| Tube amp specifically for driving ESLs - Click HERE for Original Thread |
| bigwill |
Would it be difficult to design a tube amp that just outputs a high-RMS signal, with low distortion (no output transformer)? (I don't think so :))
If so, would it be nessecary charge the stators instead of the diaphragm, and drive the diaphragm instead to avoid having to produce two signals 180 degrees out of phase? |
|
|
| B Cullingford |
What is difficult?
Probably harder than a chip amp for sure.
Certainly potentially dangerous - I have repaired such a beast - the only audio power amp I've ever worked on that had a skull and cross bones warning symbol :bigeyes:
No you would not want to drive the diaphragm, charge migration ... unstable ... diaphragm sticks to sators ... arcing ... not enough space here to explain. May I humbly suggest reading up on some theory, maybe someone else will have some good links, just remember not all you read is correct! Walker's (Quad) patent(s) is/are a good read.
I've wanted to do this for years, but some how something always came first. And of course I'm not sure I want to modify my 63s, so that means building some panels first. |
|
|
| bigwill |
| I might have to get a book :) All I know is that I want to avoid an output transformer! |
|
|
| SY |
| It's a hard problem. And mistakes are spectacular, potentially dangerous, and always expensive. I've built a d-d ESL amp that is a scaled-down (900V) version of the 5kV monster that I really need (and which just awaits more available money, hah hah). But even that scaled-down one was a real ****** to stabilize. |
|
|
| bigwill |
| I definately don't have the experience to do this. I think I'll stick with a transformer :) |
|
|
| djmiddelkoop |
| quote: | | I definately don't have the experience to do this. I think I'll stick with a transformer |
Bigwill,
I've build one with a 4.8kV supply.
It's definately the most dangerous, read deadly, and tough tubeamp I ever built.
You must be very experienced with tubes and voltages above 1kV to make this happen, so like you said if you don't have the experience don't even think of trying.
What you can do is try to find an Acoustat X amp and replace the sand for tubes. Still very dangerous though.
Dick. |
|
|
| BillH |
| quote: | | Bigwill writes:Would it be difficult to design a tube amp that just outputs a high-RMS signal, with low distortion (no output transformer)? |
I've been wondering the same thing lately. I've seen a diy direct drive esl amplifier on the net. Can't find the link at the moment, will check for it later.
I would like to try building a push pull tube amp to drive a pair of diy esl's. I'm thinking it would need transformers, though. I'm not feeling comfortable working on an amp with a 4-5KV power supply.
Just a rough idea would be a pp amp with a +- 200v power supply with a 1:5 step up transformer. The transformer probably isn't an off-the-shelf part, so I've been searching the net for transformer winding info.
I'm quite rusty at tube circuit design, haven't worked with them in about 30 years, so the whole thing is just an idea at the moment. Sy-feel free to correct me if I'm going down the wrong path. |
|
|
| SY |
| The step-up may work, but again, the hard part will be stabilizing it- reactive loads are difficult. You will probably need more step-up than that, depending on your panels and bias voltage. |
|
|
| BillH |
| quote: | | The step-up may work, but again, the hard part will be stabilizing it- reactive loads are difficult. |
Would it need more stabilization than a traditional esl setup of a tube amp driving an 8 ohm step down transformer, in turn driving a voltage step up transformer to the esl's? |
|
|
| Brian Beck |
(I've copied a posting I made to Audio Asylum on this subject, especially about driving the Quads, here:)
It’s not easy and it is incredibly dangerous. But of you’re determined and experienced with the hazards of very high voltage electronics, it can be done. I built a direct drive amp for Quad ESL-63s (and with other ESL projects in mind) and in the process gained some insight into the unique drive requirements of the Quads. To drive them to anywhere near full output levels requires a lot of voltage, even more than for many other ESLs, because the Quad uses a fairly high impedance panel design for an ESL. This is evidenced by the step-up ratio of the pair of input transformers which is about 1:170 according to my measurements (at 1 KHz with the transformers in circuit). Most other ESL step-ups are in the 1:50 to 1:100 range. The bias voltage of 5.25KV is also on the high side. The 63 is specified for “programme peak for undistorted output” of 40V (corresponding to 28.3 Vrms or 100W into 8 ohms, at which level the speaker will output 106 dB SPL in the midrange, IIRC). If you do the math, that means that maximum drive at the secondary equates to 4800Vrms or 6800 V peak!
The next thought is that you don’t really need to drive the Quad to full specified power and that you can get by with less voltage. That’s the first reasonable compromise that you ought to make. Even so I wanted to go for the full Monty, if only as a challenge.
To pull this off, you’ll need very high voltage rated tubes, which probably means transmitter tubes and a very high voltage supply. You can choose among many types of transmitting tubes, although I prefer triodes for sound quality and the lower plate resistance which provides better coupling as the ESL’s impedance drops at high frequencies. You’ll disconnect the Quad’s trannies, and connect your new amp to drive the inputs of the Quad’s delay line circuit which is on the secondary side of the trannies. The protection circuitry is therefore disabled unless you wish to incorporate it somehow into the new amp design (I haven’t done so yet but probably will). Generally you’ll use a push-pull arrangement with large HV plate load power resistors to B+, the plates then connected to the Quads via HV caps. Or you can use choke loading, or use a push-pull transformer primary as a load without a speaker attached (but it will be hard to find a conventional transformer with enough voltage rating). All of these require compromises and I’m leaving out tons of details, of course. Morgan’s Jones’ book has a nicely conceived 845-based ESL amp design, although it might not make enough voltage for ESL-63s.
I haven’t mentioned my design yet because it’s so gargantuan, impractical and crazy that nobody in his right mind will want to duplicate it. I use four 833 transmitting triodes per channel (not a typo) which are run well below rated dissipation in a push-pull mu-follower bridge configuration to achieve enough current drive (ie: slew rate) for high voltage/high frequency testing of other ESL designs. The thing runs off of +/- 4000 volt supplies and uses a pentode servo amp to correct output DC levels. I run the upper cathodes directly into the Quads without coupling caps (kids, don’t try this at home). It will deliver more than 4800 Vrms to the Quads at full tilt. How does it sound? Well, pretty damned good to my ears. The Quads now sound leaner, tighter and airier, all of which were needed improvements IMO. Detail is astonishing in a very natural way, as is punch and “sock”. Was it worth the effort? For me, yes, but then I enjoyed the journey. Now I still have final packaging ahead of me. |
|
|
| B Cullingford |
:cool: Brian I doof my hat to you. I hope also be listening to direct drive electrostats one day
:bigeyes: when you say "Now I still have final packaging ahead of me." brings visions of 8000V supplies hanging out on the floor! Guests beware!! I'm sure that is not the case;) |
|
|
| Brian Beck |
Well, not quite, but very nearly so. My prior situation was a separate dedicated listening room, no kids, a wife who knows what to stay away from and cats who aren't allowed in the room. But all that changed with two moves, two renovation projects, and two hurricanes. Everything is currently in storage awaiting time and space to continue. By the way, another aspect of insanity is that each channel dissipates a bit over 1000W at idle (class A and all), and I live in Florida, not Alaska. On our chilliest days, that’s more heat than we need for that room.
