Jordan JX92s MLTL

Well, I have finally got home, and in the past week or so, managed to finish the MLTL kits for the Jordans. I basically built it stock, but replaced the speaker binding posts with some gilded copper ones (from DIY Hifi Supply) and the internal wiring was with Amber HD 2 core speaker cable. I had an opportunity (reasonably rare) to listen to them for the first time last night for about one and a half hours. The amplification was Decware Zen using all Russian tubes and the source was my iPhone. Unfortunately, I did not have the time to set up the system.

Anyway, the room was a fairly large room - 8m by 6m with 2.4m ceilings. It was late at night, so very little ambient noise, but I also had to limit the noise as well. Technically, I had the volume control 90% on my iPhone to get to good levels, and it was probably a bit too loud given the time of night. But I didn't find the volume wanting. I still had a little spare, and the amp didn't show any signs of clipping. So, the 2W amp seemed to be okay for moderate levels of listening. Well, this confirms again, that I do not need ample amounts of power, and an amp in the 8W to 16W range may well be all that I need.

Onto the sound. Well, I was expecting something special, but I didn't get that. I supposed that there should be a longer break in time for the speakers, but the first time listening to these, didn't leave me wanting for more. Yes, the bass was present, but not as strong as I thought it'd be. And in my mind, the my initial thoughts with my (very well broken in and dustcap crumpled) FE127e closed bipoles was this: The details are much more clear and precise on the Jordan. The sound on the JX92s was cold and analytical, compared with the Fostex, which was warm and engaging.

Owing to the very short session and the ad-hoc nature of the first listening test, I will not form an conclusions. I will find a way of allowing them to break in for say 100 hours or so, before I firm up any decisions. What this has spurred me to do is to hurry up and get the Lowther OB up and running with the DX3 and the Beyma Coaxes...

The journey continues.

Beyma Crossover components

Here are the components that I ordered for the crossover for the Beyma 15XA38Nd Coax. I have chosen Jantzen components, in particular copper foil wax coil and Jantzen Superior-Z capacitors. The components are the same as per the standard Beyma crossover.

Jordan MLTL

I had purchased this set of MLTL enclosures from Decibel Hifi in Queensland many years ago, and never had the chance to put them together. In terms of the Beyma box and other projects, this is actually the quickest and easiest speaker to assemble to get some high quality speakers. So, here are the progress photos. I plan to seal all glued edges with butyl silicon, and eventually vinyl wrap the exterior. More progress to come when I return permanently back home in October. 

Beyma Coax 15XA38Nd Option - Ported Box Plan

The following is the outline drawing of the box with the port and using solid panels. I have re-arranged the bracing as well as the damping and insulation. The front panel is still made of two pieces of 12mm ply wood. There are two main reasons for this. The first being that I can get pre-veneered 12mm ply, but not 24mm. And the second being that the driver depth is 12mm, and thus I can flush mount the driver without have to rebate the edge.

Note that the design also allows for the port to be closed with a port cover. So that I can convert this between a ported and closed box design. In addition, I would like to experiment with varying the port between fully open and closed, that is, by stuffing it at various amounts. The estimated weight of this cabinet is around 46 kg without the driver and the binding posts. The speaker weights about 6.8 kg, and therefore the total estimate weight would be in the vicinity of 53 kg per speaker.

Beyma Coax 15XA38Nd Option - Ported Box

The enclosure speaker design requires the use of a Linkwitz Transform circuit to get a predicted flat response down to the low 30 Hz range. As an alternative, I modelled a ported version using the same enclosure size. The ported has an area of 250 cm3, and is back mounted about 150mm from the base of the cabinet. See below on the comparison between the two on Basta. I have chosen to predict the output at 8W power amp.

Comparison Between Closed an Ported.

The ported response is quite similar until 50 Hz, where it starts to diverge from the closed box scenario. It should be noted that the efficiency of the ported box can be seen, as the cone excursion is quite a bit less than the closed box when it gets to 20 Hz. I then further took the box design and put it into MJK's MathCAD spreadsheet. The spreadsheet used is the ML TQWT dated 11/07/07. The following are some screenshots of the results. The enclosure parameters are the same as per the drawing design in the previous blog entry.

Basic predicted response. 

Predicted response with Baffle step loss accounted. 

The final result with the BSC added, with the BSC being 20R and 12 mH.

I also modelled the same BSC in Basta and got the following result. It should be noted that in all the modelling, 8V or 8W was inputted as the amplifier power.

Basta simulation of the ported box with the BSC modelled as well to match MJK's spreadsheet.

