• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

Don't use 16V SOLID Tantalum on SE/30 Motherboards

JDW

Well-known member
First, I'd like to thank @ttb, @aeberbach and others who contributed to my separate thread on SE/30 SONY Chip Voltages.  The conclusion of that thread is that while most of the 47uF capacitors on the motherboard see only 5v, there are 4 which see about 12V.  That is actually a problem for those 4 capacitors.  I know that many people, myself included, recap very often with Solid Tantalum capacitors, myself having used 16V rated parts in the past.  But there are important reasons why we need to stop using those 16V rated "solid" tantalums on the SE/30 motherboard, despite the fact that most of you and even myself haven't experienced any burn-downs yet.

Below is a video I released a few days ago.  While it is a general advice video about how to pick capacitors for recapping just about anything, I used the SE/30 motherboard as a practical example in the final half of the video, explaining why you should take extreme care with your tantalum capacitor selection.




In summary, it's cheaper to make all 10pcs of 47uF the same type of capacitor since you get a price break for quantities of 10 or more, and that choice should be something other than 16V Solid Tantalum.  Even though 6 out of 10 can be 16V Solid Tantalum, for the sake of using the manufacturer specified voltage derating, it's best to either use 25V Solid Tantalum (yes, they do come in 7.3mm lengths) or 16V (or higher voltage) Polymer Tantalum.  You can even use OS-CON caps which don't have a voltage derating, although they do have a higher leakage current than even Polymer Tantalum.  (Leakage of Solid Tantalum is about the same as fluid filled Aluminum Electrolytic capacitors, which is extremely low at around 7 or 8µA.  But since the SE/30 is not battery powered, higher leakage really won't matter.  The OS-CON caps also have lower ESR than even the Polymer Tantalums.)

Be sure to watch the video above on YouTube so you can click SHOW MORE to expand the text description I wrote under the video.  There you will find Mouser links to individual capacitors which are fully compatible and safe to use on the SE/30 motherboard, along with my notes about usage.

I fully realize that many of you have recapped with 16V Solid Tantalum for years, and I myself have two boards like that. Indeed, I've used those 16V Solid Tantalum caps for years without issue.  But the fact remains that any Solid Tantalum capacitor used at voltages higher than the manufacturer specified derating faces increased failure rates.  As shown in Table 4 on page 7 of this AVX Technical Document, the failure rate of 47uF 16V Solid Tantalum capacitors with a recommended 50% voltage derating stands at about 0.03% versus 1.1% for no voltage derating.  And when solid tantalum fails, it is not only a dead short but the cap itself goes up in flames.  Polymer Tantalum in contrast will not burst into flames and has a 20% voltage derating.  That means you can only use a 16V Solid Tantalum at a max of 8V to be within the derating guidelines, but you can use up to 12.8V on a 16V rated Polymer Tantalum.  Choose a 25V rated Polymer Tantalum, and it can handle up to 20V continually and still be within manufacturer recommended derating guidelines.

Yes, I will admit there are some manufacturers that recommend even more stringent derating guidelines of 1/3 the max rated voltage for solid tantalum.  But that's all the more reason to use Polymer Tantalum instead.

All said, what I plan to do for my own SE/30 boards is swap out C3, C4, C9 & C10 for caps with the proper voltage derating, leaving the other 16V Solid Tantalums as they are since they only see 5V.

Lastly, when you solder tantalum caps and especially the polymer types, be sure to keep the soldering iron heat to 350°C or less and never keep the iron on each side of the capacitor for more than 3 seconds.  This information is also specified in the data sheets as heating with a soldering iron can affect failure rate as well.

I hope this information helps my fellow recappers get the longest life possible from their chosen replacement capacitors.

 

just.in.time

Well-known member
Looks like I may have to double check what’s used on my boards. Also, didn’t know about those soldering guidelines (temp and time limits).

 

JDW

Well-known member
Looks like I may have to double check what’s used on my boards. Also, didn’t know about those soldering guidelines (temp and time limits).


SMD components are more sensitive to temperature and applied heat duration than thru-hole components due to the physical size of the components (larger components handle heat better) and the fact that thru-hole types keep the heat further from the main body of the component than SMD types.  

