Atari 520STFM with strange "bodges"

Troubles with your machine? Just want to speak about the latest improvements? This is the place!

Moderators: Mug UK, Zorro 2, spiny, Greenious, Moderator Team

Post Reply
atary
Atariator
Atariator
Posts: 29
Joined: Thu Mar 21, 2019 7:03 pm

Atari 520STFM with strange "bodges"

Post by atary »

I just received an Atari 520STFM, bought off Ebay, sold "as seen or for repairs". The pictures showed that it was pretty dirty, and that the RIFA cap in the PSU had let go. It was cheap enough £40, so I thought "why not".

I've just opened it today. Various screws missing or the wrong size, but no surprise that the whole thing had been disassembled in the past. It's dirty and dusty inside, with signs of rust on most pieces of metal, but nothing serious. There are no obvious hacks to the PCB.

I did notice what appears to be a factory "bodge". I wondered how common this was any why. Given that it appeared to be associated with the DMA chip, I wondered if this was the famous "DMA bug" version. BTW I have read all the postings on this subject and opinions do vary!

There is a diode possible from +5V to pin 40 (big assumption here that pin 40 is the power pin!).
Atari ST520 DMA fix1.JPG
There is also a wire on the back which appears glued down, which I why I suspect these bodges are from the factory.
Atari ST520 bodge wire1.JPG
PCB version is C103175 Rev 1.1. Appears to have 6 ROM locations, but 2 physical chips. I haven't tried to power it up yet so I have no idea what version of TOS this might be. I'm guessing that it is a later build since the DRAMs are 256k x 4. BTW the other four sockets are free from solder, so upgrading to 1Mb will be trivial.
Atari ST520 version.JPG
Any comments would be of interest.

First job is clean all the plastics, and then lightly sand the metal to remove the slight traces of rust.
You do not have the required permissions to view the files attached to this post.
User avatar
DoG
Captain Atari
Captain Atari
Posts: 231
Joined: Sun Apr 01, 2018 11:02 pm

Re: Atari 520STFM with strange "bodges"

Post by DoG »

The diode is factory for the IMP chip to reduce power to 4,5V.
atary
Atariator
Atariator
Posts: 29
Joined: Thu Mar 21, 2019 7:03 pm

Re: Atari 520STFM with strange "bodges"

Post by atary »

I did a bit more searching and found some interesting information.

It appears that at some point in time Atari moved manufacture of their custom chips (shifter, MMU, DMA, GLUE) to IMP. Initially there were problems if the Atari was built using all 4 custom chips from IMP, so they mixed up the IMP chips with the older parts. Later the problem was fixed.

There was an interesting French technical document (translated on another website - no idea if I am allowed to link) which gave some hardware fixes, one of which was the addition of the diode between the 5V power rail and the 5V power pin of the DMA chip. The lower voltage slowed down the DMA chip which increased stability. Sounds to me that something in the initial design was "on the edge". The newer IMP chips appear to have been faster than the older ones (*) which 1) could cause timing issues 2) could result in more switching noise. Apparently the DMA fix needed to be accompanies with a fix to shifter too.

In my 520STFM all 4 custom chips are IMP make, but there is no fix for shifter.

(*) IC manufacturing processes tend to get better (i.e. faster and higher yield) over time. Sometimes the chip layout is shrunk to allow manufacture on a newer process node (smaller geometry size). This too would result in the same chip having a higher maximum operating speed and different I/O timings.
User avatar
DoG
Captain Atari
Captain Atari
Posts: 231
Joined: Sun Apr 01, 2018 11:02 pm

Re: Atari 520STFM with strange "bodges"

Post by DoG »

Atari documents about Mega ST factory mods. I got a Mega ST1 with IMP on both Shifter and DMA with these mods.

modification_imp.PNG
You do not have the required permissions to view the files attached to this post.
atary
Atariator
Atariator
Posts: 29
Joined: Thu Mar 21, 2019 7:03 pm

Re: Atari 520STFM with strange "bodges"

Post by atary »

The French text can almost be understood even if your French isn't that good!

