accoustat and direct drive. big text, small question about igbt

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi

No, not another direct drive thread!

No. I already direct drive my accoustat using one fixed bias 845 resitive loaded per stator. (i`m affraid of choke loaded to drive such a capacitive load)

Last month, I had the chance to hear a system that I was looking to hear for a while. This is a killer amp, And I mean a Killer direct drive amplifier.

The amplifier I heard use one 4-400 tetrode per stator. each tube is Resistive loaded on a resistor string and attached to a 2.5Kv power supply. Every tube are running at a stunning 250mA for about 312Watts dissipation per tube. The whole amp consume about 3Kw :eek:

Another advantage of this amplifier is that he use high gain tube to avoid using a lot of tube in the signal path.
So he use a single tube to provide the gain and the phase split. (a classic long tail circuit wit a SS ccs on the cathode. Then he feed a penthode Cf (with a SS CCS on bothom). From there he feed the 4-400 power stage... The whole amplifier is DC coupled he said!

That system was a revelation:
1- it prove me that we cas use a mix of SS device in a tube amplifier with very good as long as they don`t provide a voltage gain.
2- it remove my fear about CF and pentode (I had bad experience before with both but never tried pentode CF)

My amplifier work very well but with only 65Ma idle current, my slew rate is limited. I really realised it after heard my host amplifier.
And I don`t know if the DC coupled did a big difference but since I use 3 power supply in my amplifier, DC couple every stage is a matter of few tension divider between supply. I`ll do it too.

So, because I don`t want to buy 500$ output tube to rebuild my output stage using the same resistive loaded circuit and since I don`t 3Kw dissipation amplifier, my only option is to cheat a little.

My host suggest me 3 ways and said that might provide excellent result (technically better than my 845 R loaded)

1- use a SRPP stage with a pentode on top. That way, he also suggest to remplace the 845 power tube by a 811a or a 572B and enjoy very high gain. I`m not sure about the distortion of those device intendeed for A2 but I`ll save one or 2 gain stage... Anyway, I can trade my 845 for a 805 that seem more linear than the 572b in the operating range I look for. (the 805 give a gain of 40, 7 times more than a 845...) One big trick of the circuit is to use a resistor to bypass the gain tube and bleed the current so the penthode could run at high current while the lower triode may use only a part of the current.

2- about the same as in one but just remplace plate resistance of the lower tube of the SRPP by a big choke to have a very constant current in the lower tube. This way should provide enhanced performance over the 1st choice as long as I can find a 50 or more Henry choke that can sustain 2Kv... :bawling:

3- Still an SRPP but with a mosfet on top. this will provide the lower impedance and best overall performance (technicaly). The trick, like the previous choice, is to bypass the lower tube by a resistor to run the mosfet at high current.

Even if I`ll have to use big heatsink, I`m very interested in choice 3 but finding a power mosfet isn`t that easy. Fast simulation under dynamic condition showed me that I need at least 2.2Kv mosfet. More that 1Kv simply seem to not exist. The only product I find is IBGT (designed for switching purpose)

here the ibgt
http://www.ixys.com/Appasp/pdigbt12L.asp
the 19a250 cost about 65USD.

What do you think about such a device in a power follower application? What about linearity? they don`t provide max usable frequency...


Any comments/suggestion about a ibgt or another high voltage power device will be more than welcome.


Thanks
 
Problem with big igbt's or mosfets is large internal capacitances. It's not easy to drive a nf or 2..3 fast enough. So speed is a problem, yes.

More that 1Kv simply seem to not exist.

There are a few rated at 1.5 kV. Semelab used to make the bfc40 mosfet, 1.5kV/2A with only 550 pf input capacitance. But I don't see it listed on their site anymore so it's probably discontinued. 1.5 kV is not enough anyway I fear. You might consider stacking mosfets, dividing the voltage and sharing dissipation over several devices.
 
Maybe what you heard was the difference in output Z between your amp and his? That means that it matched the capacitance of the panels at a different point.

The 4-400 do not cost $500.
They range from free (broadcast pulls) to ~$125 or so for import new stock that will work fine.

There are other transmitting tube that will take voltage too... you need voltage mostly.

Consider an 813 over ur 845?

The 811/572 is a zero bias B2 tube, not A2

Dunno why you are afraid of chokes?

Many "direct drive" esl amp designs use two tubes, stacked, similar to what you see in that respect with an SRPP, but there are varied means to drive and bias the tubes. Also, you can always AC couple the output and use regular bias on the diaphragm...

_-_-bear :Pawprint:
 
C'est domage mon ami!

(about the best I can manage in French...)

Why give up??

The Acoustat cells are pretty good. I'd not sell them until I got an equivalently good DIY cell(s) up and running!

