked perfectly, and I was allowed to take it home after we were all let go. So one day at my house, I was using it to do my resume when bingo – the SOB erased some data without being told. Bastard was right all along. I beat it like they did the fax machine in Office Space and tossed it in the trash!
Were there palettes of dead equipment in a warehouse somewhere? (I've heard stories like that from the California end of things.)
Just the "parts mules", as we called them. Joysticks from the 2600/400/800 were tossed in the trash from day one, along with the paddle controllers – we didn’t fix either one, always replaced them.
Tell me more about the coin-op repair shop next door that you got to play in. Was that an Atari shop?
Oh yeah. They were really the first group to establish here in NJ, being Atari had a strong arcade presence before the other products came along. Field guys would do board swaps and bring them to this shop, where some very good techs performed component-level repair of said boards. They had the coolest job: They got to use advanced test equipment like logic analyzers, had no real pressure to put out numbers, and had all the games set up for free play. Also, they WANTED us to come and play to beta test and replicate problems, which we happily and readily complied. Since they were only two doors down after the merge, one could get in a couple games of whatever they wanted during lunch. They had ‘em all: Along with the popular games like Asteroids and Centipede, there were the oddballs like Xevious, Tempest, Major Havoc, Battlezone and Warlords. What a blast we had!
Why did you leave the position?
Once Warner Communications sold off the company, we pretty much knew we would be shut down as rumors were rampant. Sure enough, in mid ’84, we were toast.
So what happened to all the equipment and such?
They decided to allow us to buy some of the stuff, furniture being one example. I picked up a nice metal desk, high back office chair and set of folding chairs for less than $100 total. Many of us quietly removed some hand tools and replacement parts, but nothing of great value or quantity. I had already had a number of systems at home for research and test, including an 800, (2) 810s, a 2600 and 5200, as well as a 1027 printer and all my service manuals. I sold off the 5200 system in ’04 for nearly $300, being I made sure to let potential buyers know who I was and how good this particular example was – and it was nice. I then dumped the 800 and 810 in ’06, followed by the toughest sale of all last year: The entire set of service manuals. Some dude in Germany paid nearly $300! Not sure it was a good idea, but I’ve been moving a lot of late and needed to reduce my box count. I’m sure I’ll regret selling everything someday, but for now I still have lots of parts should I want to do some repair again.
The Brunette worked at Atari? Tell me the story.
I was sitting there pumping out 2600s one day when a manager was showing this angelic vision of a brunette around the facility. This was “my 10”, so to speak, and I was fully smitten. She was hired and upon learning she was single, I immediately set out in pursuit, even though she was fresh off a breakup. (Yeah, yeah, major red flag, I know, I know…now.) We dated for about 6 months, with me trying hard to win her heart – too hard, leaning towards overbearing -- so she tired of the pressure and broke up with me. I was miserable, of course, since SHE WAS THE ONE, and now had to deal with seeing her every day at work. It was painful, but those are the risks one takes. I became far more careful with "fishing from the company pond" from then on as a result.
Do you still keep in touch with anyone from the time?
I stayed friends with one fellow tech, and we get together every couple of months. I found some of the coin-op guys on LinkedIn and wrote to them, but none wrote back, which disappointed me. Otherwise, it’s been tough finding people from that time, even with the proliferation of social media. I had one manager whom I’d love to talk to again, but being his name was Michael Jackson, you can imagine what happens when his name is Googled! I tend to revisit the searching every so often when I reminisce about that period of my life. As short a time as it was, I wouldn’t have traded it for anything.
Any final thoughts?
Another memorable moment was the surprise raise we all got during the peak period, which was almost double what I was earning at the time. My starting salary of $11K finished at $22K – not a bad ramp-up for 3 1/2 years time!
The 400-based 5200 system is worthy of mention. The console was fun to repair, and many of us thought the failure in both design and reliability of the X-Y controller led mightily to Atari’s demise. It was comical how much time we wasted fixing and modifying that damn joystick, from the flimsy rubber boot to the poorly designed fire buttons and flex circuit.
If I may get philosophical for a moment using hindsight, no home computer of that day was going to thrive and earn serious money for ANY company. As much as I admire and respect people who learn to use them, they’re an elite group for which there was never really a profitable market. As we’ve learned, it was gaming that took off, and Atari failed miserably here when they should have been the leader.
Today, while the Internet, massive increases in memory (storage, RAM, etc.) and bus/processor speed have all been vital to the growth and penetration of the PC, the platform absolutely HAD to become idiot friendly in order for widespread acceptance and use to come to being. Seriously, how many users today what RAM is or what it does? Can navigate a file tree in a directory window? They don’t get it and don’t WANT to, IT HURTS THEIR BRAINS! The market was not going to grow and thrive until the ease-of-use paradigm was solved – just look how much people pay for Apple products! Okay, getting off the soapbox now. ;-)