Guidelines on replacing Wet Tantalums with Electrolytic?

[TERRA Operative]

Hi all.
I have a HP 5065A Rubidium standard that I'm fixing up and I need to replace all the wet tantalum capacitors as some have just started to leak.
They are 100uF, 30V Vishay 109D parts, and their price is the usual 'left arm and firstborn son' on Digikey and Mouser...

So, I plan to replace them all with Vishay 138 AML series axial electrolytic caps.

My question is, what's the general rule of thumb for this sort of thing? These seem to be basically all power supply caps around the place, so nothing too critical.
I've seen some say you should double the capacitance, others say an 0.1uF ceramic cap should be connected in parallel, etc etc.

For reference, the 109D caps have the following specs:
https://www.vishay.com/docs/40023/109d.pdf
ESR: 6
ESL: 40
DC Leakage: 2uA
Ripple Current: 410mA

A 100uF, 40V 138AML cap has the following specs:
https://www.vishay.com/docs/28332/138aml.pdf
ESR: 1.6
ESL: 0.9
DC Leakage: 12uA
Ripple Current: 180mA

Obviously, going up in voltage rating will give better specs, to the point package size becomes an issue, and a 330uF, 40V electro is better in all the above specs besides leakage at 30uA.

Or maybe I'm just looking too far into this for basic power filtering caps.  

[Zero999]

The aluminium capacitor you've selected has a lower ripple current rating, than the tantalum, which is bad. The slightly higher leakage current probably isn't an issue, in this application.

Lower ESR isn't always better, as it can cause some regulators to oscillate, especially the low dropout type.  If the ESR is too low, a series resistor is sometimes needed, to avoid oscillation.

By the way ESL, stands for Effective Series Inducatance and is measured in Henrys, not Ohms, so should be nH not Ω.

[Gyro]

"Elastomer Sealed" wet Tantalum - Ughh!  I never realised non-hermetic ones even existed (apart from those weird button ones).

[Conrad Hoffman]

I've never found that tantalums have any magic properties, other than a size advantage, which may not even exist anymore with modern aluminum electrolytics. I'd replace like with like in terms of value and voltage. IMO, the good tech goes into radial caps, not the axials, so if you want special low leakage (probably not necessary) or some other property (105C or extra long life), you might want to figure out how to lay in radials. I nearly passed out the last time I looked up the price of wet tantalums!

[Zero999]

I've never found that tantalums have any magic properties, other than a size advantage, which may not even exist anymore with modern aluminum electrolytics. I'd replace like with like in terms of value and voltage. IMO, the good tech goes into radial caps, not the axials, so if you want special low leakage (probably not necessary) or some other property (105C or extra long life), you might want to figure out how to lay in radials. I nearly passed out the last time I looked up the price of wet tantalums!

The main advantage of solid tantalum, over aluminium capacitors is, they don't dry out. When used properly, they're more reliable and can tolerate low pressures and high temperatures much better, than aluminium capacitors can. Wet tantalum capacitors are obsolete.

[woodchips]

Wet tantalums, Castanet, use silver as the contacting material inside, so worth money as scrap. Other identifying mark is that the epoxy sealing is coloured red, but not 100% infaliable.

Used is high ripple and high temperature stressed environments, usually military.

Try aluminium electrolytics, modern ones are better than older ones.

[bd139]

I replace these all the time with normal aluminium electrolytics. Usually pick a larger capacitance though.

[TERRA Operative]

Yeah, I wish someone would stuff the guts of their top tier caps into an axial case (Rubycon ZL series, Unites Chemicon KY or KZ series, etc)

At the moment, I'm thinking of using either Vishay 138AML, 220uF 40V caps (MAL213837221E3), or laying down some United Chemicon KYB, 220uF 63V caps (EKYB630ELL221MJ20S).
The United Chemicon caps are much better specs than the Vishay caps, but the Vishays would look nicer being axial... 


Anyone think it's worth paralleling an 0.1uF ceramic with each one? Some seem to recommend it.

[bd139]

The Vishay ones are fairly good but nothing special. Paralleling them with 100n makes sense. That’ll reduce the ESR at high frequencies.