Precise guidelines for soldering SMD tantalum vary by manufacturer, but on average most say to keep the soldering time for each side of the capacitor to 3 seconds or less, and some suggest letting the capacitor cool before soldering the other side.

Some manufacturers recommend pre-heating the capacitor and board to lessen the likelihood of thermal shock, but that is really only practical at the time of manufacturing, in preparation of using a reflow oven.  Some manufacturers even recommend a soldering iron temperature of around 270°C, which is rather low and can be harder to flow the solder.  (The hotter the iron, the easier it is to do the actual soldering but in theory the more potentially harmful to the component being soldered.)

I don't think we need to get too worried about the affects of soldering time though because of a NASA report about soldering times for SMD tantalum.  As shown in the data presented there, soldering has an impact on capacitance, dissipation factor, ESR and leakage.  But even though the greatest change, post soldering, was DCL leakage, even that change was deemed not significant, and NASA concluded that hand soldering is actually less harmful than what is commonly believed.  That doesn't mean you should put a 450°C iron on your tantalum capacitors for 10 seconds.  It just means that if you can reduce heat and duration of soldering, your hand-soldering will have less of an impact on how the capacitor changes once your are done.

Lastly, the NASA document made no mentioned of this, but I have read documents from various SMD tantalum capacitor manufacturers which say specifically not to solder in a cap, then desolder and reuse it.  This goes back to the manufacturer concern that excess heat and time will continue to negatively impact the capacitor and reduce it's reliability over time. But since I could not find a real world test about that, it's probably more speculation, erring on the side of safety, than anything else.

All said, use less heat with less soldering time where possible, and try not to soldering in a freezing cold room to lessen thermal shock to the components, for best results.

 

ScutBoy

Well-known member
Nice video!

Interesting - and frustrating - to see the bits about the derating of the caps. What's the point of putting a voltage rating on them, when in the fine print you say it may only be rated up to a fraction of that?

Like buying a gallon of milk, but really only getting 3/4 gallon for "reasons"....

Would it make sense to just get, say, 25V caps when the application calls for 16V (and you are really only working for 12V)?

 

aeberbach

Well-known member
As long as your physical size is OK, and you don’t substantially worsen parameters like ESR, using capacitors that are well overrated for voltage is not a bad thing at all. Sometimes you might even find they are cheaper if they are a more common part than the lower just-good-enough one.

 

JDW

Well-known member
Nice video!

Interesting - and frustrating - to see the bits about the derating of the caps. What's the point of putting a voltage rating on them, when in the fine print you say it may only be rated up to a fraction of that?

Like buying a gallon of milk, but really only getting 3/4 gallon for "reasons"....

Would it make sense to just get, say, 25V caps when the application calls for 16V (and you are really only working for 12V)?


Well, I am not about to defend component manufacturers for doing what they do because I too would prefer more clarity for the layman when it comes to specifications, but the manufacturers of these components are targeting engineers who are "supposed to know these things" rather than the layman.  It's frustrating, I know, and that really was incentive for me to make a video on the subject.

If a manufacturer says a capacitor can handle 16V, it can handle 16V.  But the issue is the failure rate is negatively impacted when that maximum voltage is constantly applied to the capacitor.  The layman would argue, "Fine... just call it an 8V capacitor then!"  And I agree that would be clearer, but they just don't do it that way.  That's because applied voltage is all about risk and percentages.  The lower the voltage, the lower the risk of failure.  Because of that, putting a voltage rating on a capacitor based on failure rate would actually be arbitrary.  Because at one exact percentage of failure would one deem acceptable? 

Complicating things further is the fact that you will even find some documents arguing the use of 1/3 the rated voltage to get what they deem "an acceptably low failure rate."  But that would mean applying no more than 5.3V on a 16V tantalum capacitor, which is not practical in every application.  The good news though is that 6 out of 10 47uF caps on the SE/30 motherboard are in fact derated by 1/3rd insofar as they only see 5V.  It's just the remaining 4 caps I mentioned which see 12V that are technically outside the manufacturer recommended specifications.  This wouldn't matter so much if solid tantalum didn't ignite when it dies.  But if you use Polymer Tantalum then you will avoid a fireball altogether because Polymer Tantalum won't ignite like Solid Tantalum.