Here is a translation (thanks to Google translate):-
IMP modification

The purpose of this modification is to reduce the supply current of the IMP chip to about 4.5 volts, in order to make it less rapid and to avoid any disturbance. This decrease in current can be obtained as follows:

Printed circuit C100167-001 rev 5.0 Mega ST2 / 4

Lift the leg 40 of the DMA (U27) and solder the diode 1N4148 between this leg and the + side of the C55. The cathode on the capacitor side. Lift the tab 40 of the "video shifter" (U31) and solder the BAT48 or BAT85 diode between this tab and the + side of C65 (the cathode to C65).

Being pedantic is should be "supply voltage" and not "supply current".

A 1N4148 is a very common silicon diode and will drop a voltage of around 0.6 to 0.7V (depends on the current flowing through it). The BAT48/85 is a Schottky diode and drops less voltage (0.3 to 0.4V).
mlynn1974
Captain Atari
Captain Atari
Posts: 443
Joined: Mon Mar 03, 2008 10:33 pm
Contact:

Re: Atari 520STFM with strange "bodges"

Post by mlynn1974 »

I always thought that TTL in those days was strictly +5 Volts, not 4.5V but I now understand that these things are all relative. Why would they make an out of spec component that requires 4.5V for an already established design? Was heat dissipation a consideration?
Still got, still working: Atari 4Mb STe, 520STFM (x2), 2.5Mb STF, Atari 2600JR, Flashback 8 Gold.
Hardware: Cumana CSA 354, Ultimate Ripper, Blitz Turbo, Synchro Express II (US and UK Versions).
atary
Atariator
Atariator
Posts: 29
Joined: Thu Mar 21, 2019 7:03 pm

Re: Atari 520STFM with strange "bodges"

Post by atary »

So @mlynn1974, I could spend hours answering your question, since I worked in a related area of the IC industry.

5V was for quite some time a standard supply voltage, so ICs were designed to work on that voltage. If you check datasheets of 5V ICs you will see that they actually specified to work at a VCC (this is the power supply name) 5V +/-5% (4.75 to 5.25) or 5V +/-10% (4.5 to 5.5). They are also specified to operate over a temperature range : 0C to 70C for commercial devices (i.e. the type of parts used to build the Atari); or -55C to 125C for military parts.

When I say specified, it leans that they are guaranteed to meet all datasheet specifications over the range of VCC and temperature. Other datasheet specifications mean: input level thresholds; output voltage levels; power supply current (ICC); and timings(signal delay).

As a generalisation for many 5V type devices, either bipolar (like TTL ICs) or CMOS, they operate faster (i.e. internal delays shorten) either at higher VCC or at lower temperatures. So if a 5V CMOS microprocessor (for example) is specified in the datasheet to operate (at a minimum of) 10MHz, then it will operate *at least* at 10MHz at 70C and 4.75V (the worst case conditions). But since we know that it will operate faster at 0C and 5.25V, if we do raise the VCC above the *nominal* 5V and cool the IC (either with active cooling or fan+heatsink) then we know that the IC will operate at more than 10MHz. How much more is NOT guaranteed. This is why overclocking works.

Now although an IC might have a VCC specification of 5V +/-5% (i.e. 4.75 to 5.25V) this does not mean that it will stop working outside this range, just that it is NOT guaranteed to meet specifications, however in many cases it might well do.

TTL is common type of 5V IC. The LS type (low power Schottky) was very common in the 80's and is used in the Atari. A 74LS00 is a commercial (74) quad NAND and is specified to operate at 5V +/-5% and 0 to 70C. The 54LS00 is a military version (54) quad NAND and is specified to operate at 5V+/-10% and -55 to 125C. BUT the silicon used for both commercial and military comes off the same production line! So although it looks like a 74LS00 will only operate from 4.75 to 5.25V we know in reality a 74LS00 will at least operate over the same range as the military device i.e. 4.5 to 5.5V.

So at a lower VCC than specified in the datasheet most ICs will still operate, but they might not meet all specs, and one of these is timings, since the silicon internal delays will be longer. This is the effect that the voltage drop caused by the diode on the VCC power pin causes (and is an intended effect). Reduce VCC too low and the IC will malfunction, but the device won't be damaged and it will work OK if you raise VCC back to a suitable level. Raise VCC too far (for example some datasheets have an "absolute maximum" VCC of 7V) and you could damage the part permanently. If you don't go above the absolute maximum you shouldn't damage the part, but it's not normally something done.

