Close

Direct Coupled Transistor Logic

A project log for Hackaday TTLers

Where DIY DIP/SSI/MSI CPU makers meet and discuss other TLA (three letters acronyms)

blair-vidakovichBlair Vidakovich 06/22/2018 at 04:2213 Comments

Introduction

@Yann Guidon / YGDESasked me to do a write-up of the Direct Coupled Transistor Logic (DCTL) of the famous CDC 6600 computer. When it was released, and for some years, the CDC 6600 was one of the fastest and most powerful computers in the world. When we take a look at the logic family that it used, it will be obvious to see why:

The Basic Unit: The Inverter

There are primarily two main articles available online which deal with the electronic description of DCTL logic:

 J. E. Thornton, Design of a Computer: The Control Data 6600 (Scott, Foresman and Company, 1970), Illinois.

and

James B. Angell, “Direct-Coupled Logic Circuitry” (1958) Proceedings of the Western Joint Computer Conference: Contrasts in Computers, 22-27.  https://www.computer.org/csdl/proceedings/afips/1958/5052/00/50520022.pdf

This is not a great deal. On Bitsavers, there is no folder for the engineering documents of the 6600, as opposed to the 1406, and 3600-series CDC logic, which I also aim to cover because it is a very interesting high-speed DTL.

So bear in mind that the information I am presenting here is limited, and if you want to build your own DCTL circuits you are most likely going to have to design your own, because there are no complete design documents online for the 6600 which you would have been able to copy and modify.

Anyway here is the basic building block of DCTL:

The CDC 6600’s inverter:

The 1950s DCTL inverter:

By the time the 6600 was built, transistor fabrication had developed and improved markedly. In fact the first few pages of the chapter of the CDC-published book on the digital electronics of the 6600’s DCTL go on about how the new silicon transistors they used in ‘69/‘70 made the 6600 possible. So this explains why the 6600 uses NPNs, as opposed to the older implementation using PNPs.

As you can see, you would be forgiven for mistaking DCTL with Resistor-Transistor Logic (RTL) if you had only an inverter to look at. I agree with the speculation on the (tiny) Wikipedia article on DCTL that it evolved from RTL.

Obviously the thought process that lead to developing this logic family was “what if we had RTL but got rid of all the resistors?” The point of having resistors in RTL is to allow you to increase the voltage margins of the logic levels. It also allows you to better control the flow of current throughout the circuitry and match the impedances of the inputs and outputs.

The first problem you have with DCTL is – how do you make sure you can switch transistors without driving them too far into saturation? The solution is to use transistors with special impedances and gain ratios.

Take a look at the special transistor characteristics that the conference proceeding document outlines:

This is obviously based on an old understanding of exactly how well-designed transistors are, but you can see we’re only switching very small amounts of current, and the V(BE) of the operation of the transistors when in saturation/conduction is far lower than your standard BC548/9 or 2N3904/6.

I haven’t checked yet, but I believe transistors with these kinds of characteristics should be able to be obtained cheaply. The maximum switching time required in this specification from the 1950s is easily obtained with modern discrete transistors.

DCTL Logic Levels

The logic levels of DCTL circuitry in the 6600 are very very close together. As you can see the V(BE) of the 6600 transistors closer resembles the operation of modern transistors.

At this stage, I became very interested about how noise is dealt with in DCTL circuits, because the logic levels are nowhere near as far apart as 0-5V as in TTL.

Angell in the conference proceeding explains how it is possible to minimise noise in DCTL:

Some kinds of noise are comparatively unimportant in DCTL whereas others are potentially severe. Because of the very low impedance level of “on” circuits, capacitive pickup is generally negligible. Power-supply fluctuations are usually unimportant, because voltage translation circuits are not required and the majority of a complete logical system can, if desired, be run from a single supply. On the other hand, because of the small voltage swings which are experienced in DCTL, inductive coupling in either multiconductor cable or via ground leads may be troublesome.
It is generally felt that with germanium transistors a maximum induced noise or the order of 25 millivolts can be tolerated; this figure amounts to 10% of the normally encountered voltage swings.