Currently, I am working on an alternative cabinet using a 24mm solid thick wall, with dual 12mm front panel. This should make construction easier. In both designs, there is an assumption of internal box stuffing. Whilst Basta only allows for a percentage, MJK's spreadsheet allows for a more accurate assessment. In the design, I started at 8 kg per cubic meter. The resultant total weight of insulation was 1 kg and resulted in a reduction in the ripples in the response. I had originally place a thick layer of insulation between the bottom of the port and the base. But in MJK's simulation, it reduced the bass response significantly. And hence, the layer of insulation to the bottom of the enclosure has been removed. It also appears that over stuffing the box will result in reduced bass output.

The dilemma of practical construction will be ensure that the correct amount of stuff is placed, to reduced internal ripples and resonances but also not too much as to reduce the bass output. There will be lots of tweaking and construction sequence issues that will need to be resolved as this build progresses.

Beyma Coax 15XA38Nd Option - Closed Box

This is my design option for a closed box design for the coaxial 15XA38Nd speaker. I have purchased this for over two years, I have not yet had the chance to open them since my departure from home. As I am relocating back home, I have decided that this is the first speaker that I should be building.

Cabinet Plan.

The following is the bill of materials. I have elected to use two layers of 12mm plywood as means of construction and only allowed for dampening around the inside faces of the cabinet. By using two layers of ply with a suitably, softer (i.e., less stiff) glue, I hope to reduce the panel resonance.

Bill of Materials

The results in a box with an approximate volume of 115 litres not included any of the dampening. I have run the simulation in Basta for this design and the following are the simulation results. The crossover is the same one as the standard Beyma crossover that can be bought separate from the speakers. The design is a relatively simple 2nd order crossover with a zobel network on the 15" low frequency driver. The approximate crossover point is 1.8 kHz. The components for the crossover that I have chosen is Jantzen Superior Z-cap, Jantzen Wax coil inductors and Jantzen Superes range.

The low frequency driver has a inductance of 1.5 mH and capacitance of 15 uF. The high frequency driver has an inductance of 1.5 mH and capacitance of 2.2 uF. The LF zobel is 8r2 and 8.2 uF.

Base Scenario.

Base Scenario with Linkwitz Transform.

Base Scenario with Linkwitz Transform and Room Gain. With this scenario, the source voltage is 6V.

So far, the predicted performance is pretty good. So the process now is to order the crossover components. The next step is to order the plywood cut to size followed by the damping materials. With the LT bass boost, the Xmax of the driver limits the power output to 4.5 W, which produces about 6 V and a cone excursion of 4mm. The estimated SPL at 1m would be around 103 dB.

Re-direction Part 3 - First Verification

This will be a long entry. Last weekend I was back home for a few days, and I took the opportunity to set up my speaker system. The main purpose of this was to verify two things - listening and volume and bi-pole speakers. The set-up was quite simple and there was no effort to optimise cabling, speaker position, etc. The amps were my previously completed LM3886 chip amp and the second was the Decware Zen kit. Both of these amps have been completed for over 3 years, but have had very little chance to run due to life circumstances. The speakers were a 'closed' bi-pole FE-127e towers.

A few notes about the amps. The Chipamp was made from the stock with no upgrades or modifications of any kind, had it was not a dual mono. The Zen was significantly upgraded from the kit. All resistors and most capacitors were changed. It has been modified to use a 5U4G rectifier and has a large 40uF ASC oil cap. The tubes used were all Russian 'EV' grade 6n1p and 6p15p.

The first arrangement was the Chipamp with the speakers set up as dipole. That is, both of them moving in the same direction (even though they are mounted back to back). This would be akin to an Open Baffle arrangement, as the dispersion pattern is a figure 8. Previously, I preferred this arrangement, and my starting point was from here. The source used was by QA350 running on battery power and with a range of CD ripped or HD Tracks converted WAV files.

First time, everything worked. Which was quite a relief. I will now talk about the Chipamp and Dipole in a whole. The sound field was strange, and with the nulling effect at 90 degrees of the speakers was quite prevalent. The sweet spot was indeed a spot, both vertically and horizontally. The speakers are mounted a little lower, around 800mm off the ground, and was lower than my ear level. I found that I had to be pretty level with the speakers and in the middle for the sound to be good. But when I was in the right spot the sound was pretty good. After listening to a few familiar tracks, I realised that this particular combination didn't sound right to my ears. So I decided to switch it to a bi-pole configuration.

