High Voltage Supply Filtering

[Robert F]

choke mod- sorry for the double post

Whoops inadvertantly hit send before finishing my post, sorry.

Now George think AC... ohms law works for DC as I have explained but with imperfect components you have reactive components in the mix as I have (tried to) explain. Maybe my theory is wrong who knows but I know what I hear.

Since you have Acoustats give it a go.

Rob.

 

[moray james]

neglected point

Hi Rob: thanks in public here for bringing this sweet mod to my attention.
There comes a point where the science inside of the box does not answer all the questions that sometimes arise. In the time that it takes to argue that something like the choke mod cannot work it could be tried out to determine if in fact there is something to it. I think that there has to be some level of trust on forums such as these. If someone is sure that an improvement has been made then if that mod flies in the face of science it is probably worth while to give it a try to at least see what is going on. For those with the ability to model then this becomes a great opertunity to perhaps make a step forward in real understanding.
Tinkering around dropped the noise of the high voltage supply by 20 db which is a bonus in my book. Some say that this is a non issue as the supply is already quiet. Well I can hear the effect of minus 20 db can anybody else? Further I do not think that many if any will argue that ESL's sound better when the HV supply is disconnected or turned off. If the supply were perfectly quiet then that would not be the case.
The choke mod works because it damps the modulation of the diaphragm (an AC component) caused by stored charge in the stators dielectric. The big issue is being missed. Most all thought that this amount of diaphragm modulation from the charged stator dielectric was too small to be an issue. However when damped by a choke the improvement is clearly audible.
I was able to experimeent yesterday with the 7 Henry chokes on my buddies 0ne plus 0ne's. I placed both chokes in series to form a 14 Henry choke and ran both panels off the one supply with the 14 Henry choke. There was a small but noticible improvement in the sound. I should think that a composite choke built up from a series of one Henry chokes will yield a wide band low capacitance choke which will sound even better. I plan to try this idea out.
The real issue here is that perhaps we want to build panels with non insulated stator wires or non insulated perf metal. Only in this way will the issue of stator dielectric induced diaphragm modulation be a non issue. This perhaps confirms why Hunt, Janszen and Strickland all believed in a "leaky" dielectric.
I hope that others who have access to some chokes will try this simple and easy mod on thier ESL panel supplies. I am sure that you will be surprised and impressed with the impact of the choke on the high voltage supply. Further if you post here to recount your findings then others who do not believe that this mod really does anything will perhaps take a second look at it. Thanks for reading I hope that some of you try this and report back. Best regards Moray James.

 

[I_Forgot]

I apologize to anyone who feels my comments were "mean-spirited". My intentions were strictly benevolent. I was hoping to steer people's efforts in a direction that might make a more meaningful improvement in ESL performance. Everyone is free to choose where they direct their energy. The time has come for me to redirect mine.

My signature file says all I have left to say about any of this.

Good luck to you all.

I_Forgot

 

[etalon90]

Hi

I re-try the choke mods. This time I do it correctly and I use 30Henry choke (hammond 157G with about 600 ohms DCR)

I didn' t do comparaison, I only do the mods and listen my system.

From my audio memory (i didn' t listen my system since the last 4 days ), it seem to have a difference. the more difference are at high SPL, the bass is more precise and the medium seem to be more comprehensible. I didn' t hear change in treble.

So, I recomment to give the Morray and Robert mods. It take less than five minutes to do and cost next to nothigh (as compare to commercial tweak as main cable)


Regards

 

[moray james]

Different chokes

Martin: thanks for listening again and also for posting. From what Robert tells me different chokes can make a huge difference. Wide bandwidth low internal capacitance chokes work best. I recently found a surplus shop with one henry chokes at one inch diameter and 3/8 inch thick. A series string of these with each choke twisted 90 degrees to the next should make a fine very low capacitance choke. They are only $1.49 US each. Let us know if you try out any other chokes. Thanks again best regards Moray James.

 

[moray james]

Additional input

It has just recently been suggested that the impact on the ESL panel when fitted with a choke in the supply is as a result of a resonant circuit being formed by the choke and the capacitance of the panel.
Rob F.'s comments are as follows Well he is certainly wrong about the resonant peak, since the 500meg R
will damp any resonances big time, in fact a 1K resistor would likely be
more than sufficient to do so. In CLC type tube amp power supplies a low
DCR choke can have exactly this resonant effect. I have played lots of
times with different values on Duncan's power supply design program and
higher DCR
or larger capacitance damps any tendency to resonate.
Any comments or ideas out there? What ever the case I just can not see anybody removeing this mod after they install it, it's a keeper in my book. I look forward to some further discussion. Best regards Moray James.

