Capacitor type substitutions - IBM 5153 CGA Monitor, and other vintage gear

[badgerthing]

Hello!

I am new here, but have always liked this forum and found it useful, so thought I should join and contribute.

This week I'll have three IBM 5150's (two Rev B, one Rev A), and an IBM 5153 CGA monitor. The 5153 appears fine from a no-signal power-on test. The 5150's don't work.

I know I'll be able to get them all working. I've restored dozens of vintage machines over the past few years with complete success. My lab of test gear grows ever larger...
The latest achievement is fully restoring a Compaq Portable II, including making a new power supply module from a modern TFX-size ATX power supply board.

I plan on functionally and cosmetically restoring them all, keeping the 5150 Rev A and 5153 for my collection, and selling the two 5150 Rev B's.
The three 5150 power supplies will be converted from their 110V to 240V as per the guide by Paul Axford, which the link has just died for. The 5153 monitor is a 240V one already.

I have a few questions that I can never seem to get straight answers for (maybe there are no straight answers for them; I understand some may depend on the circumstances).

1. Is there any harm in replacing the dozens of 10uF tantalum capacitors with 10uF 50V leaded multilayer ceramic caps? It seems silly to replace tantalums with tantalums in vintage gear, unless you enjoy electronics popcorn. So far I've used 10uF MLCCs with great success.

2. This relates mainly to the IBM 5153 CGA monitor I'll be re-capping. Of course the infamous RIFA PME271M Y2 film caps will be replaced, along with all the electrolytics. I have a list of caps from someone else, so have begun sourcing them. But the list doesn't have dimensions or pitch, so I won't be ordering until I amend the list with those details (and probably publish in this thread). The list noted two caps as low ESR from his measurements. Of course, more caps may have been low ESR, but are now high ESR from age, so the list may not be accurate.
My question is, can you usually replace most general purpose electrolytics with low impedance types? I imagine most of the electrolytics are general purpose in the 5153 monitor. And most caps I can source that are rated for long life (is this what I want for vintage gear that will sit unused often?), are low impedance types, such as NCC KY and Panasonic F*.

I know I can match most of the caps with general purpose or low impedance if I go to Digikey. It'd be a good $120+ for all the caps for the three 5150 PSUs and the 5153.

I will be making a JLC+LCSC order soon. They have a coupon available right now which will get me all the Lelon caps I could need , for $0, since I'll be ordering a bunch of other stuff.
I was thinking of putting Lelons in the 5150 and 5153 that will be going in my own collection. As an experiment; what could go wrong? If they fail early after spending most of their years on the shelf, I can report back.

From my hours of googling, Lelon seems to be in the range of Teapo, and regarded to be "okay". Of course, I've read the thread about these caps on this forum, and how hard it is to qualify caps, and how much anecdotal evidence and "opinions" float about, so I'm not too sure what I'll do.

Most of the values of caps I need, are only available from LCSC of the low-impedance type. usually 4000-8000hr 105°C rated.

Think I could fill the 5153 with new low-impedance caps? Or should I stick to finding general purpose caps except for the two mentioned in this list? (https://www.dropbox.com/s/olxmp5huf8bbe37/5153.pdf?dl=0) (source: http://www.vcfed.org/forum/showthread.php?59266-List-of-Capacitors-in-IBM-5151-5153-and-8513-Monitors)

I'll go through the monitor when I'm taking down all the capacitor dimensions and lead spacings, and see if I can find datasheets of the models of cap. The monitor could have caps from late 70's to early 80's, as it was in production from 1981 onwards, and I'm unsure of the year of my particular one until I receive it.

Some of the caps are next to very hot power resistors, and often fail. Should I maybe find some 125°C rated caps for these places? Part of these are three 4.7uF 250V.
The others are in the power supply section of the monitor. This is a very inefficient switching power supply, producing a single 115VDC output. Would be nice to replace with a modern unit, but that voltage seems hard to find.
I have no qualms about replacing things like power supplies with superior modern designs, as long as it looks mostly the same externally.

3. When you can't find datasheets for an old electrolytic, so all you know is the marked capacitance and voltage, and the ESR measurement is likely unreliable, is it good practice to replace with a general purpose cap of simply similar physical size? Choosing low impedance if it's in a SMPS (or only certain parts of a SMPS)?

Sorry if I've rambled and asked some tough questions. Wondering if there's anyone watching this space with some experience or opinions.

Thanks!

[Bud]

Ceramic caps are microphonic.
Ceramic caps drop capacitance inder DC voltage.
In power supplies capacitor ESR may be part of the design.
If the power supply has converter or linear reg ICs they may require particular limit of the ESR to properly work.

