Counterfeit Z80

[Le Fauve]

So, I know ordering on Chinese websites is risky, but I have always be lucky so far, so after the news about Z80 chips about to stop being produced, I ordered 10 Z80 on AliExpress...

You can already see where this is going, but I thought I should post about it as a reminder for peoples like me who take (small) risks thinking they'll be ok.

Small because it was 12 euros for 10 Z80 (I was expecting a couple of them being dead, but hopefully 6 or 8 should have been working, compared to my previous experience, which would still have been a nice price).

So, here is what happens when you put them in a RC2014 "classic" (I used minimal setup, removing all extra boards):

- First, they pull between 140 and 160mA (with the original Z80 from Spencer, it pulls about 30mA in this configuration)

- Then, if I use the "Snake" ROM (Snake game in ASCII characters) everything seems to work fine

- But if I try using the 4.7c BASIC rom, as soon as I enter the "Memory top" it displays the total available RAM (which seems correct when I enter different "Memory top" values), then it sometimes displays the "Ok" prompt (not always) and ask again for Cold or Warm restart.

I'm a little surprised that all those chips "kind of work" but fail exactly at the same point.

Now if I look closely to them, they all have the exact same marking, but some of them are obviously different. For example 2 of them have the "pin 1 notch" going through the package (you can see it when the chip is upside down), while the other ones don't.

Also the markings on the underside are very different.

Well, I have been burn, and I know it was my fault. I'm trying to get a refund, but in the meantime, stay vigilant and use Mouser or Digikey.

Eric

PS: The seller name was "Jin Tank Store Store", but seriously, avoid Chinese stores if you need Z80s...

PPS: The exact marking on the chips is:
ZILOG
Z84C0020PEC
Z80 CPU
1240     GZ
(and there is a huge Z on the left of the marking).

[Bill Shen]

Could you provide a link to your AliExpress store?  I’m surprised you have all NMOS Z80.  Typically it is a mix of NMOS, CMOS Z80, and non-functional parts.  I’ve also acquired a batch of used/fake Z80 from eBay.  I’ll report on my test results later today.

  Bill

[Bill Shen]

When I read about Z80 end-of-life announcement, I ordered two Z84C0020PEG from Mouser and at the same time ordered 20 Z84C0020 from this eBay vendor, https://www.ebay.com/itm/392616812507 .  The eBay order was the cheapest Z80 I found at the time, 20 for $19.5, shipping and tax included.  My goal is designing a dual Z80 tester that checks every clock cycle operation of an used Z80 operation against a known good reference Z80.  The Mouser order was back ordered as expected, but I received the eBay order fairly quickly so I did some independent testings with them.

The Z80 from eBay were poorly packaged, it was basically a stack of ten Z80 wrapped in plastic wraps and shipped in padded envelop.  Despite the poor packaging, none of the pins were bent; the parts looked good cosmetically.  All parts are the same labeling which are obviously re-labeled because the back side of the parts are not all the same. The 20 Z80 are tagged as X1-X20 to keep track of their operating characteristics.

The tester is ZRCC with ZIF socket for Z80.  By changing the clock oscillator, ZRCC can run Z80 at 22MHz, 11MHz, 7.37Mhz, 3.68 MHz, and 1.84MHz.  It has built-in memory diagnostic and can boot CP/M to run zexall.com.

So here are the test results:
Out of 20 used/fake Z80
10 are CMOS that run successfully at 22MHz executing zexall.com
6 are NMOS that run successfully at 11MHz executing zexall.com
1 is NMOS that works at 3.68Mhz, but not 7.37MHz
1 is NMOS that works at 1.84MHz, but not 3.68MHz
2 are not operational.

I also tested the 10 CMOS Z80 at 3.4V on a different test platform running RomWBW.  Eight out of 10 booted successfully into RomWBW and passed the first test of zexall.com (I don’t have time to put them through the complete zexall tests).

I plan to do more tests with Sergey Z80type.com test on a RC2014 platform.
  Bill

[Le Fauve]

Thanks for the feedback, Bill!

I love your test rig, but having only a RC2014 at the moment, I have to do with it.

This ZEXALL.com program seems nice. Would you have by chance a version compiled as a ROM image? (I don't run CP/M).

I think I have found the sources (https://mdfs.net/Software/Z80/Exerciser/) so I may try to compile it as a ROM image for the RC2014 classic if there isn't one around yet.

My Z80 were packaged exactly like yours. The AliExpress link is: https://www.aliexpress.com/item/1005005111347638.html

I also attach two pictures of them, recto and verso.

The good news after reading your message is that I may still be able to use some of those for replacing 4MHz Z80 in old systems (Colecovision, ZX81...)...

How did you know they're NMOS? Is the high current pull a characteristic of those?

I'd like to test them at a lower clock, but I'm not sure how to do this (I have the original RC 2014 clock module).