In a bi-pole configuration, the speakers are wired so that they both move in the same direction relative to the speaker baffle face that they are mounted on. That is, they both move in to the speaker and out of the speaker at the same time. The net effect of this, was a dispersion pattern that is very omni-directional. This was quite obvious when standing 90 degrees to the speakers. Unlike the dipole, there was no reduction in volume.

Sitting back and listening, I found two things. Firstly, the sweet spot is much larger and more forgiving. I could imagine a few people being in that spot. Secondly, the sound stage and sound field was much larger and there was more immersion. After a little more listening, I quickly realised that this was the sound that I preferred.

After I got to this point with the speakers, I decided it was time to try the tube amp. I cranked it up with the Chipamp and did some casual measurement on my iPhone. The levels at the listening position was around the 80 dB marked, and hovered around there with peaks just shy of 90 dB. At this point, I realised that my goal of 100 dB plus at the listening position is not required for me. Therefore, I have to revised my design criteria, and it seems that I would like the amp and speaker combination to have total capacity to reach somewhere in the 90 dB range at the listening position. At this point, I am thinking specially between 96 and 96 dB. This would give some additional headroom, and would significantly open up the speaker and amp combinations.

The spec of the FE127e is 91 dB, and in a bi-pole configuration, it should be around 94 dB, which, in newly revised, should only need a 2-4W amp to reach the desired levels. I did notice that listening at loud levels, I did get a slight headache, and it was fatiguing. Hence my next move to change the amp.

Plugging in the Zen did not alter the sound dramatically. However, having only listened to these in the span of a few days, I imagine it'd be difficult to find all the differences. However there were two important differences that I noticed with the tubes. One is that it was not fatiguing at all at any levels. Second, the sound was more full, whereas the Chipamp was more sterile. So in this short sessions, I have decided that Chipamp (and solid state) was not the way for me moving forward.  Once I settled on this, I spent probably 6 to 8 hours in total listening to this setup. And it was really a delight.

Overall, the sonic presentation was natural, easy and spacious. The bass was not loud, but was present and without authority. This is one area that is becoming more and more important. That I need to really find a good speaker to supplement to provide the missing bass presence. With the Zen, two observations was made. The first, is that almost at full volume, the levels were around the very low 90 dB, maybe hitting 92 dB. This was probably as loud as I would want to listen to for a period of time (more than peaks or more than a track). Secondly, I found that classical music, which to to date, has been a little disappointing in my other setups, was very pleasurable. It didn't seem constrained and didn't feel like it was running out of steam.

So, the little effort that I made this trip provide to be very valuable. The following items have been confirmed.

1. Peak levels would be in the low 90 dB range, and thus total system response should be at least 96-98 dB at the listening position.
2. I prefer bi-poles to dipole. I didn't get to chance to wire it as a monopole - though I was thinking about, but I was just enjoying the sound so darn much.
3. I prefer the Zen to the Chipamp. And thus, I could make the extrapolation that I prefer tubes to solid state sound, or Class A to Class D. 

This has been both important and enjoyable, and now I have a few months to prepare my plan of attack before I am back at home permanently.

Beyma Modelled Design

After doing some very quick reading, I have assumed a room mode (in Basta) as per default. See the below plot which includes;

• 1100 high, 500 wide and 400 deep box with vented box tuned to 33.3 Hz at 150mm diameter. 
• Target of 175 litre volume with 40% Viso. 
• Active filter at 2nd LP at 700 Hz and 2nd HP at 20 Hz.
• Baffle Diffraction Effects on, but no BSC modelled.
• Room Gain on

The result appears to be flat to 40 Hz at around 95 dB, -1 dB at 36.7 Hz and -6 dB at 28 Hz. Pretty close to the target I was aiming for. I am keep to the 4mm max, the power limit would be at 50 W, which would have an output of around 112 dB, or 110 dB at 35 Hz! I'd say that it is a pretty good simulation result.

With various options modelled, at 1 Wrms.

Modelled at max X-max.

Same as the first plot but with room gain turned off.

With the above, the 'mic' distance was far field. When it is changed to 4m near field (i.e., listening position), 50 W power would result in an average level of 100 dB, which is at the top end of my requirements. 16 W would get me to about 95 dB. What this means is that to adequately power this speaker will require an amp of between 16 W to 50 W.

Beyma 15XA38Nd Box - Further design

Following on from previous blog entries: here and here, I have recently dug out the Martin King's MLTL spreadsheets and applied a ported MLTL design to the Beyma 15XA38Nd. From all the various modelling studies and trials in MathCAD and in Basta, the bass section of the Beyma really does need a larger box to get the most out of the bass response. In the previous scenario, by using a 300 litre box, the modelled response in Basta was almost flat down to 30 Hz with no bass lift. What that really means, is that the design will be an exercise in compromise - bass verses box size. 