The article then goes on to detail how it is possible to reduce noise. Please see the article for more.

Matching Impedances

The whole gambit of designing DCTL circuits depends on making sure different logic elements have impedances which agree. Because of the low conducting collector-to-emitter voltages in DCTL, the transistors involved in this logic family will have quite low gain. The game you play with you tangle with DCTL is attempting to have as much gain as possible to increase fan-out.

Generally speaking, fan-in will be much higher than fan-out in this logic family.

To take a look at the limits you come up against when designing DCTL circuits, consider the following interchange:

This interchange is n paralleled OR gates driving m paralleled loads.

Angell explains that two conditions are necessary in order to make this interchange work:

  1. When all the input gates are off (V( C)=0.2V), all the output loads must be on (V( C)=1.2V).
  2. The conduction in any one of the OR gates must turn off all the output loads.

The first condition is fulfilled by the following inequality:

The term on the left is the smallest possible load current. The term on the right is the leakage current in all of the gates in the interchange plus the minimum required load.

The second condition is satisfied by the following inequality:

It means that the collector current of one conducting gate must be greater than the maximum possible current in the load resistor. Here, the node voltage is equal to the specific limit voltage required to maintain the load transistors in the nonconducting state.

So this gives you the limits within which you can match the impedance of the inputs and outputs of the interchange. This will help you determine the fan-in and fan-out, and set the load resistor value.

The important factors in this particular example, therefore, are:

  1. The input impedance of a conducting transistor must not be too low. This problem can be solved by setting a maximum limit on the the ON base current for a particular base voltage. This way, no one output transistor will take too much of the available driving current.
  2. The output impedance of the OR transistor network must not be too high. One method of solving this problem is to make sure that the V(CE) of the input transistors is LESS than a certain limit for a given I( C) and V(BE). Another method would be to limit the I( C) with respect to given V(BE) and V(CE). The absolute minimum limit for either of these conditions, the lowest V(CE) or lowest I( C) is determined by the minimum transistor gain. This minimum individual transistor gain will be much greater than the overall circuit gain. However, this individual transistor gain will be much smaller than if the transistor were in the active or non-saturated region. So it is a careful balancing act of switching current from the collectors of input OR network into the bases of the output network.
  3. The “off” collecor current of a non-conducting transistor must not be too large. This means the “off” I( C) must have a maximum limit compared to its “off”, non-conducting base voltage (V(BE)).

All of this boils down to the following: In order to have compatible connections between DCTL logical elements, V(CE) “on” [Logical “0”] =< V(BE) “off” [Logical “1”] for any particular logical element. Angell explains that it is permissible to sacrifice gain in some instances in order to reduce the V(CE) of a given logic element.

Combining the two inequalities above when considering the above interchange, you can solve for the gain of an individual transistor in conduction. This produces the following inequality:

As Angell explains, the result shows that the input transistor gain must be greater than the (a) circuit gain; multiplied by (b) m, the number of output loads; and ( c) factors accounting for the tolerances of the components.

You can derive a rule of thumb to follow if the networks of logic elements is too great:

I( C) “on” / I(B) “on” [Logical “0”] > 2m

So the gain of a saturated transistor must be greater than twice the number of output loads.

CDC 6600 Rules of Thumb

The general rules CDC outlines for 6600 DCTL is:

Selected Common Logic Elements

There is therefore a lot of work required to match logic element input and output impedances. But, the advantage to all this work is that DCTL logic elements are actually very simple.

Consider the following 6600 NPN and 1950s PNP basic logic elements:

After matching all of the impedances of logic elements, you are rewarded with very simple circuitry.

Switching Times

One can assume that this circuitry switches and propagates faster than the famous and standard hobbyist DTL. I do not have exact measurements, however the CDC documentation assumes the propagation delay of an inverter is 5 nanoseconds.

Conclusion

Although fairly fiddly, this seems like an interesting logic family.

