John Stephens | TapeOp

Once upon a time, there was a company called Stephens Electronics. They went head-to-head with Ampex, MCI, Studer and 3M, building tape decks that some folks still worship as the best sounding in the world. A conversation with John Stephens, the man behind these tape decks and their unique construction, is both fascinating and frustrating. His mind darts back and forth between different stories and ideas — one minute he's telling me Einstein was wrong — the next minute he's telling me about a speaker system in his home made up of multiple 6" x 9" speakers on a piece of plywood. There may be parts of this interview that don't even make technical sense — believe me, I tried to sort all this out — but I think what we're really left with is a glimpse into a mind that could create some of the best recording equipment ever made, and an understanding of how it was able to make intuitive (or guided, as we shall see) leaps with technology.

What happened after your military stint, what would that be, in the '50s?

In 1956 I built a solid state mixing console out of transistors, which are low voltage devices. I used pins to attenuate, transfer the gain of the mic preamp. We had put a plug in the jack — just a single pin — that would change the resistance of the tree, so that's how it could attenuate.

Who was requesting that you build a console? How did you find work in that field?

It was — remember Julie London?

Oh, yeah.

The vocalist.

And what records?

Liberty. I don't know how we connected, but we did, and with the help of another person who designed a tape deck, which had the capstan, but he had some unusual things and he wanted me to do the electronics for that, which I did. Essentially, that's what I did. He did the layout on the console and I did all the electronics. So in 1956 we started up. That's where The Chipmunks came — oh that's another story.

David Seville?

Yeah, do you know who gave them the idea? Me!

To do the speed change?

Yeah, but I didn't think anything about it. I thought it was cool. My girlfriend and her girlfriend loved to sing so we started playing around with half speed.

Then clicked it back up?

Yeah and clicked it back up and [sings] "With a little bit of luck and..." You know that one. That was the one we used and it came out as The Chipmunks — I still have it.

Les Paul [ Tape Op #50 ] had done stuff before that with the guitars.

Yes, oh yeah! He was great!

He didn't play with the vocals that way. He'd layer the vocals. Mary Ford would overdub multiple, amazing layers.

I can still hear it. So, I played it for him, and Ted Keith was involved with them and I was working in the booth doing more work — doing improvements as we learned as we built. This was the first. RCA came out with the first mixer — a relatively simple one.

They had tube mixers. They probably weren't solid state until the mid- to late '60s.

Well somebody came out with one that I think beat us by a few months.

Transistors were fairly new at that point too — you had to work on a lot of limitations with the noise of the transistors and the quality of construction, too.

Well yes, that's true. Big time. I did work with silicon transistors. Not too long after I left the Navy, I found myself working for JPL [Jet Propulsion Labs] and they were flying V2s at White Sands. Wonderful, wonderful time. It was just amazing what they were doing. I was given one project — I had a chance to use point silicon transistors, where it had a barrel with a little hole in the side of it, and you had two connectors going up on a fiber — and there they had little hair springs — just like a crystal set — and you plug the thing in and adjust the pins and everything, and throw some silicon grease in there and that was it.

Sounds kind of fragile.

Well, we did shake table stuff with it. The things they'd like to do — take a miniature vacuum tube and put it on the shake table and literally break it down until there was nothing.

With technology like that, at any given time there are many people working towards the same goal that aren't in touch.

Yeah, they have no idea. I took one of them — they were doing all kinds of things on the missiles to learn more about them so that they didn't break down so quickly — before the first flight! One of the things they did was run cables from the tip — from the nose to the rear end for them, and then put sensors — string gauges, essentially — in each of the four cables, and I'd put them on, transmit them to one of the large Ampex machines running a hundred and twenty inches per second. Get it going and they're off. That's how we transmitted — then see what happens before the thing oscillates to the point where it breaks apart. So, I made one using a transistor. Then I did random noise that was not random. That was a very interesting project. The idea was, we wanted to scramble anything from a missile so the enemy couldn't tell what was going on. So what you need is a random noise. It sounded like white noise.

But then you'd pick it up and decode it, like picking out the noise.

Right. And you have a key, which is the start of where it is, and you know the clock rate and it works.