[badgerthing]

Ceramic caps are microphonic.
Ceramic caps drop capacitance inder DC voltage.
In power supplies capacitor ESR may be part of the design.
If the power supply has converter or linear reg ICs they may require particular limit of the ESR to properly work.

Thanks.

I knew about the microphonic property, but hadn't encountered an issue before. Are the 10uF caps more microphonic than the 0.1uF ones used for decoupling?
Maybe I can screech at the motherboard to see if it makes it unstable 

The 10uF tantalums are all used on 5V and 12V rails, so the capacitance shouldn't drop too much on 50V MLCCs I imagine.

So for replacing these caps, the workings of the circuit must be understood to fully and properly select a substitute capacitor then. In that case all I can do is try and report my findings. Fill the 5153 with low ESR and see what happens?
I do have schematics but I only know the general principles of their workings.

Power supply schematic for 5150 is here: http://techmattmillman.s3.amazonaws.com/wp-content/uploads/2017/01/5150_PSU.png

[Bud]

Capacitance drop depends on physical size of a capacitor and dielectric type, not its working voltage. Never use Y5V dielectric type caps, which of course are among the cheapest ones.

[David Hess]

Is there any harm in replacing the dozens of 10uF tantalum capacitors with 10uF 50V leaded multilayer ceramic caps? It seems silly to replace tantalums with tantalums in vintage gear, unless you enjoy electronics popcorn. So far I've used 10uF MLCCs with great success.

There are two places where this can cause problems:

1. If the small bulk decoupling capacitor is actually the output capacitor on a regulator, the lower ESR may cause the regulator to oscillate because the frequency compensation depended on the phase lead produced by the ESR.

2. Why include small bulk decoupling capacitors at the load anyway?  They often serve to dampen oscillations in the high Q power distribution network.  Having a very low ESR defeats this and can cause ringing on the power distribution lines.

I would not change solid tantalum capacitors unless I specifically knew they were a reliability problem and maybe not even then.  Solid tantalum capacitors can also be replaced with good aluminum electrolytic capacitors with like 4 times the capacitance but of course they will dry out with time.

My question is, can you usually replace most general purpose electrolytics with low impedance types? I imagine most of the electrolytics are general purpose in the 5153 monitor. And most caps I can source that are rated for long life (is this what I want for vintage gear that will sit unused often?), are low impedance types, such as NCC KY and Panasonic F*.

Low impedance electrolytic capacitors (but not polymer low ESR parts) are not *that* much lower so using them to replace more general purpose electrolytics is almost always fine.

When you can't find datasheets for an old electrolytic, so all you know is the marked capacitance and voltage, and the ESR measurement is likely unreliable, is it good practice to replace with a general purpose cap of simply similar physical size? Choosing low impedance if it's in a SMPS (or only certain parts of a SMPS)?

The problem here involves ripple current rating.  Besides ESR, physically larger capacitors have a higher ripple current rating because they have a lower thermal resistance to ambient.  So a modern part which has the same ESR and temperature rating but is smaller has a lower ripple current rating.  The solution is to use a higher voltage part to get a physically larger package which will have a higher ripple current rating.

Note that ESR jumps up at a voltage rating of about 160 volts due to a difference in construction or materials.

[james_s]

Tantalum capacitors are fine, ceramic caps aren't "better" or more reliable, they have different characteristics. IMHO unless they are failing with unusual frequency just leave them alone. While the failure mode may be less than desirable, tantalum capacitors when appropriately rated are fairly reliable.

[David Hess]

Tantalum capacitors are fine, ceramic caps aren't "better" or more reliable, they have different characteristics.

Ceramic decoupling capacitors can also destructively short out in a way very similar to solid tantalum capacitors but for different reasons.

[badgerthing]

Thanks for all your input!

Tantalum capacitors are fine, ceramic caps aren't "better" or more reliable, they have different characteristics. IMHO unless they are failing with unusual frequency just leave them alone. While the failure mode may be less than desirable, tantalum capacitors when appropriately rated are fairly reliable.

Interesting. Are you sure? Vintage tantalums seem to give me far more grief.
I haven't found I needed to replace a ceramic decoupling cap yet. I've seen at least a dozen tantalum fireworks displays in vintage gear.
The device usually works again after clipping the scorched leads off.
When I restored the Compaq Portable II, there were no exploding tantalums. The motherboard often hung until I replaced all the 10uF decoupling tantalums. The CGA card and Multi-I/O card didn't work at all without replacing the 10uF tantalums. The tantalums I removed included many within spec, and many way out of spec or failed parts.
I just received the replacement handle today, so will be able to put it back together and call that project complete!

Is there any harm in replacing the dozens of 10uF tantalum capacitors with 10uF 50V leaded multilayer ceramic caps? It seems silly to replace tantalums with tantalums in vintage gear, unless you enjoy electronics popcorn. So far I've used 10uF MLCCs with great success.