On a more serious note: If anyone is contemplating a high voltage project like this one or even one less potent, besides having sufficient HV experience, let me add one other piece of advice. I never tested the HV stages and power supplies alone. I always had a buddy watching and helping. Fortunately, like me, my friend is a keen audiophile and an amateur radio operator (many of whom are used to working on HV RF amps). We kept an eye on each other and forced each other to follow sometimes tedious safety procedures and check lists – like double-checking wiring, making sure caps were discharged, etc.
Another old rule makes sense whether the voltage is 40, 400 or 4000. That is to keep one hand in your pocket. Only reach toward a high voltage circuit with one hand. Keep the rest of your body away from the bench. If you should accidentally contact a live point, there is no (or less) circuit path for current to follow (oh, and wear rubber-sole shoes). A burnt finger is better than a stopped heart, I would think. I’m not saying this alone makes it safe to touch HV points, just adds a layer of safety. Avoid metal jewelry like big Rolexes dragging across power supply caps. If all this sounds too hard, please just stick with 6922s and EL34s and be careful with them too! |
|
|
| bear |
Two thoughts:
- for a directly coupled amp running an ESL you'll need EQ in the amp to get a flat response. If you look at the Acoustat X design (not a bad place to start) you'll find the EQ.
- One trick that has been published (Audio Amateur, iirc) is to use a standard tube amp and take your outputs from the plates of the tube, off the primary of the transformer, capacitively coupled is one possibility. You will need a resistor on the secondary, but it might suffice to use a relatively high resistance there (32 ohms?) too keep the transformer happy. You get less swing than you might with a transformer coupled interface, but it's a place to start.
If you used an amp that had something like 811s in the ouputs and 900volts on the plates, you'd have a swing of 1800volts which will likely drive most ESLs within 3-6dB of their max, so it might be fine.
_-_-bear
PS. not much point to think about using solid state devices, since the B+ voltages are pretty much out of the range of possibilities...
PPS... well, having said all that, if ur using a class AB or B amp then you might have a problem with the drive, since the stators want to be controlled with a voltage all the time... this leads us to the Class A situation, and as soon as you combine Class A and BIG high voltage tubes, you're talking A) derating the "output power" of the tubes as compared to the class B/AB case and B) lowering the B+ some and C) lot's-O-heat and power being "wasted"... |
|
|
| djmiddelkoop |
| quote: | | - for a directly coupled amp running an ESL you'll need EQ in the amp to get a flat response. If you look at the Acoustat X design (not a bad place to start) you'll find the EQ. |
Bear, you are correct. I've tried mine without this EQ and you get way to many highs.
Also like Brian mentioned, the sound is leaner, tighter and airier.
Dick. |
|
|
| Brian Beck |
| quote: | | for a directly coupled amp running an ESL you'll need EQ in the amp to get a flat response. If you look at the Acoustat X design (not a bad place to start) you'll find the EQ. |
For most panel ESLs this is quite true. As the wavelength shrinks compared to the panel dimensions, beaming causes a treble rise and something has to be done about that. At the bottom end you have to deal with the panel resonance(s) and dipole cancellation. These can be handled by low-level circuits before the direct drive amp, at least in theory.
But the situation is different for the Quad ESL-63 and its newer relatives. Most of the frequency response shaping has been done in the panels themselves by adjusting annular ring dimensions and in the careful design of the delay lines which also act as low pass filters for the outer rings. Take a look at the attached response curve I measured from the input terminals to the transformer secondaries (which are also the inputs to the HV delay lines). This is NOT the Quad’s total response to the air – just what the input circuits (mainly the step-up transformers) are doing to the signal before the panels see it. While it is not perfectly flat, I wouldn’t characterize it as EQ either, or at least not much compared to what has to be done for regular panel systems. The response curve shows a rise broadly centered around 10 KHz, but then a rapid fall-off in the upper octave and beyond. The bass response extends smoothly down to infra-bass frequencies with just a gentle rise. These effects are audible in the stock speaker but correspond, in my opinion, to two slight weaknesses, minor as they are compared to most other speakers.
So you can just connect a direct drive amp to the delay lines without the need for any EQ. What you get (assuming the DD amps are flat, as mine are) is the removal of the attached response curve. In other words, the response at 10 KHz is reduced, but the upper octave and beyond is restored (think SACDs). Also, the bass is tightened a dB or two. I think both of these are improvements to the stock speaker. The treble is now very smooth and airy. The bass end has a welcome tightening, but full bass balance remains. For cleanest treble, I run my Quads “naked”, but that’s another story… |
|
|
| bear |
You're assuming that the network going to the 63's panel provides a flat output for a flat input - I am not so sure of that. Any resistors or caps in the circuit? DCR from the coils?
According to Walker's teaching, the response you get is a function of the reactive impedance of the ESL vs. the source impedance of the thing driving it. You can match to different reactances vs. frequency and get a variation in output vs. bandwidth.
Thus the Strickland patent and the Audiostatic Euro patent which seek to provide two points of matching impedance from a nominally single source yielding a nominally flat, wide bandwidth result with reasonable efficiency.
At least that's how I recall it...
I'd expect that a given high impedance tube amp's direct drive output would work exactly the same way as a transformer coupled interface of the same impedance. But maybe with NFB, the ESL's response could be held flat?? Hmmmm... not sure about that.
_-_-bear :Pawprint: |
|
|
| Brian Beck |
| quote: | | You're assuming that the network going to the 63's panel provides a flat output for a flat input - I am not so sure of that. Any resistors or caps in the circuit? DCR from the coils? |
Not at all. The whole network is anything BUT flat.
In the case of stock Quads, think of there being two cascaded “black boxes”. The first box includes the input blocking capacitor, series resistors, the protection circuit and the step-up transformers including all the transformers’ leakage reactances and resistances in this one “black box”. The second black box has all the delay line impedances as well as the panels’ impedances factored into its input impedance. Where the first box connects to the second box is where I made the measurement shown in the chart, and also where I connect my DD amp. All the effects of the output impedance of box 1 interacting with the input impedance of box 2 are already captured in the chart response. So if you delete box 1 and replace it with a voltage source (the new DD amp) you will get a system response like the stock unit MINUS the response shown in the chart, no more and no less, because the impedances were already factored in. So the new system reduces the original 10 KHz bump, etc. And that’s exactly what did happen.
My original point was only that box 1 has mild effects on the response and really couldn’t be considered to have EQ like you might see in other ESLs. The significant shaping is done in Box 2 electrically and acoustically. This makes it possible to directly drive Box 2 with no EQ required in the amp.
| quote: | | According to Walker's teaching, the response you get is a function of the reactive impedance of the ESL vs. the source impedance of the thing driving it. You can match to different reactances vs. frequency and get a variation in output vs. bandwidth. |
Of course, that's true for any networks in EE, not just because Walker said so. And the effect is contained in the chart.
| quote: |
I'd expect that a given high impedance tube amp's direct drive output would work exactly the same way as a transformer coupled interface of the same impedance. But maybe with NFB, the ESL's response could be held flat?? Hmmmm... not sure about that. |
If you use triodes for the DD output stage the plate impedance can be pretty low relative to panel impedance, so the amp behaves like a voltage source for all but the highest frequencies. In my design, I used mu-stage bridge finals which have output impedances more like a cathode follower. The output Z of my DD amp is below 500 ohms. That may sound high, but hold on a minute. This is at the high voltage/high impedance side of the circuit. Since the Quads use a 1:170 step-up ratio, that equates to an impedance transformation ratio of almost 29,000. So 500 ohms is like 0.017 ohms source impedance in an amp driving the primary, which is a damping factor of over 450 at an 8 ohm reference. My point is simply that my DD amp, and probably many others triode DDs, can be considered a voltage source when directly connected to a high Z panel. That’s especially so for Quads since the delay lines block some of falling capacitive reactance at higher frequencies. By the way, you can add HV resistors in series between the DD amp and the delay line (or panels) to change the voicing. |
|
|
| Calvin |
Hi,
the major Probs regarding a DD-ESL-Amp are the high voltages needed, the reactive, strongly varying impedance of the ESL and the enormous power You need for sufficient bandwidth.