Even so, Polymer Tantalum too has a voltage derating.  Like I said, it's 20%.  So a 16V part can safely be used with a constant 12.8V on it.

Folks who dislike derating and/or want a stock "can" look need only use OS-CON which has no voltage derating at all and lower ESR too.  The only downside to OS-CON is that leakage current is about twice that of Polymer Tantalum, which is much higher than Solid Tantalum or fluid-filled Aluminum electrolytic; but again, on an SE/30 motherboard, that leakage current won't matter much and will just be a few milliamps of loss through your wall socket.

 

JDW

Well-known member
As long as your physical size is OK, and you don’t substantially worsen parameters like ESR, using capacitors that are well overrated for voltage is not a bad thing at all. Sometimes you might even find they are cheaper if they are a more common part than the lower just-good-enough one.


Choosing a capacitor of any type that has a higher voltage rating than your stock capacitor is actually a benefit in terms of ESR.  The higher voltage rating usually ensures a lower ESR.  However, if you choose a high Hour-rating, then your ESR usually is a bit higher (which is less desirable).  For example, when comparing fluid-filled aluminum electrolytic caps, a cap rated for 10,000 hours often will have higher ESR than a cap rated for 2,000 hours. I tend to choose the high-hour cap anyway because the higher ESR still isn't too high AND you get the benefit of longer life.  Even so, most manufacturer data sheets will show you that even if you choose the highest hour-life possible for a fluid-filled aluminum electrolytic, they tend to put a ceiling on the Life at 15 years.  We know Al. Electrolytics can last longer than that, but usage beyond that for caps with fluid electrolytes are beyond manufacturer recommendations.  And that is one big reason to opt for caps with a SOLID electrolyte, whether that be Solid Tantalum, Polymer Tantalum or Organic SOLID polymer OS-CON caps.  Just take care with Polymer HYBRID caps though because even though the boast extremely low leakage current (DCL), they do have a fluid electrolyte and therefore are subject to the same life limitations as regular fluid-filled aluminum electrolytics.  Go with solid electrolytes where possible so the cap won't spill its guys all over the board 20 years hence.  You can't do that on an Analog Board or power supply, but for most all motherboards you can.

 

MrFahrenheit

Well-known member
Interesting discussion.

I got into recapping in May of this year (2020).  I didn't know what I was doing, and I certainly couldn't find definitive information about exactly which tantalum caps to buy.  Many said to use the next step up in voltage, so with the 47uf it would be 20 or 25v (can't remember which).  To me, that didn't make sense to do.  Why not match the specs?  So I bought several types of caps and voltages. 

From what I read (and I'm not an electronics expert by any means), Tantalum caps have a resistance (ohms) rating, and a 'wobble frequency) which electrolytic don't have.  I had a working Quadra 610 that I wanted to recap.  I removed the old caps, and this Mac had all 47uf 16v caps.  So I decided to try out my 20/25v caps.  When I switched it on, I noticed the startup chime was a bit garbled.  I had no mouse/keyboard input, and SCSI did not work.  The flashing question mark was running about 1/5 the normal speed, like in slow motion.  I powered it down, and removed those caps, and replaced with 47uf 16v tantalums, with a different resistance rating.  Boot up as expected.

I know this is a bit of a different discussion than what you're having, but in the end I wanted to relay my story, but also ask your opinion of:

- what happened to me

- what ohms rating you recommend when increasing the voltage

- is there a ceiling that too much extra voltage is bad

 

aeberbach

Well-known member
Having a voltage rating that is too high is never bad, it just has to be enough. It’s like using 14 gauge wire for a LED that will see 3V at 10mA. While the measurable resistance of that wire might be very slightly better than a 30 gauge length of wire you won’t really see a change. When I started doing electronics it was not uncommon to salvage huge “computer” electrolytics from power supplies sometimes rated at 450V - I could use those in 9V applications with no problems, in the days before I could afford a stock of more reasonably sized parts.

