Can anyone program some EPROMs for me

[Super_7b]

Hi All,

Yesterday I found out that my HCR EPROM programmer can't handle a particular type of EPROM, a WS57C43C.

After some googling, I found that a Stag P301 can handle this with version 8.03 software and doubtless other programmers can too.

I searched the thread "EPROM Programmer collectors roll call" (viewtopic.php?t=7406) in this forum and a couple of users seem to have Stag P301's.

If anyone out there has a programmer that can handle the WS57C43C and would be willing to program 3 for me, please either respond here or PM me.

I will, of course, re-imburse any costs.

BR

Mick

[Ukwebb]

A 4k EEPROM? Just curious, what's the use case for these?

[Super_7b]

A 4k EEPROM? Just curious, what's the use case for these?

These are for a non-Beeb application. They will contain lookup data.

BR

Mick

[daveejhitchins]

My Logical Devices ChipMaster-6000XPU has that part listed under Waferscale.

WS57C43C

PM me if you want me to try and program it for you. - Cost you return postage.

Dave H.

[maniacminer]

I've seen these WaferScale fast EPROMs before, they can be programmed in a normal programmer at 12.5V, their main claims to fame were very fast speed (so colour lookup tables, CPU jump vectors, CPU instruction decoders/microcode etc.) ~ 15ns and their low power, about 20% of similar high-speed bi-polar EPROMs. I don't recall ever having to need a "special" programmer for them - maybe there are features like marginal verify that need different voltage levels on pins to get that particular function?

As for ukwebb's "why?" well, there are cases where 4k bytes is enough to do the job. Why incur extra expense and complexity. Things like the Music 500 has a 2k byte SRAM yet only uses 64 bytes of it - that's just down to simplifying the BOM. I don't know what the application is here for this particular EPROM, maybe its historical? Maybe it's a familiarisation of the part? Maybe, like me, it's because I've got a shit-tonne of them

[Super_7b]

Hi @maniacminer,

Over on this thread viewtopic.php?p=424577#p424577, Hoglet said the following:-

Just looking at the HCR software disassembly, the 2532/2564/2716/2732 parts (all 24 pin devices) share common programming code which uses a single 50ms pulse. That's much longer than allowed by the WS57C43C and may well cause damage. So I don't think it will be possible to program these with the original HCR software.

Dave

I did try programming these chips at 12.5V on my HCR Programmer with no success, so Dave's suggestion might be one reason. I also looked more closely at the specifications and the 2732 needs "VCC = 5V ± 5%; VPP = 21V ± 0.5V" (which the HCR must be hitting as I get good results) whereas the WS7C43C needs "VCC = 6.25 V ± 0.25 V, VPP = 12.75 ± 0.25 V". The VCC might be a bit higher than what the HCR outputs and the VPP would be right on the margin when selecting 12.5V mode on the HCR.

Your analysis of the use case is accurate; I wanted fast results and I only need 512 bytes. These were the best suited chips for my application, even though I am not using anywhere near their capacity.

I also have quite a few, but not quite a shit-tonne

BR

Mick

[maniacminer]

Yeah, I saw that thread. Oddly, I've used the same IC, I've never had Vcc up to 6.25V (the absolute maximum is 7V over VIL) and then there's the VPP that needs to spend some time at VIH before jumping to VPP (yeah, there's a pin called VPP and a programming voltage called VPP) and then back down to VIH and then to VIL. It's all about the timing of this, perhaps I was lucky? I suspect issues can occur if the programming voltage isn't at VPP at the right moment, leaving the cell only partially charged and the sense amplifiers triggering depending on the voltage on Vcc leading to either inoperable ROM or one that has errors and/or "works" sometimes and not others... (our favourite failure mode )

[1980s_john]

Hi,
My Dataman 48 can program Waferscale WS57C43C according to the (long) device list.

PM me as needed,
Regards,
John

[Super_7b]

Hi John,

Thanks for the offer. Dave Hitchins already offered and has programmed the chips for me.

BR

Mick

[Super_7b]

Yeah, I saw that thread. Oddly, I've used the same IC, I've never had Vcc up to 6.25V (the absolute maximum is 7V over VIL) and then there's the VPP that needs to spend some time at VIH before jumping to VPP (yeah, there's a pin called VPP and a programming voltage called VPP) and then back down to VIH and then to VIL. It's all about the timing of this, perhaps I was lucky? I suspect issues can occur if the programming voltage isn't at VPP at the right moment, leaving the cell only partially charged and the sense amplifiers triggering depending on the voltage on Vcc leading to either inoperable ROM or one that has errors and/or "works" sometimes and not others... (our favourite failure mode )

When I got the chips back from Dave Hitchins and tried them in my breadboard test setup, they appeared blank. The same when I tested them in my HCR Programmer.

