Coupling capacitor values for MK-IV build.

Greetings, all.
This is my first post to this forum but I have been following it with great interest.I have a MK-IV kit, on order from Dynakitparts, which will power a JBL 2245H 18" subwoofer.
I have just completed a Dynaclone ST-35 with Mundorf Silver-in-oil coupling caps and am very pleased with the result. I am going to use the same output caps on the MK-IV. I have read, however, that the MK-IV bass response (same as ST-70 ) can be improved by using .47 MFD caps instead of the .1 MFD specified. Anyone experienced with this modification? Can it effect other performance parameters?
Great forum you folks have here.
Thanks so much.
Stevedel

If you know the load ahead {usually a resistor going to ground} of the capacitor then you can figure the lowest value in MFD that you can use. I found the below calculator. Remember you will be down -3db at the hz you put in. To be flat you need to put in one octave lower. If you want flat to 20 hz put in 10 hz. My SCA has 470k resistors after the cap. so a .03 mfd is all that is needed. The .1 mfd that Dynaco installed were more than enough to get full response. If I remember correctly my Stereo 70 has a 270,000 so I need at least a .06 mfd again Dynaco installed a .1 mfd. I use to use way oversized Caps. now I think a little over is better. The only exception that I can think of is preamps because the designer does not know if the amp. you hook it to has an input impedance of 20k or 400k. So he has to put in a cap for the 20K.

http://www.v-cap.com/coupling-capacitor-calculator.php

If I didn't make this clear please let me know.

I just calculated for the Dyanco Mark 4. The .1 caps with the 270,000 load resistors will take you flat down to 12 HZ. So .1 should be fine .2 will lower it to 6 HZ.

Thanks guys.
I guess that there is no real reason to second guess the MK-IV designers in this instance. I will stick with the original .10 MFD value.
I'll figure out something else to change. Maybe to super-duper low noise resistors, or some such thing.
Any suggestions are more than welcomed.
Regards, Stevedel

sailor wrote:I just calculated for the Dyanco Mark 4. The .1 caps with the 270,000 load resistors will take you flat down to 12 HZ. So .1 should be fine .2 will lower it to 6 HZ.

You might want to double check your math on this one.

The formula for the corner frequency (-3 db point) is:

159155/(resistance in Ohms * cap value in uf)

The "magic number" 159155 is derived from 1000000 / (2 * pi). The 1000000 eliminates the need to state capacitance in farads, we can use the much more common microfarad (uf).

The resistor is the usually the grid resistor of the next stage (if we are talking about output tube coupling caps).

So if you plug in the numbers you get 159155/(270,000 * .1), which is 5.89 Hz. Keep in mind that's the point at which the frequency response is down 3db - so there is some drop at frequencies immediately above our calculated number. So if indeed the corner frequency had been 12Hz there would likely be some loss of deep bass.

I went to the referenced V-Cap site to see their calculator for cap value. My advice is take the idea of a corner frequency that is 1/10 of the lowest frequency of interest (as they call for) with a large grain of salt. A corner frequency too low has some problems just as a frequency that's too high has some. But if you want the lowest frequency of interest to be 20 Hz, then the V-Cap recommended low frequency corner would be 2 Hz! That's quite problematic IMHO for a number of reasons.

My rule of thumb is the low frequency corner should be 1.5 to 2 octaves below the lowest frequency of interest. In the Mk IV it is 6 Hz - just a bit over 1.5 octaves below 20 Hz. In a Harman-Kardon Citation II it's 7.2 Hz - again about 1.5 octaves below the lowest frequency of interest (20 Hz).

One reason Bob Latino might not have heard a difference when he changed coupling cap values is due to the global negative feedback used in the Dynas/VTA amps. The feedback tends to minimize differences such as those. If the amp was run with no feedback it's likely Bob could have heard a very minor difference - but the amp uses NFB, so he didn't. As well, the stock 6 Hz corner frequency is plenty low enough.

Follow Bob's excellent advice - stick with the coupling cap values he recommended. Too LARGE a coupling cap can cause at least as many problems as it cures - for instance, look up "blocking distortion" when you get a chance...

