Datron 4700 ; advice on capacitor replacement strategy needed

[voltampere]

I am now the owner of a Datron 4700 Multifunction Calibrator.
The unit is in very nice "cosmetic" condition.

It had power supply problems due to a shorted ( and burnt, see attachment ) dipped tantalum capacitor.
For the time beeing i replaced it with a similar value aluminum electrolytic that i had.

Inguard Power Supply unplugged from motherboard measured okay, but the unit still complained after power on. Low +15V/-15V rails.
Now my lack of experience caused a bit of smoke...

Since the PS was okay when unplugged from board i thought the voltage regulator LM7815 was broken. So i desoldered the regulator and attched my bench supply with 15V set.
After switching th unit on the failure was still present.
Now i was just stupidly watching that display of the bench supply ...
... 1,5 V --> 1,5 A ( limited ).
And just could not imagine what was going on.
... until smoke started to appear on the ac assembly board and gave me the nice hint!!!
I switched the unit off but the tantalum capacitor was kind of burning already and did continue so for some scary 30 seconds...

Okay, lesson learned! If the regulator output voltage drops there is some probabilty of a component eating to much current!

I replaced this one with an aluminum as well and the unit is working 100 percent now.

I read a lot of information here and elsewhere about problematic tantalum capacitors and especially with problems in Datron equipment.
Which makes me thinking i should replace all of them.
On the other hand there are roughly 100 of them in this unit.
So that would be a lot of work ( and Dollars ).

But i dont want to have one of those light up in my precious machine again

Right now my plan is to do a complete renovation of the power supply boards.

About the rest of the boards i am unsure.
Should i try to measure them in circuit and find the shorted/leaking ones?

Thanks for your thoughts

regards

Joerg

[TiN]

Prolly more time effective is just replace all caps everywhere. 

[DC1MC]

Jörg, let me add my small voice here as well: this tanatalums are the spawn of the  , they WILL fail, in the exact moment of time to cause the most damage and embarrassment as possible. The unit that you have goes for 1000s of EUR in DEFECTIVE !!! condition, and I can only dream to have one, don't cheap out for another 100EUR and risk it having destroyed.
If you can't afford all of them now, replace 20EUR of them per week as a week-end project, or just start a Sparplan, I did for some projects  .
But do replace them all  or suffer later, your choice !!!

 Cheers,
 DC1MC

P.S> On the other side, if you're willing to part of your unit if it's too much hassle to replace the evil caps, I'm pretty sure you'll get here a good price for it 

[montemcguire]

Just to restate the standard good practice, the way to get solid tantalum capacitors to work reliably is to severely de-rate their rated voltage to the circuit's applied voltage. This means that you want to use a part with a faceplate voltage rating of 2x the circuit requirement at a bare minimum, and preferably a much greater margin. The problem will be that it's hard to find high voltage solid Ta caps, and they become large and expensive, but if you can find parts that are 3x the required voltage rating (or higher), the caps will probably not fall victim to voltage induced failure. As an example, for ±15V circuits, 35V rated solid Ta supply bypass parts will probably be sufficient, but if you can find higher rated parts, such as 50V or 63V, those would be preferable. For lower voltage circuits, it's not such a problem finding 'relatively high' voltage replacements, but for power supply bypasses for 15V rails and beyond, you have 35V parts, and increasingly limited capacitances with increasing voltage ratings.

The choice of solid Ta cap brand also can matter, and I have found that some brands are not as robust as others. In general, the Kemet solid Ta caps are very high quality, very repeatable, and very reliable. However, you still need to de-rate their 'faceplate' voltage severely if you want to have high reliability. There are other caps beside the Kemet parts that also work well, but I don't have much experience with the bad ones, and have been very happy with the Kemet parts, so I recommend them as a safe replacement, again, as long as they're de-rated by 2x and hopefully much more.

Best of luck with your Datron - that's a nice instrument, worthy of some TLC!

