Front panel update: It's in San Francisco, which I think means it'll be on an international flight soon. I was getting worried - 9 days to get from Seattle to SF?
B+ VOLTAGE REVISITED The 6DJ8/6922 datasheets listed typical plate voltage as 90V. My understanding is that as the Aikido uses dual-triodes in a totem-pole arrangement the voltage on each plate in a single tube will be half B+, making the required B+ voltage 180V. This seemed like a good place to start as it sat in between the recommended voltage for two 6DJ8 tubes (150V) and the 6CG7/6DJ8 combo (200V) as listed in the manual. The PS-21 PSU can be configured with the two series resistors R6 & R7 to output anything from 50-300VDC. Cutting a long story short, I decided to set the output to 210V. The table in the PS-21 manual showed that this meant a total of 42Kohm resistance would be required (a 12K & 30K specifically, but I went with 20K & 22K). Or, you can make use of this helpfully supplied formula: Vout = 1.25(1+[R6 + R7]/R5) So with R5 being specced at 249 Ohm, 1.25(1+[20000+22000]/249) = 212 if I'm adding the 1 at the correct point. Resistor R12 on each side of the Aikido board would then drop the incoming 210V to the 180V specified for each channel. But to find R12’s value I’d need to work out the total idle current of the tubes first - the sticking point.
The Aikido board begins to be populated with Takman audio resistors. I've used these a few times because I like audio woo and wasting money on inaudible upgrades . The salmon coloured ones are carbon-film and occupy positions like grid-stopper and grid-leak, while the teal metal-films are cathode resistors in the main. This is mostly colour coding for clarity's sake. The heat-shrinked jumpers J1 & J3 configure the heaters in parallel for the required 6.3V operation. J4 & J5 jumpers connect tube pin 9 (the internal shield) to ground. Needing, but not having, an 88Kohm resistor for R10 means joining a 68K & 20K in series instead. And here they are, with a little heat-shrink. In front of each of them is the 3 watt resistor R12 which drops the B+ voltage to 180V. Could I have just set the PS-21 to output the required voltage directly and then jumper the R12 position? Not sure. I thought that at least this way may spread the work a bit, allowing less stress/heat on the components. The tube sockets can also been seen soldered to the board here.
IDLE CURRENT, AT LAST Using the following formula from the Aikido instructions, I 'reverse engineered' what idle current the recommended resistor values in the manual would produce. Iq = B+/2(rp+[mu+1]Rk) or translated: Idle Current = B+ Voltage/2x(Plate Resistance+[mu+1]xCathode Resistor) Ouch, right? Working with this formula was initially hampered by the example equation in the instruction manual looking to be, well, wrong. The sum just doesn't seem to work. Also, in this equation B+ is assumed to be the full 180V, but I was more interested in working with the 90V half value, so the formula was rewritten as: Iq = B+/(rp+[mu+1]Rk) Anyway, if my sums were correct, at the input position the 6DJ8 example was set to draw around 4mA current, and the 6CG7 example 6mA. The output 6DJ8 tubes were set at 8-9mA depending on B+. Since I wanted the option to swap out the first stage 6DJ8s with 6CG7s the idle current in the first position would have to accommodate both somewhat; maybe higher than necessary for the 6DJ8, or lower than warranted for a 6CG7. Deciding on 6mA input and 9mA output for my build meant 15mA per channel, 30mA total. Using the aforementioned valvewizard PDF and a E88CC (6922) datasheet I was able plot the values for mu (roughly defined as gain) and plate resistance against these bias points. Now to confirm the cathode resistor values for both 6DJ8 & 6CG7 input tubes, plus the 6DJ8 output tubes. Another formula - a rearrangement of the previous idle current equation (with B+ again being 90V): Rk = ([B+/Iq]-rp)/(mu+1) The cathode resistor values: 330 ohm and 200 ohm. Going back to the idle current formula showed the 330 ohm value of the input tubes resistor should also allow a 6CG7 to operate at just below 5mA if swapped in, which isn't too bad at all provided it works out that way. With all that in mind, the R12 resistor value needed to drop 210V to 180V could now be worked out with this formula: R=V/I, or: Resistance equals Voltage divided by Current. Ohm’s Law; The first time I had ever used it. 30V / 0.015A = 2000 Ohm, so a 2K resistor was required. Pretty much gave that away in the last post.