Because of the fallig impedance of a capacitor, You´ll need some current to drive the ESL at higher freqs, i.e a lot of wattage. Additionally the amp has to work stable with highly capacitive loads.
Using high efficiency panels like MLs that don´t need much voltage to sing is the first way to a solution.
Simple triode amps can do the job then -when You´ve got the right tube, i.e a KR845 running on 2kV plate voltage and Class A. A triode with an active load (current source) will be better, especially with regard to higher load currents and greater bandwidth and linearity and low output impedance. Feedbackless amplification with exceptional good sound and dynamics is then possible ;)
Using a good panel You only need to equalize the phase cancellation which is easily done with an active filter and which doesn´t stress the amp because of the low load currents in that frequency range.
Nearly all other concepts work with pentodes (Class B, or AB) and rely on heavy feedback and compensation to get the thing linear (Beveridge, Acoustat, AudioExclusiv). The sound of these devices is relatively poor with regard to cost and is imo no way better than using a good but much cheaper and safer trannie.
One of the biggest probs is to find the right power devices (e.g running a 6HB5 or PL509 on 5kV can´t be good, when the tube is rated for 700V continous plate voltage, even when 7kV pulses are allowed!) If You´re using triodes for the power stages the driver stages will be quite demanding, but highvoltage FETs could do the job.
There are just a very few special highvoltage transistors around that aren´t intended for these applications, so in most cases tubes do the job, but not normal audio tubes. You won´t find 5kV, 500mA babes in the standard audio department.
Another prob is to get a good and clean power supply. Best thing is probably to get one from those highvoltage companies, who sell supplies up to several hundred kV and kW.
The whole beast will work and sound deliciously good but it´ll cost! It´ll be rather inefficient! It´ll heat up the listening room!
It can be a very dangerous part to work with!
But what the h... ;)
Yeah
Calvin |
|
|
| Brian Beck |
| quote: | | One of the biggest probs is to find the right power devices (e.g running a 6HB5 or PL509 on 5kV can´t be good, when the tube is rated for 700V continous plate voltage, even when 7kV pulses are allowed!) |
The best approach in my view is not to think of audio tubes, but to think like a radio amateur and find high voltage RF transmitting tubes, of which there are many choices. Amateurs have been building 1KW plus RF power amps for decades. Borrowing a few tricks from those designs is useful.
I ended up selecting the 833 RF triode for my design. This is but one example of how a transmitting tube can be used for an ESL DD amp. The 833 was originally designed for RF amplifier duty up to 30MHz and also for audio amplification duty delivering as much as 2700 W in class B PP (for use in AM modulators). But the main rating of interest is the 4000 V DC plate maximum. When biased this way and used in typical service the plate can reach peaks of almost 8000 V. The 833 turns out to be quite linear too, and can be driven comfortably to full output with a 6SN7 driver stage. There are many other choices besides the 833, such as the 3-500Z or the 3-1000Z.
Another approach to high voltage devices is to stack tubes in a series arrangement so that all tubes evenly split the voltage between B+ and load, and the grids are on a resistor divider ladder. There are lots of implementation problems with this approach but it can be made to work. I toyed with this for a while before choosing the 833 for my design.
| quote: | | Nearly all other concepts work with pentodes (Class B, or AB) and rely on heavy feedback and compensation to get the thing linear (Beveridge, Acoustat, AudioExclusiv). The sound of these devices is relatively poor with regard to cost and is imo no way better than using a good but much cheaper and safer trannie. |
Agreed. If that's the choice, stick with a good trannie.
| quote: | | Another prob is to get a good and clean power supply. Best thing is probably to get one from those highvoltage companies, who sell supplies up to several hundred kV and kW. |
I wouldn’t advise going that route. It will be expensive and hard to find a unit with the right specs for your exact application. You can do better with DIY. It will be big and heavy, but not very hard to design or build. Much easier than the amp stages themselves. Again, start with well known ham radio approaches to power supplies for those big RF amps (buy an old copy of the Radio Amateur’s Handbook from ARRL at a flea market to see design ideas). The power transformer is the hardest purchase. I use a surplus trannie about the size of a car battery that came from an old Collins transmitter. By using series diodes in bridges and series filter caps you can rather easily make sturdy well filtered HV. In my case one trannie with twin CLC filters produces + and – 4000VDC. Ham fests or electronic swap meets are great places to get these kinds of parts for relatively cheap.
| quote: | | The whole beast will work and sound deliciously good but it´ll cost! It´ll be rather inefficient! It´ll heat up the listening room! |
Right on! |
|
|
| 405man |
At the moment if you drive your ESLs from a tube amplifier the amplifier output transformer secondary is connected to the ESL transformer primary. What would be involved in producing an output transformer to drive the ESLs. This would at least remove one transformer and would remove the necessity to operate the amp at very high voltages.
Stuart |
|
|
| Brian Beck |
| Yes, I’ve seen this suggested somewhere before, but I’m not aware of anyone building one (someone probably has though). I figure that if you had an amp with a 5K primary, that’s a step down to 8 ohms by the ratio of 25:1. Say you then wanted to step up for the ESL by a ratio of 1:75, as a reasonable example. What you really need then is a single transformer that has a 1:3 ratio, from plates to ESL stators. This will have high impedances on both ends, but I’ll bet a better transformer can be made at 1:3 than the two others combined. |
|
|
| B Cullingford |
This (1:3 or similar xformer) could also be done using hv transistors or mosfets - might be better than what is usually done. May be we need to take page out of Susan's book Zero feedback Impedance thread and make our own.
Bill |
|
|
| Brian Beck |
Ouch, why use MOSFETS when you can use still use tubes? ;)
You don't need to switch to MOSFETs just because the voltage requirement dropped to within solid state range. Granted, it may work. In my experience (Zen et al), even these simple MOSFET amps suffer from rather high levels of phase intermodulation distortion (think FM) due to the highly non-linear internal FET capacitances which vary wildly with signal swings. That's why we love tubes! I'll bet that starts some responses, like poking a stick into a hornet's nest! Have at me... |
|
|
| hermanv |
Yep, I did that; picked up a 1KW ham transmitter we had built 1,400 Volts on the plate cap if memory serves. Touched my metal watch to that cap while holding on nice and tight to the metal chassis :eek:.
Wow, best described as having your arm hit with a sledge hammer. Left some little black burn holes on my wrist. Did a good job of making sure I never did it again though:smash:.
There are some spectacular new transistors, IGBT, and MOS-FETs that reach 1200V and well past, not cheap of course and maybe too slow but a lot more drive current than those 833's and you probably only dissipate 500watts in the room instead of a KW :cannotbe:.
Powerex is the first high volatge solid state co. that comes to mind, there are others.
