diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Planars & Exotics (http://www.diyaudio.com/forums/planars-exotics/)
-   -   High Voltage Supply Filtering (http://www.diyaudio.com/forums/planars-exotics/57310-high-voltage-supply-filtering.html)

moray james 15th May 2005 07:12 AM

High Voltage Supply Filtering
 
Don't know how many of you out there will be interested but this may give cause for consideration. I just completed modifying a friends Acoustat one plus one's high voltage supplies. The mod that I did is docummented on the Izzy Wizzy Audio site (http://www.izzy-wizzy.com/audio/spkr.html). This discribes the addition of an extra high voltage filter cap on the HT supply multiplier section. This simple mod does everything stated and then some in my opinion. This mod can be applied to most similar multiplier supplies.
An additional mod/diognostic tool is the mod shared by Sheldon stokes several years ago. This involves a neon bulb which is bypassed (paralled) by a small value cap (to catch transients at lamp turn on). The combination lamp/cap is then placed in series with the output of the high voltage supply after the large megohm value load resistor. When the panel looses charge and draws upon the HT supply the lamp lights up to conduct the HT to the diaphragm. As soon as the diaphragm is fully charged the lamp goes out. The neon lamp presents infinate resistance to the supply under these condition and so effectively decouples the diaphragm from the supply. It is as if you had unplugged the supply and the speaker operates in constant charge mode. As soon as the charge on the diaphragm starts to disapate the lamp turns on and reconnects the supply to the diaphragm. In normal operation the neon lamp will flash on and off. The cool thing is if you should see the lamp on all the time then you know that there must be a steady drain on the diaphragm somewhere (probably dust or bugs between the resistive coating on the diaphragm and the stator on that side of the panel. A very cool diagnostic tool as well as a means to decouple as much as is possible from the supply while maintaining automatic charge management of the diaphragm. Thanks to Sheldon Stokes for that one. Best regards Moray James.

SY 15th May 2005 12:09 PM

For whatever reason, I can't seem to reach that page.

I really like a lot of the stuff that is on that site. I tried a variation of the rigid frame last year with much less success; I subbed concrete-filled PVC pipe for the copper. Bad idea.

moray james 17th May 2005 09:36 PM

Izzy Wizzy explaination
 
I_Forgot has suggested that he thinks that the Izzy Wizzy mod does what it does not because of additional filtering of the bias supply but because the panel (older Acoustat 0ne plus 0ne) is probably a bit leaky as far as bias supply goes. So if that is the case the the additional 0.01 uf cap stiffens the supply and makes it better able to get more voltage on to the panel. I think that this makes sense. So to find out I will remove the 0.01 uf cap and decrease the resistance value of the load resistor (500M) to say 250M. If voltage on the diaphragn is really the issue then this mod should have an identical result. Any thoughts or suggestions from the forum would be welcome.
I_Forgot has told me that 10 to 50M ohms is plenty to do the job and to keep the speaker working in constant charge mode. This just makes me want a variable supply all the more. Will keep you interested Acoustat owners advised of how this goes. It would be my guess that the reason Acoustat chose such a high value resistor for this job was probably a combination of reduceing current in the event of a shock and a good price from some vendor. Best regards Moray James.

bear 20th May 2005 04:57 PM

I use model 3s these days in a homebrew space frame.

The vibration (energy) is virtually all in one plane, so the stiffness needs really only to be very high perpendicular to the panels.

I suggest a mig welder and square tubing. Fill with sand or foam for damping as required.

I used threaded rod/carriage bolts to mount the cells a distance from the main frame... it is possible to adjust the angle of the outer cells. Be sure to use insulation where any metal goes through the cells for mounting, or you'll find things getting charged up... :D

You can apply felt treatment to the frame's surfaces if you are worried at all about reflections at HF.

I found the dumpsters at the local malls to be a treasure trove of premade shapes and after the application of a hacksaw, just stock of chromed & painted square and round steel tubing, btw... ;)

The main effect was to clean up the highs compared to the stock frames. Also the interfaces have long ago had the fuse holders removed and the signal caps replaced with all polypropylene (essential), the internal wiring upgraded and the BPs changed for real ones... the interface sits on the base of the frames at the rear...

_-_-bear

PS. the drop on a neon bulb is ~100v... I'd expect it to "bleep" fairly often while playing? And neon is also a bit noisy, maybe the cap helps with that? Maybe an OB2,3,4 tube?? :- )

moray james 20th May 2005 08:41 PM

Thanks for the reply Bear. You are righr about needing strong frames for the Acoustats. I have seen Two plus Two frames cracked right in half in the middle due to stress from the panels. I have also seen the lag bolts used to mount Acoustat panels pulled out of the Fir strapping they are mounted to by the force of the panels.
What do you think about the Izzy Wizzy mod to the high voltage supply? What do you think about I_Forgot's thoughts on what this mod is doing? I have the feeling that I_Forgot is right. I hope to have the time this weekend to check this out on the 0ne plus 0nes.
I have found that filling the outside edge louver cubes with Duct Seal helps deaden and damp the styrene louvre frame reduceing resonance and cleaning up the sound. It is easy to do , cheap and stays put. A layer of black felt over this row of cubes cleans up the look and makes handeling cleaner. Regards Moray James.

bear 21st May 2005 04:02 PM

Quote:

Originally posted by moray james
Thanks for the reply Bear. You are righr about needing strong frames for the Acoustats. I have seen Two plus Two frames cracked right in half in the middle due to stress from the panels. I have also seen the lag bolts used to mount Acoustat panels pulled out of the Fir strapping they are mounted to by the force of the panels.
What do you think about the Izzy Wizzy mod to the high voltage supply? What do you think about I_Forgot's thoughts on what this mod is doing? I have the feeling that I_Forgot is right. I hope to have the time this weekend to check this out on the 0ne plus 0nes.
I have found that filling the outside edge louver cubes with Duct Seal helps deaden and damp the styrene louvre frame reduceing resonance and cleaning up the sound. It is easy to do , cheap and stays put. A layer of black felt over this row of cubes cleans up the look and makes handeling cleaner. Regards Moray James.


