Low leakage

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

dogears

Well-known member
Joined
Nov 15, 2017
Messages
548
Location
Texas
Forgive the basic question, but I've looked and failed at finding this myself. When talking about a low leakage capacitor (for example, for use in a timing circuit) is a film capacitor considered low leakage? How about an MLCC? Where on a datasheet would this kind of information be?
 
Leakage should be specified on data sheet. (Usually in terms of C and applied voltage).

Yes film caps are better. This is well studied for use in sample and hold circuits. Besides leakage, dielectric absorption (soakage) can also be a factor.  IMO DA is overstated as audible mechanism in audio path DC blocking capacitors, but in timing circuits, especially when charged and discharged using different time constants it matters (IMO).

My favorite low leakage film cap is polystyrene, but they are not very robust for use in mass production. Teflon dielectric is well regarded for S/H circuits but expensive. Polypropylene is good, and even Mylar (polyester) is decent.
======
For today's TMI back in the 80's when I made a kit CX record NR decoder, I observed that the encoder (designed by Urie) used a tantalum cap for the primary time constants. Tantalum is notorious for high DA, and in my judgement it could matter for the Urie encoder. In a case of obvious over design I used a tantalum capacitor for my playback decoder kit time constant. To reduce cost, I used a 1uF tantalum cap and scaled resistors to realize the same time constants (Urie used a 10uF in their encoder). 

JR
 
I put this elsewhere, but it should probably go here too.

Several recent tangential observations about old caps. 

After looking at many many amps of the 70-80 year old age group which lived in radio service for decades, it is normal for ESR to be very high in all caps.  A new cap may have ESR of an ohm or fraction of an ohm, with the old one being 50Ω or more.  The piece may still sound fine, but would probably sound better with new caps.  There are those that don't want anything touched, which leads to quandaries.    Comparing against NOS equivalent era caps, the NOS will be much much lower, but still higher than a modern cap.    Like 2.5Ω on the NOS versus 0.5Ω on a new polypropylene Orange Drop.  No one can say what the original ESR would have been when new, but certainly a newer type would be lower in virtually all cases.

This paper is interesting:

https://www.illinoiscapacitor.com/pdf/Papers/impendance_dissipation_factor_ESR.pdf

An important observation is the Fr parameter. Fr is the self-resonant frequency. Defined as the frequency where Xl and Xc are equal.

At this frequency the impedance is equal to the ESR.
  Below self-resonance the Xc component is dominant and the capacitor behaves like a capacitor. Above the self-resonant frequency the inductive component is dominant and the capacitor behaves more like an inductor.


I just went through a couple of late 1930's RCA line amps.    ESR on the coupling caps was almost 80Ω.  The capacitance value of 0.5mfd  looked fine at 120Hz and 1kHz.    It's use caused a large treble rolloff compared to a new cap, no difference in lows.  As much as I wanted to keep using the groovy old hermetically sealed paper in oil film cap, it was past the expiration date.  This is the only time I've ever noted this behavior, but then, how many 80 year old amps that are unmodified do any of us ever get to restore?

In this case the amp was on-air daily for 50 years, and then sat in an abandoned unconditioned space for 22 more years. 


I also recapped a 1960ish Collins 212Z SS remote mixer for someone, which had already seen a recap sometime in the '70's.  Most of the recap was wet slug tantalums.  They all still had very good ESR readings and proper capacitance readings.  What I don't recall is if the wet slug types also fail shorted like the typical tantalums, I suppose that might be a case of DECREASING ESR with age?      The standard electrolytics that were there were all pretty high, 12-16 ohm readings on 300-500 mfd caps.  Still working, but.....


Carry on.....
 
dogears said:
Forgive the basic question, but I've looked and failed at finding this myself. When talking about a low leakage capacitor (for example, for use in a timing circuit) is a film capacitor considered low leakage? How about an MLCC? Where on a datasheet would this kind of information be?

