Jump to content

Recapped a Performa 5300 - now doesn’t power up


Recommended Posts

I really want to test the emulation speed of the 603 processor, so I picked up a 5300 locally yesterday. It was able to boot up. 
 

Of course, I wanted to see the condition of the motherboard before I speed tested it. I had a hard time removing the board (someone before had managed to get it in and bent a bunch of the metal shielding around it to present it from coming out - forcing me to disassemble it just to get the motherboard out. 
 

By the find I had the board out I could see if was very dirty and needing a recap. So I started. 
 

I took pictures of before during and after. It was pretty straightforward, about the 70th Mac I’ve recapped. The only issue I had was that it used a 22uf 25 volt cap near the processor. The Apple IIe PDS card as well as the IIci cache card also use these, but they also interchangeably use a 22uf 35v capacitor. Because Of this I only have 22uf 35v caps. It has never been a problem before and many people say you can always use a larger voltage cap in place of a lower voltage one. 
 

After the recap, the board will not power up. I hear a click like it’s trying but nothing happens. 
 

I took the 5300 board and tried it inside a Quadra 630 case. Same thing. I took the Quadra 630 board and tried it in the 5300 enclosure. Nothing. I then retested the Q630 board inside the 630 case. Booted fine. 
 

I retried the 5300 board, and it clicks but that’s it. In both the 5300 and the Q630 case. 
 

What are the chances that the 22uf 25v is specifically required ?

 

Photos attached. 

41A6E250-981D-46DC-AF77-1CA85E878B64.jpeg

A82BA397-1ABD-46FF-B2BB-A1C7FB52EFB6.jpeg

65ACC580-63CA-46DB-AD85-A5E25834494D.jpeg

EEB94C7A-D24D-4805-A4C4-031BDA88A582.jpeg

833058AF-31F0-41D7-B634-91C13AB7A700.jpeg

Link to post
Share on other sites

A click and then nothing sounds almost like one of the power rails has somehow got shorted, or something along those lines, and it's tripping the PSU.  Might want to check that all the power rails are going where they ought to be and nowhere else and you haven't accidentally introduced a solder bridge or anything.

 

(This is fresh in my mind because I did exactly this this weekend on a rather unwell IIsi, and it did exactly this)

 

I would be incredibly surprised if it were due to the voltage rating of the cap, I can't think of any reasonable way that would be the problem.

Edited by cheesestraws
Link to post
Share on other sites

It's definitely not related to the capacitor voltage rating. Higher is always safer. Not necessarily always better, as the other characteristics change with the voltage rating like value derating(That 22uf is only 22uF at a certain range of temperature and voltages), and Equivalent Series Resistance. The latter is why I generally don't replace aluminum "cans" with tantalum caps. Tantalums have higher ESR and also when they blow it's pretty for a second, then not so pretty afterwards. Still people use them and they probably work fine under most if not all situations. I just like to use the form factor equivalent Poly-Aluminum caps. They look the same as the aluminums but have a "dry" electrolytic. They are... Well... Expensive. But, nothing's too good for my vintage babies! :D

 

Now, if the Q630 doesn't boot in the 5300 chassis, then you likely have two problems! The Q630 and all of the "slow bus" "Power-formas" have roughly the same voltage rail/power requirements. In fact the 52xx/53xx/62xx/and 63xx except for the 6360, were all based on a motherboard design that literally "shoehorned" in the 603 chip where the 68LC040 was. Because of that and a crazy bus split in the architecture, they are horribly slow! The PCI based 5000 and 6000 series are MUCH better. They were designed from the ground up to use the 603e and have the full bus width utilized uniformly. Its soty of like comparing the original 16Mhz Color Classic with the 33Mhz CC2/P550. It "looks" like it's only twice as fast, but because of the bus width, it's about 4 times as fast. And the later 475/575 logic board despite having the same 33Mhz/32-bit bus as the CC2 is I think is quite a bit faster than the CC2. The 5400 series vs the 5200/5300 is more like the comparison between the LC575 and the Original Color classic.

 

So what am I saying? Don't waste any precious time on that 5300 logic board! Find a PCI based slide in board from a 54/55/64/6500 machine! But first get the chassis working with your Q630 board. I suspect when you crack open the 5300 chassis, you'll find evidence that is was stored in an unworthy environment(Garage, Damp Basement, Hot Attic). The bare metal bits inside aren't designed to be super rust resistant, so if you see lots of rust spots and pitting instead of bright, yet, dull gray metal, theres a good chance that storage killed the whole machine...