I've not seen the cracked frames, or the bolts pulled out...

The mod is interesting, but I think not so important... the problem, iirc with the extra cap is that it pulls the voltage down, and doesn't really requlate or smooth very much... I like the neon idea best so far!

If ur going full bore and no-holds-barred why not just buy a regulated HV supply? They're small and often adjustable - sometimes you can scavenge them from copiers and printers... or buy them surplus. That solves the entire problem in one shot.

I like the idea of filling the outside cubes with goo... :)
I'll have to give that a try! Thanks for the the thought - good one!
Sil-Pruf silicone might work too... (no acetic acid release). Nice one.

_-_-bear

I_Forgot 22nd May 2005 04:18 AM

Acoustat bias supply ripple
 
I ran some simulations of the stock bias supply and the "izzy-wizzy" mod version. The Izzy-wizzy modification reduces ripple by about 20 dBV and somewhat increases the average DC output of the supply. 20 dBV sounds like a big reduction, but read on...

I put the results in an excel spreadsheet here:
Acoustat bias supply ripple I used excel because screen shots of the simulator output were just too big. I ran the simulation using SwitcherCad III (a free download from Linear Technology). I added the model of the 1N6533 rectifier to the library to run the simulation. The 1N6533 is a 5 kV, 50 mA rectifier that should be similar to the original parts in the Acoustat bias supply. The SwCadIII schematic file is here: Acoustat bias supply simulation file
Here is the spice netlist: Acoustat bias supply net list

The data includes the spectrum of the ripple so you can see what is going on.

With regards to audibility, I'd say a simple experiment is in order- place your ear close to the speaker with no music playing. If you hear buzzing, then the ripple makes a difference. If not, it doesn't. I have never heard any buzzing from any ESL, regardless of the bias supply configuration, so I'm inclined to think that the ripple really doesn't matter, at least at the levels that voltage multiplier circuits typically produce. The speaker is of balanced construction, so the ripple will cause the diaphragm to be pulled toward each stator, essentially ensuring that the diaphragm doesn't move due to the ripple.

Otherwise the average voltage out of the Izzy-Wizzy mod supply is a little higher, and the difference between the stock supply and izzy-wizzy supply increases as the speaker panel leakage current increases. That means a real leaky panel will show the largest improvement and one that is not leaky may show little or none by making the modification to the power supply. The difference in the output voltage is about 80V for a panel with 500 Mega Ohm leakage. I don't know if 80V is going to make an audible difference or not.

There should be no frequency dependent effects at all, higher voltage just means increased sensitivity (i.e. louder volume) at all frequencies.

I should note that the added capacitor in the bias supply makes servicing of the supply quite a bit more hazardous than the stock supply. If you do the mod, be careful- that extra cap storing the full output voltage (about 5200 V) will pack a real wallop. The fact that it has no bleeder resistor increases the hazard because charge on that cap may persist for weeks after turning the power off. Be careful!

I_F

moray james 22nd May 2005 09:29 AM

I_F: wow thanks for the detailed explaination, that's a wack of research. Much appreciated, I would have never have learned that on my own! Very interesting and well worth a folder on my hard drive. Do you think I should put a bleeder resistor on that (Izzy Wizzy) cap? If so what is a good value?
This weekend has turned out to be a bust with respect to getting over to my buddies to experiment with the cap mod and your suggestion of decreaseing the load resistor value. I will post back with findings as soon as I gan get over to do the work.
What do the simulations look like with a regulated supply? I would think that there would be a larger difference than that seen with the Izzy Wizzy mod? Thanks again I_F. Best regards Moray James.

405man 22nd May 2005 10:22 AM

The idea of the neon indicator is used in the Quad ESL63 where if you remove the grommet at the rear you can see the neon and check its flash rate

Stuart

I_Forgot 22nd May 2005 03:46 PM

Quote:

Originally posted by moray james
I_F: wow thanks for the detailed explaination, that's a wack of research. Much appreciated, I would have never have learned that on my own! Very interesting and well worth a folder on my hard drive. Do you think I should put a bleeder resistor on that (Izzy Wizzy) cap? If so what is a good value?
A bleeder would have to be a very large resistor (maybe 1000s of Mega Ohms) or it would pull down the voltage of the bias supply and cancel the effect of adding the cap. Such a large resistor would result in a very long time constant, so its value as a bleeder resistor would be questionable.

Quote:

[i]What do the simulations look like with a regulated supply? I would think that there would be a larger difference than that seen with the Izzy Wizzy mod? [/B]
There's regulated and then there's regulated. A regulated supply will maintain a constant average voltage, but the ripple might still be about the same as a charge pump multiplier, maybe even worse. When you look at the ripple in the stock supply, it's level is about 80-100 dB below the average output level. That's pretty good performance, even for a regulated supply. You can find the numbers in the excel file on the ripple spectra page- look at 0 Hz. That's the main output of the supply at DC (0 Hz). Now subtract the value of the 60 Hz output from that.

A regulator that could improve on that performance would be hard/expensive to make, and would not really add to the performance of the speaker, which is why no one does it.

I_F


All times are GMT. The time now is 12:01 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2