Non-Ferranti ULA 6C001E-7?

pkersey · December 2014

Hello!

I don't like Iss2 boards that much but after revising one Speccy like that for a friend, the results were so positive that I decided to have another try at my own boards. I decided to do it slowly and see what I would find, and asked a more tech-minded friend to be sort of a guide as I went, because I know he likes these intricacies and knows a lot more, so I could learn from him.

Some interesting stuff popped out. For instance:

Anyone ever seen a 6C001E-7 ULA with no indication of "Ferranti" manufacturing on it? Date code: 8821 (??), "PS" rather than "Ferranti" inscribed on.

Another thing ... I have here at least two of those 6C-7 ULAs in Issue Two boards, which aren't supposed to be originally built there. Do you think both boards were refurbished with new ULAs?

Spirantho · December 2014

I know the Plessey ULAs, but I've never seen one actually inside a Spectrum. I think they were made in very limited numbers in 1989 and 1990 when Plessey bought out Ferranti, and Amstrad needed a few more for repair shops. The latest I have seen are dated 1990, Ferranti ones are up to 1987, and I've never seen a 1986 one.

Plessey ones are interesting because there are few different logos, the ones saying "PS" and the ones with a little logo.

I suspect your Spectrum was therefore taken to one of the last service centres in 1990 or so.

Before anyone asks, no they are not available still anywhere - in my searches for all the last ULAs in warehouses around the world there were no huge stocks like there are of ZX81 ULAs. I think Amstrad or whoever only commissioned a very small production run.

That said, I live in hope that a large stock will turn up someday. :)
All that is just my impression from my own ULA hunt, which spanned USA, Asia and of course Europe.

Spirantho · December 2014

For your reference, this is my collection of different Spectrum ULAs. :) They actually come from different sources, some don't work, some are used, many are unused and new - some are from my preciously small reserve for my repair service (no, I can't sell them separately I'm afraid! Too many times have people asked me for a ULA when their ULA is just fine - it often gets the blame when it's not at fault, because it runs hot - and people think that means it's dead.).

There are other small variations of the design but these are the main types. The long ones at the bottom are of course Toastrack and +2 ULAs (I do have brand new ones of these for sale on my shop), and the bottom right square one is the +2A/B and +3 ULA (which again, I have brand new ones for sale).

Some points of interest:
The ULA5C102E - The Issue One ULA requires the "dead cockroach" mod. This was used for the first 20,000 or so machines, even though the Issue One was only made to about 13,000. They appear in early Issue 2 machines along with the mod.
The ULA5C112E at the top left has a date code after the ULA5C112E-2 next to it! This probably means that the two were deemed equivalent by Sinclair and were made in different factories, but could also mean that the ULA5C112E is actually a -2. As with other ULAs missing the torch symbol, this is from the Philippines.
The ULA5C112E-3 is the last of the 5C series before the ULA6C001E-5 arrived, and requires an Issue 2 Spectrum for the colour tuning.
The ULA6C001E-5 (left-hand side) is the rarest of these ULAs. I only know of datestamps from 8320 to 8324. They were used in some of the last of the Issue 2 machines before Sinclair realised they don't work in Issue 3 machines. If you put one of these in an Issue 3, it'll usually boot but will be terribly unstable and crash. The last Issue 2 machines used a ULA6C001E-6, I believe.
The ULA6C001E-6 is the most common in rubber key Spectrums. They also die a lot - probably because the Spectrums they're part of are the main "workhorses" for many people, unlike Issue 2 machines which were more prone to dying in the 1980s, and therefore being replaced. They were in use for all Issue 3, 3B and I believe 4 and 4S machines.
The ULA6C001E-7 was used for the last 48K Spectrums, both the Pluses and the last rubber keyed ones, but are much more common in Pluses because Sinclair didn't make many rubber-key ones at that time.

I have never seen an '86 code ULA6C001E-7, but they probably exist in fairly small numbers - at this point Sinclair were making the 128 so I imagine most of their resources went there, and as they were about to be bought out by Amstrad, they probably didn't want to spend a lot on a new production run.

The last five ULA6C001E-7 chips are all spares for service shops, I believe. The 8711 and 8720 have different logos on them which suggests they may be different factories. All the ULAs with a small torch logo are embossed "UK" in the circles on the back - the 8720 has no stamp at all, in common with the "PS" marked Plessey ULAs.

Of the Plessey ULAs, the "PS" ones have a batch code printed on the reverse. The ones with the "Plessey" logo is embossed "UK" in the same way as the Ferranti ones. Given there are two Plessey ULAs with the same datecode (8936) this means that whoever commissioned the production run must have used at least two factories! Yet these chips are incredibly hard to find. This suggests there may still be a warehouse somewhere with the chips, but given how hard it was to find these that I have, there probably isn't.