I don’t know what the wobble frequency could refer to (ripple current?). I also can’t really guess what happened to your Quadra except to suggest that maybe something was installed incorrectly and it was corrected when you changed all the caps the second time? If looking at a machine doing that I’d start by looking at things connected near a crystal, sounds timing-related. 

 

JDW

Well-known member
...Tantalum caps have a resistance (ohms) rating, and a 'wobble frequency) which electrolytic don't have...I decided to try out my 20/25v caps.  When I switched it on, I noticed the startup chime was a bit garbled... removed those caps, and replaced with 47uf 16v tantalums, with a different resistance rating.  Boot up as expected.

- what happened to me

- what ohms rating you recommend when increasing the voltage

- is there a ceiling that too much extra voltage is bad


All capacitors have resistivity called ESR (Equivalent Series Resistance).  That is not exclusive to Tantalums.  In fact, Tantalum capacitors have a STABLE ESR across their entire working temperature range.  Fluid-filled Aluminum Electrolytic capacitors like you find on the stock SE/30 motherboard, have increase ESR the lower the temperature, especially when you reach 0°C and below (which is a bad thing).

The "wobble" you mention is likely referring to "Ripple Current."  All capacitors deal with Ripple to some extent, but the Ripple Current rating really only applies to capacitors used in switching power supplies.  Ripple is merely the repeated act of the capacitor charging and discharging.  If you drain the capacitor to zero and then quickly charge it repeatedly, it will warm up fast in accordance with its ESR.  More specifically, the higher the ESR, the hotter the capacitor will get when hit with high ripple currents.  This is why you don't often find aluminum electrolytic capacitors on the output stage of switching power supplies, even though they may be used elsewhere in the circuit.  Typically ceramics are used instead because their ESR can be as low as 0.005Ω.

As to what happened to you, well it doesn't make sense that 20/25V caps would have garbled your sound.  That makes zero sense.  More likely it was a soldering flaw you didn't know about but later rectified when you soldered in different caps.  You also need to be aware that if you did not clean off your board well after removing the stock capacitors, leaked fluid could have caused some problems in the audio circuit, and that would remain true even with your newest caps in there.  The board should be cleaned well with 99% ISO alcohol and given plenty of time to dry in direct sunlight before you power it on.

There is no "ohms rating" you need to look for when "increasing the voltage."  The stock capacitors will have higher ESR than Solid Tantalum, much higher than Polymer Tantalum, and much, much higher than Polymer Electrolytics like OS-CON.  But even if you replace with fluid-filled aluminum electrolytic capacitors, ESR is nothing to worry about.  Those caps mainly hold up the voltage and prevent voltage dips.  There is no switching power supply on the motherboard, so ripple current and ESR don't matter much at all.

When you ask about a "voltage ceiling" you are really asking if it's possible to select a capacitor with such a high voltage rating that it would be bad in the circuit.  Yes.  Choosing too high a voltage rating will make the capacitor so big it won't fit.  Physical size is tied to the voltage rating.  The higher the voltage rating the bigger the cap and the lower the ESR and potentially the longer the life.  But in terms of the SE/30 motherboard, you cannot fit an SMD capacitor replacement longer than 7.3mm onto the pads of the 47µF caps, which means 25V is the highest voltage 47µF Solid Tantalum capacitor that will fit.  So going with a high voltage cap is fine.  Just make sure it physically fits.

 

MrFahrenheit

Well-known member
Thank you so much for such detailed replies. I have been looking for definitive answers to my questions. 
 

The issues I had with the Q610 related to sound, ADB, SCSI, and clock speed. The board was well cleaned with IPA99 with no signs of prior electrolyte leakage. I clean the entire board in IPA99 by hand. The difference between the 16v and 20/25v caps mostly was higher ohms rating on the higher voltage ones and lower on the 16v. 
 

So are you saying that the stock electrolytic caps have a much higher ohms rating than even what most tantalum caps have ?

How does the farad rating work in this mix?  Is that also a size issue?  Is it safe to use a different uf rated cap than original?  If not, why would Apple vary the uf rating of caps for the same machine?  I have LC Macs that at C9 either gave a 2200uf or a 47uf cap. And replacing with a 47uf works just fine. 
 