Didn't the old Beveridge ESL have direct drive? |
|
|
| B Cullingford |
When I said might be better ... I meant better than using bjt or mosfets into a standard xformer - I was NOT saying better than tubes. But it does not hurt to consider all the possibilities!
It seems that for most of us that direct drive electrostsats are dream and somewhat impractical, this leads to a transformer. Just because most amps out there are designed for 8 ohm loads the commercial electrostats are designed to interface with them (with a few exceptions - I once repaired a Accoustat, it used questionable (even for the time) op amps to drive tubes seemed a strange choice of circuitry). We DIY folks should consider what might work better, tubes to transformer to transformer to panels never made any sense at all, an extra lump of iron in there. And silicon (as opposed to silicon dioxide) to iron to panel we should consider what the optimum configuration looks like. Our amps can be special purpose and be all the better for it.
Well I've explained my thoughts a liitle better, but I expect pulling the stick out of the hornet's nest might not help to much!
Of course then there is problem of sourcing the iron - so now you might have to build the panels, wind the xformer, and build an unusual amp ... triple the fun! |
|
|
| Brian Beck |
| The 1:3 turns ratio (or so) transformer idea for tubes-to-ESLs really does make a lot of practical sense. I never claimed that a DD ESL amp was very practical; I did it to learn and just for the fun of making something like an ultimate statement. I would seriously like to work with a transformer winding expert, which I am not, to make a decent low-turns ratio HV design. It ought to be possible to achieve very high coupling, high primary inductance and low leakage inductance. In other words high bandwidth, although the capacitive nature of the load will be difficult. But in any case better than the two transformer case, I would think. |
|
|
| hermanv |
I didn't mean to make any value judgments, there is a good reason tube amps are still popular.
On the other hand it is perfectly possible to make first class amplifiers out of sand it just may be harder.
Either solution with a step up seems possible and makes more sense than first stepping down and then up again. i.e. If I wanted to built a transistor model I'd start with a couple of hundred volts instead of 50V for my main supply.
One big problem is that the panels have anything but flat reponse and the commercial speakers most likely were tuned with transformers in place.
Of course if it was all easy we wouldn't need the forums:). |
|
|
| 405man |
One of the main advantages of tube amplification for direct drive of ESLs must be that the primary voltage swing on the output transformer is of a similar order of magnitude to that required to drive the ESL. Attached is the schematic of the Quad ESL63 from which it can be seen that the input is connected to 2 step up transformers with primaries in parallel and secondarys in series effectively providing twice the voltage swing. If each transformer was replaced by tube amplifier and the 2 amplifiers fed in antiphase then each amplifier would only have to swing by half of the required voltage which for conventional ESLs may only be in the order of 1.5-2 times the primary voltage and even for the ESL63s may still only be of the order of 3 times. This then has the advantage that each of the amplifiers only has to deliver half the total output power. The other advantage of tube drive with the ESL63s is that the protection circuit could be retained though you would need 2 of them. I hope that all makes sense
http://www.euronet.nl/users/temagm/.../esl63_sch2.jpg
Stuart |
|
|
| hermanv |
Typical British product, obscure and possibly clever all at once.
After a quick look I'm guessing the purpose of the 555 is to protect the power amp from ES panel arcing by momentarily placing a short accross the amplifiers outputs? I didn't see where the trigger for TR3 comes from.
I thought I'd read in an earlier post that the required swing was still beyond most typical tubes, something over a kilovolt? So a step up might still be required for any conventional power amp design. i.e you couldn't very well adapt an existing 6550 or KT88 design for this job.
My point being that once you take the step of getting a custom transformer designed and built, other design options become available.
Again I make no value judgements for or against tubes it's more to say that either tube or transistor design may well be equally difficult. The cliche, "no free lunch" seems to the point.
Really old TV sets (1950 to 60's) had very high voltage windings on their huge power transformers, we got that 1400V I mentioned from a vacuum tube voltage doubler so that transformer probably ran 800 VRMS unloaded. Wonder if any of those are still gathering dust in someones attic? |
|
|
| 405man |
The trigger for TR3 comes from a wire antenna close to the diaphragm and detects ionisation which occurs prior to arcing.
Stuart |
|
|
| BillH |
So, if you wanted to design a 1:3 transformer, how much power would it have to supply to the esl? Just for discussion, let's say you wanted 2kv p-p audio to the esl and had a 20 meg resistor in series with the high voltage. I believe we can assume a low value for the impedance of the high voltage supply.
P=V squared/R
=(2000)squared/20,000,000
=0.2 watts
That would mean quite a physically small transformer and a low wattage amplifier.
This seems too good to be true, can anyone check the math and my assumptions? |
|
|
| B Cullingford |
20 meg resistor in series
?? there is a high value resistor in series with the membrane for constant charge but not in series with the plates - it would form a low pass filter. Don't know what the capacitance of a typical (what ever that is) esl panel is but you need enough current to charge/discharge it (obviously) to follow the signal voltage and I do not think that going through a 20M resistor is going to get you there.
Anyone know off the top of their head what a typical capictance of an esl panel is? Guess I could get out the text and look up the formula and constants and...
Bill |
|
|
| hermanv |
We can make a rough guess;
My ML ReQuest's are spec'd at 1 Ohm at 20KHz. Also at 88dB SPL (from memory I could be off a dB or so)
So 2.83 VRMS (@8 Ohms) to get 88dB SPL will then require 2.83 Amps at 20KHz or 8 watts.
Allowing for a little more volume and some head room seems like 100 Watts would be about the minimum for decent levels, probably 200W would be better. (All this is at 20KHz usually little music power up there, I think the power whould drop linearly with frequency)
I doubt the transformer absorbs much so its not unreasonable to assume almost all the power is to charge and discharge the capacitor.
If I knew the tranformer ratio I could work backwards from the current to get the capacitance. |
|
|
| BillH |
A reply from flemming to an older thread at diyAudio:
My els is 40cm X 200cm and the capacitance is 1200pF. |
|
|
| Brian Beck |
ESL capacitances can vary all over the place due to stator spacing and element size, but 1000pF is often used as a design bogey for a mid-sized panel. Imagine the difference in capacitance between a big Sound Labs speaker and a little tweeter cell! The transformer must drive reactive currents into the ESL due to this capacitance, resulting in lots of VA rather than dissipated power (watts). Therefore the transformer will be larger than expected. Plus it will need a lot of primary turns to build up enough primary inductance to maintain bass, and with that big step up ratio, that means a huge number of secondary turns. So it will be much bigger than expected for just that component of power needed for acoustic energy. It will also need hi-pot HV windings on the secondary. If we were to go for a 1:3 being discussed here, the trannie can probably be made smaller. By the way, cone speakers waste even more power in losses having nothing to do with sound, so the ESL just has its own unique challenge.
You are right that the large resistance (20 Meg or so) applies only to the diaphragm bias supply and has no bearing on the audio drive to the stators. BTW, there is a resistive component to the ESL impedance corresponding to the radiation resistance which is the part that dissipates power, in this case dissipated as sound leaving the speaker. Surprisingly, ESLs will appear inductive near the low-end resonance due to mechanical EMF. Professor Hunt’s famous work analyzed these impedance components in addition to the large capacitance. |
|
|
| BillH |
Thanks, Brian. I'm beginning to understand now. The capacitive reactance is a real problem at high frequencies. The 1000pf esl has a reactance of 16 ohms at 10 khz. 4kv into 16 ohms is a lot of current drive needed.:hot:
The use of a step up transformer seems to make the amplifier work even harder as it needs to supply an output current of speaker current x turns ratio. Aaah, you just don't get something for nothing.