[Gyro]

I've never found that tantalums have any magic properties, other than a size advantage, which may not even exist anymore with modern aluminum electrolytics. I'd replace like with like in terms of value and voltage. IMO, the good tech goes into radial caps, not the axials, so if you want special low leakage (probably not necessary) or some other property (105C or extra long life), you might want to figure out how to lay in radials. I nearly passed out the last time I looked up the price of wet tantalums!

The main advantage of solid tantalum, over aluminium capacitors is, they don't dry out. When used properly, they're more reliable and can tolerate low pressures and high temperatures much better, than aluminium capacitors can. Wet tantalum capacitors are obsolete.

I think the main reason in this case is the 125'C temperature spec. This may have been relevant inside a hot Rubidium standard, depending on HP's required lifetime specification.

[Zero999]

I've never found that tantalums have any magic properties, other than a size advantage, which may not even exist anymore with modern aluminum electrolytics. I'd replace like with like in terms of value and voltage. IMO, the good tech goes into radial caps, not the axials, so if you want special low leakage (probably not necessary) or some other property (105C or extra long life), you might want to figure out how to lay in radials. I nearly passed out the last time I looked up the price of wet tantalums!

The main advantage of solid tantalum, over aluminium capacitors is, they don't dry out. When used properly, they're more reliable and can tolerate low pressures and high temperatures much better, than aluminium capacitors can. Wet tantalum capacitors are obsolete.

I think the main reason in this case is the 125'C temperature spec. This may have been relevant inside a hot Rubidium standard, depending on HP's required lifetime specification.

Modern, solid tantalum capacitors can work up to 125°C with no problems. I don't see why anyone would use a wet tantalum capacitor nowadays.
https://www.mouser.co.uk/datasheet/2/40/TPS-2066671.pdf

[David Hess]

Going by the ripple current ratings *and* dissipation factor, which isn't listed in your datasheet, about 4 times more capacitance is required for comparable performance and that is what I usually do.

Keep in mind that those specification are maximum and do not represent typical parts which can be significantly better.

[edpalmer42]

Since this is for a 5065A, you might want to PM 5065AGuru and either ask him directly or get him to respond here.  I suspect he knows more about those beasts than the designers.

And don't forget that there are a few of those capacitors inside various enclosed modules, not just on visible boards.

Ed

[TERRA Operative]

In the end I went with 220uF, 63V United Chemicon KYB series, bypassed with an 0.1uF ceramic.
I 'converted' them to axial with some clear heat shrink, so they should hopefully work ok seeing as they are essentially just power supply filter/bypass caps, pending operational tests... 
They are good for 1.3A ripple current, 0.055 impedance, with a measured 0.13 ESR @ 40VDC (Cap tester limits), and 320uA leakage @ 63VDC. (measurements taken without ceramic cap in place).

[T3sl4co1l]

FYI, apparently wet-slug have one of the lowest leakage currents among large values; this showed up in a Jim Williams low-noise amp appnote, the number of which I forget; and also got mention in AoE2/3 I think.

Be sure to clean up very, very well after that leakage; it's pure battery acid (give or take gel?), AFAIK.

Wet tantalums, Castanet, use silver as the contacting material inside, so worth money as scrap. Other identifying mark is that the epoxy sealing is coloured red, but not 100% infaliable.

MIL aluminum electrolytics do that too -- best to check the part number (these are M39018 series).  M39003 are hermetic dry slug tant; I forget what wet is, and with whatever sealing method.

Tim

[David Hess]

When I replaced the axial capacitors in my DM502, at first I bent one lead back to fit the capacitor in the original spot but then I decided it was better to wire the leads more directly where that was possible.

[TERRA Operative]

In my DM502, I found the caps passed testing with flying colours, ESR, dissipation and leakage were better than datasheet specs, so I just left them alone for now.
Sometimes you get lucky with the old Vishay 30D and 500D series caps.

I did replace all the old crappy carbon composite resistors with 1%, 100ppm metal film resistors though, some of mine had drifted over 150% of their original value....

[edpalmer42]

From what I can see, your traces are in good condition.  When I repaired my unit, a couple of traces on that board were corroded so badly that you couldn't tell there had ever been a trace! 

If there was any corrosion under the capacitors, it might be worth the effort to do some reverse-engineering of the schematic to check for missing traces.

Ed

[David Hess]

In my DM502, I found the caps passed testing with flying colours, ESR, dissipation and leakage were better than datasheet specs, so I just left them alone for now.
Sometimes you get lucky with the old Vishay 30D and 500D series caps.