I do have a TMS9928 module though, and I think I remember it has a jumper to use its own "3-4ish MHz" clock for the RC2014 bus.

I'll give it a try, but I'm afraid I may lost serial communication with this clock (I also have the original serial module which uses the bus clock to determine the bauds rate, and I'm not sure at all it will still work if the new clock isn't exactly half of the original one).

If someone has any idea for slowing down the "classic", I'm interested :o)

I'm still very surprised that the Snake game seems to work very well at full speed. Does this mean that you can overclock if you only use a subset of Z80 instructions?

While looking at my CPU board compared to the new one, I noticed two differences:
- There is no decoupling caps on the old one

- The INT line uses a 3.3k pullup, while the new one seems to use a 10k.

Would it help if I change mine for a 10k? (and add a decoupling caps?)

Thanks again for all those information! I learn a lot from your message (like every time I tried something new on the RC2014 :o) ).

Eric

[Spencer Owen]

This is a really interesting thread.  While I hope it will be many years before we _need_ to resort to counterfeit or used Z80 chips, it is certainly worthwhile evaluating and grading what is currently available.

Regarding the question of running the RC2014 Classic II at a lower speed, there are a few options.  The most flexible is the Dual Clock Module https://rc2014.co.uk/modules/dual-clock-module/ which has a selection of speeds which still give a common serial baud rate clock.  You can also use the circuit from that to divide down your existing 7.3728MHz clock with just a couple of chips or so.  Another option is to replace the crystal in your existing clock module, although that is less convenient to keep swapping around.

Spencer

--
You received this message because you are subscribed to the Google Groups "RC2014-Z80" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rc2014-z80+...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/rc2014-z80/1c040946-d53b-4078-8d35-6d88f55cf8cen%40googlegroups.com.

[Bill Shen]

Eric,
I determined CMOS vs NMOS based on power consumption on my test hardware; with 3.68MHz clock running memory diagnostics at nominal 5.0V, Zilog CMOS draws 80-90mA, NMOS series B draws 150-160mA, and NMOS series A draws 190-200mA.  The differences in power consumption are very distinct.  These are Zilog parts, I don't know about other Z80-compatible.

I'm not aware of zexall for ROM.  Having the source code, you should be able to assemble it for ROM system but zexall makes CP/M BIOS calls, so you need to replace it with your specific I/O calls.  If your RC2014 can configure for the CP/M environment (ROM paged out), then you can run the original zexall even if you don't have CF disk.

Z80 should operate the same (except slower or faster) within its operating frequency range, so you can slow down the clock and reduce the serial baud proportionally to run your processor slower, e.g., 3.68MHz and 57600 baud should work on your set up.  Conversely RC2014 is conservatively designed, so you should be able to run it at 14.7MHz and 230400 baud.

You should always have decoupling caps.  10K pull up is rather slow, I would use 3.3K pull up.
  Bill

[Nathaniel Lockhart]

I did the same thing recently! A couple of them were legit Z80s, but the others drew a high current, exactly like was previously mentioned. And they ALL had the large Z on them with the same date code and everything! I swear you can see the brushstrokes from them being painted over.

Nate

--
You received this message because you are subscribed to the Google Groups "RC2014-Z80" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rc2014-z80+...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/rc2014-z80/2aacae1e-c6ad-4ea2-a989-ad9e133425dan%40googlegroups.com.

[Le Fauve]

Thanks for the advises :o)

Now, for the first time ever, instead of just reimbursing me, AliExpress has sent me a return label to return the "fake" Z80s, so I probably won't be able to test them much.

For the clock, I was thinking to plug the clock module into a breadboard and to add a flip flop to divide frequency by 2.

Now, I had made tests using a breadboard for the Z80, but I wanted a more reliable solution.

I did find a 40 pins ZIF, so what would be the best way (economically) for ordering a couple of CPU modules PCBs?

I had a look at Z80Kits.com which keep telling me shipping will be free, but being in Australia I kind of doubt it :o)

(shipping is always the issues... unless it comes from China :o( ).

Eric

PS: Looking into my parts bin, I found 5 other Z80 I ordered years ago, which looks suspiciously like the ones I just received, so I'll probably test those instead.

[Mark T]

Steve Cousins has the gerber data for his designs published on his website, so you could download these and have a batch manufactured at jlcb or similar. The one I linked includes a clock circuit, so you could build this with a cut down ic socket for the crystal if you want to experiment. You might need to mount some components on the opposite side of the board to fit the zif socket and I’m not sure if the cpu socket holes are large enough to suit the zif socket.

There may be other pcb options in Steve’s project pages.

https://smallcomputercentral.com/sc706-rcbus-z80-cpu-module/

[Lionel Peterson]

Interesting thread.