I've been trying an output transformer calculator made by one of the diyAudio members. It's a .zip file containing a windows installer. Plugging in different values of transformer wattage and low frquency cutoff really shows the tradeoffs that have to be made in transformer design. You're right about the large number of windings. It shows anywhere from about 800-2000 primary turns, depending on cutoff frequency, core size, etc. |
|
|
| Brian Beck |
| Let's see, I get 16,000 ohms at 10,000Hz for 1000pF. With 4,000V that's still a quarter ampere, which is 1000VA! Fortunately, as others have been saying, musical energy is much lower up there than in the lows and mids |
|
|
| hermanv |
I'm sure this has been worked out in some detail by others but as I think about the drive requirements for ESL panels I arrive at some assumptions and conclusions that the finshed ESL speaker panels probably can not appear purely capacitive.
Let me attempt to explain my thinking: If you hook a voltage source (an amplifier) to a capacitor the current and therefore power delivered into the load doubles for each octave increase so unless an ESL panel becomes less and less efficient at coupling to the air, the resultant SPL would rise with frequency.
Enter the crossover or compensation network, this network would most likely add inductive reactance to reduce or prevent the rise in accoustic output with frequency. By its very nature this would make the capacitor look more resistive. So the phase angle between drive voltage and ESL panel current is most likely not a simple 90 degrees.
Of course if the compensating network is ahead of the power amp this isn't true, with the exception that the amp would deliver decreasing voltage into the panel with increasing frequency resulting in a more or less constant total wattage being delivered into the ESL panel.
It almost seems that ESL panels should be driven by current sources not voltage sources.
Did I miss something obvious? |
|
|
| B Cullingford |
yep its does not behave just like a cap - as Brian posted earlier here. Add to that other stuff in the speaker - I'm most familiar with the quad 67 which is different from most esls but is very instructional to read about no matter what esl you are considering. My quick comments are explained further and much better at the link.
If you look at the Quad 67 - the cap and resistor in the input is to tailor the low responce (the R alters the Q of the bass response - cap shorts for highs) and the delay line not only delays the signal to the rings which makes the diaphragm move as if the source is behind the speaker - as opposed to a piston - but rolls off the higher fequencies - helping with reflections in the mylar controlling the directivity, and I believe therefore reducing the capacitance load at higher frequencies, that is adding some series inductance (sort of).
All very clever - it is explained fairly well at quadesl.org
Bill |
|
|
| BillH |
I'm still trying to figure out what output tube could be used push-pull to drive a step up transformer and esl.
Since my math skills are failing me, I need to simplify the process. Here goes...
If I'm happy with a 10 watt amplifier with an 8 ohm output transformer driving an 8 ohm step up transformer, then I should be just as happy with that 10 watt amp driving a single custom step up transformer instead of the 2 transformers. Assuming no losses in the transformers for the time being.
So the question for anyone that's used a tube amp to drive esl's: How much power do you feel is enough?
Thanks again to all for helping this tube noob. |
|
|
| Brian Beck |
If you’re happy with 10 watts now, and then you replaced the step-down/step-up pair with one step-up transformer with the same equivalent turns ratio, you’d still be happy with 10watts. In general, most ESLs with their step-up transformers have about the same *sensitivity* as cone speakers. (ESLs are usually way more *efficient* watt for watt, but that’s not counting reactive VA, another story.) My Quads make around 86dB SPL at 1kHz at one meter at 2.83 V rms (equivalent to 1 watt into 8 ohms). This is average for cone speakers.
To “Hermanv”, if you drive a capacitor with a current source you get an integrator, which has a 6 dB per octave downward slope from very low frequencies all the way to the highest frequencies. Maybe not what you want, but your point of voicing an ESL with a series R is potentially useful to counteract directivity gains. |
|
|
| hermanv |
That was exactly my intent.
As I proposed, the accoustic power ouput, all else being equal, should rise 6 db per octave if driven with a voltage source, if the integrator drops 6 dB per octave you end up with a flat response
I know its not that simple. ESL panels have resonances resulting in peaks and valleys. Some are a series of different size panels in parallel partially to spread out those resonances. Film tension and total area enter the equation. As must conductivity of the conductive coating and the holes, slits or vents needed in the stators
At some point the wavelength exceeds the membrane width so you get roll-off at low frequencies etc..
Just musing out loud. |
|
|
| el`Ol |
| If it was possible to get info about the Audio Exklusiv active electrostats it may not be necessary to reinvent everything, but all I found googling around was that they are transformerless. |
|
|
| Calvin |
Hi,
The AE´s come in passive and active versions.
The active version utilzes an amplifier that is nearly an exact copy of the Strickland amp. Using the same design, same transistors and Tubes (6HB5, to my knowledge a relatively unstable guy in this application, running on app. 2.5kV continous plate voltage while rated for just a couple of hundred continous volts. Think the PL/EL509 should be a bit better). Both designs feature a SRPP similar structure and rely on heavy compensation and feedback to get things linear)
yeah
Clvin |
|
|
| SY |
| I can tell you from personal experience that running EL509 types at 2.5kV is not a good idea. |
|
|
| SY |
| Spooky! The earthcurrents link describes almost exactly what I've been working on for my direct drive ESL amp. I thought it was original... rats! |
|
|
| sasha |
Hello guys,
Let me jump inn.
I am sorry to tell you that you need about 10KV pp for full output to your ESL.
Math is simple: 100W amp vill produce 28V RMS. 28x1.414=39.5V p
and double that = 79Vpp. If you use 1: 100 step-up you have close to
8000Vpp. With little head-room close to 10KVpp.
Accoustat uses 5KV B+ in bridged config. = 10KV ideally but in practice
maybe 8KVpp.
I was thinking to build AB push-pull tube amp in bridged mode with
5KV B+ using 4(one channel) 6HV5 tubes good for 5.5KV plate voltage.
This will give me 8KVpp with not much plate dissipation, meaning
500W or so power supply.
Another solution is solid state DD amp. Yes, it is possible.
You can use cheap horizontal output transistors used in tv's and
with 1200 B+. This in AB class will give 1000Vpp and two amps bridged
2000Vpp. Now you just take as many of the bridged amps as you need connected in series and you can get as much swing as you need.
The trick to do this is that every amp (block) has it's own independent power supply and than you can connect output of one block to ground
(floating) of next block and output of that block to ground of next block and so on. Good thing about bridging the amps is no matter
how big ripple is in your power supply it will be canceled on the
output because it is in phase. For power supply You can use 12V / 50W transformer put backwards and you can usually increase 12V
to 16-17V without overheating to get more voltage on the other end. I was able to get over 800V DC from one 12V/110-220V (dual primary) with 12V increased to 17V and with use of voltage doubler.
Capacitors used 4 33mf/400V ripple 2%. You can use 7 transformers
with 12V primaries connected in series to 110V main. Well, you got
the picture. Main thing is to have individual power supplies for every
amp block. My amp block design was very simple.
Two high voltage NPN transistors working in push-pull AB class
biased to couple of mA and for drive I used two small signal
transformers connected out of phase to transistor base.
Primaries in parallel in phase. Use of transformers is needed to
insulate low voltage input side from multi KV output side.