I've been reading a book on spy technology, so I've actually come across some of these things that you're talking about, strangely enough. One of the excerpts about code dealt with creating something where you fill in with noise where there seems to be nothing, but then you have your own proprietary filtering to bring the signal — to clear it away.

That's exactly what we were doing. Essentially what we did — I had a number of stages — I think about eight — so you'd have one, three, five, seven, nine and use that and it's white noise.

You ended up back in this area making consoles for Julie. Was this directly for Victory — for the label's studio?

I just did one of those. I needed something that was fun to do. Do it once, well that was fun. Now what can I get into?

What year did Stephens Electronics start? Was it with the tape decks and consoles?

The '60s — well the '50s in a way, because very shortly after building Julie London's, I started building amplifiers — one was a 3-track, 1/2 inch Ampex. I did a lot of work changing the electronics.

Taking the stock version of the Ampex deck and putting new electronics inside — was this a transition from changing it from tube to changing it from solid state?

Mmmm hmmm.

So you were getting into repro and record electronics?

While I was building that console I built one for Vegas. I forgot the name of which one it was back there, but it was back in the '50s or '60s.

One of the large casinos?

Yeah. A second copy went down to a place in Texas — a little studio there. In those days we didn't have the miniature thing because the transistors were still fairly large, so we used the industrial relay plastic cans with octal sockets. I built a couple of intercoms for aircraft.

Were you hired as a freelance designer on all these different projects?

Yeah, I was building magnetic amplifiers at one time. Last minute, I was asked to rebuild a twenty-eight volt, four hundred watt supply and it had to be just so large. They started it and it didn't pass — I took the same case and did it. That went out to Cape Canaveral, to supply power to one of the missiles.

Were you doing a lot of listening tests for things? I'm always curious about that end of it, as far as designing electronics is concerned.

We have an open air theatre in Burbank. The conductor of the orchestra of the theatre [and I] got to know each other very well and I said, "Can I record your orchestra?" — because I was there singing as a group — light operetta and stuff like that. I took the microphone [he had designed one earlier] and started building them at night there and trying to couple it out — in fact there was one company that hired me to try to do it and I couldn't do it at that particular time — to couple it well enough to have a decent signal-to-noise ratio. As you go through this stuff you just learn things. In those days, it was not stereo. Everything was mono. You had the seven inch or five inch reels to 1/4 inch reels — I still have it. I went out and found one with a capstan, with a good flywheel on it. I've always worked in machine shops. I have my own mill and lathe — I could machine stuff and that's what I did. I modified the players so that the wow and flutter were very, very low on it and then put [on] a larger capstan — double the size. I wanted to go up to fifteen [ips] and I did. I did mastering stuff on that machine because I went in and tore the guts out of it. It was a vacuum tube, and I learned very quickly feedback loops are good and a feedback loop around the record head is great. Nobody uses it.

Right — it's one of the novel things about your tape decks.

Yeah, and I do it with the repro the same way and it knocks — it does something to it I don't understand. So anyway, that's what I was doing. I was recording it every time they were doing their show.

We all know about the Stephens tape decks. What years were those being made?

Where's the manual? I don't think there are any dates in here. I'd have to go back and see. Well, you can start with the tape recorder. In that era I was starting to do audio — big time — especially with speakers. I never built one for anybody — that would be early '60s, and that was when I rebuilt this reel-to-reel, 1/4 inch machine, which I used a lot.

I assume the 2-inch machines would be early '70s.

It would be.

Because they're descended from the video recording machines, I guess — initially the transports that Ampex was doing.

I have to go back — I was building audio. I didn't have a deck — 3M did. They were willing to sell me their one inch decks.

The whole machine — audio electronics and everything?

No electronics, just the transport. We went that way for two or three years until my sales were going faster than theirs and they said, "John, we have to stop this. You can order in as many as you want and that's it." Well, I knew it was comin.' You know, their isoloop has a real problem. I checked out their patents and there was only one that would affect me, and that was the differential diameter of the capstan. I came out with an improvement and made it sinusoidal because the path — where the diameter changes — after you run the tape over that type of capstan it wears away the oxide at that point. I built one and put a patent on it. I started playing around with it — and had DC motors on it — and improved the way it worked with the whole thing, realized it always crapped out, found how to take care of that, etc. It had a sixty- cycle motor, which was noisy. One of the problems with it was the tape speed could change — it's not very much but it does. So [I thought], okay if I put an encoding disc on the reverse outlet I can weed that out, change the motor to a DC motor and that's what I did. Before you start, have a loop and maintain the right speed of the path. That started the whole thing. Then I started playing with controlling the speed by the take up motor — finally discovered a way of doing it. I used to show — when I was selling them — at the AES shows and one year when we were in New York, Ampex came to me and wined and dined me and said, "We'd like you to come with us and we'll give you anything you want, etc., etc.," and the next day I said, "No thank you."