When I read about the Z80 being discontinued, unlike the folks in this thread I decided to order another RC2014 kit (a mini with CP/M expansion board) to replace the kit I bought, built, and quickly lost track of a couple years ago - and quickly found again about 20 minutes after buying the 2nd kit.

No complaints, glad to have a spare kit, but I wonder how many other folks took the opportunity to get another kit?

Ken

[Spencer Owen]

On 13/05/2024 11:10, Le Fauve wrote:
>
> I had a look at Z80Kits.com which keep telling me shipping will be
> free, but being in Australia I kind of doubt it :o)

Hi Eric,

Yes, there is a charge for shipping to Australia from z80kits. However,
I am curious why it tells you the shipping will be free. Based on your
IP address it should show you £7 or £12 (depending on standard or
tracked).  Are you using a VPN or anything that might confuse the
website?  I am a bit worried that something might be amiss there.

Cheers

Spencer - z80kits

[Sergey Kiselev]

Counterfeit Z80 CPUs (or any other ICs) is not a new phenomenon.

My impression is that Chinese recyclers will sort desoldered ICs according to their type and then relabel them with whatever seems to be the popular IC type and the highest bin/frequency. E.g., most Z80 CPUs will be labeled as 20 MHz part. Sometimes NMOS Z80 parts will be added to the mix ;) Remember that the fastest NMOS Z80 - Z80H were rated at 8 MHz...

That also seems to be the case with other ICs, e.g. 8087 FPUs will get labeled as 8087-1 - the 10 MHz rated part, where the original chip might have been a 5 MHz part. 

From the ICs production perspective, especially for long running legacy products, such as Z80 / Z84Cxx, the process technology was continuously improved, likely until some time in 90's, that includes the switch from NMOS to CMOS and several node changes (die shrinks), with every such transition, the yield (number of working dice on a silicon wafer) and the performance (including frequency rating) are improved, and the power consumption is reduced.

As the result, by mid-90's, it is quite possible that most Z84C00 produced would work at 20 MHz, or even higher frequency, regardless of their actual marking.

There are a few reasons to why most Z80 ICs wouldn't be sold as 20 MHz parts, but rather as parts with lower frequency ratings:

1. Market demand. Selling only some parts as higher frequency, allows charging higher prices, while keeping lower prices for lower frequency, generally more popular parts. This can motivate the manufacturer "downgrade" parts.

2. Testing/binning. When ICs are made, they are tested and "binned" according to the frequencies they can work at. If, due to #1, the IC is going to be sold as a 8 MHz part, there is no reason to test it at 20 MHz. Skipping some tests makes the production cheaper. It is important to realize that while most CMOS parts will work at 20 MHz, they were not tested and might not work reliably at that frequency... which most likely will result in incorrect results of certain instructions (that use the longest path in the ALU).

While Z80 has been in production, I've been a huge proponent of buying new parts from Zilog and its official distributors. I even contacted Zilog back in 2012, and asked them to make Z84C0020PEG parts available for purchase in single quantities (vs. 100).

It is still not too late to put an order for the new Z84Cxx parts, so if you're building your own systems or selling kits, that would be a great time to stock up on some...

With Z80 series being discontinued, I fully expect that the supply will switch to these counterfeit/relabeled CPUs from Chinese sellers.

This raises the question of testing the CPUs, which would be particularly important for folks that sell kits: Last thing you want to do is to remotely troubleshoot someone's not working system, figure out that you've sent a faulty part, and then spend money and time sending a replacement. Sounds like a lot of time and money wasted plus unsatisfactory customer experience.

All this brings the topic of testing the Z80s...

There is Z80 NMOS CMOS tester a project, a few iterations of it actually, last revision seems to be here: https://github.com/djtersteegc/z80-cmos-nmos-tester (Tindie - PCB only: https://www.tindie.com/products/tersteeg/z80-cmos-nmos-tester-pcb/)

It does check whether the Z80 is CMOS or NMOS variant (or East German U880, not that Chinese will ever recycle that). It also allows you to run that CPU at a certain frequency, and measure the power consumption (using an external DMM).

Which might be good enough of a test...

As Bill mentioned, I am working on developing the z80type utility (https://github.com/skiselev/z80-tests) - It currently can determine whether the CPU is CMOS or NMOS, and also distinguish between several manufacturers of CMOS CPUs. There is more research going, and it seems that I might be able to add detection of a few more CPU types. With that being said, this utility is not designed to do an exhaustive CPU test...

There are utilities like zexall/zexdoc that test all Z80 instructions with multiple inputs and verify the correctness of the results. These utilities are mostly intended to check the correctness of emulators, not to test actual Z80 CPUs. It also takes a while to run them - a few hours, depending on the frequency.