But the question in the choice between bass response and box size must include room gain. As I read in a few sites and books, modelling flat down to 20 Hz does not guarantee a good clean bass. The room modes will inevitable add to the bass lift, and if it is flat, it may need to be equalised to reduce the bass response, so that the overall bass response in the room is better. The below shows the difference between a 140 and 250 litre cabinet in Basta. The differences according to the MJK's MathCAD models is less apparent. The difference is basically a cabinet depth of 400 mm verse 650 mm - quite a change. Obviously if I played with the width as well, the depth increase would be marginal. As modelled, the width is 438 mm and the height is 1,100 mm, quite big already. 

Basta with 140 l cabinet.

Basta with 250 l cabinet

Based on the Basta model, here is the various points at which the predicted response hits 90 dB (-6 dB) and 95 dB (-1 dB). Generally the other frequencies sit between 96 and 96 dB.

140 litres = 34.5 Hz and 52.6 Hz (∆ 18.1 Hz)
160 litres = 32.8 Hz and 48.8 Hz (∆ 16.0 Hz)
180 litres = 31.7 Hz and 45.2 Hz (∆ 13.5 Hz)
200 litres = 30.9 Hz and 41.8 Hz (∆ 10.9 Hz)
220 litres = 30.4 Hz and 39.1 Hz (∆ 8.7 Hz)
250 litres = 29.8  Hz and 36.3 Hz (∆ 6.5 Hz)
300 litres = 29.1 Hz and 34.0 Hz (∆ 4.9 Hz)

I have also added the difference in Hz between the -6 and -1 dB points. It interesting to note that there is quite a difference going from 180 to 200 litres, as compared with the other steps in volume. What the delta represents is the sharpest of the corner, obviously as the volume gets larger, the corner point - per se - becomes more and more sharp, and eventually becomes a hump. At 1000 litres the hump is about 101 dB, but the 90 dB point is at 28 Hz.

Looking at the various graphs, I would have to say that as a pure modelled frequency response 240 litres does look to be very appealing. But as mentioned before, I must consider room gain, and then determine the volume based on that parameter. So now, the next step is to estimate the room gain in my two potential listening rooms....

Re-direction Part 2 - Speakers

Described by many as the heart of the system. Finding and/or designing the right speaker is probably the most important, as the approach I am taking is from the ears back. There is so much discussion and research on speaker design, and most of which is well beyond my capability to completed comprehend. However, from my research, there are number of key issues in obtaining the best quality speaker seems to be.

Crossover points - avoid if possible 250 Hz to 4 kHz, about 4 octave range. This is the most critical range of musical information and many describes it as the most critical point to avoid any crossover.
Use of neodymium or Alnico magnets seem to lead to higher quality of sound.
Use of cone materials that prevent breakup well outside (i.e., 2 octaves or more) the frequency range being played on the speaker.
Different speakers were designed to different enclosures, make sure the enclosure fits the speaker. General rules of thumb like using high Qts on open baffles, etc.
Important to have a smooth frequency response with good dispersion.

If we take the above first two items in our next design assumptions, we would be left with three speaker system. One for the LF work under 250 Hz, and one for the mid-range work from 250 Hz, and a HF unit crossing between 5kHz to 10 kHz that would take it to 40 kHz. The choices for the various drivers would be limited in the MR and HF areas. But once we take into consideration all the criteria, the range of speakers that can service in our requirements become somewhat limited.

However, in recent review of components and equipment, I have realised that I have amassed a range of speakers. FE207e, Lowther DX3, Saba green cones, Beyma Coax, Joran JX92s, FE127e, FT17H, and Alpha 15A. To be honest, to get good speakers cost money. And living in Australia, the cost of shipping of most speakers are quite sizeable. So the vein of being economically responsible, I will have to design a speaker system that uses one or more of the existing speakers in my collection. It appears from what I have, that it is mainly full range speakers, with the exception of the Beyma Coax. All the other speakers will mate well with a sub crossed at below 300 Hz or so, and can all probably do with a super tweeter (Fostex FT17H).

Therefore, it leads me to designing two speaker systems. One, the full range with sub and super compliment, and the other around the Beyma Coax. My next post will deal with the design of the Beyma Coax as a possible contender.