The completion of the Aikido board with capacitors large and small. I really enjoy the look of those black beauties! In the reverse-angle view you can see the output coupling capacitor position is still empty. The reason for that is coming up. Nice clean underside.
TRANSFORMERS Once all of the electrical characteristics of the circuit were worked out, it was just a matter of finding a suitable transformer (or transformers) to power it. Transformers transform the AC voltage from your wall into the different AC voltages your power supplies require, which in turn convert that AC to DC for your circuit. The transformer input wires are the primaries and the output wires are the secondaries. For the B+ supply input on the PS-21 PSU board John Broskie advises a 250VAC 100mA transformer output for the configuration I'm using. This was in addition to the two 7-8VAC secondaries needed for the heater supplies, plus something with enough juice for the 5VDC 1A requirements of the volume control PSU. 4 power supplies. Separating out the heaters and giving them their own transformer seemed to be pretty standard practice, so as long as another could be found with a high voltage 250V output paired with a suitable low voltage secondary, two transformers would prove ideal. I did waver between Toroid or EI type transformers, researching their pros & cons, however in the end I went with whatever I was able to obtain cheapest. These were available with free shipping to my country from Mouser and looked to be suitable: Hammond 266L14 - 28VA 2x7VAC 2A Hammond 369AX - 40VA 250VAC 100mA / 6.3VAC 2A
With the board soldering completed, it was time to turn attention to the chassis layout. My plan was to order the chassis from Hifi2000 in Italy, having them drill all the necessary holes for me so I could get away with as little metalwork as possible. However this would require being absolutely certain of the internal layout before supplying them with the files. Grabbing a piece of MDF I started measuring and drilling the hole positions for a test fitting of the components. And here are the bones of the initial layout. (For some reason my photos are getting worse as we go along!) This is why the output caps weren't soldered to the board; they are just too large at 70mm long and 40mm diameter. These Miflex KPCU's were a ridiculous choice really - overly large and expensive - but I did fall in love with the idea of them and sometimes audio-jewellery-phoolery wins out. It took an age to find suitable and attractive brackets to mount them to the chassis, but I got lucky when I found what I believe are brackets made for drone armatures.
On to the preparation of the transformers. The required input and output leads were twisted into their respective pairs, and the remaining unwanted ones I separated and bundled together. The dual 120V primaries of each transformer had to be connected in series to configure the transformers for my country's 230/240V operation. I decided to use a terminal block to do this. The transformers have also been swapped from their initial positions in order to better feed the wiring to the PS-21 power supply board later on. The extra, unused wires out of the transformer had to be safely terminated. I crimped an insulated connector to each wire and covered them with heat-shrink. I then heat-shrinked them again so they are bound together, unable to come free. They would eventually be mounted against the side walls, out of the way.
In an effort to try and contain the stray magnetic field from the EI core transformers, I wanted to add a steel shield between them and the rest of the circuit. Not sure that I could find a local place to bend such a sheet for me (and not convinced it would look very pretty anyway) I fell upon the idea to use a steel side-panel from a different Hifi2000 case. My first piece of metalwork was to drill a couple of holes large enough for the transformer wiring to thread through. With the largest needing to be about 20mm diameter I was at a bit of a loss, until discovering what a step drill was. Now I love step drills! Rubber grommets - also nice.
OOPs, FORGOT A RESISTOR Looking back, it turns out the 88K resistor mentioned earlier in the thread is also one that varies in value. Paraphrasing the Aikido manual, resistor R9 & R10 form the part of the circuit designed to eliminate power supply noise from the output. This appears to be the hallmark of the Aikido circuit - low noise. It's formula is: R10 = R9([mu+1]/[mu+2]) My notes are a bit fuzzy on this point, but mu appears to change slightly depending on the voltage/current applied to the tube. Mu was listed as 33 on the 6DJ8 datasheet, but at the operating point I had chosen was more like 31-32. With R9 being 100K the formula above meant R10 would be 87.8K at mu 31 and 88.2K at mu 32. I just split the difference. Mu appears to be roughly explained as gain or amplification factor. In my notes I had also written this more technical explanation: "mu is the efficiency ratio of the grid compared to the plate in controlling current. So, 33 mu = 33x more efficient."