Jim, I never said 12 HZ. was the -3db frequency. I took the number you came up with 5.89hz and moved it up one octave and rounded it up to 12 Hz. The exact answer is 11.78 HZ a .2 cap would be 5.89 for flat response.
Why make this more difficult than it needs to be. A cap blocks AC the same way if it is in the circuit of an amp or in a speaker. No one would ever say I need my tweeter to roll off at a certain frequency which say requires a 4 mfd cap but I think it would sound better if I changed it to a 8 MFD cap.
I think most people, including myself up until a year ago, have installed larger capacitors than necessary because most people don't know what happens when you increase the MFD value of a cap. After all bigger must be better.
The site above makes it simple to understand and easy to figure for everyone.
I also use calculators on the internet to build crossovers, to figure Ohms law and a dozen other kinds of electronic problems.

Ok, I used mfd instead of uf. It's the same thing. MFD is the industrial symbol for microfarad.

sailor wrote:Jim, I never said 12 HZ. was the -3db frequency. I took the number you came up with 5.89hz and moved it up one octave and rounded it up to 12 Hz. The exact answer is 11.78 HZ a .2 cap would be 5.89 for flat response.

Thank you for clarifying - it wasn't clear from your prior post (or I simply misunderstood your statement), so I wanted to clarify it in mine.

Also - a single C-R section rolls off the signal at a rate of -6db per octave, so if the "flat" frequency - the beginning of the rolloff - was 12 Hz, then the signal would be down 6db (not 3db) at 6 Hz, 6 Hz being one octave below 12Hz. If the rolloff started at 12Hz the -3db point would be about 8.5 Hz.

Why make this more difficult than it needs to be.

If I may... why be so defensive? I was only trying to help by sharing some useful info. I'm sorry if you felt I was calling you out for some reason - that was certainly not my intent. I just wanted to share some things I hoped would be interesting and worthwhile with all the forum readers.

But it IS more complex than it first appears. For instance, what if the original designer used a smaller cap value to limit the low frequency energy to an output trafo that can't handle a lot of deep bass? Or in an amp with more than two stages, maybe the amp won't be stable unless the two poles (another term for corner frequency) are staggered a good ways apart in their -3db frequencies? One rule of amp design is that the farther the multiple poles are apart the more stable (all else being equal) the amp. Often that requires some compromise to get adequate spacing to ensure stability.

A cap blocks AC the same way if it is in the circuit of an amp or in a speaker.

I think you meant DC, not AC, since a cap will pass an AC signal but block DC.

In any case, as long as the cap is part of an identical network you've pretty much got it. There are some differences due to the inductive nature of the load seen by the C-R circuit in a speaker crossover network, but for our purposes here we can ignore that for now.

No one would ever say I need my tweeter to roll off at a certain frequency which say requires a 4 mfd cap but I think it would sound better if I changed it to a 8 MFD cap.

Actually that sort of thing happens all the time - not only the frequency but the rate of the rolloff has to be carefully considered when you work with a tweeter. Remember, a 1st order (single C-R section) high pass filter only rolls off at 6db per octave. So the tweeter will see significant energy a full octave lower than the frequency at which the rolloff begins. The tweeter may be fine at 3K - but an octave lower (at 1.5K) it could have performance and/or durability problems. There are also phase issues involved.

I think most people, including myself up until a year ago, have installed larger capacitors than necessary because most people don't know what happens when you increase the MFD value of a cap. After all bigger must be better.

There was a time I did too!

The site above makes it simple to understand and easy to figure for everyone.

Yes, the site calculator does the math for you very well. But the recommendation for determining the low frequency corner (1/10 the lowest frequency of interest) might well cause trouble in a variety of ways. So use the calculator, but I would recommend you not use the "1/10th" recommendation.

I also use calculators on the internet to build crossovers, to figure Ohms law and a dozen other kinds of electronic problems.

That's great, the web is FULL of useful tools like that. But use the results carefully to get the best results.

Incidentally, if you want to read a NICE piece explaining blocking distortion, amp designer Randall Aiken has a great page on it here:

Blocking Distortion Explained

sailor wrote:Ok, I used mfd instead of uf. It's the same thing. MFD is the industrial symbol for microfarad.

It is one of a number of abbreviations for microfarad, but again just FYI - it is often considered obsolete. Kind of like the term micro-microfarad (abbreviated mmf) is now picofarad (pf), the abbreviation now is the Greek letter mu followed by a capital F. But on the web it can be difficult to insert the symbol for mu, so the "uf" or "uF" abbreviations have kind of become de facto standards.

