2. tech-talk
  3. 130415-99123

Mark Levinson has odd choice for cap upgrades in one of their service bulletins. Why?

Pursuant to one of the Mark Levinson service bulletins, for Model 331, 332, 333 etc, they have outlined one of many things that should be performed on any amp that comes in for service.

One of the line items has me baffled. They recommend replacing four caps on each voltage gain / input board. These eight caps are PP type of .01uF @ 160v. They recommend replacing them with Ceramic X7R .01uF @ 200v. These are axial configurations.

This is not an expensive upgrade but I thought to myself that polypropylene caps had low failure rates and good longevity compared to other types especially if they were used in a proper operating envelope.

I just finished watching and reading some information on the perils of using ceramic caps in certain applications. For one, they tend to drift heavily with temperature changes. In a monster like the Model 333 there will definitely be a large temperature swing. The ceramics also tend to exhibit piezo effects with vibration. While vibration is only inducing small voltages, I can imagine the sum of many caps being subject to vibration not being a good recipe for an audio signal.

ML has stated that the ceramic replacements should be installed with spacers to keep them lifted from the circuit board. I am guessing that this could address temperature concerns, vibration or parasitic capacitance issues. They do not provide any reason.

I would really like to learn a little more behind their reasoning as it seems this particular "upgrade" is counter-intuitive. Can anyone shed some light on this?
generatorlabs
georgehifi's avatar
georgehifi
10,312 posts
 
generatorlabs
  Mark Levinson recommend replacing them with Ceramic X7R .01uF @ 200v

The XR7's are great for HF filtering, I use them on dac I/V stage voltage rail decoupling.That's why ML are recommending them.

Cheers George 
generatorlabs OP
25 posts
 
@imhififan Some pp cap has >50% voltage derating when operating temperature over 85 ºC , that could be the possible reason Mark Levinson recommend replacing them with Ceramic X7R .01uF @ 200v
I could see that. So you prompted me to look up derating tables for Ceramic. Got a little lesson on cap classifications as well. Here are some quotes from TDK:

A Class I capacitor* (C0G, C0H, C0K, etc.) is made from ceramic materials that are not sensitive to temperature changes, thus the capacitance value of a capacitor measured at a low temperature (example -25°C) will not significantly vary from the same capacitor measured at a higher temperature (ex. 75°C).
EIA refers to these as “temperature compensating,” and are measured in ppm/°C.
These capacitors typically have low capacitance values because of the ceramic material used to manufacture them but they exhibit nearly perfect capacitance stability regardless of their temperature, making them an excellent choice for applications in which frequency control is needed such as in radio or television tuners.
A Class II capacitor2 (X7R, Y5V, Z5U) is made from ceramic materials that are derived from a barium titanate base which is temperature sensitive. Thus the various temperature classifications stating the extent of the sensitivity over a given temperature range. These capacitors allow for larger capacitance values in the small surface mount packages.
Additionally, capacitors made from barium titanate bases (Class II, III, and IV) are ferroelectric and therefore susceptible to “aging” in which a capacitor’s ability to hold capacitance will decrease over time if left in an unheated and/or uncharged state. Class I capacitors are not ferroelectric and therefore do not age.

So I am not trying to beat a dead horse here. If we are going to dialogue about this then lets do just that. Today I learned things that I did not know yesterday and that is always a good thing.

I tried to see if the caps I need could be found with a C0G designation but caps of such small capacitance were not available with that designation.

I started to prep my monster for the eminent surgery. I still have the option of going PP or ceramic. I still also have an open mind about it. There is no real rush. So you know when you go to a fancy restaurant starving and they bring out the entree and it is a smaller than your appetite? That’s how I feel when I look at the cap comparison in this photo:

https://drive.google.com/file/d/0B0yvjhlC62G7aGVRZzYzLTVaUEU/view?usp=sharing

It just looks like something is missing :)
I bought 15 of these Vishay ceramics X7R from Mouser. They were really inexpensive.
Mouser also has a 250v PP Vishay cap that will work which is about $1.22 each.

Here is what will really get your grey matter baking....There are identical PP caps sprinkled on the current gain board yet ML did not make those cap changes mandatory. The current gain board is the heat generating furnace yet they don’t seem concerned that those PP caps are steadily being warmed. They only focused on the voltage gain board. Things that make you go hmmmmmm.

imhififan's avatar
imhififan
2,213 posts
 
I tried to see if the caps I need could be found with a C0G designation but caps of such small capacitance were not available with that designation.
I found some radial C0G
http://www.mouser.com/Passive-Components/Capacitors/Ceramic-Capacitors/_/N-5g8m?P=1z0wpzoZ1z0wgtrZ1z...


imhififan's avatar
imhififan
2,213 posts
 
There are identical PP caps sprinkled on the current gain board yet ML did not make those cap changes mandatory. The current gain board is the heat generating furnace yet they don’t seem concerned that those PP caps are steadily being warmed. They only focused on the voltage gain board.
Due to voltage rating of those caps? Are they working on same voltage or the caps working at lower voltage on current gain board?


erik_squires
16,398 posts
 
By the way, important to note that thermal stability ant temperature/longevity are not the same thing.

You can have a very stable cap that has a low thermal rating.

Also, who cares. These are RF filtering bypass caps. You could use 1 or 4 and the audible effects would be 0.  A cap with 20% variance in value is just fine here.
More to discover