# How many 6SN7 tubes did it take to store a bit?

One of the most important components of a computer is a circuit called a flip-flop, which has two stable states (that it can flip-flop between, hence the name); it is used for temporary storage of a single bit, in a way that can be accessed quickly.

The first generation of computers, before transistors, had to build flip-flops out of vacuum tubes. Apparently this was done using 6SN7 tubes.

The wording of that article and the one on the ENIAC that links to it, suggests 6SN7s were used as flip-flops, which suggests a one-to-one correspondence. But I'm used to a flip-flop being made of six transistors, which suggests it would need six triodes, and a 6SN7 contains two triodes (the vacuum tube equivalent of a very simple integrated circuit), which suggests it should take three of them to make one flip-flop.

But I'm not very familiar with vacuum tube technology, and for all I know, maybe it doesn't work the same way.

How many 6SN7s did it take to make one flip-flop?

But I'm used to a flip-flop being made of six transistors, which suggests it would need six triodes,

Not really, a basic flip flop does not need 6 transistors. maybe there's a mix up with RAM cells? Two will do it quite fine (*1,*2). Similar the two triodes of a 6SN7 is all what's needed - well, plus two rather equal resistors between them and another two to restrict current flow.

Going to the wiki main page about flip-flops will not only show transitor based examples, but as well Eccles and Jordans patent drawing to do it with two valves. An en detail article can be found at the IEEE Spectrum web site.

Of course, more complex flip-flop may need a higher number of active (valve/transistor) components, but that is for additional functionality beyond storing a bit.

*1 - When thinking about basic (binary) electronics the first rule is that active elements are only need for inversion (and intermediate amplification). Everything else can be done by resistors and diodes. Before TTL there was RTL and DTL. The fact that we use transistors for almost everything is simply that they did become dirt cheap and can replace the two other elements, so only one design to care about.

*2 - What a FF needs is circuit that is stable in one of two states. The most simple way to do so are two inverters connected in a 'circle' that is the output of one is the input of the other. It will stay quite stable - until the output of one gets overwritten by external 'force', making it flip into the other state.

Two inverters, that's two active elements, no matter what technology is used, relays, valves or transistors, all the same.

A conventional flip flop requires the following:

1. Two inverting switching elements that will strongly pull their output in the direction opposite what's required to turn them on.

2. Two elements that can weakly pull the output of whichever switching element isn't on in the opposite direction.

3. Two elements that, in response to external stimuli, can pull those outputs in the same direction as the switching elements themselves would, strongly enough to overpower the elements described in #2.

In a modern CMOS memory, all six roles will be played by transistors. In an ENIAC-era machine, role #1 would be played by a pair of valves (the circuit elements within typical vacuum tubes), and role #2 would probably be filled by a pair of resistors. Use of resistors here will burn a lot of power, since they will not only oppose the output of whichever switch is off (which wouldn't consume any power), but also the one that's on. I don't know what kind of element would have been used for #3.

• "valves (the circuit elements within typical vacuum tubes)" - not exactly. The word 'valve' is the British English name for what Americans call 'vacuum tube'. Apr 14 at 19:39
• @MichaelHarvey: I know that valve is more of a British term, and "vacuum tube" is more of an American term, but if one doesn't use the term "valve" to describe the constructs that a 6SN7 and 6360 each have of, what term should an American use? A 6SN7 does not contain two "vacuum tubes", but rather two collections of metal bits within a single vacuum envelope. So what are the metal bits called? Apr 14 at 20:15
• I have always called the separate elements in a multiple tube/valve its sections. "Another early type of multi-section tube, the 6SN7", "Differences in u-Factor between sections of a dual triode will show up as DC offset." An ECL80 or 6AB8 has a triode section and a pentode section. Apr 14 at 20:27
• @MichaelHarvey: I was referring to what I guess you call "sections", though that term seems a little generic since it says nothing about what the things do. Further, while something like a magic eye tube would often have two sections, only one of them would typically be used to allow or block current flow, while the other would be used illuminate phosphors. Describing a ECL80 as a vacuum tube that contains two valves would seem much more descriptive than anything else that's comparably concise. Apr 14 at 21:04
• I suppose if you want to use 'valve' in its older sense of a device or contrivance that controls a flow of something such as a gas or liquid, then I would agree. And it would keep Brits and Americans happy. Apr 14 at 21:06