Link to post
Share on other sites
14 hours ago, macdoogie said:

It's definitely not related to the capacitor voltage rating. Higher is always safer. Not necessarily always better, as the other characteristics change with the voltage rating like value derating(That 22uf is only 22uF at a certain range of temperature and voltages), and Equivalent Series Resistance. The latter is why I generally don't replace aluminum "cans" with tantalum caps. Tantalums have higher ESR and also when they blow it's pretty for a second, then not so pretty afterwards. Still people use them and they probably work fine under most if not all situations. I just like to use the form factor equivalent Poly-Aluminum caps. They look the same as the aluminums but have a "dry" electrolytic. They are... Well... Expensive. But, nothing's too good for my vintage babies! :D

 

Now, if the Q630 doesn't boot in the 5300 chassis, then you likely have two problems! The Q630 and all of the "slow bus" "Power-formas" have roughly the same voltage rail/power requirements. In fact the 52xx/53xx/62xx/and 63xx except for the 6360, were all based on a motherboard design that literally "shoehorned" in the 603 chip where the 68LC040 was. Because of that and a crazy bus split in the architecture, they are horribly slow! The PCI based 5000 and 6000 series are MUCH better. They were designed from the ground up to use the 603e and have the full bus width utilized uniformly. Its soty of like comparing the original 16Mhz Color Classic with the 33Mhz CC2/P550. It "looks" like it's only twice as fast, but because of the bus width, it's about 4 times as fast. And the later 475/575 logic board despite having the same 33Mhz/32-bit bus as the CC2 is I think is quite a bit faster than the CC2. The 5400 series vs the 5200/5300 is more like the comparison between the LC575 and the Original Color classic.

 

So what am I saying? Don't waste any precious time on that 5300 logic board! Find a PCI based slide in board from a 54/55/64/6500 machine! But first get the chassis working with your Q630 board. I suspect when you crack open the 5300 chassis, you'll find evidence that is was stored in an unworthy environment(Garage, Damp Basement, Hot Attic). The bare metal bits inside aren't designed to be super rust resistant, so if you see lots of rust spots and pitting instead of bright, yet, dull gray metal, theres a good chance that storage killed the whole machine...

I’ve set it aside for now but I’m thinking perhaps the 22uf cap might not be soldered on properly as it’s the cap right next to the CPU. 

 

As for your comments about tantalum caps, most Macs used both electrolytics and tantalums. Tantalum caps are totally safe provided you use good brand ones, don’t overload them, and solder them on properly (not backwards). Nobody is running a vintage Mac 24/7 anymore so the life of a tantalum cap should be fine. 
 

The main difference between tantalum caps and others is they have a frequency and a resistance. I believe the 22uf that I have are 900mohm and 900hz. I can’t remember. But I do know these figures also play into how replacement caps work. 
 

I recapped a Quadra 610 with higher voltage tantalum caps. I believe it used 47uf 16v caps (all 10) and I put on 20v caps. Those caps also had 3ohm resistance. The Q610 booted but everything was slow. Moving the mouse cursor had interruptions and it took forever for the flashing question disk to appear. When it did it only flashed like once every 5 seconds instead of about 1-2 seconds. SCSI didn’t work. I removed all of those caps and put on 47uf 16v 900mohm or 1.2ohm (can’t remember) and it booted up normal right aways. I had another Mac where I put one of these higher rated caps on the SCSI and SCSI didn’t work until I put a lower spec one on. 
 

So cap choice can definitely affect the machine after a recap. These caps I have are for Apple IIe PDS cards and IIci cache cards. Both of those interchangeably used 22uf 25v and 35v so I went with 25v. 
 

In regards to your statement about ditching this board because it will be too slow:  I bought this model on purpose. I want to test and compare for myself the speed differences in emulation and native code on a variety of OS versions. This board is the key to doing those tests. I want to determine for myself how bad a 603 chip was in this form factor. I already have 80 vintage Macs I’m not limited to any one machine. I have it all from Plus/SE/Classic through to G5 dual 2.0, and everything in between. Once I can get this working I can pull test data together for all of the models I have. 
 

Lastly, a Lc475 is many times faster than a CC2 because the 475 uses a 68LC040 and the CC2 uses 68030. The CC2 has the wider paths and its processor is also twice as fast. Yes that equates to about 4x as fast. Crazy. But the 040 chips were yet at least 2-4x faster than 030 chips, depending on the model and platform used. My LC475 beats my IIfx in most tests. 40mhz 030 vs 25mhz LC040. I’ve even replaced the LC040 with a regular 040 in one and clocked it to 33mhz. The LC475 platform was quite good, my testing puts an overclocked 475 on par with a Q650/Q800 for all of the tests. So anyone who didn’t need an Internal CD, or NUBUS slots had a good option in 1993/4 with the 475/Q605. 
 