I was thinking about improving the Z80 tester (linked above), to do more thorough testing specifically. Some ideas:

• Hardware:
  • Programmable clock divider for the CPU frequency. Ideally set to lower frequency after a reset. I can't seem to find a good part for it... One possibility is to just implement my own in a CPLD...
  • Watchdog
  • A read/write register to do NMOS vs CMOS test. The original tested doesn't have read access, so while it makes the CMOS vs NMOS visible to user, it is not visible to the software...
  • SRAM - will allow for more flexible software vs. just using CPU registers
• Software:
  • Test the CPU type. For NMOS CPUs - there is no reason to even try anything higher than 8 MHz.
  • Run the CPU while gradually reducing the frequency. E.g., run at 20 MHz, if it gets stuck or produces incorrect results, reset (by watchdog in case of a hang CPU), and try lower frequency
  • More exhaustive Z80 tests: test all the Z80 registers, exercise various Z80 instructions, etc.

Is anyone interested in such a tester and wants to collaborate?

--Sergey

[Bill McMullen]

Sergey: I understand your logic and the manufacturing / marketing issues but I think there's one point you might have overlooked for the small distributor of kits.

I'd like to be able to include new Z8S18033 processors from authorized distributors and in fact I have done so on several occasions.  However, those chips cost me very significantly more (4-5 times) the cost of used chips.  That leaves me the choice of either raising kit prices by at least 25% or testing and including used chips.  After the Covid supply issues / cost increases, I had to raise prices about 12% and the sales dropped considerably.  As it is, my profit from kits is VERY minimal on a per hour rate.

I've chosen to provide used chips (with disclosure) BUT I also personally test every such chip.  PLCC chips are convenient to test and I do so on my reference system at full rated PHI and ASCI speeds.  The QFP chips are more difficult to test without a clamshell but I only include those in assembled and tested systems which have run for at least 24 hours.  Have I found some bad chips?  Absolutely and some of the issues were subtle but those chips immediately went into the garbage can.

So I think the real question becomes whether kit purchasers are willing to pay an extra $15 or more for new chips.  Of course I could reduce that to perhaps +$7 or so BUT then it becomes an inventory issue and whether I'm willing to pay $7,800 up front for 660 chips (FYI, I'm not willing).  I've dealt with some overseas ISO certified non-authorized distributors and been quite satisfied but it's one more hassle and the couriers get to enjoy my increased shipping costs.

Given my experience that the quality of the used chips is definitely declining, it's likely that I won't be making Z180 kits and systems in the future but will simply stick to eZ80 systems which give much better price-performance.

[Sergey Kiselev]

Bill,

I haven't overlooked that point. I was and I am still selling kits. I've been buying new authentic Z80's since 2012 when I first encountered the issue with counterfeit parts.

Yes, it makes everything a bit more expensive. But it saves time and headache, and it supports the manufacturer and not the counterfeiters.

With that said, I am not making money selling kits. In the best case it pays for the parts my future projects. I consider it more of a community service...

Of course the Z8S18033 in the PLCC68 package is no longer produced, and it might be difficult to procure new old stock parts.

--Sergey

[Bill Shen]

I haven't sold any Z80 kits ever, but I do know my designs have been downloaded and built by a number of people.  So even for non-kit designers like myself, it makes sense to survey the used components market periodically.  Designs will need to evolve and adjust according to market conditions and availability.  For an example, I'm currently working with Andrew Lynch on text-to-speech design based on SP0256-AL2.  Just a few weeks of hands-on experiences and reading about others experiences have demonstrated that working SP0256-AL2 is expensive and hard to find; that non-working fakes are everywhere, so design approach will need to be quite different, if it is ever realized.

Z80 and its peripherals, fortunately, are abundant and used parts are cheap and mostly working, so the RC2014 community is in a good place, for now.
  Bill

[Bill McMullen]

Sergey: I fully understand and respect your desire to use new chips and support the manufacturers.  If I could get Z180 processors in the $7 range for low quantity, like the basic Z80's from Mouser, I'd definitely be using them.   However the Z8S18033's are/were about triple that price and being North of the 49th I also have to deal with extra taxes and brokerage fees.

Ironically the eZ80F91 at 50 MHz is only 2/3 the price of a Z8S18033 and offers a LOT more capabilities and about 4.5 times the performance.  Given the more reasonable pricing, I only use new eZ80's from authorized suppliers.

FYI: I've successfully used an ISO certified Asian distributor in the past and I note that they have a significant stock of new PLCC Z180's.  I just have to decide how serious I am about producing more Z180 kits / systems and whether those chips are worth the extra $40+ courier fee.

[Tadeusz Pycio]

I am in a comfortable systuition because I do not sell my modules, so I have no problem with the quality of the electronic components on offer. I also represent a different approach to this problem, as I only use second-hand components. Firstly, they are definitely cheaper, although here I often have problems with fakes, and secondly, I 'save' such chips from disposal by giving them a second life. Retro kits are not mission-critical systems, so no reliability guarantee is required. I often browse the NOS chip sale adverts and use them, so exotic chips are sometimes used in my projects.