[ap]

I would agree with the apprach to replace them. I happen to know another owner of a 4700 whose unit on a regular basis stops operation due to a burnt Tantalum. Annoying. Now, interestingly enough, other of these units are very stable (ie., no such problems). It may have to do with the type of Tantalum used (which may have changed), age/usage and it may also have to do with ripple currents in each specific unit. Also, Tantalus deteriorate when they are not powered over a longer time. There are applications (mostly government/areospace/MIL) where stocked hardware spares are subject to power cycles on a regular basis. Re. ripple, if the main capacitors loose capacity, and the ripple voltage on the power lines increases, the ripple current in the Tantalums increases. These however were not built for this. Using Tantalums as blocking capacitors on power rails is not good. It is actually a use beyond specifications in most cases, but back when these were popular because of their AC bahaviour (low ESR...), it seems many engineers just ignored that or did not know. Today something like an Oscon is a much better choice, if you do the repair. Or low ESR Elcos. I would not use Tantalums again.

[voltampere]

Seems like i really should spend the time and money to replace all of the suspicous tantalum caps.

Now ap did suggest to replace them with a more "modern" type.
I checked these "Oscon" types and they seem to be about the price of good quality tantalums.

So these would be more "failsafe" than new stock tantalums ?

What are the cons of these types?
Will they behave as good ( or better ) in terms of noise/ripple blocking?

Another thing is that i want to measure the noise/ripple of the power supply in the current condition ( old caps ).
What would be the correct setup?
I found some information but it looked quite complicated.
Grounding of the Scope probe. Loading the PS ( i dont have a variable load, only some resistors)
Any pointer to some good information would be appreciated.
So i could have some before/after data of the PS noisewise.

Regards

Joerg

[ap]

You should measure the ripple on the supplies in the calibrator as defined in the service manual (hope you have). I have not done any checks, but usually the allowable ripple is at least specified. Be carefull that you may have to use bandwdith limitations on your scope and sometimes common mode noise inflates the reading, so be sure you do not measure this.
The Oscons are supperior in filtering, there are some white papers available. Thats why they are not cheap. I used them on a Datron 4910 that blew up all its Tantalums because I wanted to be sure that the filtering of the new parts would at least be as good as in the old ones.

[voltampere]

I worked my way through the manuals.
I have pdf copies of the user manual of the 4700 and the service manual of the 4708.

The user manual has a spec of the DC Voltage output noise.
I attached as No 1.

I can not find any spec of ripple of the PS in the service manual.
And also no specified procedure to measure these.

I only found a procedure for "Common Mode Null Adjustments"

On the web i found this article about ripple measurements:
https://www.eetimes.com/document.asp?doc_id=1273282

So i wll try my best to measure the ripple before cap replacement.
Although it will be very difficult. If i pull the boards from the mother assembly i have only the unloaded situation.
Is that sufficient for the ripple measurement? ( i dont think so )
So i would have to simulate the load which is kind of difficult for me without the proper equipment.

Measuring while the boards are installed will be extremely hard because of the tight space situation.
Especially when i want to keep the probe tip and the ground lead of the scope as short as possible. ( as mentioned in the linked article.)

regards

Joerg

[ap]

You need to measure wit the load (i.e. boards) installed. They will draw current, and if the caps are low, they will not be able to deliver the charge needed during discharge cycles and the voltage will drop. IIRC these PSUs are not switchmode, so the ripple will be lower frequency. You could measure that with cables soldered to the test point (e.g. ccross a Tantalum connected to the supply rail). This will give you a first indication. If you detect something then you can think about how to improve.

[quarks]

So far all Tantalum in my gear failed directly after powering on. I never had one exploding when the gear was on for a longer time. Does anyone have other observations?

[DC1MC]

So far all Tantalum in my gear failed directly after powering on. I never had one exploding when the gear was on for a longer time. Does anyone have other observations?

This paper:
https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20160001192.pdf

may give you some insights. IMVHO, it could be a similar issue with other ionic fluids devices that needs polarization to keep the ions in their proper place and don't cause self-corrosion or other unwanted chemical reactions, like lead-acid batteries and so on.

