Electrolytic vs Tantalum for Supply ByPass?

[industrialarts]

Hi

I just had an Arp Omni on my bench for repair that has an impressive amount of tantalum caps - 50-75, maybe - and the majority of them had failed. This is something I encounter often in gear of this vintage.; tantalum caps from this era seem to fail often and usually present as a near or dead short. This unit was DOA but after quickly finding the shorted supply bypass caps, it was pretty much up and running.

A large portion of tantalums were used in the keying circuit, so replacing with electrolytics was fine. I did read a lot of posts regarding this and didn't come up with a solid answer - what are the issues with replacing all the supply bypass caps with electrolytics as well? I understand that one of the major benefits of tantalum is large amounts of capacitance in a small package but if space is not an issue then what other factors could be problematic? In fact, the replacements for the keying circuit were actually physically smaller than the original tantalum caps. Plus, my research suggested that you should overate the voltage of the tantalum caps by 2 - 3 times the working voltage, whereas electrolytics are usually in the area of 1.5 or so. Plus, tantalum caps cost about 10 times as much as an equivalent electrolytic.

One thing I was thinking about was in-rush current if there is a considerable amount of caps. Is there much of difference in that respect? Is there anything that makes a tantalum that much better for supply bypass?

Thanks


BTW, in case anyone is curious, I have been using the Wurth electrolytics for anything in the 1uf-47uf, 35-50vdc range. You can get them from Mouser, they have great specs and look cool as well

 

[Bo Deadly]

I understand that one of the major benefits of tantalum is large amounts of capacitance in a small package but if space is not an issue then what other factors could be problematic?

My understanding is that the only real requirement for using tantalum is low leakage. If the cap has substantial DC across it and either side is high impedance, then leakage might be an issue. But that would be highly unusual. It would really have to be an esoteric circuit to require such low leakage. Otherwise, just swap in an electrolytic of the nearest larger capacitance and nearest larger voltage.

For critical signal paths utilizing 10uF or less, you can use film if you have the cash.

If the cap does not carry high fidelity signal or is generally not critical to signal integrity, you might consider MLCCs. MLCCs can be high impedance at low frequencies, they're microphonic and generally sensitive in other ways but are just fine for supply bypass. They are even smaller than tant for their capacitance, they have recently become available in very large capacitances for cheap and they last longer than electrolytic.

 

[Newmarket]

My understanding is that the only real requirement for using tantalum is low leakage. If the cap has substantial DC across it and either side is high impedance, then leakage might be an issue. But that would be highly unusual. It would really have to be an esoteric circuit to require such low leakage. Otherwise, just swap in an electrolytic of the nearest larger capacitance and nearest larger voltage.

For critical signal paths utilizing 10uF or less, you can use film if you have the cash.

If the cap does not carry high fidelity signal or is generally not critical to signal integrity, you might consider MLCCs. MLCCs can be high impedance at low frequencies, they're microphonic and generally sensitive in other ways but are just fine for supply bypass. They are even smaller than tant for their capacitance, they have recently become available in very large capacitances for cheap and they last longer than electrolytic.

Just a word of warning that the MLCCs with highest capacitance per volume are quite poor in terms of temperature coefficient.
eg Y5V and Z5U dielectrics.
I don't have comparative figures for Al Electrolytics or Tantalum to hand. But generally I wouldn't go worse than X7R.

On the Tantalums - yes they did have a bad reputation. I believe they are a lot better now. The only problem I've had with them is where the PCB Assemblers have placed them with incorrect polarity. They do smoke / smell / make a mess when they 'go'.

 

[industrialarts]

Thanks for your replies. Funny you mention the MLCCs - they are not a technology I was familiar with until researching the tantalums, and they just started popping up on the Mouser website. I guess they would be a viable alternative and have similar packaging to the tantalum but since there don't seem to be any specific advantages I'll probably just start using electrolytics for the most part, with space requirements being the the only consideration.

One other thing I suspect - and has been alluded to from reading and a couple other techs - is that tantalums don't like spikes much and the power supplies in the older gear I work on produce a lot of turn-on/ turn-off spikes. I know I could install TVSs to prevent that sort of thing but it's just more work

 

[12afael]

Don't use tantalum on power supplies. They tend to fail in short circuit so they can take your transformer if you are not careful.

 

[Bo Deadly]

... but since there don't seem to be any specific advantages I'll probably just start using electrolytics for the most part

Like I said, MLCCs are a large capacitance in a small package for low $ and their performance will not deteriorate like electrolytics. So there are specific advantages. And there are specific disadvantages which have been described above. But for a power supply bypass MLCCs are just fine. But get X7R just because you can for the same money. The only catch to this might be if the bypass is for an amp that is sourcing an sinking a lot of current at low frequencies to the point where the higher impedance at low frequencies might be an issue. I don't know anything about the circuit so I can't say if they would be appropriate. Usually tants in old gear are used in the signal path which would not be ok for MLCCs. You would have to use electrolytics.

PS: The polypropylene comment was a joke just in case you didn't pick up on that. Even a 10uF PP cap is about 2 inches long and costs $12 USD.