You should be aware that the sound you get from direct drive is still related to the "matching point" between the amplifier impedance and the reactive impedance of the capacitive cell(s) it is driving.

Same thing if you "match" with a transformer(s).

Do you NOT have the Acoustat interface box - model 121??


If you do, why not run them with the interface for a while...?
Build a big tube amp (add an output tranny to what you already have??) and see how you like that, it's pretty darn good.

Also, at least one person has sucessfully run ESLs off the plate line (capacitively coupled) of a standard tube amp that was lightly loaded on the secondary with a resistor!

A tube that is far smaller than the 4-400 would work, like a 4-125 or a 4-65 - again the key is voltage swing, not current. They're less expensive too... these tubes will handle in the range of 4kv on the plates. The 805 you mentioned and the 811/572b generally don't run that high...

Lacking the Acoustat interface, it is fairly simple to use a single transformer to match the Acoustat cells from ~200 Hz up. Just use a transformer of similar turns ratio, to the one Acoustat used for the "high frequency" transformer, block the LF with a cap at the input, and away you go...

_-_-bear :Pawprint:
 
DIY-Panel

Hi,

a tip for really really good...no..not good....excellent DIY-panels: ;)
http://www.capaciti-diy.com/
- excellently built
- with tight tolerances
- long time stable
- absolutely safe to handle
- good efficiency
- very good set of measurements
- very good sounding
- very priceworthy

Imho the best DIY- as well as professional panels You can put Your fingers on!

I had the opportunity to listen to the 80/23 in combination with a 8"-driver in CB, actively crossed over and driven dy D-amps.
It was fu***** brilliant :)
See the right half of their website under ´Produkte´

jauu
Calvin
 
HI
Do you NOT have the Acoustat interface box - model 121??

No, I had the medaillon interface but sold it on Ebay 2years ago.

For the interface, once you listen a direct drive amp, you don' t use interface anymore. Just never...

As for the swing to the panel, As I said, current and low impedance is the key.

When I compared my directdrive amplifier to a original acoustat-X amplifier, I realised that the high of the acoustat-x was lacking.
To swing 1kV into 1700Pf at 10Khz need a lot of current.
Their SRPP low current arrangement provide enough swing but lack ''guts''

My amplifier, event with 65Ma per stator lack current drive to play. Sure it does play well but not like I would like.
After hearing a power 4-400, I realised how my sparkle could be...

Because Build such a powerfull amplifier require too much power and money, the only remaining option is to reduce the load

(size of the panel and the capacity. One acoustat panel is about 425Pf. My amp will drive this easely without any mods.)

And with basshorn, I' ll gain on the dynamic side.


-------------------------

Calvin

THx, I'll retain your link

I was thinking about a custom panel by ER-audio
http://www.eraudio.com.au/

It will be very hard to choose between them
:bawling:
 
Why or Why not

Hi,

since I haven´t heard the ERs yet I just can compare technical aspects of them and the Capacitis.

ER / Capaciti
construction differences
- full range ESL-Kits / hybrid panels
- flat metal stators / thin wire
- 3 mechanically segmented panels / one electrically segmented panel
- standard tranny with1:90 / toroid with 1:130
- HV-cascade or fully regulated / HV-cascade on good PCB with timer and automatical switch on
- Kit to be end constructed / completed and tested panel + electronics
pros and cons (with regard to ER´s ESLIII construction manual)
- The hole sizes of the ERs seem fairly large to me. This could lead to lower efficiency and uneven force distribution on the diaphragm.
- The insulative coating doesn´t seem to be very thick (the holes are hardly ´rounded´) and this is a very problematic point with every sheet metal stator. Like in the Acoustats the Capacitis use a thickly insulated massive copperwire -but of smaller diameter of ~1mm- that is safe against sparking.
- 3 flat panels angled gives the venetian blind effect with distinct lobed distribution characteristic whereas bowing one panel like ML does or electrical segmentation leads to a smoother and more even character
- The ER has its 3 panels just wired in parallel -You might use some series resistor for frequency correction and segmentation though- whereas Capacitis are corrected for frequency response and distribution character by segmenting electrically into 3 or 5 strips.
- Even the smaller strip of the Er-panels is far too wide to give a good HF-distribution (70mm). Capacitis reach much better values that even come close to a good normal tweeter. You are not fixed to a sweet spot like with the ERs
- The build quality and the precision and consistancy of the capacitis is state of the art (tolerances of just 1/10mm in stator-flatness)
- The best way to built a ESL with good resolution and high dynamics is to build a hybrid. As a sideeffect it can be build much smaller and optically pleasing.
- the segmentation of the capacitis leads to a much easier to drive load even though the transformation factor is higher than the ERs
- small d/s of just 1mm gives good panel efficiency and dynamics
- measurements of the capacitis are very good for ESLs
- price is extremely affordable for that niveau of quality