Cheers,
DC1MC

[quarks]

the original Tantalums in my Datron/Wavetek 4808 are
            
AVX   TAP104M035F   6x   0.10 µF   35 V
AVX   TAP334M035F   3x   0.33 µF   35 V
AVX   TAP105M035F   28x   1 µF           35 V
AVX   TAP225M016F   2x   2.2 µF   16 V
AVX   TAP225M035F   6x   2.2 µF   35 V
AVX   TAP475M016F   9x   4.7 µF   16 V
AVX   TAP106M016F   16x   10 µF   16 V
AVX   TAP106M025F   42x   10 µF   25 V
AVX   TAP106M035P   5x   10 µF   35 V
AVX   TAP226M016F   4x   22 µF   16 V
AVX   TAP226M025P   16x   22 µF   25 V
AVX   TAP336M025P   9x   33 µF   25 V
AVX   TAP476M006F   7x   47 µF   6.3 V
AVX   TAP107H006F   3x   100 µF   6.3 V

that should be similar in a 4700.
If you want to replace them I would go for the original TAP series and where possible use higher voltage ratings.

[voltampere]

Thanks for the list.

Should i really replace it with tantalum type again?

What about the "OSCON" type that ap mentions?
Any good reason not to use them?

I dont like they way these tantalum caps say goodbye.
I watched one of them burning and i don´t want that happen when the case is closed and i am not around!

I did the ripple measurements of the In-Guard PS.

Common 2 supplies :
+15V(2)  --> 14,75V    23mV Pk-Pk
-15V(2)   --> -15.06V   23mV Pk-Pk
-10V(2)   --> -10,05V   21mV Pk-Pk

Current Option supplies (unregulated) :
+22V(c)  --> 19,3V   200mV Pk-Pk
-22V(c)   --> -19,3V  200mV Pk-Pk

Reference Divider Common 4 supplies :
+36V(4)  -->  35,68V
+15V(4)  -->  14,87V   34mV Pk-Pk
-15V(4)   --> -15,40V   34mV Pk-Pk

Common 2 +/- 8V supplies:
+8V(2)   -->    7,99V    27mV Pk-Pk
-8V(2)    -->   -7,97V    27mV Pk-Pk


Looked again very carefully but couldnt find ripple/noise specs or procedure in the service manual.

Do these figures look alright?

Regards

Joerg

[DC1MC]

Oscon are Panasonic's hi-quality stuff:
https://industrial.panasonic.com/ww/products/capacitors/polymer-capacitors/os-con?list=1

And a bit of overkill, if the power supply isn't a SMPS, but a standard analogue one, modern brand name elcos should be enough and  close to the original tantalums ESR. Some while ago this was the cool thing to do because the elcos of the day were sucking pretty bad on the ESR and being big. Personally, I will not use tantalums today for anything that doesn't have significant high frequency ripple and/or fast switching.
And your device is not such a thing.

 DC1MC

[ap]

Electrically, a good ESR elko will do, sure, provided it fits mechanically (may sometimes not be the case). ESR should be selected related to original parts spec, to be safe. ESR is important not only if a switchmode power supply is used as source, but also to dampen transient loads. That depends on specific use. I would not want to take the effort and analyze, and be safe by using a low ESR type anyhow. Elko  typical aging failure mode is leakage. A Tantalum aging failure mode typically is short circuit/burn, as you well know. Oscon aging failure mode is open circuit. Finally your call. There is no single true answer.

[voltampere]

Thank you so far for all the input.
I appreciate it.
I am not ( yet  ) an electronics engineer ( as you may have noticed ).
But i want to learn as much as i can while repairing/restorating equipment.

So far i learned that tantalum electrolytic capacitors have been used because of their very low ESR value.
Though nowadays it seems like there is some competition.
If i compare these two :
https://content.kemet.com/datasheets/KEM_A4009_ESW.pdf
http://datasheets.avx.com/tap.pdf
Then it appears to me that the Tantalum is not superior, at least regarding DF and ESR values.
( and the Tantalum costs quite a bit more ).
For example a 22uF/25V would be:
Tantalum --->  DF = 8; ESR = 1,5 Ohm @ 100kHz
Aluminum -->  DF = 10; ESR = 0,7 Ohm @ 100kHz ( hope my calc is correct )
So there must some other parameter that makes this tantalum preferable?

