Replacing all capacitors with solid polymer ones inside amplifier

[rjsalvi]

I haven't come across any mid-70s vintage gear using LDOs, but I haven't worked on every mfrs. offerings. I have however used polymer load caps on a few vintage early 2000s NAD amps which use the LM3xx regulators and they're stable, but again, it's a limited sample size. Also, no issues with gear using the 78xx regulators.

 

[Sansui_guy]

Hi, I'm newbie here although I've been reading a lot of (good) stuff for a while now . So, resurrecting this topic.
Due to its better characteristics overall, I am tempted to use aluminum polymer capacitors, preferably hybrid when available, on low voltage, say till 50 Vdc, exclusively for filtering/smoothing in some voltage rails in some circuits (not in signal path, coupling, decoupling, bypass, etc.). Just asking you guys your opinion where price, size and shape aren't a concern, but applicability and effectiveness. Any thoughts and experience are very welcome

 

[HPMguy]

Do you mean "supercaps"? Check the impedance before you make the change, those are usually used for replacing batteries in circuits with low current flow (memory in remote controls, solar watches, etc). Not sure they can deliver the current needed for source filtering.

 

[Sansui_guy]

Nope, I mean the regular solid aluminum polymer capacitor’s applicable mainly on low voltage lines on some amp circuits.

 

[chazix]

Any thoughts

I'm no expert, and certainly not very experienced, so these are only thoughts. I gather that Al-polymer caps have less tolerance for "surge voltage" compared to Al-wet-electrolyte, so I would think when replacing the latter with the former, you'd want to bump up a step in voltage rating.

There's also a small chance of the Al-polymer's lower ESR being too low in applications like the output of a regulator. If this is a worry at all, it's probably only a worry for three-terminal regulator chips with lots of "loop gain". I don't think I'd give it a second thought with a discrete-component regulator.

 

[tarior]

I'm not sure what anyone is hoping to gain by changing out their electrolytic caps with Al-Poly, solid polymer, or whatever caps. New, high quality, 105 degree C. electrolytics, and film caps where appropriate, will last longer than any of us will.
These units were designed to use certain types of components, with the limitations of said components in mind.

 

[chazix]

One thing I find attractive is not being at the mercy of the end seal in a wet electrolytic. Might be OCD, but I worry about bending leads that are passing through a seal, as is often hard to avoid when replacing caps in vintage gear. When I can, I match lead spacing to the original, but that's only going to get harder in the future. When I can't match, I often hand-form leads so they won't stress the seal, which is a PITA.

With a solid polymer part, I think I just wouldn't worry about lead bending. That would save time both in selecting new parts, and in replacing the old parts.

Again, I'm just hand-waving, not speaking from experience yet.

 

[GrisPato]

Hi, I'm newbie here although I've been reading a lot of (good) stuff for a while now . So, resurrecting this topic.
Due to its better characteristics overall, I am tempted to use aluminum polymer capacitors, preferably hybrid when available, on low voltage, say till 50 Vdc, exclusively for filtering/smoothing in some voltage rails in some circuits (not in signal path, coupling, decoupling, bypass, etc.). Just asking you guys your opinion where price, size and shape aren't a concern, but applicability and effectiveness. Any thoughts and experience are very welcome

Read the data sheets well.
Example:
Kemet A759MS107 Organic polymer, 100uF-50V.



Amazing ripple current rating from 2830 mA, compared to an Elna RFS same values with just 380 mA.



But . . . See the datasheet of the Kemet, compensation factor of 0.05, or 152 mA at 100 / 120 Hz. The almost 3 Ampere is at 500 kHz!
So the Silmic audio capacitor has twice the ripple current rating at the important frequency.
ESR is for sure better, 38 mOhm for the Kemet and 85 mOhm for the Silmic ( measured my self at 100 kHz).