I will cover CDC’s 1604 and 3600 DTL logic in the next write-up.

Discussions

Tim wrote 01/04/2020 at 22:35 point

Thank you for the interesting write up!


One point a am still a bit confused about: Why do you claim that the CDC6600 uses DCTL logic instead of RTL logic? The main difference between DCTL and RTL seems to be the base resistor, which is omitted in DCTL.

Every circuit diagram in Thorntons book has the base resistor. There is also a paper by him from 1980 ("The CDC6600 project"), where he straight out refers to RTL (p.341). I would therefore conclude that the CDC6600 actually uses RTL.

DCTL looks to me like something that only works on paper as there is basically zero noise margin. If there is a mismatch between transistors there will be a lot of issues, same if there are transmission line reflections.

  Are you sure? yes | no

roelh wrote 08/05/2018 at 07:40 point

I designed a DCTL ALU, that has only 12 transistors, and all functions, see https://hackaday.io/project/160177-alu-in-dctl-technology.

  Are you sure? yes | no

Yann Guidon / YGDES wrote 08/05/2018 at 10:11 point

That's awesome !!!

  Are you sure? yes | no

Yann Guidon / YGDES wrote 07/29/2018 at 11:25 point

From https://en.wikipedia.org/wiki/Direct-coupled_transistor_logic  :

A similar logic family, direct-coupled transistor-transistor logic, is faster than ECL.[4]

OK now I'm curious...

  Are you sure? yes | no

Yann Guidon / YGDES wrote 07/29/2018 at 13:33 point

I googled a bit and found this description :

The direct-coupled transistor-transistor logic (DCT/SUP 2/L) family consists of a multiple-emitter AND gate and a NOR gate similar to direct-coupled transistor logic (DCTL). High speed for low power is obtained by limiting the voltage swing and using a low voltage power supply of about 2 V. Using a conservative, standard Schottky process, the DCT/SUP 2/L NOR gate has a delay of about 1 ns for 4-mW gate power. A computer-aided analysis shows that this is faster than the basic gates of emitter function logic (EFL), emitter-coupled logic (ECL), or Schottky transistor-transistor logic (T/SUP 2/L) with the same process and gate power. A comparison of actual arithmetic logic units shows that Schottky DCT/SUP 2/L is smaller and faster than ECL and Schottky T/SUP 2/L. The higher speed and density of DCT/SUP 2/L makes it a better large-scale integration (LSI) concept than the other logic families.

That's promising but I need to see an actual circuit/schematic :-)

  Are you sure? yes | no

Julian wrote 07/29/2018 at 09:54 point

Those are beautifully simple circuit designs.  Perhaps the simplest design of anything I've seen using bipolar transistors.

  Are you sure? yes | no

Yann Guidon / YGDES wrote 07/29/2018 at 10:11 point

I agree :-)

  Are you sure? yes | no

Julian wrote 07/29/2018 at 11:01 point

Somewhat related, I've just posted a log discussing the design of the CDC6600's Peripheral Processors, and how to apply some of the same ideas to a somewhat smaller system: https://hackaday.io/project/159826-io881/log/150136-inspiration-from-the-design-of-the-cdc6600-peripheral-processor

  Are you sure? yes | no

Yann Guidon / YGDES wrote 07/29/2018 at 11:26 point

Oh my ! :-D

  Are you sure? yes | no

Ed S wrote 07/28/2018 at 17:29 point

  Are you sure? yes | no

Yann Guidon / YGDES wrote 07/28/2018 at 17:33 point

Thanks !

The text is updated :-)

  Are you sure? yes | no

Blair Vidakovich wrote 07/29/2018 at 00:34 point

thanks! i really enjoy doing such reviews!

  Are you sure? yes | no

Yann Guidon / YGDES wrote 06/22/2018 at 20:16 point

Wow ! that's a really interesting study ! Thanks for taking the time to write it !
I took the liberty to adjust a few links. I hope you can provide a link to the Angell paper :-)
I'll probably try DCTL along with other topologies :-)

  Are you sure? yes | no