Of the 128K ULAs, of course, both are effectively the same chip despite the markings - although I've heard that the Ferranti one is slightly more compatible.

balford · December 2014

The ULA6C001E-6 is the most common in rubber key Spectrums. They also die a lot - probably because the Spectrums they're part of are the main "workhorses" for many people, unlike Issue 2 machines which were more prone to dying in the 1980s, and therefore being replaced. They were in use for all Issue 3, 3B and I believe 4 and 4S machines.

Not sure about this, I believe that issue 4A boards and above require the 6C001E-7 ULA to match improvements in RAS/CAS memory signal timing on those boards.

Details are in Chris' book, unfortunately I don't have my copy to hand.

B

Spirantho · December 2014

I've certainly seen -6 ULAs in Samsung Spectrums, and they're 4S.
I see very few Issue 4 rubber-keys, next time I see one I'll take another look, but I don't remember seeing -7 until issue 6.

All of what I wrote is just my own braindump, though - so none of it should be taken as gospel truth.

balford · December 2014

I've got 3 issue 4A boards, they all have -7 ULA's with date codes that match other components on the board, for what that's worth.

Excellent brain dump by the way :)

B

Spirantho · December 2014

There we are then - Issue 4 was the transition period. :)

balford · December 2014

Oh, I've also got a -5 ULA happily running a soak test in an issue 3 machine. More later :)

B

Spirantho · December 2014

I'm guessing you did something funky to the board to make it work. :) Unless maybe some Issue 3s are different to other Issue 3s...?

balford · December 2014

Well the machine I tested it in is fully socketed and has had all new lower RAM (not by me though), but appears otherwise standard:

Soak test ran for about an hour without issue:

Maybe the -5 ULA's weren't as uniformly stable as was hoped, leading to their quick supersedence by the -6 (this one's a bit deaf and requires plenty of volume to load from an audio source).

B

Spirantho · December 2014

I suspect if you change those 200ns RAM chips with the 300ns RAM chips Sinclair was using at the time, you may find it's less reliable.

If you try it, let us know what happens. :)

balford · December 2014

Just tried this with some STC RAM pulled from an Issue 4 (although I think this is still 200ns RAM):

And the results:

Not conclusive I know, if you can let me know exactly what type of RAM is disagreeing with your -5 ULA I can see if any of my machines have some I can pull for testing.

Also, I present to you, for your entertainment and exhilaration: :)

B

Spirantho · December 2014

Very interesting! I'll have to find out what my Issue 3s that I tested with used. I can't remember how many I tried with, but I do know that mine doesn't work in at least one perfectly working issue 3.

So who else has other ULAs that haven't appeared in this thread so far? :) We could make a gallery of ULA types. :)

balford · December 2014

Forgot to mention earlier, I have a few PS marked ULA's and have wondered about their origin - now explained by your good self, cheers :)

B

Glenn · December 2014

Spirantho wrote: »

The ULA6C001E-6 is the most common in rubber key Spectrums. They also die a lot - probably because the Spectrums they're part of are the main "workhorses" for many people, unlike Issue 2 machines which were more prone to dying in the 1980s, and therefore being replaced. They were in use for all Issue 3, 3B and I believe 4 and 4S machines.

Looks like my issue 2 has had a ULA swap in the past then.

balford · December 2014

I'd guess your machine is a late '82 build, judging by the datecodes on the 74LS157's, so it would have had a 5C112E ULA originally - but as Ian states, the very last issue 2's would have had a 6C001E-6 fitted.

So yours is still sort of correct :)

B

pkersey · December 2014

Here comes a pic of the PS ULA. Sorry for the late post.

https://flic.kr/p/pneaFF

Spirantho · December 2014

Glenn wrote: »

Looks like my issue 2 has had a ULA swap in the past then.

Yes, the chip is right but the date code is 1984, which is too late (and the other chips are 1982 :) )

The 6C chips are far, far easier to calibrate the colour on than the 5C chips, so it's not at all uncommon to see a 6C in an Issue 2.

balford · December 2014

Here's another small oddity I have:

Note the lack of a -6 or -7 suffix, I assume it's a -7 due to the late 1984 date code though.

Edit: only markings on the back are:

"ULA 6C001E
4621"

B

Spirantho · December 2014

Does it work? I presume it's made in the Philippines... Should be a -6 but I've not seen one like that before!

balford · December 2014

Yep, works fine in an issue 4A board so I'd assume it's a -7.

B

Jakeyw · December 2014

Here's my short list of known ULA datecodes and types. I've added info from this thread to it.