In my estimation, especially when noticing the amount of machines that run, you didn't make as many machines obviously, as say, Ampex or MCI.

We were number three, which blew me away.

Where was your factory?

The first place was right next to the Burbank Airport. That was a two-story building — and I have intuition and I have a guide and you know, it doesn't kind of talk to you, it just kind of nudges you quietly. I don't know how to put it, but it's been there all my life. That's why I said no to Cal Tech. I could have worked in the NASA program, which I would've loved, but I said no. So many times I would say something and not know why, because there's a path. Anyway, the first shop — here's what happened — intuition — change where you're living. Find another place. Rent one with a machine shop underneath it. I find one and it's right near Burbank Airport — perfect. I said, "Is this available upstairs for rent?" He said, "No. It probably won't be. He's been there for years." I said call me if it is and gave him my phone number. He called me the next day and said, "He's moving out."

Lucky.

It's not luck. After a while you say, "Thank you, thank you, wherever you are." So that's where it started, and I worked by myself for a long period of time. I worked with some motion picture stuff. The great thing is, the landlord, or the machine shop guy, made a deal — fifty- cents an hour to use his equipment. So I started building. That gave me the ability to work with the lathe and really build stuff. The motors I used came from Eton and what they would ship is a brush ring. The rotor had no shaft — it's a hole in the middle — and the stator [the stationary coils], which is a magnetic structure that they put together, and I built the casing.

Because you wanted it a certain way?

I wanted to have a hub on there. The bearings are huge because I knew the first time it's out, it's going to land on those hubs. Somebody is going to knock it over and it's going to land on those hubs. I don't want that shaft to bend.

Or to bend from tension.

Well, tension wouldn't do it, but that's what I did. I built them all. That was the final thing that happened there. I tried various ideas. One of the problems was the encoder. I could not find an accurate encoder for encoding the signal. The encoder was a photographic disc that was mounted on the base of the capstan drive — the one in front.

You're talking about the thing that's on top that you can see?

No, underneath. It just spins in there with lots of spokes.

It tells you the speed.

Yes. We take that pulse to handle the speed. So the more accurate that pulse is and the less jitter it has, the tighter the servo will be.

Because you're creating an electronic loop inside of there to keep it on task, basically.

Yeah, and up to a certain point, gain — the change of the gain of your servo will help. After that point, it's now affected by jitter. You can crank the gain way up on the servo and it will still remain stable and it won't improve it at all. It was mind blowing. I didn't realize that. I finally took a big sheet and built a strobe — made one out of tape and so forth, and by that time I was doing — I had a dark room and ended up using Kodalith [Kodak Lithographic film — you can project patterns on it in the darkroom and the contrast is very high, so it comes out as clean, solid black on clear] and learning about it and how much of a disaster it can be and what we were trying to do. As that disc rotated, the light was totally homogenized across that surface. You will find you have a slight amount of wow, even though the Kodalith is a sharp, cut-off type of film.

But the light would still be uneven in there.

Yeah, it will still come in there. Finally, I took some more light as high as I could and made sure the tip of the glass was good, so when it got down here it was going to be that way. I had to take a longer exposure time, but that took care of that problem. We had all kinds of problems with Kodalith. We had to go and get powder, shake it up real good — it's got to be relatively fresh. Don't go for any soup because there could be problems. I got a list. Twenty commandments.

There were a lot of obstacles to overcome to push the tape deck towards the higher perfection you were looking for.

Yeah, and before I went to that, I went to gears. The reason I went to gears is because, we had a company not too far away from the airport called Grove Gear and I went to them because they were building precision gears for missiles.

Right. So the tolerance is really high.