 

[ccaudle]

tend to fail in short circuit

Not only that, tantalum caps can ignite, and the tantalum pentoxide and manganese dioxide break down and provide oxygen to the combustion, so once it starts you can't put it out, you just have to wait for the entire capacitor to be consumed. Had that happen a few times on a backplane design that had a lot of tantalum caps. The tantalum polymer electrolytic caps don't have that problem, but the traditional tantalum electrolytic that are just tantalum and manganese dioxide do. Once I knew what to look for I found some app notes from cap vendors that basically said don't use these caps for power supply bypass on a low impedance supply unless you have a slow start circuit. You can also do some things during manufacturing to help prevent the problem, but the short of it is use tantalum polymer if you just absolutely are space constrained, otherwise use aluminum.

 

[jetbat]

Changing out all the tantalums for aluminum electrolytic in a piece of equipment will change its sound. Since this is a keyboard it may not matter as much as a preamp, EQ, or such. You may still want to retain the tantalum caps in the signal output section and power supply where they originally were. Obviously the caps worked for a long while before failing and did not take out anything else when they did fail.

 

[Bo Deadly]

Changing out all the tantalums for aluminum electrolytic in a piece of equipment will change its sound.

How so? Please be specific.

 

[MaxDM]

tantalum caps are not perfectly linear and have a 'sound'

I think they have better ESR than electrolytic caps, especially as they age, so they don't sound as 'soft'

 

[Bo Deadly]

tantalum caps are not perfectly linear and have a 'sound'

I think they have better ESR than electrolytic caps, especially as they age, so they don't sound as 'soft'

Note that a lot of folks here don't embrace these sorts of claims. Surprisingly even the most technically accomplished builders have expressed belief in some form of inexplicable sound phenomenon. I guess when you work in the industry and your friend builds something they're proud of, you're obligated to agree to impart some magic dust on it just to conform to the usual tribal social protocols. But when people start talking about tropical fish capacitors and how some op amp sounds "3D", usually someone (like me) will pop up and knock that down to keep things grounded in reality.

It is entirely possible that changing some old capacitors changes the "sound" but if one took the time to take high-res FFT captures before and after, I think you would find that you simply got a little more bass back or the circuit started to work better as it was designed to. Or, 9 times out of 10, there was actually no change at all and it was just a "I worked on this and so now it sounds good" type of psychoacoustic effect.

ESR of newish electrolytic and tantalum are very similar and not more than a few ohms at low frequency which is completely insignificant for 99% of analog circuits like DIYers might build. Impedance is higher and more important but still probably not even visible in a spectrum and certainly not audible.

 

[MaxDM]

Note that a lot of folks here don't embrace these sorts of claims. Surprisingly even the most technically accomplished builders have expressed belief in some form of inexplicable sound phenomenon. I guess when you work in the industry and your friend builds something they're proud of, you're obligated to agree to impart some magic dust on it just to conform to the usual tribal social protocols. But when people start talking about tropical fish capacitors and how some op amp sounds "3D", usually someone (like me) will pop up and knock that down to keep things grounded in reality.

It is entirely possible that changing some old capacitors changes the "sound" but if one took the time to take high-res FFT captures before and after, I think you would find that you simply got a little more bass back or the circuit started to work better as it was designed to. Or, 9 times out of 10, there was actually no change at all and it was just a "I worked on this and so not it sounds good" type of psychoacoustic effect.

ESR of newish electrolytic and tantalum are very similar and not more than a few ohms at low frequency which is completely insignificant for 99% of analog circuits like DIYers might build. Impedance is higher and more important but still probably not even visible in a spectrum and certainly not audible.

ESR of old electrolytics can be quite evident in the top end, and I have personally heard significant changes when recapping.

Tantalum is more stable and has an ESR which is consistent as it ages, as well as having a much lower ESR, by comparison.

As far as not being able to hear the difference in types of caps, that is probably due to the subtlety of the sound difference, the individual's ability to hear distortion and mental bias, one way or another.

I hear a difference in different capacitor types, so I don't worry too much about the tech aspect.

Here's an old article which seems to describe some difference.

http://diyaudioprojects.com/mirror/members.aol.com/sbench102/caps.html

 

[MaxDM]

Included is Tantalum:

http://diyaudioprojects.com/mirror/members.aol.com/sbench102/caps1.html

 

[Bo Deadly]

Here's an old article which seems to describe some difference.

"The signal level was held constant at about 70 volts RMS at 600 Hz across the capacitors. (for about 26mA signal current). This is probably more than you would normally expect, and serves to show the results better."

Distortion of caps is proportional to the voltage across them. Small signal audio circuits only use electrolytic and tantalum for coupling and the sizes are specifically selected to have very small voltages across them. Putting 70 VRMS across a cap does not show that "there is a distinct sound to different types of capacitors". It shows that the author does not understand E very well.

If you're going to post links like that, I'm definitely going to knock them down.

 

[Audio1Man]

Tantalums were used to save space and provide very low ESR in the 197x and beyond. The ESR of old electrolytic was very poor then.

The Tantalum is more stable and has low ESR which is consistent as it ages. The big problem is that Tantalum’s DON’T like ripple currents, and self destroy themselves and the PCB’s.