Back to caps, I’m thinking of removing and replacing that 22uf to see if it solves the problem. Alternatively I’ll try and get a different 22uf cap with different specs to try. 
 

Whats strange is it all worked fine before I recapped the 5300 board. Even the monitor/enclosure worked fine with the 5300 board. I hadn’t tried the 630 board in it until I had the failure. The 630 board does nothing in the 5300 case. Not even a click. The 5300 board acts the same in both the 5300 enclosure and the Q630 enclosure. It clicks and immediately shuts off. 
 

As for replacing the caps on a board with caps that look the same, to each their own. I like the look of tantalum caps on the board. They just look like other chips. If you’re trying to be original, you’d use the aluminum electrolytic capacitors instead of modern replacements all together. 
 

I bought a Mac that was sold as recapped on eBay. When I got it I couldn’t be sure it actually was. I mean it seemed like it had blobs of solder on the cap contacts but they looked original. I removed them all and put tantalum caps on. At least I know for sure it’s been recapped and anyone else looking at my Macs will know that too. 
 

And tantalum caps don’t always just blow up when they fail. I have a IIfx that was being slow and crashed a lot. I recapped the board. Worked perfectly after. 
 

From all of the research I’ve done, I’m very happy with replacing my caps with  tantalum caps. It took a lot of effort to find the right hz frequency and ohm rating. That may be what I’m facing with this one (the Mac might want a particular frequency or ohms near the CPU). 

Link to post
Share on other sites

It's true that Apple designs have used both Tantalum and Electrolytic. Although, in more recent years, they switched from Tantalums to "PosCaps" i'm guessing largely due to Tantalum being a conflict, or "blood" mineral. Still, on older boards, you'll see a mix. One obvious reason is cost. Tantalums are certainly more expensive than the old Aluminum electrolytic. However, the function of the capacitor it also a big factor. The basic uses for capacitors in a circuit are Bypass, Filtering, Decoupling, and Bulk. Some of these characteristics overlap of course and all are frequency dependent with any type of capacitor as the fundamental reactance is frequency based. The larger value electrolytic caps 22uF-100uF used on these old mac boards largely play the role of bulk capacitance. Bulk capacitance is a sort of local charge storage designed to keep the voltage rail clean and steady when large current transients occur. Because of that, caps with higher ESR are poorly suited for use as bulk caps as the current "outrushes" will cause a temporary voltage drop per Ohms Law. In a digital circuit, when the voltage drops, the noise margin(ability to tell a '1' from a '0') narrows and high performance busses in particular are more likely to suffer data corruption(ie stop working pr cause a hang or crash).

 

To start with Tantalum caps have a larger ESR than Aluminum caps, usually in the 100's of milli-ohms to a several ohms as you documented. And, I believe more-so than in their aluminum counterparts, have a larger increase in ESR with higher voltage tolerance parts. So with a Tantalum, increasing the voltage tolerance can affect the circuit operation more than the same voltage tolerance increase on an aluminum cap. Aluminum caps in general have lower ESR by about a factor of 10, so typical ESR values for those caps would be in the tens of milli-ohms to low 100's. So when I say that the PolyAL caps "look" more like the old AL "wet" cans, it's not just about appearance, its also about how the cap "looks" electrically to the circuit, especially during transients. This is even more critical with mosern equipment which runs at lower core voltages and slimmer noise margins as well as higher frequencies and larger and faster swings in current draw. Think of modern GHz class processors that have sub 1.0V core voltage and fluctuate between a few amps when "sleeping" (not machine sleep, CPU sleep which are common very short "naps" between bursts of high intensity work) and nearly 100Amps when running full bore in "turbo" mode. The reason we can get away with tantalums for bulk caps on our older machines is that they aren't that fast, and have much higher 5V and later 3.3V core logic. Still if you start adding extra stuff to the machine like daisy chained SCSI hard drives, CPU accelerators or even some of our overclocking tricks, it's possible to tax the system in ways that are less perceptible. Its possible(cant say for sure without anaylzing a system under test), that "slow" running issues you see could be bus errors being retried(probably not as likely) or Clock and PLLs actually running slow or having difficulty maintaining lock due to voltages being out of spec.

 

Yeah The 68040(475/575 class machines) is MUCH faster then the equivalent "external" frequency 68030. Two primary reasons: 1.) The 33Mhz 68040 runs internally at 66Mhz and not completely unrelated or separate 2.) Caches! Even my 216Mhz ARM Cortex-M7 class MCUs seem to crawl when I forget to enable the caches!

 

Full disclosure: I'm a BSEE with 25 years experience including ASIC and Logic Board level System design. Worked 12 years for Apple during what I consider to be their best years(from just after coming back from the brink of death until just after Steve Job's passing).

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...