Really tight. I said, "Do you think you could do this? This is the number of teeth I want and this diameter, and I want it open for this type of bearing," so on and so forth. "It shouldn't be any problem." I said, "Okay, I'll take three." And I brought them back. "I mean, they're good but," they said, "How'd you figure this out?" I told them. "Can you do that for us?" So I made them a little thing that they could read out on a scope.

To see the tolerance?

And they said, "Oh, he's right!" Then they came back and gave me three gears and they worked. One guy in Beverly Hills still has that type.

Where is the gear?

As a strobe — and you rubbed the edges of it so you have a light source and a pickup like this.

Kind of like a Hammond wheel.

Yeah. It worked — not great, but it worked. It was close enough. It had wow in it, but most machines did.

Yeah, it's kind of the nature of the mechanism — there's going to be some percentage.

Here's another curious thing, I found a way — I'm going to FETs. FETs are a different game. I'm still working with them because they are unusual animals.

They can be a little tube-like, in a way.

Yes, they can. They have some interesting characteristics that can be used. And that's what I did — and now I'm modifying all my tape recorders with this new assembly — in the later models that whole assembly adjacent to the encoder disc unplugs and the sensors are mounted like this, looking at the edge of the disc. It's sliding into your light source pointing upwards and you have three of them. Two of them are one hundred and eighty degrees away from each other and their signals are mixed together, so if there's any wow or anything, it cancels. Then you have one electronically ninety degrees apart — that's the third one — for direction information. I found that if I use the FETs instead of transistors, I can tighten up even more — the cut-off points and the cut-in points of the edges of the spokes.

Okay, so just reacting electronically.

It's absolutely amazing. I've seen it on a scope where it's doing this and now it's just a little bit and all I did was do that. You have to treat it in an unusual circuit for it to work, which took a while, but I did it for this guy in Beverly Hills on his machine and it's one of the best machines I have.

Is it twenty-four?

No, sixteen. When I did this isoloop, I tried to find somebody that was doing it and there was. There was a company in Texas. They were using that design for an earthquake recorder and I got their permission, because they had a patent on it, to use it as long as I don't build any earthquake machines. It was just a handshake.

Why do you think that no other companies — I mean, they were definitely checking out your machines — many people say that your machines sounded better than anyone else's analog tape decks.

Who knows?

One thing with the design of these decks — I've noticed there's nothing in it!

At the shows, that was one of the problems.

They're looking for the parts — where's the hidden box?

I got a call from somebody who said, "Do you know what Ampex has in their lab?" "Can I guess?" "You'd be guessin' right!"

I thought you said your friend called you and he told you they said they'd taken it apart and they couldn't put it back together.

No, they couldn't figure out how to build one. It was too expensive or it was too precision or it was too something. I think there's certain things that were going on. I wasn't playing around with just audio, I played around with physics — and if you notice, the power supplies are positive ground.

How do you do that?

You use NPNs instead of PNPs, etc., or turn them upside down so your high voltage is negative. What's interesting is, that gives you a better signal [to] noise ratio of your first stage. [God, now I know why these things were so weird to work on. -CM]

Okay, for an amplifier stage.

Because, you know — we better stay here, but we'll get into physics and what that whole group of people have done to us. Everything is symbolic. Everything they say is a symbol of the real thing. You have to know the code and understand why the code is very, very necessary for you to understand it like that — snap of the fingers.

You were doing this stuff with the loop — you know, the feedback loops.

Which made it simpler. Everything became real simple.

Would that eliminate parts?

Yeah. The ATR-100s — that's what they came out with. What do they need all of that for? How much power is that drawing? Oh my God!

What year did Stephens Electronics close and stop making tape decks?

Around 1974.

So you had a run with the two inch deck — only really a run of three to four years?

Oh, it's got to be longer than that. I'd have to go back and check.

What year is my machine from?

I don't know. On the back plate there's a serial number. I have it in my book.

There are user groups — there are a handful of guys in L. A. that keep in touch with each other who are all using these tape decks for various sessions — like Mark Linett, Rob Schnapf and producers like that.

Really? Is this just for Stephens? I'll be darned.

They haven't heard a fully modded deck either. I think my deck is the most modded of any — I mean, when you did the low frequency extension on it, it got down to what — you said it was flat to thirty cycles.

Something like that. I have in the curtains, a frequency response that is twenty-eight, twenty- nine cycles to let's say, twenty thousand, plus/minus point one. There is no bump.