Revised Assumption

Having read and reviewed more about my desired listening levels, I have come to the conclusion that 100 dB with 6 dB of overhead is probably a little too much. I believe that a value of 90 dB at the listening position is more sensible. Therefore, my assumption would change from a required combined speaker and amp able to deliver around 96 dB at 1 W at 1 m. Given a small buffer, the target range should be 95 to 98 dB. Of course it would be nice to have additional headroom of say 6 to 9 dB, but given my basic premise, I believe that the revised SPL levels are more realistic.

Verification of thought process

Since I will have a short trip home in the coming weeks, it is my plan to verify my second assumption, that is the 104 dB levels. My goal is to take out my pair of closed FE127e towers (yes, those with the front dust cap crumpled) and play them using my Decware Zen and then LM3886 Chipamp in the 100 cubic meter room. To hopefully negate the effects of the crumpled dust caps (yes, they are both now crumpled…) I will rotate them around and use the 'back speakers'. Wonderful thing about dipole and bipoles! The towers are made from 19mm thick Tasmanian Oak and have light filling. But the top and bottom are loose, leaving plenty of opportunity for tweaking.

The drivers are all terminated separately so that I can experiment with dipole, dipole and the parallel and series. I will listen to my favourite music, and see what levels are acceptable. If, and I stress if, they are suitable with the Zen, then I would see so many options opening up in both speaker and amp combos. If they are no loud enough with the LM3886, then I really need bigger amps and/or higher sensitivity speakers.

Through this, I will also experiment (hopefully) with the monopole, dipole and bipole arrangement. Which should give some further guidance as to my preferred arrangement. The source will be my QLS350 straight into the amps. I might try bringing in my HT sub, to see the effects of low end reinforcement. I will consider this my most important experiment. Speaker configuration, low-end reinforcement and power level combinations.

Note - to be honest, I am keen to see if I can get these to work well with the Zen, and then I can upgrade them to the F120A or even the new Jordan full rangers!

Re-direction Part 1

After much pondering about my audio system, especially on my imminent return home, I have been through many of the same processes as I have done in the years prior to my departure from home. At the end of the day, it always came back to the same things as I have covered in my previous blog entries. Namely, back and forth between an all tube amp system or a tube plus SS system (for more power) and that of the speaker system. Which has oscillate between a high efficient system using horns, or a lesser (but still quite) efficient system using a traditional full range speaker driver.

Thus, I have come to a different process of deciding. A little background about my gear generally. In all my interests and hobbies, I plan and want to just have one - what is perceived by me as the ultimate, and then that is it. Not that it has to be the most expensive or best, but something in my mind that is no compromise. That is case for my cameras, computers, watches, cars, bass guitar, etc.. I have found that going around from low end to high end, ends up costing more money and time. But in the audio field, as I started with very little knowledge, and wanted the DIY route, it was natural that I take a longer journey.

In terms of headphones, I have I have almost reached that point. With my (soon to be built) C3g amp with HD650. Even with my existing range of built amps (about 6 in total) and headphones (over 4 high-quality pairs), I am more than a happy camper.

With speakers however, it has been a (very) long journey. Mainly due to moving houses, moving country and having children. But right now, I am in a better spot to move towards that end goal. So, instead of choosing an amp and speaker, I am going backwards, and choosing what I want in terms of sound, and then working back the audio chain. That is, desired sound, sound levels, speaker, amp and then source.

What I want is a system that is able to reproduce audio between 20 Hz to 40 kHz with high fidelity and ability to engage me beyond what can be 'objectively measured' - that is, it gives me enjoyment. What would I mean by high fidelity, I would assume low distortion and the frequency response to be generally smooth to within 3 to 6 dB in the frequency range, with a drop of say -6dB at 20Hz being acceptable.

Being honest, my listening room will never ever be larger than 7.0m by 5.3m by 2.7m high, so about 100 cubic meters. But at the moment, the listening room is 4.7m by 4.0m by 2.7m, about half the volume at 50 cubic meters. The system should be built for those two rooms.

The next question for me was how loud? I have been looking around the Internet, and measuring the values that I typically listen to. It seems that a value (for peak levels) of around 103 to 109 dB was about right, with most program levels being around 80 dB to 90 dB at the loudest normal levels.

In the larger room scenario, the listening position would be about 4m from the speaker, and the smaller room, it'd be 3m from the speaker. Which is not all that different. From this information, I would then determine the appropriate levels of speaker and amp performance to meet those dB levels. A listening position of 4m is about a 6 dB loss and at 3m, it'd be about 5 dB. In either case, the differences are marginal, so let's assume 6 dB loss due to speaker distance.