This transformer is not critical just be sure it has good bandwidth
and good insulation between primary and secondary (more than 10KV). This can be 1:1 ratio and also if you cant find such transformer it can be built easy. 50 turns or less on each side and
that is it. You gonna need two transformers for each amp.
So if you use 7 cascaded and bridged amps you need 28 transformers for one final mono block. Also to get higher impedance
on main input you can connect transformer primaries in series
and say drive it with gianclone. Amp is without neg feedback so distortion will be high but no more than 1% in total. Use of emitter
resistor is necessary and it is 220 Ohm with 0.2 mfd capacitor across to bust voltage on output on high freq. Well, emitter resistor is kinda
neg. feedback.
Why use this design and not tubes? It is simple, proven to work
cheap parts are available and it will be in the future, stable, no
maintenance, and after all it is another way of doing DD amp.
Regards. |
|
|
| SY |
Yes, for full-range use with Acoustat-type panels, you do need that kind of swing. Or swing less and accept a loss in attainable SPL. The beauty of the autoformer and direct-drive-with-dummy-load shown in the earthcurrents link is that the amp can be configured to run class AB or B. This makes the output tube choice much easier.
BTW, you really don't want to use 6HV5s, unless you love oscillation and smoke. They were the weak point in Acoustat's d-d amps. |
|
|
| Brian Beck |
Sasha,
Good plan. I’m not a solid state fan, but your design makes good sense. I have also floated supplies before for various applications. In addition to being concerned about the insulation rating of the power transformers, you must be concerned with primary-secondary capacitance and perhaps secondary-to-frame (usually grounded) capacitance. The preceding amp stage has to drive not only the bases/gates/grids, but also these capacitances all added together. I chose split-bobbin transformers in such applications to minimize capacitance between primary and secondary. I measure transformer capacitance by shorting the primary leads to each other, and also shorting the secondary leads to each other, and then connecting a cap meter across these two nodes.
Your point on high drive voltage requirements is well taken. An output stage in a bridge configuration is almost essential since you get double the drive voltage for a given power supply configuration compared to a single ended drive. And the balanced nature of the ESL almost demands it anyway. In my wacko transmitting tube design, I used bipolar supplies at +4000 and – 4000 volts (it’s all DC coupled with level shifters and a pentode servo amp). Ideally this would give 16KV peak-peak output which is about 5.6KVrms drive. In practice the outputs won’t quite reach the rails, but I do achieve my design goal of 4800 Vrms (13.5V peak-to-peak). This is all my Quad ESL-63s can/should handle anyway.
I made my own high voltage scope probe which is capacitively compensated for flat response. It’s impressive to watch music waveforms dancing on the scope at thousands of volts while listening at normal levels.
My friend refers to all this as “Lethal Fidelity”. |
|
|
| sasha |
Hmmm,
He is claming that Accoustat is AB class but correct me if I am wrong
it looks A class to me? |
|
|
| sasha |
Brian,
Capacitance of primary/secondary in my driver transformers is not an issue because it is only small transformer used to insulate input and output and each amp block has it's own driver transformer.
True primaries are connected together but still this is only little transformer easy to drive. Maximum of 7 transformer primaries is connected together in series probably better than in parallel to reduce capacitance or maybe some combo parallel/series. |
|
|
| Brian Beck |
| Yep, you're in good shape then. |
|
|
| zelter |
Is this discission dying put?
Zelter |
|
|
| Brian Beck |
| You've got the microphone... |
|
|
| Urban |
Hi. Jumping in a bit late i the discussion. Couldn't a Class D amp be used ?
The output stage of a digital amp is (basically) a pair of transistors generating a square wave of (say) +/- 50V. To make this more "analog" the square wave is put through a choke and then loaded by a capacitor, before being sent to the speaker.
If we could use (some sort of) transformer instead of the choke, we could produce a high voltage squarewave loaded by the ESL panel (being a capacitor in itself).
As the transformer would work at the sampling frequency (I suppose) say 96kHz, all the problems that are frequency related, as the transformer having to cope spanning from 20Hz to 20kHz, would be gone. Also, the efficiency would be much higher, i.e. it could be made smaller and cheaper.
Still haven't built those panels...
/Urban |
|
|
| el`Ol |
| I ask myself whether it is useful to set the transformers between the output transistors, like McIntosh do it in their AB amps. And there are very interesting 400V switching transistors with a lot of power. |
|
|
| sasha |
Hello all,
I am close to finishing Acoustat SC-X3 DD tube amp from 1979.
I am having problem determining what is the part on the schematic
marked VR1 and VR2 with number on it V68Z2. I believe it is a trim-pot
but I am not sure. This question would be best answered by
djmiddelkoop since he has experience with this amp. Anyone else
is welcome to help. Also, I need four 0.1mF(micro Farad) / 5000V
for this amp and if somebody has it I am willing to buy it.
Best regards!
Sasha.
P.S.
I can't wait to hear 10KVpp singing thru my ESL's.:happy1: :happy1: :happy1: |
|
|
| hermanv |
VR1 and VR2 would normally be contractions for Voltage Regulator 1 and Voltage Regulator 2. With a name like V68Z2 it sure could be a 68 Volt zener especially since 68V is a standard value for a zener. The less common symbol sometimes used for a zener diode is a little rectangular box with a streched "Z" in it usually turned sideways, it could look like a trim pot symbol.
If I'm right, your biggest problem is in detrming the correct wattage zener to use. Find the supply voltage, between that and the zeners (in series?) will be a resistor. Supply Voltage minus zener voltage divided by resitor value equals zener current. Current times zener votage equals watts; typically one would use a zener around double this wattage for margin.
One other component often called a VR is a transient suppressor like a metal oxide varistor. These also come in 68V versions. They are designed to absorb a high transient energy and have relatively loose voltage threshold accuracy i.e. they will trigger somewhere in the neighborhood of 68V. |
|
|
| sasha |
Thank You for Your replay hermanv.
I am posting part of that schematic. Now it will be more clear
what is this part and what is the value.
Sasha. |
|
|
| hermanv |
ESL panels are known to arc over from excessive excursion or dirt.
I would guess that VR2 is meant to protect Q3 from large voltage spikes. There is AC feedback from the panel back to the amplifier through the 4.7 pF capacitor C37. If VR2 was a zener it would kill this feedback unless the volatge divider of R57 and R56 is arranged so that the zener normally does not conduct.
Without knowing the cathode Voltage of V3, I can not calculate this.
As an arc suppresor, a standard 1/2 watt zener would be fine since R57 is 6.8 Meg. A larger zener or an MOV would probably have too much capacitance and would affect the signal from C37. Note that C37 might be there to stop the amplifier from oscillating and not for linearizing feedback.
I can help more if you know the idle state voltages or even the AC swing of the output, but I think the design is such that normally VR2 does not conduct. |
|
|
| sasha |
Maybe it is a thermister used to thermally stabilize circuit.
I heard that this amp has some temp. related problems?:scratch::hot: |
|
|
| hermanv |
The traditional symbol for a thermistor is a resistor inside a circle.
Othet than that as I said earlier I'm guessing, but I am an EE and I do have 30 years design experience. I think my guess is good but unfortunately I can't guarantee it. |
|
|
| sasha |
Another guess:
Multi turn trim-pot for tube bias adjustment. I don't see any other
way to adjust bias in this circuit and I think bias adjustment in
tube amps is a must. Since this is only one half of the one side
of the amp it is important to adjust equal bias on both sides to
achieve minimum voltage offset on the output.