 

[petes'picks]

Polymers appear vastly inferior for power supplies if their ripple current ratings are reduced that much at 120hz. For pwrsupply when compared to Nichicon PM and Panasonic FC, a 100uf 50v FC has a .65 compensation factor at 120hz (615ma of current) In other words those polymers cant handle hardly any current at 120Hz.

 

[Sansui_guy]

Yes, I see. Haven’t noticed that before, just curious why they states the opposite at the beginning of the DS?

 

[GrisPato]

Polymers appear vastly inferior for power supplies if their ripple current ratings are reduced that much at 120hz. For pwrsupply when compared to Nichicon PM and Panasonic FC, a 100uf 50v FC has a .65 compensation factor at 120hz (615ma of current) In other words those polymers cant handle hardly any current at 120Hz.

Here again, factor is 0.65 ( 615 mA is at 100 kHz) so 400 mA at 120 Hz, a measly 20 mA more then a Elma Silmic.
Is the problem I encounter with a lot of talk about components and there parameters.
I did not study electronics, little experience. Mostly statements like, low ESR is better, high ripple current is better, etc, etc.
Then you wonder, what is a low ESR value, how low is low enough, is there a specific value needed in some specific situation, etc. Then nobody knows. People will recommend a high ripple current capacitor ( like Panasonic FC, Nichicon UPW ) in places where the ripple current will be in uA, while the ripple current rating for even a general purpose capacitor is already in the 100s of mA. Or a coupling capacitor with lower ESR although ( difference is in mOhm) while it is in series with a resistor of 5-600 Ohm.
People talk about ESR as if the resistance slows down the signal but in the NFB network there is often a resistor of around 10 kOhm. The NFB signal is not slowed down as far as I know!?
That is why I say, read the data sheets, and not the marketing on the top but the data.

 

[mroboto]

Solid Polymer alone sounds like crap.

Aluminum Electrolytic Capacitors sounds smoother.

Multiplying capacitors is the key, I like to add Ceramic or Film capacitors.

Ceramic sounds better at high audio frequencies, the highest values like 10uf 50V everywhere you can for DC filtering of sources components.

The highest voltage you go the more difficult is to find high capacitance Ceramic capacitors.

 

[mroboto]

People talk about ESR as if the resistance slows down the signal but in the NFB network there is often a resistor of around 10 kOhm. The NFB signal is not slowed down as far as I know!?

The best you can use is non inductive resistor.

 

[GrisPato]

People talk about ESR as if the resistance slows down the signal but in the NFB network there is often a resistor of around 10 kOhm. The NFB signal is not slowed down as far as I know!?

The best you can use is non inductive resistor.

As far as I know for the resistor used in the NFB network this is not an issue. It is not wirewound because it is mostly around 1/4-1/3 watt, wirewound is normally for high wattage resistors. And even then the values are small. For example, emitter resistors don't need to be non-inductive when there is a inductor behind ( this is the case in all my amplifiers ).
What I am trying to say is that there is a lot of talk about parameters of components but not about what importance and effect they have in the specific implementation.
All I can say is: inform yourself extensively, read about both sides.

 

[GrisPato]

Solid Polymer alone sounds like crap.

Aluminum Electrolytic Capacitors sounds smoother.

Multiplying capacitors is the key, I like to add Ceramic or Film capacitors.

Ceramic sounds better at high audio frequencies, the highest values like 10uf 50V everywhere you can for DC filtering of sources components.

The highest voltage you go the more difficult is to find high capacitance Ceramic capacitors.

You can do what you want to do, if your happy and your ears are happy.
Here again, ceramic is better for high frequency, true. But audio is low frequency.
Ceramic can be better then film for high frequency, but.... , here we mean really high frequency > 1 MHz and into the GHz.
Higher capacitance ceramics use a type of dielectric that is really unstable, lot of distortion and the capacitance drops like a stone. No problem in some locations but for sure proven bad in the audio path with no benefit at audio frequency.