I replace them with modern low ESR types and axial 0.1uf ceramics as needed.

Duke

 

[Newmarket]

ESR of old electrolytics can be quite evident in the top end, and I have personally heard significant changes when recapping.

Tantalum is more stable and has an ESR which is consistent as it ages, as well as having a much lower ESR, by comparison.

I'll suggest that it's not ESR that is of particular importance here - although if any ESR figure is not stable then it will affect the signal fidelity.
Capacitance in Al Elecs does decrease over time so lf response may be altered. But as noted they should be sized for negligible voltage across them at the lowest audio frequencies. This will degrade with time / use.
This is for signal ac coupling. Different considerations apply for PSU Decoupling / Energy Storage / Bypass.

 

[JohnRoberts]

If you can hear and measure a difference caused by capacitor dielectric that design may be faulty.

Tantalum capacitors in PS decoupling are unlikely to be audible unless they short and catch on fire.

Back in the 70s there was full throated criticism of tantalum's DA*** (Dielectric absorption) by the audiophools du jour. This too is extremely unlikely to result in audible artifacts. DA was a well know phenomenon making tantalum capacitors inappropriate for sample and hold circuits.

There are very linear modern capacitors for use in audio frequency filters where nonlinearity could be seen (measured, heard). Properly used in DC blocking applications modern electrolytics will be neutral.

I don't argue with people about what they say they hear on the internets, but I always invite them to present objective measurements showing distortion or artifacts.

JR

*** I've shared this story before. Tantalum DA can affect behavior in side chain time constant circuits. Back in the 80s I designed a CX decoder (long story). The CX encoder used to make CX records (designed by Urie) used a tantalum cap in it's side chain, so I used one in my decoder more accurate tracking. This was arguably over engineered for an inexpensive audio kit.

 

[Monte McGuire]

Like I said, MLCCs are a large capacitance in a small package for low $ and their performance will not deteriorate like electrolytics. So there are specific advantages. And there are specific disadvantages which have been described above. But for a power supply bypass MLCCs are just fine. But get X7R just because you can for the same money.

Just to fill in a few blanks, modern MLCCs do offer large capacitance and small size, but their capacitance can often be a strong function of applied voltage. At 3.3V or 5V, the "sag" caused by an applied DC voltage will not usually be important, but at 15V, some MLCCs can deliver as little as 30% of the part's faceplate capacitance, while other parts can deliver 80-90%.

Also note that this voltage sag is not a strict function of the dielectric type, it is also the result of other factors about the cap's construction (dielectric thickness I assume). So, to be certain, you must go to the manufacturer's web site and look up the specs for that specific capacitor part number and see what capacitance will be left at the DC voltage you plan on using the part at. Usually they provide a nice little graph of C vs. applied DC and you can move the pointer over it and read out the change at the desired DC bias. But remember, this must be verified for that specific part - make no assumptions.

As for not deteriorating, after an MLCC cap is "fired", or brought above the Curie temperature of the ceramic, the lattice will slowly rearrange itself and the charge domains will slowly go away. X7R decays at about 3% per decade-hour of time, which is the lowest of the high K dielectrics. X5R is a little worse IIRC. It's not a huge effect, but for some regulators that will become unstable without a specific load capacitance, it is something to consider. For example, I'm using the Analog ADP7142 and ADP7182 which require 1.5µF load capacitance, so I'm using 3x Kemet C1206C105K5RACAUTO on the output, which will start at 3µF, get derated from the 10% tolerance, lose capacitance due to 17V applied DC, and then lose capacitance over time, end up at 1.97µF at 6 weeks, 1.85µF at 11 years, and end up at ~1.5µF after 114 million years (10^12 hours) Here's a link that goes into some details: Does the capacitance of ceramic capacitors change over time? Are there any points to be aware of regarding the changes over time? | Murata Manufacturing Co., Ltd.

And, an even more comprehensive paper from Kemet on a lot of the details about how capacitors work: https://sh.kemet.com/Lists/Technica...capacitor drops to 37 uF- 30 uF- or lower.pdf

 

[MaxDM]

Just changed some 70's electrolytics in a compressor today

There's a noticeable degradation in sound when dealing with old electrolytic caps.

Manufacturers admit that shelf life is not eternal.

Tantalum does not go soft and fuzzy, by comparison.

If you are only concerned with power rail filtering then maybe tantalum is a bit of a risk, because they react badly to high current peaks. Some manufacturers install current limiting resistors as a safeguard against this weakness.

Electrolytic caps with film caps in parallel sound better IMO, generally speaking, but some of the sound of a lot of 70's discrete gear is due to Tantalum caps.

As far as distortion caused by polarized caps, SSL was certainly aware of that aspect when designing their stuff. Everything distorts. Perfection is more of a Marketing thing than if is reality, IMO

 

[Newmarket]

Just posting this link for anyone wanting TMI on electrolytics
(link text refers to placement but document from Cornell Dubilier covers a lot more)

https://www.eevblog.com/forum/testg...er-supply-unit/?action=dlattach;attach=148540