At this point, I will ignore room gain and other incidental losses, as they compound the complexity. For peak levels, let's assume for simplicity sake 104 dB. Therefore, the entire system of amp and speaker should be able to hit 110 dB. Therefore the first assumptions that are quantifiable are:

1. Frequency Response range of 20 Hz to 40 kHz. -6dB at 20 Hz and within 3 to 6 dB over the range (that is plus or minus 1.5 to 3.0 dB).
2. Combined maximum SPL levels of system should be 110 dB at 1 W at 1 m.

With the above in mind, we can now start to tinker with combinations of speakers and amps that meet the above requirements. The first for me is the philosophical combination of the two devices. So, if the speakers were to be 101 dB efficient, the amp would need to be at least 8 W in power. If the speakers were 92 dB, the amp would need to be at least 64 W in power. So, as the speaker efficient increases, the amp options open up for low powered tube amps. And as the speakers become less efficient, the ability to use tube amps rapidly decrease. The table below summarises the combinations.

Speaker Sensitivity // Amp Sensitivity
110 dB // 1 W
107 dB // 2 W (45 Amp)
104 dB // 4 W (2A3, AD1 Amp)
101 dB // 8 W (300B, KT88 SE, F2a Amp)
98 dB // 16 W (6C33, 211, 845, GM70)
95 dB // 32 W (211, 845, GM70)
92 dB // 64 W
89 dB // 128 W

Given my general interest in using Class A Tube amps, where generally SE tube amps limited to below 16 W or so, and PP tube amps around 32 W for Class A, it would appear that unless the speakers were to be at least 95 dB efficient, the use of tube amps would be not possible. Even with the rough limits mentioned above, the voltage and/or current requirements are quite large, limiting the ease of DIY. However, the choice in the number of amplifiers used has not yet been addressed.

Therefore, if I want to use tube amps (and I do) then I would have to find highly efficient speakers. This would generally rule out most of the speakers available. Focusing only on the mid-range (or pseudo full range) would limit the choice to high efficiency speakers such as 'traditional' full range family (e.g., Lowthers, Fostex, Supravox, etc.), horn loaded compression drivers, and other very efficient mid-range (generally) PA speakers.

Well, a longish entry, Part 2 of this will cover the issues dealing with speaker selection.

Relaxed session

Well, I had the first session in a long time with my EM4 in a somewhat high quality arrangement. I was using the O2 headphone amp and the D-zero DAC from my Macbook, using Fidelia as the source. Short commentary.

Source does matter. Some files were better, and some were worse. But I must say that hearing Bach's Cello suites through this setup was very intimate. In the stillness of the afternoon, and using IEM, I really felt like I was in the room with Janos as he plays the cello suites. On another thought, 1812 Overture was just flat, even on 176.2 kHz source file. I think that complex music just doesn't play well with the setup. Not sure why, but it was just not as inspiring as other kinds of music.

But, it was a refreshing and enjoyable session of one and a half hours.

EL84 PP Idea

Well, I was on the train home and was thinking about the Beyma. I was considering have it passive crossover. Using copper foil inductors and Jantzen caps, it turned out to be around US$350. I was then thinking what if I just built two simple EL84 PP amps instead, and then I can experiment with a cheapie PA active crossover. This was, I can tune them and then once the sweet spot found, I can do a hardwired active crossover. Would that be cheaper?

Anyway, more incoherent thoughts later, see below. 2.6W in SE mode, and thus is Class A PP, it'd be 5.2W. 2nd harmonic should cancel out (mostly) and the 3rd harmonic should sum, but the predicted 3rd harmonic is pretty low. The PSU below using a choke input. It could be a simple, 'cheaper' build. But it could also be quite good. The output transformer is 8k in SE operation.

OB Design

Here is my idea for my convertible open baffle. The overall size is 1,100mm high and either 550 or 600mm in width. It will have a 90mm hardwood return frame that may have bracing and sound damping applied. The back will allow the addition of a boot or shroud that can make it a U frame, closed or ported box. The main issues will be how to make it rigid enough for the closed and ported box.

The lower driver will be an Eminence Alpha 15A. Simply because I have a pair. They could be changed in the future, but for the moment, that at it. The main speaker panels will be removable to allow me to change different drivers. I am not sure whether to have a separate panel for a super tweeter,or to have that sitting on the top.

But if this does work, I will have a very modifiable and tenable OB speaker that will fit my Fostex F207e, Lowther DX3, Sabe Green cones and potentially also Visaton B200, Supravox 6" or 8" Alnico, PHY or anything else I can get my hands on.

This, my Beyma Coax ported box and the Jordan TL will form my main speakers.