I then had a closer look at the datasheet and so much for their statement at the top of the sheet "Pin Compatible with 4K x 8 Bipolar PROMs". I had already noticed that pins 19 & 21 have A11 and A10 vs A10 & A11 on a 2732, but that was irrelevant for me as I only need the bottom 512 bytes, so A9-A11 are tied low. Worse still was Pin 18 which is CS2 on these chips and /E on the 2732.

The opposite sense!

So, the chip was disabled and thus appeared blank! No wonder the HCR couldn't program this chip without a new, custom, entry in the device selection table. If that is taken care of, maybe the HCR could be "persuaded" to program this chip.

Oh well! A simple swap to the signal on Pin 18 and the chip sprang into life.

Thanks once again to Dave Hitchins.

BR

Mick

[maniacminer]

Ah, the minute in the library saves a week in the lab I'm guessing it's an error on the WS57C43C datasheet having A10 & A11 swapped over, the rest of it is standard JEDEC/Intel 2732. I don't recall the A10/A11 swap though.

[Super_7b]

Ah, the minute in the library saves a week in the lab I'm guessing it's an error on the WS57C43C datasheet having A10 & A11 swapped over, the rest of it is standard JEDEC/Intel 2732. I don't recall the A10/A11 swap though.

I did wonder if it was an error with the address lines...

That Chip Select, however, is not an error; it really is in the opposite sense to the 2732. Why would they do that? Particularly as they claim pin compatibility.

BR

Mick

[1024MAK]

The active logic level for chip select, chip enable and output enable for some devices can be specified at order time. This is especially true of mask ROM chips.

Hence some manufacturers also had EPROMs that were available with a selection of different active logic levels for their control inputs.

This reduced the complexity of the address decoding, as you could parallel two ROMs/EPROMs, each having a different logic level combination for their control inputs.

The final version of the QL does this. I'm sure there are earlier examples if you go looking.

Mark

[wiggy]

That Chip Select, however, is not an error; it really is in the opposite sense to the 2732. Why would they do that? Particularly as they claim pin compatibility.

Yes, they are claiming pin compatibility - but importantly not with the 2732.

The 2732 etc are NMOS (or, later, CMOS) UV EPROMs... The Bipolar PROMs are (were) a whole other animal: noted primarily for their speed, and based on blowing fuses rather than the trapped-charge-second-gate cells of the erasable devices. The Bipolar PROMs are one-shot by design.

FInding a 4kx8 part to compare was a challenge - most manufacturers stopped at the 82S191 (or clone) 2kx8 part in around 1980. AMD, however (who kept making bitslice DIY-processor parts longer after everybody else stopped - classically where these would be used for microcode storage) have one in their 1984 databook: the AM27S43 on page 2-73... turn the page and, yes the pinout is the same as the WS57C43C.

As to why different manufacturers adopted different pinouts for their parts - its probably a side effect of parallel evolution, not helped by them being based on different technologies (and/or different markets). Where one manufacturer "gets there first", other folks tended to copy (Op-Amps, say). But otherwise contrast, say, RCA's 4000-series CMOS with TI's 74-series TTL logic (not to mention all the preceding RTL/DTL stuff...) , or even the 2516 EPROM with the 2716 EPROM, which is close but not quite the same, hence the different number...

[atsampson]

I'm guessing it's an error on the WS57C43C datasheet having A10 & A11 swapped over, the rest of it is standard JEDEC/Intel 2732.

It wouldn't really matter even if it's not an error - you can permute the address lines (and the data lines) on an EPROM or SRAM however you like, provided you do it in the same way when writing and reading. It doesn't matter which physical location in the chip you're using for each bit, as long as they're all different!

[maniacminer]

I'm guessing it's an error on the WS57C43C datasheet having A10 & A11 swapped over, the rest of it is standard JEDEC/Intel 2732.

It wouldn't really matter even if it's not an error - you can permute the address lines (and the data lines) on an EPROM or SRAM however you like, provided you do it in the same way when writing and reading. It doesn't matter which physical location in the chip you're using for each bit, as long as they're all different!

It does matter if you have a program that needs to be read from an IC that needs A0 -> A0, A1 -> A1 and D0 -> D0 etc... I agree, if you're rolling your own, all you need to remember is what maps to what. I've seen some comedy problems involving this over the years.