 

[etalon90]

choke again

HI morray

I listened my system for long time yesterday and try several HV bias scheme.

for the choke, I try 3H,7H,30H,150H. (all cheapos hammond iron)

the 150 henry give the best result and it perform even better when used without load resistor at all. (Just an CL)

But after played with choke, I just switch to 25Meg load resistor alone to see wich improvement I gain with the choke...

WOW. the choke give a nice effect (coloration?) on the voice and give some "sparkle" on trebble (not the extreme treble) but scrap the bass. In overall, the resistor sound is better.

With 5Meg resistor, I nearly reach the voice of the choke but both extreme are not good. 25Meg seem to be the best compromise here sisce it`s a small improvement over my previous 33Meg.

With choke, is there a audible distortion (charge migration?) in very low bass (x<40hz) and I can`t accept this... The resistor didn`t exibit this problem.


Next time, I`ll try to mix resistor and choke to find the best compromise.
I bet that 10Meg+40henry will be a good start.

Keep posted.

 

[etalon90]

UPDATE

UPDATE

I try several compinaison of choke+resistor this afternoon with mitiged result.

Then, I suddenly remember that I have a pair of gridchoke (with nikel lamination) design for audio in my inventory...
they are 4000 Henry and can`t take any current but I know they are fine for few mA. (I accidently feed about 43mA for 2 minutes without any damage)

So I hooked them on and, and they provide the best overall sound.
that give the beautiful medium of a small choke, the trebble is just fine and the bass didn`t suffer any audible distortion (to me).

I think that 4000H will be very very hard to beat.
The secret is possibly the wide bandwith of an audio product instead of a power supply iron.

Use audio choke and give us some news

Thanks again Morray and Robert for the idea, I`m glad to have improved my system again...

best wishes

PS the choke are Magenquest BCP16Ni at US$150 per pair. MQ have also a M6 gridchoke that give about 1000Henry and cost about US$100 per pair.

 

[moray james]

Choke Quality

Rob F. has told me time and again that the quality of the choke is very important. High quality wide bandwidth chokes will have verylow internal capacitance. This is achieved useing multiple chamber windings to form the choke. The chokes That Rob uses are six chamber or six section 10 Henry choke. I will be building up some ten section ten Henry chokes built up from ten one Henry chokes that will be oriented such that each choke is at right angles to the next. Will post what I hear as soon as the chokes arrive and I get the chance to try them out. Best regards Moray James.

 

[moray james]

Ron has been kind enough to contribute to this discussion and as I mentioned before believes that the change in the speakers response/performance can be best explained as a result of a resonent circuit being formed by the inductance of the choke and the capacitance of the speaker panel. I have included Ron's comments in hopes of generating further interest and experimentation by others interested. Best regards Moray James.

Hi Moray:

Those 500 Meg ohm resistors are too high to have any effect on the Q of a tuned circuit. Even 10 to 20 Megohms is pretty high. We could do a circuit simulation to show what effect the resistors would have but it maybe just as easy to connect the componets and measure the result.

The resonant frequency of a tuned circuit is equal too:

Fr = (2 * Pi)/(LC)^1/2

Where Fr = the resonant frequency
Pi = 3.1416
L = the circuit inductance
C = the capacity

The term (LC)^1/2 power is actually the square root of the (LC) product.

Hope this helps.

Ron

 

[moray james]

Correction

Hi Moray:

The equation in my last e-mail for the resonant frequency is wrong. The following is the correct one.


Fr= 1/((2 * Pi)*(L * C)^1/2)

Sorry about the mix up.

The Q of a tuned circuit is a function of the resistance. Generally it is the resistance of the chokes winding that controls the Q. Adding and external resistor will do the same thing. With that in mind it maybe that the bias resistance of 500 Megohms or the smaller 10 to 20 Megohms may make the circuit a low Q. In effect dampening out the resonant rise of the tuned elements.

That sort of contradicts the statement in my last e-mail but it is a possibility. I would try the choke without the resistor to see what happens. Then add the resistor to see if it changes the Q of the circuit.