 

[mroboto]

As far as I know for the resistor used in the NFB network this is not an issue. It is not wirewound because it is mostly around 1/4-1/3 watt, wirewound is normally for high wattage resistors. And even then the values are small. For example, emitter resistors don't need to be non-inductive when there is a inductor behind ( this is the case in all my amplifiers ).
What I am trying to say is that there is a lot of talk about parameters of components but not about what importance and effect they have in the specific implementation.
All I can say is: inform yourself extensively, read about both sides.

Yes, since they are 1/2W, forget about, that was just an example.

You can do what you want to do, if your happy and your ears are happy.
Here again, ceramic is better for high frequency, true. But audio is low frequency.
Ceramic can be better then film for high frequency, but.... , here we mean really high frequency > 1 MHz and into the GHz.
Higher capacitance ceramics use a type of dielectric that is really unstable, lot of distortion and the capacitance drops like a stone. No problem in some locations but for sure proven bad in the audio path with no benefit at audio frequency.

Transistors are Mhz rated, why not capacitors ?

The flagship BA-F1 is a great exemple of very fast amplifier at 600Khz frequency response.

Film caps I did try in my mods tends to be slower, missing details power and dynamic in high frequencies, (With tubes)

 

[GrisPato]

Transistors are Mhz rated, why not capacitors

BA-F1 is a great exemple of very fast amplifier at 600Khz frequency response

Good question, same idea, fT for transistors.
Because I own 2 Sansui BA-F1 and several other integrated amplifiers from this model range I studied this a bit.
The fT is the frequency where the transistor went down to unity gain, gain of 1. So because you want to be at maximum 0,5 dB down at 20-30 kHz with some margin Sansui used transistors that have fT of >25 MHz.
And fT is a curve.
Example: Sanken fT 50 MHz ( typical )
compared to Onsemi MJ21193 fT 4 minimum.


Between 100-300 mA of emitter current fT is around 25.





For the Onsemi it is 5-6 MHz.

And to repeat, I am not an EE, I am just a novice. People with limited knowledge ( as I am) should be really careful with stating as facts what we think we know.
it is nice and fun to discuss it but we have to keep in mind that we do not know much (yet).

 

[petes'picks]

Lets say your 100uf 50v polymer is rated at 2900ma of current and you derate it .05 for 120hz then that is only 145ma of max recommended current at 120hz. I think I am gonna go with the 400ma rating from the electrolytic for my amps power supplies.

You can see how you can get into trouble fast if you try replacing the larger value caps in a power supply with polymer because by the time you get to 330uf 63v polymer can only handle 200ma of current while the FC electrolytic of same value can handle 830ma of current.

I am sure there are places where polymer will start being used but doubt you will see much being used in power supplies of amp. And higher DC leakage associated with polymers sounds like a bad thing in the audio path. Creating a new amp circuit using just polymers may be doable but replacing all electrolytics in an existing circuit that was designed with electrolytic parameters, well I hope someone does it and posts before and after distortion figures and frequency responses.

 

[Sansui_guy]

I can see there are some higher ripple available like this KEMET PHH227MLP4110ME4 but not only. Coefficient is between 0.2 ~ 0.3 for 100hz to 200Hz. Ripple will fall between 5A to 9A. I guess it would be far enough for most filtering circuits.

 

[GrisPato]

I can see there are some higher ripple available like this KEMET PHH227MLP4110ME4 but not only. Coefficient is between 0.2 ~ 0.3 for 100hz to 200Hz. Ripple will fall between 5A to 9A. I guess it would be far enough for most filtering circuits.

At least there is a better ripple current rating although I can not see a place in an audio amplifier where you would need this at this capacitance.
Interesting is that this Kemet has leakage on par with normal electrolytic capacitors.
So I went searching more, and there are indeed some hybrid polymer that have the same or slightly better ripple current with little more leakage ( 20-30% ). Still would have to be a solid argument that you really need the 50-60 mOhm lower ESR to justify the cost.