What really stops me from using the tantalum type as a replacement is this information that i found:
http://www.kemet.com/Lists/TechnicalArticles/Attachments/154/1999%20CARTS-Europe%20-%20Replacing%20MnO2%20with%20Polymer.PDF
They are explaining that the self healing of the capacitor needs a bit of time.
And if the current during self healing is not limited to a certain amount then you could have a thermal runaway. ( at least that is my interpretation of this paper )
This means it should be forbidden to put a tantalum between + and - of a power rail without some current limiting resistance in series.
That is exactly the place where the 2 burnt pieces have been in my Datron 4700.

So for me the bottom line is right now:
I will use aluminum type for the places where they are installed between + and - of power rail.
Since these are only a few i could choose the more expensive oscon type here.
All other places i will take a wild guess if the very low ESR is really required.
Until somebody convinces my about the necessity for tantalums i also prefer aluminum here, but maybe the less expensive "standard" low ESR.

Sound like a plan? 

Best regards

Joerg

[Kleinstein]

One downside of AL electrolytic caps is that ESR goes up at low temperatures quite a bit. This can be a problem with some voltage regulators for outdoor equipment, but is not a problem for lab equipment that is normally not used at freezing temperatures.

[quarks]

about replacing Tatalum capacitors, I generally try to keep my gear as original as possible, when replacing components, but that is only my personal preference.
But what I do try to change to higher voltage Ratings when these physically fit in the same place.

But if you go for other good quality replacement types it will most likely work just fine.
Only in case it does not work, you must be prepared to eventually spend more time/money and rework everything again.   

BTW: There is a new YouTube Video from Afrotechmods, which might be interesting for others



or

http://afrotechmods.com/tutorials/2018/03/17/why-electrolytic-capacitors-are-actually-kinda-shitty-%f0%9f%92%a9/

see the att. comparison chart (2:21min in video)

[Pipelie]

I use Radial MLCC to replace the Tantalum capacitors in all my Datron equipment, such as 1081/1271/1281/4000/4912/4920/4950/4708 etc. they all work fine.
only the Tantalum capacitors at power rail, not the one in the signal path. and sometimes I use Polymer solid electrolytic capacitors.

and look out the DC Voltage Characteristics
The capacitance value changes by applying DC voltage
to high-dielectric series type MLCC. This is peculiar only to
this capacitor type and does not happen in other types, such
as an electrolytic capacitor and temperature-compensation type
MLCC.

[Pipelie]

for reference:
http://rohmfs.rohm.com/en/products/databook/applinote/ic/power/switching_regulator/cera_cap_appli-e.pdf

and this is part of MLCC I used.

[montemcguire]

The only caveat I'll add is that the Oscon caps ought not be used right at their faceplate voltage ratings, since they can go into a leakage - runaway condition. There's no need for the extreme 2-3x derating that a Tantalum cap needs, but I have personally found that a genuine, fresh stock 25V Oscon is not safe at 22-24V. So, choose a slightly higher rating and you should be fine. For example, a 25V part should be fine at 15V. I don't know what happens when you let the leakage continue to run away, but I did not want to find out, and shut off the lab supply during prototyping!

If you don't need the extremely low ESR of a polymer Al cap, some conventional modern wet Al caps work really well, and are extremely robust and long lived. The Panasonic FR series and the United Chemi-Con KZN series are hard to beat, very small, and very inexpensive. They will also have much lower ESR than any Tantalum, but more ESR than polymer parts. IMHO, this might be a better fit anyway - sometimes a Ta cap is used because of its ESR: some of the old 3 terminal regulator ICs will become unstable if their load capacitors have insufficient ESR. The ESR of the Ta cap is actually used to stabilize the regulator's feedback loop for some regulator ICs, so if they're used as PSU rail bypasses, check the rails with an oscilloscope to look for odd oscillations. These look like odd trapezoidal oscillations at frequencies that are well below 1MHz, not a typical high frequency sinusoidal oscillation such as that from an unstable op amp. At any rate, my main point is that Ta caps were sometimes used because they were not low ESR caps, and modern super-low ESR aluminum and polymer aluminum caps may cause some regulators to become unstable. Check with a scope to make sure, just in case the circuit actually requires the Ta cap, but in a lot of cases, it doesn't.