From what you have found, is it recommended to use higher voltage caps on other Macs as well, or is it more important on the SE/30?  Should all caps be of higher voltage or just the 47uf?

I typically solder at 400’c, should I be lowering that?  I found solder flowed a lot easier at 400. I use a good quality 60/40 tin lead solder with flux core, and a good quality gel flux. 
 

My process is to apply flux and heat each soldered point on all old caps for a few seconds each. Then I carefully push and twist off the old cap. I’ve done nearly 500 caps this way between damaging pads so for me this method has caused less damage. 
 

Next I apply more fresh flux and fresh solder and I remove the old solder and cap leg and clean off the pad. 
 

I then clean the entire cap areas with IPA99 and qtips. I use a toothbrush and IPA99 where necessary. I also spot clean the whole board  with either the toothbrush or qtips. The cleaning process by hand takes about an hour per board. 
 

Then I solder down the new caps and clean again with IPA99. 
 

I’m pretty methodical about the whole process. Maybe I should take those caps and try them again in another Mac to see if I can repeat the issues I had. 
 

I have yet to start recapping my power supplies. I’m not happy with how I desolder through hole caps and I need better tools for that. I don’t have a desoldering station.  

 
Last edited by a moderator:

JustG

Well-known member
JDW,

Thank you for another informative and thorough video.  I came across your channel when I first started getting back into old Macs and was pleasantly surprised to find out that you're a member on this forum too!  I greatly appreciate all the hard work and time you put into your productions.  There are several channels I watch when it comes to vintage computing, yours is the only one that my wife will sit down and actively watch with me.  She puts up with my hobby but has very little interest otherwise, you should take her interest as a huge compliment!

Hopefully this isn't sidetracking the subject matter (I have a SE/30 board sitting on my bench that I need to order capacitors and the timing couldn't have been more perfect).

Please keep up the great work.

 

JDW

Well-known member
The difference between the 16v and 20/25v caps mostly was higher ohms rating on the higher voltage ones and lower on the 16v. 
 

So are you saying that the stock electrolytic caps have a much higher ohms rating than even what most tantalum caps have ?

How does the farad rating work in this mix?  Is that also a size issue?  Is it safe to use a different uf rated cap than original?  If not, why would Apple vary the uf rating of caps for the same machine?  I have LC Macs that at C9 either gave a 2200uf or a 47uf cap. And replacing with a 47uf works just fine. 
 

From what you have found, is it recommended to use higher voltage caps on other Macs as well, or is it more important on the SE/30?  Should all caps be of higher voltage or just the 47uf?

I typically solder at 400’c, should I be lowering that?  I found solder flowed a lot easier at 400. I use a good quality 60/40 tin lead solder with flux core, and a good quality gel flux...  


Normally, the higher the voltage rating the lower the ESR unless the Hour/Life rating is very high, in which case the ESR could be the same or higher than lower voltage rated capacitors.

Solid Tantalum should have a slightly lower ESR than the stock capacitors but I wouldn't say it is substantially more.  I am of course talking about when the stock capacitors were new. But Polymer Tantalum is much lower than Solid Tantalum.  OS-CON caps are lower still.  Ceramic caps are even lower -- as low as you can go.  But you would not want to use ceramic caps in most recapping projects because their voltage derating requirements are more severe than Solid Tantalum.

Lead solder is perfect, but soldering at 400°C is not.  I strongly recommend using an iron at 350°C or lower.  I use 350°C most of the time.  Bumping up to 400°C very rarely on ground planes is fine, but that's really too hot to be soldering in SMD capacitors, as per component manufacturer's guidelines.

If a stock cap was 2200µF, you would not make a wise choice if replacing it with a 47µF capacitor, even if in your testing that 47µF cap seems to work.  The voltage rating of a replacement capacitor is flexible, but unless you are an engineer with access to the original schematics, you cannot so easily say that a substantial chance in Capacitance would be a good idea for the circuit as a whole.  The best rule of thumb is to never change the Capacitance when recapping, especially when in doubt about how much capacitance is actually necessary.

Using higher voltage replacement caps is nothing exclusive to the SE/30.  Using higher voltage SOLID Tantalum capacitor is critically important, for "derating" reasons my video explains in detail.