The idea of such a tester should be particularly attractive to me, but here I have a bit of a dilemma as to whether the form of a dedicated device is justified. Great idea with the changing clock and control via watchdog. I would see this device more as an RCBus module that will use the available CPU and memory modules. If we are talking about a CPLD then it would be tempting to add other functionalities like single-step, bus monitor,... which will certainly be of interest to a larger audience.

It's a good idea and to the best of my modest ability I will support it.

[Tadeusz Pycio]

I don't want to suggest anything, but I'm starting to see such a tester as a module with a DS1077 programmable clock and a PIC16F877 microcontroller (because this one has an 8 bit port - PSP to communicate with the Z80 bus and can be used as a watchdog) and glue logic. It is just an idea, without checking its correctness. This is not a retro solution and there will certainly be better ideas.

[Le Fauve]

On Monday, May 13, 2024 at 6:33:35 PM UTC Spencer Owen wrote:

Hi Eric,

Yes, there is a charge for shipping to Australia from z80kits. However,
I am curious why it tells you the shipping will be free. Based on your
IP address it should show you £7 or £12 (depending on standard or
tracked).  Are you using a VPN or anything that might confuse the
website?  I am a bit worried that something might be amiss there.

Thanks for the reply Spencer!

I didn't use any VPN. I just added two CPU PCB, and I got this (see image below; I didn't completed the order though, so perhaps it's supposed to adjust the postage later, but I cannot chose any international shipping option):

I'm a little annoyed at the idea of paying 7 GBP of postage for 10 GBP of goods, but I'll probably get the CPU module kit instead, at least I'll get two real Z80 (assuming they'll be available again).

In the meantime I'll try to use my wirewrapping tool to do something a little more reliable :o)

About the counterfeit chips, I returned them this morning, and already got reimbursed around 2pm. At least this part of the deal worked fine.

I'm just annoyed that I had to return them, because they will surely try to sell them again...

Also, I didn't mentioned it but I also had ordered 5 PIA and 5 SIA/2 at the same time. They were delivered.
The PIA are probably fakes (they all have the exact same marking on top side but look slightly different on the bottom side). I'm not sure how to test them yet, but I'll see if I can try something.

The SIA/2 were a little more expensive (about 3 euros each) but at least they looks like genuine used (and not rebranded) parts (they all have different marking and production date).

So my next step is probably to order a few Z80, PIA and SIO/2 from mouser or some serious place.

Is there any other "Z80 family" chip I should order? I mean I don't have any project at the moment, but I'd like to have the parts to build at least a minimal functioning Z80 based computer or some common peripheral (parallel port, ...).

Eric

[Spencer Owen]

Hi Eric,

On Sat, 18 May 2024 at 10:09, Le Fauve <lefa...@gmail.com> wrote:

Thanks for the reply Spencer!

I didn't use any VPN. I just added two CPU PCB, and I got this (see image below; I didn't completed the order though, so perhaps it's supposed to adjust the postage later, but I cannot chose any international shipping option):

 Ah, it seems like it defaults to UK shipping unless you are logged on to your account. You can click the Calculate Shipping link and set the destination to Australia if you are not logged on. 

I'm a little annoyed at the idea of paying 7 GBP of postage for 10 GBP of goods, 

That "one size fits all" covers items up to several hundred pounds.  Please feel free to calculate what it would cost to post a 50g parcel from Australia to the UK and I will be happy to charge you that amount.

Spencer 

[Lionel Peterson]

It literally says "shipping options will be calculated during checkout", why should/how could the shipping cart list all available shipping options BEFORE you tell it the shipping address?

Inferring shipping address from IP address? I guess, but honestly that would likely be hit-or-miss, better to specify shipping info then see actual rates/options IMHO.

Ken

On May 18, 2024, at 04:09, Le Fauve <lefa...@gmail.com> wrote:



On Monday, May 13, 2024 at 6:33:35 PM UTC Spencer Owen wrote:

Hi Eric,

Yes, there is a charge for shipping to Australia from z80kits. However,
I am curious why it tells you the shipping will be free. Based on your
IP address it should show you £7 or £12 (depending on standard or
tracked).  Are you using a VPN or anything that might confuse the
website?  I am a bit worried that something might be amiss there.

Thanks for the reply Spencer!

I didn't use any VPN. I just added two CPU PCB, and I got this (see image below; I didn't completed the order though, so perhaps it's supposed to adjust the postage later, but I cannot chose any international shipping option):

<Screenshot 2024-05-18 at 18-44-32 My Cart - Z80 Kits.png>

I'm a little annoyed at the idea of paying 7 GBP of postage for 10 GBP of goods, but I'll probably get the CPU module kit instead, at least I'll get two real Z80 (assuming they'll be available again).

