Sinclair SCL/ULA

[David G]

There has been lots of threads about the ULA, but all seem to focus on specific questions or issues. But what about general info? I went down the rabbit hole and am creating this topic to consolidate the information I have learned


The SCL is the main logic chip of the Sinclair motherboard -- the "glue logic" chip that connects the processor (the Z80) to the RAM & ROM memory, keyboard, video and expansion connector

From [edit:page Chapter] 25 of the ZX81 user's manual "How the computer works":

The last big chip is the logic chip, or SCL (Sinclair Computer Logic) chip. Again, this was specially designed & made for the ZX81, & it connects the other chips together in rather a clever way to make them do more than they normally would.

In the photo it is marked "SCL (dogsbody)". Dogsbody is not a Dilbert reference. I looked it up and it means "someone who does menial or drudge work". So this SCL does all sorts of small tasks for the ZX81 and replaces all the various chips of the original ZX80 design. It has many small circuits inside. One chip replaces many. But it doesn't get the glory that the Z80 CPU does, it's more of a lackey that does much of the work but gets none of the credit. However, it was very important and innovative -- it has been said that the SCL was the reason Sinclair was able to offer the first computer priced under 100 pounds/200 dollars

In the photo the paper label on the chip says "Sincliar Computer Logic". But in production ZX81 the chip is without an add-on label and the top of the chip has printed text "ULA" (Ferranti Uncommitted Logic Array) or only a datecode number

[David G]

SCL vs ULA - what's the difference?

In short:

• ULA is a type of customizable chip used in many different devices back in the 1970s and 1980s

• SCL is the Sinclair Computer Logic chip, customized on the ULA product line. The ZX81 used one, and the Spectrum used a different one

The SCL Chip is designed around a Universal Logic Array (ULA) made by Ferranti. Note that "ULA" is a Ferranti trademark from the early 1970s -- eventually similar chips were eventually made by other semiconductor companies, but were called something else. The generic type is Gate Array. This has since evolved into the Structured ASIC, the CPLD, and on to today's user-programmable FPGA. The ULA was not user programmable, but had to customized before final manufacturing -- a custom semiconductor mask had to be created and used in the creation of the chip at the Ferranti chip fab

Although Sinclair documentation called it the Sinclair logic chip, ULA does mean Uncommitted Logic Array.
With these, all the semiconductor structures are already designed and defined by the chip manufacturer. But the final top metal interconnect layer can be designed by the customer (meaning the industrial buyer, that is for example, Sinclair) to suit their application. The chip manufacturer then adds this custom metal interconnect layer making the chip a semi-custom design.

Once the design is finished, the ULA is dedicated to the particular application, making it a type of application-specific IC (ASIC). Hence the Spectrum ULA cannot be used in a ZX81 and vice-versa

Back in the day the ULA technology was a game changer, allow huge cost savings while delivering big capability

[David G]

The ZX81 SCL/ULA comes in three versions. All these SCLs interchange, but the last one has an improved video signal circuit

• 2C158E is the earliest version

• 2C184E is the middle revision. No significant difference from 2C158

• 2C210E has the video back porch signal, for better compatibility with more expensive color TVs

The ZX81 Issue 3 mainboards were the first to get the 2C210 chip

Revisions include:
* 8102 [1981 2nd week]
* 8109
* 8111 [2C158E]
* 8119
* 8120
* Ferranti ULA2C184E 8130
* Ferranti ULA 2C184E 8146
* Ferranti ULA 2C184E 8150
* Ferranti ULA 2C184E 8211 [back: STI 2C184E PHIL. 0821]
* Ferranti ULA 2C210E 8231
* Ferranti ULA2C210E 8233
* Ferranti ULA 2C210E 8245
* Ferranti ULA 2C210E 8250
* Ferranti ULA 2C210E 8310 [possibly NOS for Timex/Sinclair 1500 (improved ZX81)]
* Ferranti ULA 2C210E 8439


Without the Back Porch signal (without the latest ULA 2C210E):
* Works fine with all B/W TV
* Works fine with most colour TV, although older expensive color TVs may have a dark picture
* Works fine with some LCD TV
* Some modern LCD TVs will show a dim image with white parts shown as dark grey

If your TV shows a poor picture, it is better to add a Composite Video modification than search for an updated ULA, which will end the end still have a fuzzy RF signal instead of a crisp, clean, sharp video signal

[David G]

• sinclairzxworld site search

• Deep dive on the 2C184E SCL explains all the functionality and includes equivalent circuit diagrams

• New/replacement ULA where to find. Are they still available?

