Why do SMD electrolytics have higher ESR than Through-hole equivalents

[diydidi]

Hi. I recently purchased some 10uF 16V capacitors and some 3,3uF 16V capacitors. I have noticed for a while now, that these metal can SMD type caps exhibit a higher ESR reading than their through-hole brothers. Any idea why this is so???

[ReadyTeddy]

SMD electrolytics are sold in many different grades--some have high ESR and some have low ESR. For example, check out some of the offerings from Panasonic to see the wide selection available.

[Mick Bailey]

What I have noticed is that the older SMD caps are now failing where the equaivalent leaded component would have been fine. Some circuits need a complete recap. Generally the caps will go high ESR then OC.

[Steve A.]

I have noticed for a while now, that these metal can SMD type caps exhibit a higher ESR reading than their through-hole brothers.

Are they the same size physically? And would that even make a difference?
I was wondering how you measure ESR for your readings.

Thanks!

Steve Ahola

[The Dude]

The SMD's smell worse too, if you've ever recapped a board full of them. I will often replace them with a "non-SMD" style cap if there's room. They seem to last longer and, in some instances, you can tip them away from high heat sources. It's always amazed me how some boards are designed with electrolytics right up against say a regulator heat sink or some other high heat source. That's just asking for failure. Whenever possible, I leave the leads out of the board a bit and tip the cap away from the heat source.

[oc disorder]

this is related..from down under !

http://members.ozemail.com.au/~tabbler/Main/Home.html

DA-88's ARE GOING DOWN!
TASCAM DA-88

The failure of surface mounted electrolytic capacitors looks set to bring the life of the TASCAM DA-88 and its decendants to an end.

I have worked on Tascam DA-88 machines for years and must have seen hundreds of them. These days, these machines are rarely used except for the purposes of exchanging audio between studios, particularly in the film industry. Therefore, I rarely see these machines any more.

The DA-88 was a well-constructed machine, using expensive double-sided boards with high quality components and discrete capacitors - mostly! Early DA-88's had a few problems, such as clutch and braking problems and at one stage a crook batch with an incorrectly manufactured cam follower that ultimately caused the mechanism to fail. TASCAM revised these parts and if these machines were updated, these and later machines proved to be very reliable. Many people preferred the DA-88 and its derivatives to ADAT machines due to their superior reliablilty, superior sync performance when slaved to other machines or systems and longer playing time.

I have just looked at a DA-88 for the first time in more than a year. This machine had multiple problems, . The mechanism and drum assembly appeared to be in good condition. I checked the obvious things like the power supplies and for leakage around the backup battery, which I have known to have caused problems with some machines due to track corrosion. In this machine, all of these things were OK. On face value, all of the problems seemed to be unrelated electronic problems. In the past, I have rarely had to work on an electronic fault in a DA-88, except when someone has blown one up by attempting to make their own incorrectly wired sync lead!

[Mick Bailey]

Are they the same size physically? And would that even make a difference?
I was wondering how you measure ESR for your readings.

Thanks!

Steve Ahola

They're smaller physical size - perhaps electrolytics don't scale down too well. I measure ESR with a home made tester that measures in-circuit. That little meter gets a lot of use!

[olddawg]

I think they are small, cheap and simply dry out. I used to repair a lot if camcorders. We had to replace buckets of them. Learned early on if one was bad the rest were going and you would have a call back if you didn't.

[diydidi]

I used my trusty DIY ESR meter. It uses a 100kHz square wave for the tests. Just noticed a different reading with smd caps.

[km6xz]

Which dielectric type are you referring to, tantalum electrolytic or aluminum electrolytic capacitors?
Are you seeing them failing in one model or in everything? And what failure mode?
Early tantalums were more problematic but improved a great deal in the last 20 years, while electrolytic smd caps are subject to the same conditions that cause early failure as any other electrolytic capacitor.
The tantalums tend to fail due to shorts, which is related to the type(pulse, peak current, etc) of voltage across them. Electrolytics have higher esr naturally and can dry and drift to lower capacitance and higher esr.

How does your tester eliminate effects of the rest of the circuit if testing in-circuit? The two cap types are usually used in different applications so have different components attached.

The only rash of tantalum cap failure I have seen are very old leaded caps from early synths. Nothing modern, such as the DA-88 era and later, that I have seen with unusual failure rates. When one fails, no need to testing, the fail short and sometimes burn. Any with SMD versions are modern enough to have higher reliability styles. If you have a database of past repairs, it would be interesting to see whether there is a pattern in the last 500 or 1000 repairs compared the same number from 10 years ago to see if same model cap failure rates increased with time. There were a few specific models which had high rates of cap failure, the Alesis mixers for example or Behringer rack mount effects/compressors from the late 90s and early 2000's. ADATs and DA-88s did not seem to have any problems with high rates of cap failures.

I probably did more ADATs than anyone in the world, so have an extensive database of failure modes. We reworked our first 5,000 ADATs before the factory did and have a nice plaque presented by Alesis at the 5,000th warranty claim. Many of those ADAT warranty repairs were not for failures but in factory required modifications. Later we took over all their "factory" service and got all Alesis products returned to the factory re-shipped to us on pallets daily from the factory and they shut down their internal shop. For a couple years all Behringer factory returns in the western half of the US were shipped to us and the eastern half went to New York to dbm. That was a LOT of broken gear and one of the most common failures was shorted tantalum caps on rack effects units, or 4560 or 4580 SMD IC's on the front panel pcbs. The rest of the models and brands did not see any higher rate of cap failure than expected. I still have a copy of a database that covered 1994 up to 2004 when I gave the business to the employees when I moved here. That db totals 60,000 and after converting to MySQL 5.1 I have played with distilling useful data from it like the distribution of specific defects or fixes to the total repairs by model, brand or types. Tantalum cap failure was concentrated on some models and brands but overall they were rare when considering all the thousands of models that used them. Since I have gotten back into repairing, I sometimes search the db for clues as to higher frequency of defects or solutions by model.