Just a guess.
Regards. |
|
|
| rcavictim |
| quote: | Originally posted by sasha
Hello all,
I am close to finishing Acoustat SC-X3 DD tube amp from 1979.
I am having problem determining what is the part on the schematic
marked VR1 and VR2 with number on it V68Z2. I believe it is a trim-pot
but I am not sure. This question would be best answered by
djmiddelkoop since he has experience with this amp. Anyone else
is welcome to help. Also, I need four 0.1mF(micro Farad) / 5000V
for this amp and if somebody has it I am willing to buy it.
Best regards!
Sasha.
P.S.
I can't wait to hear 10KVpp singing thru my ESL's.:happy1: :happy1: :happy1: |
Sasha,
You are the fellow that fixed my RPTV and got to hear the speakers in my avatar aren't you?! I remember our discussion about your efforts to make an ESL amp.
I have quite a large supply of some expensive NOS, Plastic Caps Inc. metal can oil caps rated at 0.5 uF/5 kV. Be happy to sell a few.
I think I have some dry, wound foil and plastic 0.08 uF/15,000 VDC axial caps also.
Did you get anywhere with my multi channel surround amp?
Cheers,
Rob |
|
|
| sasha |
Hi Rob,
Yes that's me. Actually last night I found on Ebay just what I need and
very cheap too.. I have already placed the bid but if I don't' vin I will contact You. Regarding Your amp, it was bad DSP board so I stopped there and used it for parts.
Regards. |
|
|
| rcavictim |
| quote: | Originally posted by sasha
Hi Rob,
Yes that's me. Actually last night I found on Ebay just what I need and
very cheap too.. I have already placed the bid but if I don't' vin I will contact You. Regarding Your amp, it was bad DSP board so I stopped there and used it for parts.
Regards. |
OOPs. Well you weren't supposed to do that without making an arrangement with me since I had not gifted you the amplifier, just gave it to you to evaluate and repair if practical. Perhaps you could save me the back panel that contains the half million rca receptacles. I need to make a multi source stereo audio selector for my big system downstairs.
Can you send me your phone number on PM and I'll try to call you.
Cheers,
Rob |
|
|
| sasha |
Hi Rob,
Sorry I misunderstand agreement for Your amp.
I think You told me to fix it for cheap or use it
for parts. Anyway the rest of the amp is in garbage, but don't worry I have nice Yamaha amp for You.
It looks very nice cosmetically but it is not working so you can use it. I'll see You soon. I have your PN somewhere. |
|
|
| bigwill |
| I love how this thread is still going :D (I'm the starter) |
|
|
| djmiddelkoop |
| quote: | I am having problem determining what is the part on the schematic marked VR1 and VR2 with number on it V68Z2.
I believe it is a trim-pot but I am not sure. This question would be best answered by djmiddelkoop since he has experience with this amp. |
Sorry for the late reply Sasha, didn't notice it before.
VR1 and VR2 are Varistors(=resistor which acts like a zener above a specified voltage) with the orginal P/N 6BZ2.
They are used for protection of Q1 and Q3.
In case you cannot source the parts, replace it with a 1N4007 diode from the Base (Anode) to the Collector (Kathode or white band on diode) of these transistors.
This will give protection also.
Hope this helps,
Dick. |
|
|
| sasha |
Thank You so much Dick,
Now I can finalize my amp. Is there anything else important I should know about it and it is not in the schematics before I flick the switch??? Are all values in the schm. correct?
And bravo to You hermanv, You were spot on regarding this mysterious part.
Thank You all guys.
If You don't here from me soon I am probably fried by 5KV.
:hot: :hot: :hot: |
|
|
| djmiddelkoop |
| quote: | | Thank You so much Dick |
No problem, you're welcome.
| quote: | | Are all values in the schm. correct? |
Yes, don't worry. The filament xformers for the upper tubes must be capable of handling 5kV isolation between primary and secondary windings.
When switching on for the first time make shure the HV is disabled and check the voltages you can.
| quote: | | If You don't here from me soon I am probably fried by 5KV. |
Be shure to stay away from the tubes and surrounding parts when the HV is on.
Try to get a HV probe for your voltmeter.
Dick. |
|
|
| sasha |
Thanks again,
I am using 6HV5 tubes instead of 6HB5 originally on Acoustat and I think this is much better choice since according to specs sheet
for 6HB5 it can't handle 2,5 KV A-class which I believe is main reason for
frequent problems on this amp. Right? |
|
|
| djmiddelkoop |
For the 6HB5 I have good results with the GE, National and Realistic brands. I know that Sylvania will cause trouble.
I was also looking at specs of other Compactron tubes and also found the 6HV5 as a replacement candidate.
I havan't had the chance to try them yet, but I don't see why it shouldn't work.
Give it a try and please let us know.
Dick. |
|
|
| RetroAudio |
| what kind of panels ya driven' there sasha? |
|
|
| sasha |
| It is Audiostatic ES100 clone. |
|
|
| RetroAudio |
From the Audiostatics I've seen around (I don't know much about them frankly) it looks like a smaller speaker so probably has not as much capacitance as say a larger Martin Logan. That being said, the alternative tubes listed here just may work ok, you'll have to "experiment". Keep in mind that the 6hb5s look like they have the most current capability which is what is needed according to the Acoustat lover's society when driving multiple panels. It's sort of funny that the other tubes are rated for more power dissipation but look like they have less current capability according to the specs I see. I don't know, maybe I'm wrong. After Googling some, it doesn't look to me that specs are the strong point with some of these tubes.
As for the IN4007 being installed from base to collector, I don't believe that this approach will provide much overvoltage protection. When it does kick in, you'll be a day late and a dollar short. Look at the circuit again and just do some elementary analysis. Perhaps you'd be better off staying with the original game plan and using a suitable voltage limiting device somewhere in between 0 and 150 volts. Also, there is only one bias to each side of the circuit, and that's the variable resistor denoted as such.
I have to compliment you Sasha on your work if you indeed have built one of these beasts. If it's not too much of a secret, I for one would enjoy seeing your work along with my compliments to the chef...:) |
|
|
| djmiddelkoop |
| quote: | | After Googling some, it doesn't look to me that specs are the strong point with some of these tubes. |
Looking at the specs myself I don't see any problem in using the 6HV5. Actually, I believe it will perform better because the Ra is lower.
| quote: | | if you indeed have built one of these beasts................. |
If you want to have a look at my amp, well here it is:
HVamp
Dick |
|
|
| RetroAudio |
| Sorry I wasn't more clear here,....just meant that from what I was able to come across that info seemed to be somewhat limited. I wonder about pinout compatibility, but it may be a moot point with your guy's custom setups. Dj - Nice looking OTL.....:) How easy is it to work on though if something went wrong? Do the panels fold out or anything? It definitely looks like you put some time into this one! |
|
|
| djmiddelkoop |
| quote: | | It definitely looks like you put some time into this one! |
Indeed, even though I can say I'm pretty well experienced.
| quote: | | How easy is it to work on though if something went wrong? Do the panels fold out or anything? |
For servicing this approach is not very handy, nothing folds out.
I had to put everything in this 19" case as I don't have room for 2 seperate ones. Plus everything has to be safe for my children.