• Making a new SCL from a CPLD chip and Andy Reas's improved ULA from a CPLD

• ULA heat issues. Should I get a heat sink? Or a better power supply?

[1024MAK]

Interestingly, in the photo from the manual, although it clearly is an issue one printed circuit board (PCB), the ‘ULA’, the ROM and the Z80A (an NEC µPD780C-1) have all been placed in their sockets the wrong way round. Pin 1 of each of these chips is towards the bottom left of the board (as shown). Whereas on a real working board, pin 1 of all the chips is towards the side of the PCB where the edge-connector is.

It’s also possible that the ‘ULA’ in the photo may not actually be a real ULA. The paper label covers the chips markings, so it could be any 40 pin chip, such as another Z80A. This may have been done if supplies of real ULA chips were not yet available.

This PCB does not have a ‘silkscreen’ (the white printing that identifies component numbers) and the keyboard connectors are just the metal contacts without the plastic housing. Both unlike production boards.

Here’s a link to where I describe the ZX81 Variations.

David, you may want to change your reference to the manual, you say page 25 instead of chapter 25.

Mark

[David G]

Thanks! Now edited to say "Chapter 25" instead of "page 25"

you have eagle eyes

[Moggy]

You state that most 2c158 ULAs were soldered to the motherboard.

In over 40 years of collecting and using, in my experience most of these early ULAs were socketed as were those offered in ZX81 kit form and it was the final 2c210 ULA used on issue three motherboards that was soldered directly. I have yet to see an issue three socketed ULA direct from Sinclair as opposed to something changed by the user. Sinclair being extremely penny pinching probably wanted to save a penny on the socket as the 81 was nearing end of life..

Also the 2c158 chip has its type number on the underside along with the word PHIL, the top side bearing week/year date.

The heat sink issue has been discussed ad nauseam in threads various and suffuce to say the answer is no.

The chip in question has been tested as running at around 62 deg C. Its max working temp is 125 deg C so do the maths as they say.

[Lardo Boffin]

There are quite a few recent zeddie photos here : https://sinclairzxworld.com/viewtopic.p ... 785#p34785 for reference.

[1024MAK]

To be honest, with regards to which chips are socketed or directly soldered, it’s best to not make any assumptions. There are many variations, from all socketed to all directly soldered. The kits were supplied with sockets, so they normally have all chips in sockets.

My summary would be this:
ULA - issue one boards: sometimes soldered directly, sometimes in a socket, issue 3 boards, more likely to be soldered.
CPU - issue one boards: often in a socket, but sometimes soldered, issue 3 boards, more likely to be soldered.
ROM - issue one boards: often in a socket, but sometimes soldered, issue 3 boards, more likely to be soldered.
RAM - both issue one and issue 3 boards - normally soldered, some issue one boards may have it in a socket.

Some things that are definitely certain:
The early issue one boards did have the ULA soldered directly. These normally have the 2C158E ULA.
A lot of later issue one boards had the ULA, CPU and ROM in sockets with the RAM soldered directly (the two that I have to hand were like this).
I have an issue 3 board to hand (which is from a TS1000), it has the ULA, CPU and ROM in sockets with the RAM soldered directly.
I also have other issue 3 boards where all chips are soldered directly.

I’ve not seen enough TS1000 machines to come to a conclusion.

I’ve not analysed Lardo’s photos. Ideally boards need the solder side checking, to see if any of the soldering is rework (e.g. traces of flux) which may indicate the fitting of a socket when a chip was replaced.

Mark

[David G]

thanks. I have edited that post to remove the socket information. It was old info collected from prior posts on this forum