Let me know what you find out.

Ron

 

[Brian Beck]

Moray et al,

I wonder if the capacitance across a single large resistor such as your 500MEG might be a factor. Perhaps it provides a leakage path for higher frequency garbage from the high tension P.S. (in effect forming a capacitor divider with the ESL) You might try replacing a single 500MEG resistor with a series string of, say, 22 separate 22MEG resistors, kept spaced apart or oriented in a straight line. Or to replace a single 22MEG resistor, try ten 2.2MEG resistors in series. This will break up capacitances into series elements with much lower total effect. I'd be curious if you hear a similar improvement to the inductor. Just a hunch. I respect what you're hearing, but the prior explanations seem to be grasping at straws to me.

 

[moray james]

You bet

Brian: you are right about the capacitance of the noninductive high Meg ohm resistor. Your solution is exactly what I plan to do. That said the capacitance of a excellent 6 chamber choke like Rob uses or even the ten stage 10 Henry choke that I will be building up (a composite choke made from ten one Henry chokes in series) still will both have more internal capacitance than the noninductive load resistor. So that begs the question what the heck is that choke doing??? While I too thought that the idea of the choke damping out an AC modulation of the diaphragm caused by the capacitance stored in the stators dielectric was a bit of a long shot I find myself wondering if that may in fact just be the real deal. I would love for someone who has practical experience with supplies like this to give us all the definitive answer. I'am all ears and waiting for a better theory to be put forward. All the best for now Moray James.

 

[djmiddelkoop]

Morey, this is all very interesting.

As said before, I also can clearly hear the difference between a 22Mohm resistor and a 150Mohm between the HV and the mylar. However I changed the 22M to 44M to get a tighter low end response.

We must consider the role of the choke together with the capacitance of the speaker itself.
Also, the type of HV powersupply is important.
General there are 2 types:
boosting or cascading the mains frequency to the required HV
or a switching powersupply working at a much higher freq like 20kHz as in some of the Acoustats.

In the last the choke together with the capacitance of the speaker will act as a huge filter to filter out any ripple.

Maybe it's important to mention the type of powersupply your friend has. Just to get a complete picture.

 

[moray james]

Supply Type

DJ: I was not aware that Acoustat built any speakers with switching supplies. Do you know which modles? The 0ne plus 0ne speaakers that my friend has are cascade supplies.
I was able to get the composite ten chamber chokes built up and cryo treated. These are made up of ten one Henry chokes which were surplus. The chokes are one inch diametre and 3/8 inch thick, they are in series spaced apart and each is at 90 degrees to the next. We were running two Hammond seven Henry chokes in series for a total of fourteen Heneries. The new composite chokes at ten Heneries sound better than do the Hammonds.
The unaccepted theory is that the choke damps diaphragm motion (back EMF) that is the result of charge build up in the stator wire dielectric which then modulates the diaphragm. I think that this makes sense since there is no current to speak of coming out of the supply (steady) except when the diaphragm bleeds off some and the supply has to top the diaphragm up. Perhaps you can give this modification a try and let us know what you find. Best regasrds Moray James.

 

[djmiddelkoop]

The switching powersupply can be found in the Spectra models 11 and 1100.

If I correctly understand Moray, you use the choke without any series resistor, and before the choke there is a cap to ground.

In general, how do we use a choke?

1 for reducing ripple , like in a CLC powersupply
2 for smoothing out current , like in LC powersupply
3 as a frequency depended component , like in a speaker crossover filter

Where can we make the link to the results of the choke experiments ?

Number 1 does shure work here due to the high value of the choke :it’s equivalent to a high value series resistor for AC, or ripple voltage. So ripple and mains garbage will be reduced.
Number 2 also works here due to the high value of the choke for AC : it acts as a constant charge to the panel as it’s equivalent to a high value resistor for AC. For DC or very low frequencies <5Hz, the choke is a low value resistor, the value can be measured with your ohm meter.
So it will keep the speaker constant charged. If there is some leakage the charge voltage of the speaker will not drop as with a resistor. Here the low value DC resistance of the choke applies.
Number 3 doesn’t apply here, because of the high value it works at the whole audio spectrum the same.

This explaines why a larger choke sounds better, as larger means it’s equivalent to a higher resistor.
The multi chamber choke works better because of the much lower leak capacitance. This can be seen as a leak C parallel to the choke which wil make it less perfect for AC.