Check this video also,
really good explanations here:

https://www.youtube.com/watch?v=67M7fsbLUIU
 
EmRR said:
I put this elsewhere, but it should probably go here too.

Several recent tangential observations about old caps. 

After looking at many many amps of the 70-80 year old age group which lived in radio service for decades, it is normal for ESR to be very high in all caps.  A new cap may have ESR of an ohm or fraction of an ohm, with the old one being 50Ω or more.  The piece may still sound fine, but would probably sound better with new caps.  There are those that don't want anything touched, which leads to quandaries.    Comparing against NOS equivalent era caps, the NOS will be much much lower, but still higher than a modern cap.    Like 2.5Ω on the NOS versus 0.5Ω on a new polypropylene Orange Drop.  No one can say what the original ESR would have been when new, but certainly a newer type would be lower in virtually all cases.

This paper is interesting:

https://www.illinoiscapacitor.com/pdf/Papers/impendance_dissipation_factor_ESR.pdf


I just went through a couple of late 1930's RCA line amps.    ESR on the coupling caps was almost 80Ω.  The capacitance value of 0.5mfd  looked fine at 120Hz and 1kHz.    It's use caused a large treble rolloff compared to a new cap, no difference in lows.  As much as I wanted to keep using the groovy old hermetically sealed paper in oil film cap, it was past the expiration date.  This is the only time I've ever noted this behavior, but then, how many 80 year old amps that are unmodified do any of us ever get to restore?

In this case the amp was on-air daily for 50 years, and then sat in an abandoned unconditioned space for 22 more years. 


I also recapped a 1960ish Collins 212Z SS remote mixer for someone, which had already seen a recap sometime in the '70's.  Most of the recap was wet slug tantalums.  They all still had very good ESR readings and proper capacitance readings.  What I don't recall is if the wet slug types also fail shorted like the typical tantalums, I suppose that might be a case of DECREASING ESR with age?      The standard electrolytics that were there were all pretty high, 12-16 ohm readings on 300-500 mfd caps.  Still working, but.....


Carry on.....

ESR readings on old capacitors tell you very little, unless it is way off. The best for old audio equipment is to measure capacitance and the leakage current at the rated voltage, both should be within certain limits depending on capacitor size, rated voltage and type of dielectric used. I own a couple of capacitor analyzers, I have a Sencore LC77 capacitor/inductor tester, and I also have an old EICO capacitor tester with an "eye" tube to indicate if the capacitor is leaky or not, the Sencore is more precise but the EICO is great for a quick go-no-go test. Both instruments can test electrolytics and film caps, however, I usually don't bother with electrolytics, I just replace them,  but sometimes it is good if you have a fault to check if an electrolytic is the one causing it.
 
Electrolytic capacitors are used in almost every piece of gear to prevent DC from going into audio control potentiometers.
These capacitors' leakage is one of the main causes of potentiometers getting scratchy.
So I guess using low leakage capacitors extend the durability of the equipment, right?

What would be good practice to choose the capacitors used to prevent DC from going into pots?

I've read this in eevblog.com/forum :
Low leakage and electrolytic capacitor don't really go together, but given that leakage goes up with temperature, capacitance and percent of rated voltage applied, you can obviously arrange circuit conditions - if not environmental - in your favor: use the smallest capacitance with the highest voltage rating possible.

I guess checking the datasheets for low leakage is also something worth doing.
I guess using good potentiometers is also something important.
I've seen 60 years old pieces of gear with good pots and Siemens electrolytic caps not doing a single scratch..
 
Good quality electrolytic capacitors are generally adequate for DC blocking with potentiometers.

There are more sensitive low leakage applications, like timing circuits and sample and holds. Film capacitor dielectric is generally very low leakage.

JR
 
Can we assume that the more the high-pass cut-off frequency (for a volume control i.e.) is low the sooner the pot will start to be scratchy ?
What about the statement that higher voltage-rated electrolytics will have less leakage ?
 