The English word "few" typically means 3, and therefore if you are heating each side of a cap for 3 seconds or less, there's nothing to be concerned about.  Will a cap die if you heat each side for 5-6 seconds?  No.  But as per manufacturer recommendations, excessive heating, especially of Tantalum caps (including the Polymer types) could affect Failure Rate in the same manner is ignoring the voltage derating guidelines would.  Running a 16V solid tantalum cap at 16V doesn't guarantee failure.  It merely increases the probability of failure over time.  And when SOLID tantalum fails, it burns.

As to your other comments, I don't see any issues with your process of soldering and cleaning.  As to the need for a desoldering station, I honestly use mine very little.  I prefer a copper desoldering braid.  If you want to see how desoldering stations can fail where another technique works, check out my video on recapping a 512K motherboard.

Thank you for another informative and thorough video...


Thank you for your very kind words!  I have a video on recapping the SE & SE/30 Analog Board and the SONY power supply, as well as floppy drives.  I don't have an SE/30 motherboard recapping walkthrough video, but I did put links to recommended caps in the text description under my latest video, which is linked in my opening post.  That should point you in the right direction.

 

techknight

Well-known member
Like buying a gallon of milk, but really only getting 3/4 gallon for "reasons"....


You mean like a bag of chips that is half air? yea... 

you never are supposed to run caps up close to their rated voltage anyways. like using 6.3V caps in 5V circuits. I see it alot, and I also see the ramifications alot. 

 

ttb

Well-known member
What a thorough video, amazing.

I can't be sure, but I believe that the Console 5 kits you mention at 48:00 likely include both the solid tantalum caps and Nichicon VX axial caps, hence their description. Someone who's actually bought the kit can chime in, though.

 

ScutBoy

Well-known member
What a thorough video, amazing.

I can't be sure, but I believe that the Console 5 kits you mention at 48:00 likely include both the solid tantalum caps and Nichicon VX axial caps, hence their description. Someone who's actually bought the kit can chime in, though.


This is true - I've bought that kit before, and it includes the axials along with the tantalum caps.

 

JDW

Well-known member
What a thorough video, amazing.


Thank you for your kind words, @ttb.

I can't be sure, but I believe that the Console 5 kits you mention at 48:00 likely include both the solid tantalum caps and Nichicon VX axial caps, hence their description. Someone who's actually bought the kit can chime in, though.


This is true - I've bought that kit before, and it includes the axials along with the tantalum caps.


Thank you @ttb & @ScutBoy for correctly pointing out that the 2 Axial caps are the Nichicon VX caps in that Tantalum kit.  I have just added a new temporary Comment under my video to announce the formal correction along with my apology for the foolish oversight.  I have also submitted an edit of that video, cutting out several seconds from that section to exclude my incorrect remarks, and that will take a day or so for YouTube to render the video (at the same video URL). I will later delete my Comment, since the video will then be edited so no correction in the Comments will be necessary.  

Since no part numbers for the Tantalum caps are given in that Console5 product description, my recommendation that people ask Console5 for that info prior to buying the kit remains.  It is important to know the voltage rating of Tantalum caps before you decide to use them.  If they are 16V, for reasons specified in my video, I would not recommend their use. @ScutBoy, since you said you have purchased that kit before, if you know the markings on the 47uF caps, then we can know the voltage rating.  If you are able to provide that info, I will further update my Pinned Comment to add that.

Thanks again for the correction!

...you never are supposed to run caps up close to their rated voltage anyways. like using 6.3V caps in 5V circuits. I see it alot, and I also see the ramifications alot. 


That reminds me of the DiiMO PDS accelerator card that I recapped.  The stock caps on the 5V rail were in fact 6.3V rated aluminum electrolytics.  I did a recapping video on that where I replaced the caps with 10V Niobium Oxide caps which have a 20% voltage derating -- meaning, you can apply a constant 8V on the new caps without worry.  Niobium Oxide are better than Solid Tantalum in that Niobium Oxide doesn't fail as a short and won't burn.  The big downside to them is that they only come in low voltage ratings of 10V and lower.

 
Last edited by a moderator:
Top