In the meantime I'll try to use my wirewrapping tool to do something a little more reliable :o)

About the counterfeit chips, I returned them this morning, and already got reimbursed around 2pm. At least this part of the deal worked fine.

I'm just annoyed that I had to return them, because they will surely try to sell them again...

Also, I didn't mentioned it but I also had ordered 5 PIA and 5 SIA/2 at the same time. They were delivered.
The PIA are probably fakes (they all have the exact same marking on top side but look slightly different on the bottom side). I'm not sure how to test them yet, but I'll see if I can try something.

The SIA/2 were a little more expensive (about 3 euros each) but at least they looks like genuine used (and not rebranded) parts (they all have different marking and production date).

So my next step is probably to order a few Z80, PIA and SIO/2 from mouser or some serious place.

Is there any other "Z80 family" chip I should order? I mean I don't have any project at the moment, but I'd like to have the parts to build at least a minimal functioning Z80 based computer or some common peripheral (parallel port, ...).

Eric

--

You received this message because you are subscribed to the Google Groups "RC2014-Z80" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rc2014-z80+...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/rc2014-z80/e313d89f-dbf5-41f9-9ef9-c8c358fb7806n%40googlegroups.com.

[Dirk Prins]

Some time ago I created an emulator for the RC_2014 to debug my programs.
As a part of that I used the ZexLax (ZexDoc) program to verify the emulation.
Seeing this thread I thought it interesting to make a rom version of the program.
It proved to be not quite that simple, since the program is designed to run from address 0x0100 I have to copy it from ROM to $8000 and after disabling the ROM back again to 0x100.
I encountered also a few snags on the way.
But it works now (see attachments):

(I did put this post up earlier but deleted it again because I noticed I responded to the wrong message, also I added some more options to the program)

Op zondag 12 mei 2024 om 05:49:56 UTC+2 schreef Le Fauve:

[Dirk Prins]

My rom settings for this image:

32 K starting at 0x0000 (16K at 0x0000 should be ok too).

Op maandag 20 mei 2024 om 14:07:00 UTC+2 schreef Dirk Prins:

[Bill Shen]

Here is another report with re-labelled used Z80 from eBay:

I have three SBC designs that use Z80 PLCC44 package.  The price of Z80 PLCC44 package is going up lately, so I've purchased 20 PLCC Z80 from this vendor to stock up, https://www.ebay.com/itm/353635270036 .

Received the parts 3 weeks later and observed all 20 PLCC Z80 have the same date code (1336), so more likely than not they are re-labelled parts.  I tested them on ZRCC rev1.3 which is a testbed for overclocking Z80 to 29.5MHz and higher.  Here are the test results at room temperature and 5.0V :

• 16 passed memory test, booted into CP/M and passed the full zexall test at 29.5MHz
• 3 failed to boot at 29.5MHz, but boot OK at 25MHz
• 1 will not boot at all, not even at 3.68MHz.

I tested the three Z80 that failed at 29.5MHz in Z80ALL which has operating frequency of 25.175MHz.  They all passed memory tests and zexall test.

The characteristics of this batch of PLCC Z80 are consistent with my previous experiences that PLCC Z80 either does not work at all or works at 25MHz or faster.
  Bill

On Saturday, May 11, 2024 at 3:25:29 AM UTC-6 Le Fauve wrote:

[Le Fauve]

On Saturday, May 18, 2024 at 12:41:37 PM UTC Spencer Owen wrote:

That "one size fits all" covers items up to several hundred pounds.  Please feel free to calculate what it would cost to post a 50g parcel from Australia to the UK and I will be happy to charge you that amount.

Thanks Spencer for the kind offer, but I think I'll just try ordering complete CPU board kits instead of just the PCB (hopefully they'll be back in stock soon?). If I found another board or two, the regular shipping price will make sense :o)

On Monday, May 20, 2024 at 12:07:00 PM UTC Dirk Prins wrote:

Some time ago I created an emulator for the RC_2014 to debug my programs.
As a part of that I used the ZexLax (ZexDoc) program to verify the emulation.
Seeing this thread I thought it interesting to make a rom version of the program.
It proved to be not quite that simple, since the program is designed to run from address 0x0100 I have to copy it from ROM to $8000 and after disabling the ROM back again to 0x100.
I encountered also a few snags on the way.
But it works now (see attachments):

(I did put this post up earlier but deleted it again because I noticed I responded to the wrong message, also I added some more options to the program)

Thanks Dirk, that will be helpful. I cannot just test it at the moment because my ROM module only support 8KB roms (and I cannot easily bank it off) but I'm sure I'll manage something (in any case I'll have fun trying ;o) ).

Eric

[Dirk Prins]

Hi Eric,

Did a little check but just the pure zexdoc binary will be over 8KB, so no possibility to create a smaller version.

Dirk.