As is usual in life things have multiple factors going on. JR's comments are of course perfectly correct but there are other factors at play, like how smooth and 'clean' the potentiometer track really is. whether there is any DC voltage present on either the incoming signal or the destination the wiper is feeding. Then how much 'disturbance' is permissible. For example if a high gain amplifier follows the pot (anywhere in the following chain) then any disturbance may be noticed. As you may be changing the 'source impedance' of the amplifier the wiper feeds into you may be altering the noise contribution from the resistance of the pot if there is a lot of gain being applied.
 
Thank you for the valuable mentions.
I've used a circuit of this kind in a homemade monitor controller, with 10K pots, and the pot became scratchy after few years of intensive use..

1682605260695.png
I did not use high end pot, but good quality ones though, like this Bourns
1682605387626.png and then this smaller one : 1682605404288.png

but I used good electrolytic capacitors
 
Thank you for the valuable mentions.
I've used a circuit of this kind in a homemade monitor controller, with 10K pots, and the pot became scratchy after few years of intensive use..

View attachment 108376
I did not use high end pot, but good quality ones though, like this Bourns
View attachment 108377 and then this smaller one : View attachment 108378

but I used good electrolytic capacitors
In a circuit like that you can use much higher-valued resistances in order to be able to use film capacitors. First, the input resistance R62 can be raised considerably, it will increase DC off-set, but you have a unity-gain buffer afterwards, so the off-set voltage output will not be very large. You would also have too raise the value of the pot, a combo of something like 1uF poly caps with 100-200k pots/resistors is not a bad choice. Yes, larger valued resistors increase noise, but you have only unity-gain amplifiers, and you are using the circuit for line-level signals, so the noise floor will not rise to problematic levels. The main noise culprit if you use large resistors will be op-amp noise current, if you use a FET op-amp, this problem is largely reduced, and the main sources of noise will be op-amp noise voltage and resistor thermal noise, but, with unity gain amplifiers as the ones in your circuit, the total noise output will be very low. Whether you prefer the 5532 to something like the OPA2134 is something for you to decide, though.

As for DC off-set, in the absolutely worst case scenario you'll have 245 mV of DC at the output of the second op-amp (R62 set at 200K, NE5532), which can be easily dealt with.
 
Last edited:
We tend to think that film caps last forever, but I had more than one instance of polypropylene caps in synthesizers from the 80s used for tuning S/H circuitry getting leaky to the point of audible tuning instabilities. All of these happened to be orange colored for some reason... :geek:
 
We tend to think that film caps last forever, but I had more than one instance of polypropylene caps in synthesizers from the 80s used for tuning S/H circuitry getting leaky to the point of audible tuning instabilities. All of these happened to be orange colored for some reason... :geek:
Nothing is perfect for sure, I can also claim that I have measured Mica caps from very old tube gear which have extremely low leakage. In either case, poly caps are much, much better than any electrolytic. I try to avoid electrolytics as the plague (or COVID these days?) but you can't escape them.
 
Nothing is perfect for sure, I can also claim that I have measured Mica caps from very old tube gear which have extremely low leakage. In either case, poly caps are much, much better than any electrolytic. I try to avoid electrolytics as the plague (or COVID these days?) but you can't escape them.
Yes, mica are different. When I recapped my OB-8 the mica caps used in the low pass filter were the only ones I kept.
 
Mica is a film.

Mica is a solid mineral, different mechanical properties than a thin plastic film. I think mica capacitors are usually enclosed in an epoxy molding. Could be better seal than the typical plastic film construction, but I'm not sure about that.
 
From my reading of Douglas Self books, I always try to minimise Johnson noise as much as possible, but I see you point of view
Minimizing Johnson noise is good, but you have to apply engineering principles, don't just say "noise bad" and that is it.
 
Back
Top