Tantalum vs Ceramic, Cap as short term "battery" power for PICAXE.

Shafto

Senior Member
Was messing around today, just being curious. I made a circuit where an 08M2 was powered by 3.5V, through a 10ohm resistor and diode, with a capacitor on the PICAXE side of the diode, so when power is removed, the capacitor charge keeps the PICAXE going for a couple of seconds. The PICAXE can register the power removed as an input.

I was playing around with different capacitors, and I noticed something odd. When using two 6.3V 330uF tantalum caps, for a total of 660uF, the circuit would keep running for roughly 2 seconds.

When I switched to four 6.3V 220uF ceramic capacitors, for a total of 880uF, the same PICAXE circuit would run for roughly 1 second. When I added two more, for a total of 1320uF, it finally ran for roughly 2 seconds.

I am purely curious, why does it seem to take twice as much ceramic capacitance being used as an energy store to run a PICAXE, than tantalum? I tried looking into this, but all the information I can find comparing these caps is mostly related to frequency/leakage/temperature. The only thing I found is that ceramics generally have much lower ESR than tantalum, but it increases at lower frequencies. Could it be that at the ceramic ESR becomes very large at no/low frequency (energy store mode)?
 

premelec

Senior Member
That's interesting... do you really mean 'ceramic' rather than electrolytic? One thing to try - assuming you have a high resistance input multimeter - is to charge each set to a fixed voltage and then see what the voltage reads with the meter after some hours... could just be an internal leakage situation [ESR is series resistance]. Usual back up cap would be an Ultracap or supercap of 1uF which will keep things running for a while... Q=CV=IT : Charge, capacity, voltage, current, time.
 

Shafto

Senior Member
These are the ceramic caps I'm using:

https://search.murata.co.jp/Ceramy/image/img/A01X/G101/ENG/GRM31CR60J227ME11-01.pdf

And these are the tantalum, F950J337KBAAQ2:

http://datasheets.avx.com/F95.pdf

If the ESR was high enough, then the current demand from the picaxe would drop voltage over the cap, making it less useful, as I understand it? When I read that ESR is generally low for ceramics, but gets higher at low frequency, I thought that could be a possible cause. There's no graph in the datasheet for ESR vs frequency though, so I'm not sure.

Did you mean 1MF super/ultra cap? I checked out some chip package EDLCs like this one:

https://media.digikey.com/pdf/Data Sheets/Seiko Instruments PDFs/CPH3225A_CP3225A.pdf

I would need two in series to handle the voltage, the ESR is quite high. Seems to be geared towards low uA draws, not mA. In going back to look at this one though, I see digikey has a new EDLC from seiko in stock that looks much better, with only 25ohm ESR:

https://media.digikey.com/pdf/Data Sheets/Seiko Instruments PDFs/CPX3225A752D.pdf

I'll probably get a couple on my next order and give them a try. I don't really need the picaxe running long after power is removed though. I don't really need it to do anything in this case, I'm just curious about what's happening between the tantalum and ceramic. I just tried this again to confirm it, in a bit of a different way. Using four 220uF ceramics for 880uF, getting a little over 1 second run time, add 440uF more ceramic and it runs slightly longer for about 2 seconds, take away the 440uF ceramic and add 330uF tantalum to the four original ceramics, runs well over 2 seconds.
 

Janne

Senior Member
I have not looked into the details of your parts, but the capacity (F's) of some ceramic caps are subject to big deratings. Temperature and voltage are parameters that can cause your 1µF cap to be more like 100n..
 

westaust55

Moderator
Ceramic capacitors when used at anything approaching their rated voltage in many types exhibit a situation where the available capacitance diminishes significantly.


Such information is available in many ceramic capacitor data sheets as a curve of capacitance versus applied voltage/rated voltage.

I recall having provided similar comment in past threads on this forum along with links to some datasheet a that show the situation.
 
Last edited:

AllyCat

Senior Member
Hi,

Yes, page 14 of the muRata data sheet seems to give that information. I think we can assume these are "High Dielectric Constant type" capacitors which generally have relatively poor tolerances (+/- 20%) and maybe further "drift" due to temperature, soldering and aging, etc..

The "nominal capacitance test conditions" appear to be with a moderate ac (1 kHz) amplitude and little dc bias. The curves on page 14 show a 50% fall in capacitance when the rated voltage is applied and a (further) 30% fall if the amplitude of the ac signal is small.

Cheers, Alan.
 

techElder

Well-known member
Seems like an "unknown capacitance" problem to me, too. If you are going to try to characterize a part, then you will have to normalize everything else around that part. Reduce the number of variables in your problem to as few as you can.

Another curiosity I had was what program are you running in the PICAXE to determine "the circuit would keep running"?
 

premelec

Senior Member
@Shafto - The double layer cap you reference in post #3 is the sort of cap I was thinking for holding up a circuit for a while if that's what you want to do - if you go to Digikey URL and search "double layer capacitor" and use the filters you can find 5 volt one farad types for instance. iF you could stand 2 volts drop [from 5v] amd had .01 amp draw then the hold up time with one farad would be at least 200 seconds - practical situation more as current usually drops as voltage goes down... [t = CV/I]. As an incidental note double layer capacitor assemblies are being used to provide short high current pulses when put across 12v batteries in delivery trucks that start and stop a lot - even with a weak battery the capacitor will charge up to give the necessary starting current pulse... I wouldn't think ESR is the 'problem' - however another interesting characteristic of ceramic capacitors is mechanical variation in their capacitance with mechanical force... microphonics... In your example I still suggest the static leakage test...
 
Last edited:

Shafto

Senior Member
Thanks Alan! Somehow I missed those graphs on page 14. That would explain what I'm seeing. Also thanks to Janne and Westaust55 for hinting at the issue as well.

Tex, I'm not really trying to characterize a part, I was just curious as to why I was needing twice as much capacitance for the same job from ceramic rated at the same voltage, at the same temperature, than the tantalum. To see how long the PICAXE kept running, I just watched the serial data screen, and how long it went after registering the power being removed.

I suppose tantalum would seem to be the best for a very small SMT "chip size" energy store, other than possibly those new low ESR seiko units I posted. Seems it wouldn't be problematic to put two in series to handle the voltage, as many coin style EDLCs seem to be just that, like this one:

https://www.digikey.ca/product-detail/en/eaton/KR-5R5H104-R/283-2818-ND/1556246

Those 1F 5V EDLC caps would likely work great, they have a nice and low ESR 1/5 of the seiko, but they are also quite huge in comparison to a few 1210 parts. The ESR is quite important when using a capacitor as an energy store, equivalent to the internal resistance of a battery cell, as I understand it. The small chip sizes are generally too high in ESR for anything but very low uA draws, but the seiko units looks promising. I'll post back after I try them out, might be a while though.
 

premelec

Senior Member
I notice that data sheet shows both 100MF and 100mF :-0 Millifarads is what you want... ESR is relevant to the extent the I^2 X ESR is significant as that's the power loss... on charge it doesn't matter so much as discharge in backup use.
 
Top