Op zaterdag 25 mei 2024 om 10:29:25 UTC+2 schreef Le Fauve:

[Le Fauve]

On Saturday, May 25, 2024 at 11:12:56 AM UTC Dirk Prins wrote:

Did a little check but just the pure zexdoc binary will be over 8KB, so no possibility to create a smaller version.

How much over? 3D Monster Maze was over 11KB but I still managed to make it fit in a 8KB ROM with some compression.

Your bin file is 10KB so unless this is very tight code I think it may fit. On the other hand, I'll have to find a way to make it run at $8000 or above, because I have no easy way to disable my ROM.

By the way, I don't know your ROM module, but the old one only grab the bus when reading from memory, not when writing. If that's the case with yours too, you should be able to read from ROM/write to RAM without disabling the ROM, which would save you the intermediary copy at $8000 (don't quote me thought, I didn't try this yet ;o) )

Eric

[Dirk Prins]

Maybe it could be fitted but it would take a lot of work I think.

Also if you were to run it from $8000 you would have to change the zexdoc program itself because all the crc's are calculated from the presumption it will run from $0100.

So you would have to run an adapted program and take down all the crc's and incorporate them in the new program (to run from $8000).

A lot of work :-(

My ROM module is the standard pageable ROM: https://rc2014.co.uk/modules/pageable-rom/

Interesting thought about the ROM/RAM tho.

Copying directly to RAM should be possible yes, but it will only take a little bit of code to play safe and do it in 2 steps.

Dirk.

Op zaterdag 25 mei 2024 om 13:37:31 UTC+2 schreef Le Fauve:

[Bill Shen]

If you examine zexall.com loaded in memory, you'll see large blocks of 0's that take up about 1/3 of the code.  The program is 8.7K ($100-$2300), so it should be possible to compress to fit 8K even with a fairly simple compression method.
  Bill

[Michael A. Druckenmiller Sr]

Most interesting,

I also ordered a set of 10 from eBay.

Here's my question as I have never actually finished building my RC2014,
and I am clueless as to how to go about testing these CPU's.

Is there a step by step, for dummies/newbies as to how to use the
ZEXALL/ZEXDOC .com files?

And, how to get legible results back?

Thanks,

Mike Sr.

[Dirk Prins]

Mostly I got my info from:

https://jeffavery.ca/computers/macintosh_z80exerciser.html

After that it was mostly stepping through the code with my emulator.

So just see what the code does (in your favorite emulator/debugger and you get a good idea how it works.

It only reports successful or failed tests.

So it a test fails I rerun it in debug mode and save the result in a text file.

Then I compare it to a text file output (also in debug) of a successful test so I can see where the error originates (first occurrence of different crc's) and what the starting machine state and the resulting machine state is.

So if a see for instance a flag that is not set I can look into that.

A lot of work ;-)

Dirk.

Op zondag 26 mei 2024 om 11:17:31 UTC+2 schreef Michael A. Druckenmiller Sr:

[Mark Durham]

I wonder if the markings that Eric described in post #1 are in common use amongst AliExpress suppliers.

I was stocking up on some DIL PLCC sockets and thought I'd take a chance with an AliExpress seller/shop called "Patterned Painting Store" (https://www.aliexpress.com/store/3092054?spm=a2g0o.store_pc_home.pcShopHead_8004688.0) and ordered 4x Z84C0020PEC (CPU) and 4x Z84C3010PEC (CTC). I wasn't holding out much hope since the CPU was 0.45 GBP and the CTC 0.64 GBP each.

Well they arrived here in the UK - 2 sets of 4 chips stacked together - and wrapped in some sort of transparent cling film. The CTC chips didn't look very good. The outer chip of the bundle had all sorts of crease marks on it as if the seller had only just rebadged the chip and hadn't bothered to wait for the coating to dry before wrapping them in cling film.

The Z80s all had the same markings as Eric described.

I've got a small setup using Steve Cousins boards - SC108 + SC110 + SC129 running off a 7.3728MHz clock on the SC108 board - which I can use to see if the CTC and Z80 are even Zilog devices. I'd be happy if they worked at that speed but I would also be interested to know if the Z80 could really handle 20MHz. I can replace the oscillator on the CPU module but I'd need to add in a 7.3728MHz one on the SIO/CTC board (SC110) to get the baud rates right whilst testing.

Hopefully attached are some images of the Z80s and the CTCs. I did try rubbing a couple of the CTCs with alcohol (IPA) and most of the gunk came off along with the markings as can be seen in one of the images.

[Les Bird]

I've done a lot of testing of these Ali Z80 chips. I ordered 100 Z80s from the same supplier as Mark, Patterened Painting Store. I have a Z80 tester that I can set to different clock speeds and at the highest setting of 20MHz not a single chip worked. I then preceeded to test each one by dropping the speed divider in half one notch at a time (10MHz, 5MHz then 2.5MHz). I ended up with a good collection of 10, 5 and 2MHz chips and a few 0MHz chips.