I did replace all the old crappy carbon composite resistors with 1%, 100ppm metal film resistors though, some of mine had drifted over 150% of their original value....

The original problem with mine was shorted rectifiers which I did not expect.  The three aluminum electrolytic capacitors that I removed all tested good, but I replaced them anyway along with the two epoxy dipped tantalum capacitors which I originally thought would be the problem, and I replaced the socket for the dual tracking regulator with collet pins.

Then on the back I added TVS protection to the three supply rails as shown.

[TERRA Operative]

In my DM502, I found the caps passed testing with flying colours, ESR, dissipation and leakage were better than datasheet specs, so I just left them alone for now.
Sometimes you get lucky with the old Vishay 30D and 500D series caps.

I did replace all the old crappy carbon composite resistors with 1%, 100ppm metal film resistors though, some of mine had drifted over 150% of their original value....

The original problem with mine was shorted rectifiers which I did not expect.  The three aluminum electrolytic capacitors that I removed all tested good, but I replaced them anyway along with the two epoxy dipped tantalum capacitors which I originally thought would be the problem, and I replaced the socket for the dual tracking regulator with collet pins.

Then on the back I added TVS protection to the three supply rails as shown.

Mine had a faulty LD110 chip, it would lock up as it warmed up. An ebay part seemed to be genuine enough that it now passes all calibration and performance tests with flying colours.
I also stuck a DIP heat sink on that dual regulator using thermal epoxy, it seemed to get a little warm, so I figured a heat sink couldn't hurt.
I contemplated replacing the IC sockets, but it seems to be working fine with the originals after a dose of DeOxit.

Nice mod with the TVS diodes too, was that to fix a problem or for a 'just in case' thing?

[TheMG]

If you want to go the path of "no expenses spared", look at solid tantalum M39003 Tantalex capacitors. They are in my experience nearly indestructible, we have some transmitting equipment with linear power amplifiers that run quite hot, 24/7, over 30 years old, has several of those capacitors literally mounted right up against the heatsink, zero failures of the capacitors they all measure within spec, even ones where the outer plastic wrapping melted due to fan failure!

[mawyatt]

In the late 70s we needed a very large capacitance that could handle signals of a couple volts of either polarity, and needed to be very small. The signals of interest could be less than 0.1Hz but needed to be preserved without significant distortion. We tested all sorts and OEM electrolytics, dry & wet slug tantalums, single, back to back, with diode coupled and so on.

The single wet slug tantalum was selected from an OEM (I can't remember) and proved to be reliable in an environment that ranged from below -20C to ~60C without any issues. This was in a production unit that had an initial build of 10,000 units, and never heard of a problem.

Anyone else use wet slugs tantalums in a bipolar application like this?

Best, 

[David Hess]

Mine had a faulty LD110 chip, it would lock up as it warmed up. An ebay part seemed to be genuine enough that it now passes all calibration and performance tests with flying colours.

I have a DM501A that I am working on with a damaged LD120.  The input multiplexer of all of these Siliconix converters is susceptible to damage from overload resulting in high leakage current.

I also stuck a DIP heat sink on that dual regulator using thermal epoxy, it seemed to get a little warm, so I figured a heat sink couldn't hurt.

I contemplated replacing the IC sockets, but it seems to be working fine with the originals after a dose of DeOxit.

The regulator is ceramic packaged so I was not worried too much about temperature however the edge wipe sockets from Texas Instruments are failure prone, especially where temperatures are higher.

Nice mod with the TVS diodes too, was that to fix a problem or for a 'just in case' thing?

I do not completely trust the fault protection built into the RC4194 dual regulator and the logic supply is not fault protected so I added the TVS diodes to prevent damage to the difficult to replace LD110 converter.  I figured better safe than sorry.

[matthuszagh]

Were you not worried about the lower temperature rating here? I wonder if HP chose these precisely because of the 125C rating. I'm currently facing a similar task: replacing at least one bad 109D in an HP 5065A and am contemplating the Nichicon BX series. These are rated to 150C. A few others that I looked at were the BW and BT series (also high temp). The ripple current on the BX is lower than on the ones you used, but still about double that of the original 109Ds.

[TERRA Operative]

I'd be worried if your lab was getting that hot....