IIRC, the reason for the change was that MFD was sometimes misinterpreted as millifarads (1 millifarad = 1,000 microfarads). But your post was clear, no problem knowing what value you meant.

When I was in high school taking electronics in a vocational program my teacher wouldn't accept mf, MF, mF, etc... It had to be µF. How to go about it on a computer? Press and hold the alt key then type in 230, followed by capital F. Tony

Tony, thanks a lot for the tip!!

What is the input impedance of a MK-IV? I thought it was 470,000 Ohms.
Forget what I said about accepting the designers specs.
I checked the V-cap website and I want to play.
Thanks all.

Another piece of information. An octave is one half or double a given frequency. Example, if your frequency is 40 HZ, to go down an octave is 20 HZ or one half, to go up an octave is 80 HZ or double.
I posted the V-cap site because of the calculators. It was meant to help the people on this board more easily understand capacitors. How they work and the minimum size necessary to do the job. Much simpler to use than trying to remember the formula. I also don't accept everything they wrote on the page but that was not why I posted the site. Also I have never bought a V-cap.
Jim, I don't disagree with what you have posted. Unfortunately when I write I tend assume I can leave out a few steps and people will follow along.
I know it is confusing and I skipped some steps of my thought process. I will try to explain. You are right, it is a 6db per octave roll off however at the corner or crossover frequency the signal will always be -3db down. That is not what people are trying to find. They want 0db down or flat. That can be found one octave up before the cap starts to roll off the AC. So I took the answer 5.89HZ which is -3db down and moved it up one octave to get the wanted flat response answer of 11.78HZ. I never read this anywhere I just figured it out on my own. I guess I should call this simple equation "Sailors octave up adjustment for flat response equation". LOL
Actually I did mean AC but blocking would be the wrong word. I should have said roll off. Yes, capacitors block DC but they also in essence can block [roll off] some AC.
As far as uf and UF goes, I learned MFD in my fathers TV repair shop 50 years ago [look at the caps in your old Dynaco equipment]. I will continue to use it even though it is outdated.

sailor wrote:You are right, it is a 6db per octave roll off however at the corner or crossover frequency the signal will always be -3db down.

That is the definition of corner frequency - 3db down.

That is not what people are trying to find. They want 0db down or flat. That can be found one octave up before the cap starts to roll off the AC. So I took the answer 5.89HZ which is -3db down and moved it up one octave to get the wanted flat response answer of 11.78HZ. I never read this anywhere I just figured it out on my own. I guess I should call this simple equation "Sailors octave up adjustment for flat response equation". LOL

Okay - but so you know, the -3db down point is where the voltage is about 70% (70.8% to be exact) of it's original value. Conversely, an increase of 3db is 1.413 times the original voltage value.

Also - I have a utility sheet I use that allows you to solve for capacitor value, resistor value or corner frequency. If people here want it I'll post a link to it, you can keep it on your hard drive.

What???? Jim you need to let go of this. In this case I know I'm correct. You only need to look at a graft of a 2 way first order high pass crossover to know that I am correct. At one octave up from the crossover point the signal is about 98% of flat. As far as your +3db math????? Where did I say anything about +3db. I said flat. There is no +3db hump anywhere???

sailor wrote:What???? Jim you need to let go of this.

I just want to make sure the information a reader takes a way from this is correct, that's all.

In this case I know I'm correct.

I'm sorry, you may have a "rule of thumb" that works for you, but in absolute terms you aren't correct about some of the questions/issues raised.

As far as your +3db math????? Where did I say anything about +3db. I said flat. There is no +3db hump anywhere???

Look - the FACTS are that decibels are ratios - and the facts are that when measuring amplitude ratios (voltages) the -3db point will be .708 times the reference number and the +3db point will be 1.413 times the reference number. In other words if the original number was 10 volts the -3db point will be 7.08 volts; and if the original number was 7.08 volts then the +3db point will be 1.413 times 7.08, or a hair over 10 volts (the difference is due to rounding).

Whenever you say a voltage or frequency (both measures of amplitude) is "-3db" you need to be able to determine "-3db from WHAT"? If the -3db point was 5.89 volts, it was down 3 db from a bit over 8 volts. That's the correct math. As I said, your rule of thumb is fine but the math needed clarification IMHO.

That's all I'm going to say.