Oarta output transformer

It looks like the Lundahl LL1623 rated at 90mA would be the best choice for this amp. The amorphous core price is less than half of the Tamura, but more importantly, it is available. The interstage transformer would be the next choice. This would probably be the most important choice, given the different voltage swings that needs to be accommodated by the different tubes.

I think at this point, it would be important to at least try to limit the range of tubes to be used. This has to be correlated with my current stash of tubes. In this, I would say that there are four families that I would like to run:

1. 45, 145, 245 and 345. I have a number of pairs and have been universally hailed as an excellent audio tube.
2. F2a. I have three pairs of these long life tubes.
3. KT88/EL34 family. I have two quartets of the KT88 and EL 34.
4. 300B. I have one pair. 

Looking at the above, it would make most sense that I provide more effort towards the first three families given the volume of tubes on hand. With the 300B narrowed out, the voltage swings required are now limited to 30 Vpp to 60 Vpp, a little more sensible. This would mean that a tube with a gain of 20 would be plenty. At this point the 5687 and C3g seems the right direction. If I use a higher gain tube, I can use a step down interstage, which should be a good think overall. The other option, is to allow a different tap on the interstage to be used to connect to the grid of the output tube, which can select between say 1:1 and 2:1. But this may affect the performance and bandwidth of the interstage.

More questions... 

Oarta Output Transformer

It seems that the Tamura F5003 is in limited supply, and may not be available. So I may have to revert to another brand of amorphous core. Maybe Lundahl?

Revised Beyma Box

I have been playing around with the box model for the Beyma co-ax, and since then, I have made the following changes. 

1. 300 l vented box - a nominally 1.2 H x 0.55 W x 0.6 D box.
2. Vent tuned at 30 Hz.
3. HP at 20 Hz, 2nd order.
4. LP at 700 Hz, 4th order.
5. No Linkwitz transformer.
6. Source voltage at 14 Vrms, which is around 25 W amp.
7. Disabled baffle step correction.

At this condition, I was able to get the following predicted response. Note that the driver excursion is limited to below 4mm. At 25W power, it is giving around 100 dB, with f-3 at around 27 Hz. Fairly decent result, with less tinkering with EQs. Bigger box, but that may have to be the sacrifice. 

Beyma Ported Cab

To design a simple box for my Beyma 15XA38Nd, I have employed Basta. Using a 200l ported box with two 75mm diameter 147mm length ports (tuned to 25 Hz). The box is also fairly standard at around 1m in height, 0.5m in width and 0.4m in depth. Something easy to construct.

My speaker box model also includes a Linkwitz Transform at 20 Hz along with a 4th order low pass at 1,800 Hz in tandem with a 2nd order high pass at 15 Hz produces a near flat frequency as model between 30 Hz to 700 Hz. With -3dB points around 20 Hz and 1 kHz. See below.

Crossover at 1.8 kHz and maximised bass response

There is quite a lot of filtering in there to achieve this. The 1.8 kHz crossover was set from the standard Beyma passive crossover (2nd order) tuned to that frequency. I think that 1.8 kHz seems a bit high for a 15" driver, so if I set the model to 1 kHz instead, it gives a much more 'flat' predicted response. Note also that the Xmax of this driver is 4.0mm and according to the model we hit 3mm at around 20 Hz with 1W of input. I may have to tune the high pass from 15 Hz to something higher to limit the response. The below shows using a 4th order HP at 20 Hz, which limits the Xmax to under 2mm at 20 Hz. 

Crossover at 1 kHz and slightly reduce bass response to limit Xmax.

Given my listening room is only 5x4m in size, I think that a flat response down to 30 Hz may well be plenty enough. My question now is how to experiment with the various cross-over points? The easiest solution would be to employ a PA style crossover, something cheap from Behringer or similar. Which would allow me to test and tweak to my heart's content. 

Another option would be to make the box closed, and with the same settings as the second scenario above. The response is still pretty decent. 

Closed box with same parameters as above.

To be honest, if the speaker performs anywhere near what the model shows, it'd be great. Even if I were to use a closed box or vented with the standard passive 2nd order filter, the response is still pretty decent. See the two simulations below.

Passive crossover ported box, no EQ.

Passive crossover closed box, no EQ.

Oarta PSU

One interesting point is that if the PSU is cap loaded with a 10uF, at 200mA, it will give around 380V. But if I remove the cap, and have it choke loaded (using the LL1673), at at loads of around 100mA to 150mA, it will give voltages of 260-280V. Perfect for 2A3 and 45. So in this case, if I can provide switchable PSU filter, I can further extend the range of usable tubes, rather than using resistors to drop voltages.

Oarta - Some Design Images

PSU Design.