DATE	ULA		ISSUE(s)

8214	5C102E		1
8217	5C102E		1
8219	5C102E		1
8220	5C102E		1
8222	5C102E		1
8223	5C102E		1
8224	5C102E		1,2
8228	5C102E		2
8233	5C102E		2
8235	5C112E		2
8237	5C112E		2
8240	5C112E		2
8244	5C112E2		2
8245	5C112E2		2
8248	5C112E2		2
8301	5C112E
8301	5C112E2		2
8304	5C112E3		2
8308	5C112E3		2
8309	5C112E		2
8312	5C112E3		2
8314	5C112E3		2
8320	5C112E3		2
8320	6C001E5		4B
8323	6C001E6		2
8324	6C001E5
8326	6C001E		2
8329	6C001E6		2
8331	6C001E6		2
8332	6C001E6		2
8335	6C001E6		3
8338	6C001E6		3
8340	6C001E6		3
8341	6C001E6		3B
8342	6C001E6		3B
8343	6C001E6		3
8345	6C001E6		2
8347	6C001E6		3,3B
8348	6C001E6		3B
8349	6C001E6		3
8350	6C001E6		3
8401	6C001E6		3B
8402	6C001E6		3B
8403	6C001E		3B
8407	6C001E7		4A
8408	6C001E7		3B
8410	6C001E7		4A
8411	6C001E7		6A
8412	6C001E6		3B
8413	6C001E		3
8414	6C001E7		4S
8416	6C001E7		4A
8418	6C001E7		3B,4A
8419	6C001E7		3B
8426	6C001E7		3B
8428	6C001E7		4A
8429	6C001E7		3B
8432	6C001EX7	6A
8443	6C001E		6A
8444	6C001E7		2,4B
8445	6C001E7		4S
8446	6C001E		4S,6A
8447	6C001E7		6A
8448	6C001E7		3B
8449	6C001E7		6A
8504	6C001E7
8543	6C001E7		2
8543			6K
8543	7K010E5
8607	7K010E5
8612	7K010E5		6U
8623	6C001E7		3B
8625	6C001E7
8711	6C001E7
8720	6C001E7		4A,4B
8816	6C001E7		4S
8821	6C001E7 (PS)
8936	6C001E7 (PS)

balford · December 2014

Just to revisit this a little, I've installed one of my PS 6C001E-7 ULA's for a soak test (datecode 8936), and one interesting detail is that they run quite a bit cooler (almost cold!) than the Ferranti series.

They do consume 15mA more than the Ferranti ULA though, so not everything was improved :)

B

1024MAK · December 2014

Err, more energy in, but less heat out :confused:

Does this ULA come in a standard plastic package, or is it a ceramic package?

Mark

guesser · December 2014

1024MAK wrote: »

Err, more energy in, but less heat out :confused:

I was thinking exactly the same thing heh

it must radiate it all as wideband RF all over 80 and 40 meters :lol:

balford · December 2014

Standard plastic package, exactly the same as the one posted further back up this thread.

I'd say it gets about as warm as an NMOS Z80, but no warmer. Compare that with a 5C112E ULA that you could fry a (very small) egg on :)

I measured power consumption using my standard regulated 9V PSU, so don't have any reason to doubt my power measurements. I did do it twice as I was sure the results would be the other way round :)

B

Spirantho · December 2014

Do you find that the Philippines-made ULAs (with no logo) die more easily than the UK-made ones (with the torch logo)? I have a stack of dead ULAs and they're pretty much all Philippines-made ones - but then they're also more common....
I just wonder if they have different heat characteristics and/or power consumptions.

spodula · December 2014

guesser wrote: »

I was thinking exactly the same thing heh

it must radiate it all as wideband RF all over 80 and 40 meters :lol:

Perhaps it gives out more current on the outputs, particularly cass/spk, PhiCPU, ect.

1024MAK · December 2014

The only thing I can think of is this: maybe the output buffer transistors have a lower volt drop when the output is high. Then there will be less power loss (and so lower heat) but a slightly higher output current depending the "load".

Mark

Spirantho · December 2014

When I get some time (ha ha - that'd be nice), I might investigate if the Plesseys are better than the Ferranti's at loading or something like that. I also wonder if the UK Ferranti's are better than the Philippines ones.
Maybe we could make a league table of the best to the worst ULA to have in your Spectrum :)

If my Plesseys really do run cool as well, I might put one in my main Spectrum (with a heatsink anyway, just to be safe). A Samsung Spectrum with TR2 replaced and a Plessey ULA with a heatsink. That sounds like the perfect 48K Spectrum!

balford · December 2014

Spirantho wrote: »

Do you find that the Philippines-made ULAs (with no logo) die more easily than the UK-made ones (with the torch logo)? I have a stack of dead ULAs and they're pretty much all Philippines-made ones - but then they're also more common....
I just wonder if they have different heat characteristics and/or power consumptions.

That's my (limited) experience also. I've got three dead ones, all Phillipines-made. One I killed myself by a misguided attempt to test a 4116 of unknown origin, the others were faulty on arrival.

Spirantho wrote: »

If my Plesseys really do run cool as well, I might put one in my main Spectrum (with a heatsink anyway, just to be safe). A Samsung Spectrum with TR2 replaced and a Plessey ULA with a heatsink. That sounds like the perfect 48K Spectrum!

That's exactly what I've done :) I've popped a Plessey ULA into my Issue 4A board which also has a switch mode 7805 replacement, Zorn's LRR and URR boards, CMOS Z80, 74HCT logic etc. Should be bulletproof :)

B

Non-Ferranti ULA 6C001E-7?

Comments