In the PC world CPU VCC has been pushed well above specified limits in order to achieve higher speed for many years, and it is known as overclocking. It is much more complicated due to having more than one power supply voltage.

In practise we don't have the datasheet for the IMP parts (AFAIK) so depending upon the specification of the VCC range, the voltage after the diode drop on the VCC pin of the DMA part could either be "in spec" or "just below specification" (which isn't really a big deal).
mlynn1974
Captain Atari
Captain Atari
Posts: 443
Joined: Mon Mar 03, 2008 10:33 pm
Contact:

Re: Atari 520STFM with strange "bodges"

Post by mlynn1974 »

Thanks for the explanation @atary. In Computer Design II at University in 1993 they didn't explain the effects of under-voltage or the tolerances.
Sinclair had similar issues with non-Ferranti ULA chips. I wonder how many ST models used these IMP chips and if they have any undesirable effects?

Reminds me of a scrolltext in Pompey Pirates menu though where they claim some STs run at different speeds:
"When Writing This Menu, I Found 4 Different ST's Ran At Completely Different Speeds... Genie's One Is Slower Than Mine, Sledgehammers's Really Crawls Along And A Half Meg Old TOS One We Borrowed Ran About The Same Speed As Genie's. Now Either Mine And Neil G.'s Run Faster Than Normal Or Motorola's Quality Control Is Sloppy!"
Could these differences explain that?
Still got, still working: Atari 4Mb STe, 520STFM (x2), 2.5Mb STF, Atari 2600JR, Flashback 8 Gold.
Hardware: Cumana CSA 354, Ultimate Ripper, Blitz Turbo, Synchro Express II (US and UK Versions).
User avatar
DoG
Captain Atari
Captain Atari
Posts: 231
Joined: Sun Apr 01, 2018 11:02 pm

Re: Atari 520STFM with strange "bodges"

Post by DoG »

IMP chips are very common. Since the factory bodge job was introduced later in the revisions, there is IMP with and without the mod. The IMP chips have a reputation of being worse. Not always justified though. There is so many board revisions and models that each Atari have there own quirks.

Atari slapped together what they had in the factory at the time. "Power without the price". I have seen ROMS without sockets, sockets fitted crooked, 68k proc fitted with bent pins and broken pins from factory just hanging on with solder. And other strange things.
ijor
Hardware Guru
Hardware Guru
Posts: 4078
Joined: Sat May 29, 2004 7:52 pm
Contact:

Re: Atari 520STFM with strange "bodges"

Post by ijor »

Except Blitter that is a special case, all ST custom chips were initially Gate Arrays. The third generation were full custom chips fabbed by IMP. Full custom chips are cheaper to produce in quantities, but developing and testing time is much longer.

Gate arrays are usually slower and then IMP chips tend to be faster. A faster chip is more sensitive to glitches and undershoots. This was, I believe, the main problem with IMP chips. Especially the DMA chip is extremely sensitive to glitches. There is also an issue of a faster chip producing a different clock skew. This probably affected Shifter.
mlynn1974 wrote: Sun Jun 13, 2021 12:09 pm Reminds me of a scrolltext in Pompey Pirates menu though where they claim some STs run at different speeds:
"When Writing This Menu, I Found 4 Different ST's Ran At Completely Different Speeds... Genie's One Is Slower Than Mine, Sledgehammers's Really Crawls Along And A Half Meg Old TOS One We Borrowed Ran About The Same Speed As Genie's. Now Either Mine And Neil G.'s Run Faster Than Normal Or Motorola's Quality Control Is Sloppy!" Could these differences explain that?
I don't think so. We are talking about speed in two different senses. The chipset speed refers to analog delays. How much time it takes for a signal to change after the active clock edge, sensitivity to glitches, clock skew, etc. The actual computer speed depend only on the frequency of the crystal oscillators.

The ST was manufactured with slightly different frequencies, including, but not only, PAL vs NTSC systems. This is documented in a few places. I'm not 100% this was what Pompey Pirates were detecting though. May be they run into issues such as wakeups or interrupt jitter that were not understood at the time.
Fx Cast: Atari St cycle accurate fpga core
Post Reply

Return to “Hardware”