What speed does that run at?

Thirty.

Really? And there's no head bump?

No head bump.

How do you get rid of head bump?

Very carefully! It took two and a half years to develop the heads. When I was developing this thing I wanted to figure out a structure where I can plug any size, any number of tracks, and I won't have to do anything. I won't have to rewire the deck and that's what we added? There was with a company in Goleta which is now gone — took two and a half years. I met them just at the right time. They had just started their outfit. They were working out of a garage at one time. I said I wanted to do this, so what we started doing was experimenting. One change and see what that change does because you ask, "How do you build these things?" "Go away boy." So, you learn — and it took two and a half years. One of the major steps we did — how many tracks. It was getting to the point where we saw 12-tracks show up and that died. The next step was sixteen. We started one, two, four, eight, sixteen — twelve died, so possibly the next step — that was twenty-four.

What about thirty-two and forty now?

Here'swhatIhadtodo-Ihadtofigureouthowbigcanwe get it (the tape) and still have the same noise ratio. That's what was amazing about the whole thing. The signal [to] noise ratio of this machine even then was so small that when 3M came out with a new, low noise tape, it was worse on our machines — and they didn't believe me so they sent a tech over and he said, "There's no noise." Yeah! When you stop the tape, the noise level goes down. What happened was, the earlier stuff was quieter, but it was below the noise level of the average tape recorder, and our machines were below that noise figure, so the earlier tape was good. I digress. We checked out forty. We could do it. The signal-to-noise ratio, crosstalk, everything else was fairly decent. We tried forty-eight — could not do it. It totally fell apart. To play that stuff I had to have special guides to keep the tape...

To keep the tape from jumping up and down?

Exactly. It was night and day and I said, "That's it! Okay, we got that solved."

When you were adding more than 24- tracks, were the electronics for each channel changing at all?

No, exactly the same, because we had the feedback loop to run everything, so we didn't have to worry about that. Mismatch was cancelled. On your paths, on repro, there's only two plug-in audio modules — one for the tape itself and the second one for the line amp. In record, one module.

Just to send the signal in?

Just to load the signal that's coming in onto the head and apply the feedback loop and so on.

Pretty minimal. There are no transformers on Stephens decks?

That's right. Before I go there, when I started building stuff — tubes — people liked tubes and you very quickly find out why — because they were warm and feely. Transistors are cold and tubes — you can plug in, but you can't do it quite as well with a transistor. Okay, we'll put in a case. Octal socket...

That's the reason for the little op amp assemblies?

Yeah, they're nine-pin and they get warm.

It also probably works well for troubleshooting, you know — "Pull one out, put it in, is the channel working?"

The other nice thing is because of the design, you can unplug a module and plug in another module in the same place and not have to align. Nothing changes.

What bias frequency are you using?

Somewhere around two hundred kilocycles.

That's a little higher than some decks use, too.

Yeah, and we did tests at Producers Workshop. I said, "I want to try something." I'm going to double the frequency of the bias of this particular — I think that's what I did on a 16-track because it was a difficult thing. It had to be synchronistic to the erase bias. It's going to be double the frequency. It's going to have to have special relays, so on and so forth. It was a mess to do but the idea was double the frequency, four hundred kilohertz.

Yeah — get it further from the audible spectrum.

Yeah — and see what happens.

Did it work?

Yes. Big time. "Yes, we can hear the difference! Wow! That's great! Can you do it to the others?" "Uh-uh." It's too complex.

Not cost effective?

Yeah, basically. You could hear the difference. It was definitely an improvement. I keep wondering, trying to figure out why you can take any product, record and play it back on a Stephens and it comes back to you better. How does that work?

It's doing something. "Euphonic", is the word for it. But it's hard to pin down what's euphonic in electronics.

Yeah. I have a strong suspicion, but I haven't done much checking out. But it's in relation to the distortion because certain things are happening — our distortion curve is almost a mirror image. So let's say we have one tenth of one percent distortion and there's something else that has one tenth of one percent distortion, and one is going to house the other's negative. They cancel, and that's what I suspect is happening.

I think Pink Floyd bought one of your machines themselves to take back to England because they were so taken after working with it in L. A. at Producers Workshop.

Grateful Dead. Pink Floyd.

TheCars.