I also found a supplier on Ali that sold Z80s in a PLC44 pacakge for $1.60/ea, Silicon Patterend Store (no idea where they get these store names from). I ordered 50 of them and found this cool PLC44 to DIP-40 adapter (https://martenelectric.cz/z80cpu_plcc44_dip40_adapter.html). These all came in the sealed package and, honestly, looked legit and brand new. So far each one has worked flawlessly at 20MHz. Seems like a nice option for a DIP-40 Z80. The PLC44 to DIP40 adapter uses the PLC44 through hole socket so it's easy to solder.

 

[Sergey Kiselev]

They might work at 20 MHz, but they are fake / remarked.

First of all, Zilog was not using this logo in 2013. Second, the marking on them suggests non-RoHS compliant parts, and Zilog stopped manufacturing them in early 2000's. It would be interesting to check and compare the bottom side of the ICs, and see if they all look the same, or they look differently.

[Mark Durham]

Hi Sergey, I was hoping that somebody would jump in with those sorts of details. I guess knowledge like you have regarding the historical manufacturers markings on the chips isn't readily / easily available on the web?

The attached photo is of the bottom view of all 4 chips.

[Tadeusz Pycio]

I once wrote this 'guide' and I plan to put it on my website in a broader form.

1,2 - this designation appears from the late 1970s and is found in processors manufactured until the mid-1980s

3 - this marking appears in productions of the second half of the 1980s
4 - my favourite logo, the Phoenix, is from the 1990s
5 - the turn of the century logo with a large Z in yellow or white, reproduced by counterfeiters using lasers

6 - contemporary products 

Processors with the Phoenix logo have their variations. Between 1990 and 1991, a different Zilog font was used, which was changed in 1992. There are also different variants of the numeral 4.

If anyone has more detailed information, please contact me. I don't know the exact date of the change of designation from the big Z to the modern form (circa 2003?).

[Bill Shen]

The bottom of chips are not re-marked, so the easiest way finding out whether a batch of identically date-coded parts are re-marked is look at the bottom.  If the bottom are different, then they are re-marked.

I have designs based on PLCC44 Z80, so I ordered some, then some more, then even more for a total of 50.  I tested them all and found they all worked and were all CMOS (judging by current draw); 90% of them ran CP/M and passed zexall at 29.5Mhz; the remaining 10% ran at 25.175Mhz and passed zexall.   I also tested a few at 3.3V, 16mhz and the few tested all worked.  PLCC Z80 are much more consistent than DIP Z80.

  Bill

[Mark Durham]

I did a quick test on my setup with an SC108 + SC110 + SC129. My reference CPU is marked just like #4 in the photo that Tadeusz uploaded but with a date code of 9303, and my multimeter shows a current of 44.8mA at the SCM prompt.

All 4 Z80s that I purchased, work at 7.3728MHz in my setup. I don't have a way as yet of changing the CPU clock easily to see if they will work reliably at higher clock speeds. What was of interest was that 2 chips drew around 49mA at the SCM prompt and the other 2 chips drew around 100mA. Clearly not 4 chips of the same type. Perhaps I have 2 CMOS devices and 2 NMOS devices.  

[7alken]

ya, this is always possible issue ... I feel that I left quite "disgusting amount of money" these days on ali for small amouts of samples of cpld/fpga/displays/sdrams/connectors, each from more vendors, so hope I will have some statistics, one day .... ))

in relation to testing QFP (or even BGAs, I try to tackle with them as it seems its in fact easier to hotair(bottom/up)  them than fine 0.5 or 0.4 QFPs or even QFNs) ... So I tried to think how to replace BGA ZIF... only for quick testing, to "push" chip somewhere for a while ... and then came idea of flex-PCB, in case of BGA to lay over say 3 layers of 0.3mm thin, processed to allow pulling contacts from balls to edge, at least for 1.27mm / 1mm (not sure about 0.8 ...) pitch, having this "sandwich" on some rubber/silicon, just something forcing the flex-PCB to the balls, even if I only "thumb" the chip on it for test ...

Till now, its only picture, though, so if anybody has time to test it or try something like that for QFP (QFNs are evil, grrr), then GO :-)
(some flexpcb on twitter confirmed they CAN do metalized vias in it and also the edges to be solderable to the PCB around ... so...)

(... the particular chip is little secret, but info IS here )) .. my holy grail of curiosity in retro ...)

7alken

[Nicolae Muntean]

Z80 Romanian clone with signature FEDERICO FAGGIN!!!!! NEW IS ORIGINAL

--

You received this message because you are subscribed to the Google Groups "RC2014-Z80" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rc2014-z80+...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/rc2014-z80/39603039-a79d-436f-ba0e-820bf8b93533n%40googlegroups.com.