The common habit of doing a "cap" job is one I do not understand, maybe a way to pad the bill or I do not know. Other than some old synths, I have never seen anything where wholesale replacement of caps improved anything except profit. Replace it if the conditions changed and stressed the part, or if there is a measured deviation of characteristics to outside of the tolerance budget of the circuit design but time alone based replacement does not improve reliability. In fact during its rated service life, the highest failure rate for parts is when new. A part that has been working for 10 years has a higher predicted reliability to last an additional year than a brand new part.

How are you limiting other factors in-circuit to determine ESR of the individual part if it is connected to a whole circuit environment? ESR meters usually work by applying a low duty cycle pulse and measuring the voltage across a divider, but that assumes the circuit lump resistance, leakage and inductive reactance is low or nulled out. You might be measuring the higher inductive reactance of aluminum electrolytics compared to the normally low inductance and ESR of tantalums. PC board leakage and simple dirt can change the apparent test results also. The best way to test most parts is how they perform in the circuit. Does the current flow at some test point match the predicted value for the circuit conditions? If yes, it is working. That goes for tubes, transistors, resistors, caps, just about anything. Simple static beta testers of transistors tell very little about the ability of that part to work in its circuit. A curve tracer that can mimic the load and potentials of the circuit will find problems that would never be seen with an ohm meter or static transistor tester. Same with any other part, those analyzers that mimic real circuit conditions tell much more than static tests with unrealistic test levels and signals. A tube curve tracer or a real chassis are the only ways I know of to test a tube to see if it is capable of working in the circuit.

[diydidi]

Hi.
I measured new caps from different makers, out-of-circuit. I had to replace smd caps on a motherbord of a Roland KR4500 piano. most of them were oozing-out dielectric onto the pcb. I had to repair some tracks under the microscope too..

[km6xz]

What series caps, and what dielectric?
How many have you seen that measured as faulty, a dozen, a thousand or a few on one PC board? I am trying to understand the scope of the problem you are having.

[Mick Bailey]

If you get a Takemine preamp where the volume is decreasing over a number of months I'll guarantee every SMD electrolytic on the board will need replacing.

EDIT: I've been measuring ESR in-circuit for years without problems. Much like in-circuit resistor checks, the exact value doesn't matter so much as it's either high or low. If (say) a 22uf electrolytic is reading 30 ohms, it would be unlikely that the cap was parallelled by a resistance of equivalent value, though working alongside a schematic or having an understanding of the circuit is useful. If the actual value measures higher with the component removed, it doesn't matter - I only need to get an indication that it's bad. The test voltage needs to be low enough not to turn on any connected semiconductor junctions. The tester I use will also check ESR with the circuit powered so this gives more realistic test conditions.

[diydidi]

They're smaller physical size - perhaps electrolytics don't scale down too well. I measure ESR with a home made tester that measures in-circuit. That little meter gets a lot of use!

Hi tell me. What sort of rule/ scale/ chart do you live by when it comes to discerning between a good and bad capacitor using your meter?
I'm still struggling with this a bit. I do of course know that anything higher than 10ohm is bad.

[Mick Bailey]

Hi tell me. What sort of rule/ scale/ chart do you live by when it comes to discerning between a good and bad capacitor using your meter?
I'm still struggling with this a bit. I do of course know that anything higher than 10ohm is bad.

It depends on the cap size and voltage rating, and whether it needs to be low ESR to begin with. A suspect cap in one application can be fine in another. If the cap is meant to be low ESR (say in a SMPS) then I'm not expecting much of a reading at all - barely off the mark, the same with large caps in transistor amp power supplies. I generally find that below 3.5 ohms is ok for most caps in tube amps, a little lower for the main filter caps. It's useful to do a quick check against a new component, or the same component elsewhere in the circuit under test. Some miniature caps are fairly high ESR from new.

With many circuits, by the time ESR becomes an issue the reading is usually very high and there's no ambiguity. An electrolytic cap is not a precision component and will be used in circuits that can tolerate drift in value and ESR up to a certain point. Think of ESR tests like you'd test resistors; you can run your meter probes along a whole row of resistors and just be looking for near zero ohms or very high resistance - the actual value can be largely ignored. A resistor that drifts in value generally won't matter in an amp - if it measures 90k or 110k instead of 100k, so what? It will fail absolutely if it's shorted or open or very high resistance, though. The same with caps.

I built my meter with an analogue movement and I can see from the needle position that below 1/3 FSD is Ok, above is suspect, and FSD (30 Ohms) is really bad. It's not a linear scale - halfway is 6 Ohms. This means I can quickly check all the caps in a circuit without actually reading off a value. Sometimes an analogue movement has the advantage over digital.

What may surprise you is how low the ESR of some old caps can be - much better than new. Sometimes customers bring amps for a cap job and won't have it any other way. Just looking through a bunch of Silverface-era Mallory filter caps and they all measure less than 0.5 Ohm. No leaks, no bulges, no hum, low leakage.