As the amp is in a closet under the stairs to the next floor, the noise of the fans can't be heard.
Dick. |
|
|
| RetroAudio |
| Dj - I seem to recall from a past thread that were into developing an all-tube approach. Do you keep the EQ intact or just run the signal straight? I can respect the proprietary nature of the circuit no matter what you're approach and have read about past amps employing an all tube circuit, but never have been able to learn much about them. I think it might be because nobody really knows except the designer....:) Wish I could read the language of your website, might learn a thing or two.........sigh. |
|
|
| sasha |
This is what I have built so far.
To assemble one board doesn't take too much time 2-3 hours.
And there is 4 boards for 2 channels. High voltage supply is cascaded
6 separate transformers each giving 800V. This is actually 220V/12V
step-down transformers put backwards and 12V side in series to
fit 110V input. Well, each transformer will see more than rated 12V
but in my case no overheating. With help of voltage doubler circuit
I was able to get over 800V from each transformer.
Power supply is around 250W, which considering max. allowed
dissipation of the 4 tubes (4x30W=120W) is OK.
To get custom made power transformer cost a fortune. |
|
|
| sasha |
This is candidat for amp case.
Old Kenwood.
It might even look good. |
|
|
| djmiddelkoop |
| quote: | | Do you keep the EQ intact or just run the signal straight? |
RetroAudio, I first tried the amp straight as I also wondered why Acoustat made that EQ filter. The amp sounded way to bright.
This was tried on several ESL (Acoustat, Final700, Audiostatic100) with all the same result.
Hence the reason for this EQ, which is a low pass filter. I made a simple filter (R-C-Pot in series) after the first stage to ground, and with the pot you can adjust the right balance of high/mid of the sound.
On the website there is no information but general stuff.
Sasha, I have a custom made HV xformer, with 2000V secondary on a 250VA core. Paid about 75 euros, =90 US$, for it.
BTW, nice pictures.
Dick. |
|
|
| RetroAudio |
Dj - that's been my experience as well with the EQ. I have run rough freq. response plots to verify (I was playing around, do my best work that way) and found the low-pass filter as well as the hi-freq boost. The hi-freq adjustment is amazingly touchy and one can tell the difference when the circuit is cleaned up with better components, at least I can. Stats seem to need this EQ in general from what I have gathered. I have heard of others sticking the servo up to other panels and reported it sounded just fine. Of course, some measuring would be needed to get it just right, a capability not everyone has. I don't know, these projects never seem to end cause there's always something left to do.
Sasha - keep up the work! You'll learn things as you go and will be the better designer for it. I think you've undertaken a very respectable project, but please be careful for everyone's sake! |
|
|
| djmiddelkoop |
The EQ has to be adjusted if an other panel is connected.
On the original servo there are 2 user selectable EQs depending on the distance to the wall at the back of the panels. The main difference is in the low frequencies.
Ofcourse these EQ where designed to suit the 3 panels that came with each of them.
The Hi-freq is mandatory or else the panels will sound to bright.
The original Acoustat servo amp suffer from to slow and outdated op-amps, introducing distorsion with todays high class music sources. Listen to piano for example.
Introducing better components can improve it , but this flaw will always remain unless you do a major circuit update.
Dick. |
|
|
| RetroAudio |
| I agree, there's much that needs to be done to make things sound better. Perhaps the stock unit sounded good in it's day, but when I first started listening to one, I wondered what all the fuss was about. It sure didn't line up with all the hype I had been reading about and it wasn't the best sound I had ever heard, but not the most offensive either. |
|
|
| sasha |
EQ in my setup is going to be via Behringer I am using anyway and
this way should be better than analog filter (hopefully):whazzat: .
Front end is TL074 (quad) with jfet input. Basic circuit, only feedback
resistors (2) and no caps at all.
Main complain about this amp was use of heavy feedback.
I will try to play with it (lower it) as much as possible.
Regardless of distortion, well up to 1% is ok.
As soon as I get my distortion meter (in the mail) I will be able to measure distortion and determine how much to lower the feedback.
In my mind this setup :
1: class-a
2: tube amp
3: no transformer in signal path
4: electrostatic speakers
can't be bad!!!
:D :D :D |
|
|
| RetroAudio |
| sasha - once you get the amp up and going, you can start modding like the rest of us and things will get more interesting yet. The sky's the limit! |
|
|
| sasha |
Thanks for the support RetroAudio.
The rest of the project coming soon. |
|
|
| sasha |
Today I put all together and it was partial success.
Amp is working, good pretty clean swing, but the bias is too low.
Probably because I am using different tubes than original ones.
Also when I play 1K sine there is sound coming from the tubes
like running water and it is pretty loud.
My main problem now is low bias, only 1.5mA instead 5-6mA.
Any advice what to do to increase bias? |
|
|
| RetroAudio |
| sasha - perhaps you're finding out that the tube might not be the best one. If your circuit is correct, then perhaps the tubes have a different transconductance that will allow only what you're finding. That's only a guess , I'm no true expert on tubes. Not sure what else to tell you at this point besides to verify your bias circuit values. |
|
|
| djmiddelkoop |
Sasha don't let your first result let you down.
The choice for 6HV5 is a good one, altough I still have to try it.
Your bias is indeed to low, ofcourse because you are using a different tube in the circuit for 6HB5.
What you could try is to increase the G2 voltage by increasing the +150V on the bottom tube. Consequently the upper tube must change to if you use this methode. This can be done by altering the 470K resistor.
The other methode is to lower the G1 voltage on the bottom tube.
This could be done by playing with the 100k collector resistor to the -150V.
Your goal is to have 6mA bias with half of the supply voltage at the cathode of the upper tube.
DJ |
|
|
| sasha |
Hi,
Actually I manage to solve bias problem by lowering G1 voltage on the upper tube.
First I tried with bottom tube but that was no good
because preamp circuit developed distortions with change of 100K
resistor. On top of that I didn't realize that tube I am using is
triode and original one is pentode as "rcavictim" pointed out.
He suggested to connect G2 to cathode and omit voltage divider and
150V to G2 so I did. Now bias is fine.
Next problem was proximity of high voltage lines and feedback line.
This gave me some peaks in output signal when close to max swing.
Silicon should solve that problem.
I remember concern from one member regarding not enough
current produced by output needed for driving speakers on high freq.
Today I watched music signal on RTA to see how much of high freq.
is actually in real (music) signal. I compared 2K and 16K and found that 16K is on average 15-20 dB below 2K which was the strongest
in the spectrum. That means 16K has about 32-100 times less swing.
I hope my math is right. This suggest we should have enough current.
I was able to put 40mA in my tests through dummy load.
As soon as silicon I applied is dry I will try speaker load.
Distortion figures are ok. Only 0.5% THD on full swing (9KVpp) and
3% fully loaded (dummy load).
The rest is coming soon. |
|
|
| rcavictim |
Sasha,
Glad to hear that my consultation was helpful and that you have solved your low idle current problem.
Rob |
|
|
| djmiddelkoop |
Good job Sasha !
Your math is right, don't worry.
Don't forget the EQ at the input also attenuates the high freq quite a bit, so the figures at the end are even much lower.
I don't have any problems driving big ESL with long high voltage cable( =even more capacitive load), so it must be right.
BTW, i measured one of those VDR from the original Acoustat X used for protection, and it kicked in at 68V.
DJ |
|
|
|
|
|