Don’t forget that if the HV is raised most ESL sound more dynamic and better.
This will be also the case with the choke as there is almost no voltage drop, as explained in 2.
Remember that every ESL has some leakage which will cause the bias voltage to drop when a resistor is used.

The back EMF theory can only work for very low frequencies as here the large choke will also act as a big value resistor.

Hope this helps.
D.J.

 

[Robert F]

dj Good theory except of course in both my application and in Moray James' the choke follows the high value resistor, in my case 500M! So I am confident in saying low DCR has absolutely nothing to do with the effect we are hearing. Ter is considerable voltage drop over the 500M R and extra DCR wont help. I suspect (know) that the 500 M is not a perfect resistor i.e it is reactive - it has considerable capacitive reactance. So maybe the chokes provide more isolation at high frequencies. That and the fact the panels are drawing current with leakage induced by movement of the diaphragm coupled with the fact that chokes store a charge and therefore supply the current draw are my best guess at the mechanism here... Could be wrong of course!

Regards,
Rob.

 

[djmiddelkoop]

Thanks Rob,
I completely missed the resistor that precedes the choke.
In this case my explanation doesn’t make sense, except for the back EMF.

I can assure you that the capacitive reactance the 500M or any high Meg resistor has, plays no role here. I have quite a lot of these kind of resistors and we are talking picofarads at the most here.

However, I can follow your :

the fact the panels are drawing current with leakage induced by movement of the diaphragm coupled with the fact that chokes store a charge and therefore supply the current draw are my best guess at the mechanism here...

This does make sense and could very well explain the experiments.

DJ

 

[Brian Beck]

I can assure you that the capacitive reactance the 500M or any high Meg resistor has, plays no role here. I have quite a lot of these kind of resistors and we are talking picofarads at the most here.

I must respectfully disagree here. The high voltage side (secondary circuit) of an ESL is an extremely high impedance environment, possibly higher than any other we are accustomed to. Voltages are very high, currents are miniscule and we’re talking about tens or even hundreds of mega-ohms either in the series resistor or on the diaphragm itself. Therefore picofarads DO matter.

If you happened to have 2pF of shunt capacitance across the large value supply resistor, there would be about 4 mega-ohms of capacitive reactance across that resistor at the upper end of the audio band. Whether that resistor is 20M, 100M or 500M, the admittance of that shunt cap swamps the conductance of this resistor. Now, the shunt C may even be larger than 2pF, and I’ve also ignored stray C to the nearby chassis.

My struggle with the choke idea (and I’m not discounting people’s listening impressions one bit) is that a perfect 10 Henry choke in series with said resistor will only exhibit 1.25M of reactive impedance at 20KHz, and lower below that frequency, even if it were a perfect inductor (no shunt C in itself). At 20 KHz the combined complex impedance of a perfect 20M resistor and a perfect 10H choke has an impedance magnitude of 20.04M, almost imperceptibly more than the 20M resistor. Let’s try a 10H in series with a 500M resistor – that gives a magnitude of impedance of 500.0015M. See my point? The magnetically-stored choke energy must be irrelevant since its only effect is this miniscule increase in impedance of the series combination. In either case, the resistor totally swamps the choke in series with it. Now let’s consider a real world choke, even a multi-segment design. It will exhibit many picofarads of shunt C that will drop its 1.25 M reactance to something much lower, making the case even stronger that the resistor will dominate. I don’t what listeners are hearing when they make the choke change, but I’m betting against the series inductance and more inclined to think that shunt C’s matter, although I’m not convinced of any argument at this point.

 

[djmiddelkoop]

Thanks for your thoughts Brian,
indeed you're right, because of the high impedance environment the shunt capacitance cannot be overlooked.
Also your interpretation shows that we are looking at the wrong explanation of what we are hearing.

BTW, I measured the capacitance of some 100M resistors and my meter reads 3pF.
I don't know if this is because of the shunt resistance or the reactance, but it's more than I thought it would be.

Now if powersupply noise is an issue here why do we hear a considerable improvement when the series resistor is made smaller ?
I'm still thinking it has more to do with energy and/or impedance.

I also do not want to discount people’s listening impressions one bit, hey I here it too.
The goal here is to find out what is going on, why are we hearing these effects ?

DJ.