KT88. 400R cathode 

300B. 740R cathode

KT66. 300R cathode.

EL34. 400R cathode - can share the same plate as KT88.

6550. 400R cathode - can share the same plate as KT88.

The Oarta - One Amp to Rule Them All

It has certainly been a quite a long time since I have done ANY DIY audio stuff. I blame this due to my relocation to Hong Kong for the last 18 months and having young kids. There just ins't time. Well, with the imminence of return back home on the horizon, I have started to dream up what I will finish up. In this long time away, I have realised that I like variety. But I don't want to build too many amps. So, I have taken a few ideas, based around a SET and merged it into one amp. Which I call - Oarta, or One Amp to Rule Them All, not exactly creative.

The basic premise of this amp is that the PSU will allow the supply of around 390V at 200mA max, with heater windings two sets of heater windings, 2x 6.3V at 2A and 1 or 2 5.0V at 2. This should allow me to use a large range of tubes that I have on hand: F2a, 300B, KT88 and EL34. And also open up a whole range of other tubes that can run at those voltages.

The second part is a universal output transformer. I have chosen a 5k tap, but may move to a 3.5k for more power. For this amp, it will be no compromise, so a amorphous Tamura F5003 or F5002 will be on order. The driver stage will be a medium mu (think around 20) that is interstage connected to the output. To accommodate the wide range of output tubes, I will probably want a higher gain stage, so C3g triode connected may well be the final choice.

The trick here is to allow the switching of the output tube. To this, I am thinking of an adaptor plate that allows the socket and cathode resistor to the changed. The bypass resistor will then be mounted on the chassis. In this way, I can have a large film cap that isn't changed with each socket. On each adaptor plate will be 5 connections; grid, plate, cathode, heater + and heater -. I need to find safe and reliable connectors that allow me to change.

The final trick is have a 6.3v and 5.0v winding, and finding a method to accommodate a range of heater suppliers. For the indirectly heated 6.3v tubes, this is no issue, but with the DHT tubes, like the 300B, PX25, etc., this will need some more work. There may need to be an allowance for something like the Tentlabs filament supply, that can run from the 6.3v to power a 5.0v and run from the 5.0v to power a 4.0v tube.

If this can be made to work, I have an amp, Oarta, that will allow me to change tubes, without having to change anything else. This will allow me to not skimp on the rest of the components, knowing that it will be universally applicable. At the moment, my current thoughts for the design are:

PSU.
Amplimo 7N1474 Transformer (here - I may need to supplement with additional heater taps)
2x 6AU4GT diodes
Cap input 10uF ASC Blue line
Two LC stages with Lundahl LL1673 200mA (here) each with ASC 50uF.
I am thinking about a separate LC stage (15H with 40uf) for each channel and an additional LC stage for the driver.

The main issue with the PSU is finding space for all the ASC oil caps and the chokes.

This will give me around 386V of B+ at 190mA current draw (simulating with PSU Designer). At this B+, the following would be the power output with different tubes (from SE Amp CAD) using F5003 all running at 80mA per tube.

300B: 7.6W with 120 Vpp
KT88:  6.8W with 62 Vpp
EL34: 8.8W with 60 Vpp
KT66: 4.5W with 46 Vpp
F2A: 5.8 W with 32 Vpp (by hand)

And I would think that this would also work with tubes like GU50, PX25 and some of the other more rare European or exotic NOS tubes. And if I don't mind burning off voltage as heat, it could be used on 2A3, 45,  AD1, etc. - again, options. Beyond these, there are still a number of issues to work through, but in my mind, the concept already has been resolved. I am now thinking about cathode bias for the driver, and most likely will head down the LED route. Given the C3g triode seems to work well with around 2.0-2.5v bias. 170V at the plate, 2.5v bias 15mA. Nice and simple.

The main course of action now, is to determine how best to build the adaptor plate, this will make or break the design. It must be safe, easy and high quality to preserve the rest of the amp design. And the last question is how to deal with the different voltage swings required by the various tubes. Which differ by a factor of 4. I do not want to simply waste it by using an attenuator, and the first thought was to have taps on the interstage, which would allow me to choose the right ratio, 1:1, 2:1 or 4:1. This I think is possible with the Sowter 9525. But I probably want to avoid changing the IT ratios as that will change the frequency response characteristics.

Or I could just vary my source output - say between +0 dBV, +6 dBV and +12 dBV. (Which incidentally is possible with the Wadia 121 DAC that I am considering, see manual). Or, I could just accept it, and have to set the volume levels differently for each tube. This though, needs more thought and consideration.