I could hear the combination of the 80/23 panel with a Visaton TIW200, actively driven with D-amps by HiFi-Akademie as can be seen on the Website (right hand) under ´Produkte´. This fully active and quite compact hybrid played extremely well for the money and is very easy to build and looks good ;-)
I can recommend this combi very very much.

jauuu
Calvin
 
Re: Capaciti/ER comparison

Roger Sanders and other ESL designers avoid stator electrical segmentation due to phase shift, analogous to a multi-way crossover. Also, ER claim their ultra-thin stator provides superior low-level detail due to elimination of tunnel resonances in the holes.

Comments?
 
Comment

Hi,

if You can´t do good, give the bad thing a good name. :devilr:
Well, every top class ESL -wether beeing a marketed product or a DIY solution- takes care about the frequency- and distribution-character of the speaker.
This could be done several ways....mechanically, by bowing the stators like ML does, segmenting/bowing electrically like Quad, Audiostatic and Capaciti do and to cross over electrically to specialized panels with a passive crossover on the primary side.
The first two methodes are well known and built to the tenthousands without anyone reporting ´phase-probs´. If it would be so You should be able to see something in the frequency- and acoustic phase-plots or group delay plots. Neither ML-style nor Audiostatic-style nor Quad-style panels exhibit anything abnormal here. On the other hand electrically segmented panels show some positive aspects like broadened sweet spot, smoother frequency response and easier to drive impedance than just flat panels. If You can do better than just flat panels than do better! ;)

ER´s right with regard to stator thickness and tunnel effects.
BUT (!!)....
....this holds true for only the following condition(s):
- the length of the tunnel (i.e. thickness of the perforated sheet) has to be of a comparatively equal or greater size than the diameter of the holes
- the ´walls´ of the tunnels have to be straight

Things change, as soon as
- the walls of the tunnel are tapered. This can be achieved with deburring/electropolishing the sheet and using a thick layer of insulating coating
- You are using wire stators. The round shape of the wire doesn´t work as a bandpass-tunnel
- You work with usual dimensions of less than 1.5mm sheet thickness and holes greater than 3mm in diametre. There´s no problem because the upper end of the bandpass (the tunnel) is above the audible range.

Looking at the pics of the ESL3 construction manual You can see standard off the shelf metal sheets with quite big holes and great distances between the holes (relatively small opening percentage). Further, the white insulative coating seems to be thin, because the edges of the holes look very sharp and nearly not rounded.
This leads to following assumptions:
- use of sheet material that is not deburred, or electropolished to give rounded hole edges. These points are prone to flash overs, because of greatly reduced thickness of insulation and higher field stress. To overcome this prob You have to use a proportionally thicker insulation that would give a better rounded surface too, but reduces efficiency.
- lower overall field strength and a uneven field distribution compared to sheets with smaller and closer spaced holes or thin wire-stators.

In conclusion:
You have to use a thin sheet here to prevent the cheap construction from becoming rather useless for a good panel :whazzat:

ML shows with their new panels where to go. Relatively thin sheets with small holes positioned very close to each other. Electropolishing the surface to give rounded hole edges. This prevents for the field stress, allows for thinner and more even insulation and thus leads to higher efficiency and resolution/openness of sound. The bowing allows for easier designs without supporting structures to reach the needed stability and widens the sweet spot and improves optics.
The other way is to build stators with thin insulated wires and use electrical segmentation like capaciti does. So we have two winners here of which just one is a DIY-solution :D
Since these better panels are still cheaper than the ER-solutions my recommendation should be obvious

jauu
Calvin;)
 
Capaciti

Hi,

No, I´m not the Capaciti guy and I´m not envolved with him in any other form as having met him life once.
I´m building my own panels....but if I wouldn´t the Capacitis were the only ones I´d surely liked to have :D
Build and parts quality, tolerances, measurement values and sonics are on the highest possible level I´ve seen yet in app. 20 years of interest in ESLs and there are some very interesting features to come from them in the nearer future that might lead to a great step forward in ESL-technology.
You can contact him at http://www.capaciti-diy.com/

jauu
Calvin
 
Hi etalon 90,

I am the capaciti guy.

Unfortunately our website is just in german language, but we are working on the english site, since more and more people from scandinavia and the US are visiting our site.

It will take another 2 weeks for the first revision in english, which will be updated from then continiously.

BTW: I am very pleased that Calvin recommends positively about our products, because in germany he is considered to be most competent regarding ESL.

Regards,

Capaciti
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.