There are two places where this can cause problems:

1. If the small bulk decoupling capacitor is actually the output capacitor on a regulator, the lower ESR may cause the regulator to oscillate because the frequency compensation depended on the phase lead produced by the ESR.

2. Why include small bulk decoupling capacitors at the load anyway?  They often serve to dampen oscillations in the high Q power distribution network.  Having a very low ESR defeats this and can cause ringing on the power distribution lines.

I would not change solid tantalum capacitors unless I specifically knew they were a reliability problem and maybe not even then.  Solid tantalum capacitors can also be replaced with good aluminum electrolytic capacitors with like 4 times the capacitance but of course they will dry out with time.

My question is, can you usually replace most general purpose electrolytics with low impedance types? I imagine most of the electrolytics are general purpose in the 5153 monitor. And most caps I can source that are rated for long life (is this what I want for vintage gear that will sit unused often?), are low impedance types, such as NCC KY and Panasonic F*.

Low impedance electrolytic capacitors (but not polymer low ESR parts) are not *that* much lower so using them to replace more general purpose electrolytics is almost always fine.

When you can't find datasheets for an old electrolytic, so all you know is the marked capacitance and voltage, and the ESR measurement is likely unreliable, is it good practice to replace with a general purpose cap of simply similar physical size? Choosing low impedance if it's in a SMPS (or only certain parts of a SMPS)?

The problem here involves ripple current rating.  Besides ESR, physically larger capacitors have a higher ripple current rating because they have a lower thermal resistance to ambient.  So a modern part which has the same ESR and temperature rating but is smaller has a lower ripple current rating.  The solution is to use a higher voltage part to get a physically larger package which will have a higher ripple current rating.

Note that ESR jumps up at a voltage rating of about 160 volts due to a difference in construction or materials.

That's good to hear. I'll get a pile of caps of same value and similar physical size then, disregarding whether they are general purpose or low impedance (no polymers as I understand they're much lower impedance).

So there's no downside to the longer life rated caps? They barely seem a few cents more, even with top brands on Digikey.

Capacitance drop depends on physical size of a capacitor and dielectric type, not its working voltage. Never use Y5V dielectric type caps, which of course are among the cheapest ones.

Really? I learn something new every day.
I swear I've read many times that it related to voltage. Maybe this comes from the fact that higher voltage parts will often also be larger in physical size.

[james_s]

Ceramic decoupling capacitors rarely fail, but higher capacitance MMLC types are less robust. I have quite a bit of vintage gear including a 5153 monitor and an original XT it goes with. I replaced a tantalum capacitor that blew up on the XT motherboard once but that's the only one I've ever had fail in anything.

[David Hess]

So there's no downside to the longer life rated caps? They barely seem a few cents more, even with top brands on Digikey.

I am not aware of any with longer lifetime parts.  Higher temperature parts might have some compromises and I mentioned voltage rates of 160 volts and higher.

Longer life parts may simply use a larger case size to hold more electrolyte but I am sure there are other differences.

[james_s]

Generally speaking, you can pick some combination of lifespan, temperature rating, voltage rating, capacitance, ESR, cost and physical size, but you can't have them all. All else being equal, longer life capacitors will typically be larger.

[badgerthing]

Ceramic decoupling capacitors rarely fail, but higher capacitance MMLC types are less robust. I have quite a bit of vintage gear including a 5153 monitor and an original XT it goes with. I replaced a tantalum capacitor that blew up on the XT motherboard once but that's the only one I've ever had fail in anything.

I see. I guess I should stop using MLCCs to replace decoupling tantalums then.

Would you suggest I replace all the tantalums? Or only the one that popped?
I feel long-term reliability would suffer, based on my experiences with the bad ones in the Compaq Portable that did not pop.

Leaded 10uF tantalums from AVX are NZD$0.35 each.
SMD-3528 10uF 16V tantalums from KEMET that'd likely fit are NZD$0.12 each.

[james_s]

Personally I would only replace the one that failed, unless several of them had failed or the particular capacitors in question are known to have a high failure rate. I try to keep vintage gear as original as possible, the only time I bulk replace parts is when it's something like electrolytic capacitors that dry out with age.

[David Hess]

Would you suggest I replace all the tantalums? Or only the one that popped?
I feel long-term reliability would suffer, based on my experiences with the bad ones in the Compaq Portable that did not pop.

I would not replace any but if I did, I would voltage derate the new parts at least 50%.  Replacing parts without need carries the risk of damaging something.

[james_s]

I've seen a lot of stuff damaged or degraded by someone trying to fix what wasn't broken and improve something without understanding what they're doing. A lot of times the engineers who designed it in the first place knew what they were doing, especially with stuff that was expensive.