A lot of time was spent during early construction tweaking the layout for this project, but eventually it had to be finalised so I could order the bottom & rear panels. Without them I couldn't proceed much further. I had made sure to obtain all the necessary internal parts in order to accurately measure and place them into the layout plan, and this is what it looked like in the end. I placed my order, and sent the CAD files to Hifi2000 in Italy. I was worried that the process was going to be awkward, but Hifi2000 were very efficient and helpful. I don't have anything but good things to say about the way they handled everything. And here's the nicely drilled finished product (apologies for the blurry photo). The additional grills were positioned so they'd sit under the regulators and tubes. The rear panel: six inputs, ground lug, two outputs and space for the power entry. Lovely!
A test fitting of the transformer section, and thankfully all the holes seemed to be in the right place. At this stage I'm cutting the wires to the correct length, crimping ring connectors to them, and prettying them up with heat-shrink. I'm not a huge fan of wire work. It can be quite tedious, and if you get it wrong and cut too short it could be awful to fix.
Before the components could start being properly affixed there was some more metal preparation to be done. Although the Hifi2000 cases are generally high quality they do need a bit of help when it comes to strengthening where the panels join. Panels tend to be attached at the ends, leaving the opportunity for flexing in the middle. Anyway, it's probably easier to show rather than tell. I bought these aluminium threaded bars to brace the panels and, importantly, to also provide a way of making sure the panels were coupled for good earthing. Again, made for either drones or robot building, they were quite expensive and the threads were not metric but in some kind of ancient measurement known only to civilisations past. Does anyone know places to purchase these from? I found them very difficult to even search for. Drilling holes for the bars and hoping to be as super-accurate as possible. Mission mostly accomplished. The tape helps protect the anodising if you happen to push too far and contact the drill chuck with the metal, and also prevents the finish being scratched by the aluminium shavings. Not a bad hole.
In this project, a major focus of mine was proper electrical safety and grounding. It had kind-of been baked into the instructions for previous builds, but this time it was mostly on me. The first step was to make sure that if any part of the chassis was accidentally electrified, that voltage would have a path of exit through the ground pin on the IEC connector (which is attached directly to the chassis itself) and not through the body of the user. This way the case effectively becomes an extension of your house's ground system. Someone please correct me if I'm wrong. To accomplish this I spent some time sanding certain parts of the anodising back to bare metal. The anodised surface doesn't transfer electricity, so I needed to make sure each panel had a conductive path to the case floor where the ground lug would be. Sanding the powder coating from the steel shield took a bit of work. Further, the air in the space I work in is high in moisture and very near the sea, so this panel started to corrode within the week and I had to do it again.
Sanded areas where the ground lugs will be located. On the left is where the aforementioned safety ground will be, while the right side is for the circuit ground. The side panel sanded, which will make contact with the taped but currently unprepared bottom panel. The bottom panel completed. Note there are a few more drill holes that have had the aluminium exposed around them.
With the sanding completed and the aluminium braces attached, I did a test fitting of the case. I got a bit overexuberant with the sanding on this side and now there's some bare metal visible where there shouldn't be. Hopefully some careful administering of a black marker will fix that. This interior shot shows the placement of the aluminium braces. Without them the middle of the rear panel would have no resistance to it pushing inward. Looks like the inside of the Death Star! Almost all of the ideas or techniques used to prepare the chassis came from part 3 of the article @Davey mentioned on the first page: https://wallofsound.ca/audioreviews...tage-project-part-3-metal-working-for-dummies I'm lucky to have found that series before starting this build, as it helped enormously.
POWER SUPPLY CONSIDERATIONS Another thing I spent a considerable amount of time on was trying to work out a sane power supply protection scheme, which meant a lot of learning 'from the ground up'. Unwilling to get too complex with relays and timers, I initially mapped out a plan for a 2-stage manual turn-on process that would make use of a 3-way rotary power switch. The plan was to switch the heaters on in the first switch position, before then applying B+ voltage in the second. Some people feel this is a good option so the tubes don't get blasted with full B+ when cold; and then others tend to disagree completely. I was talked out of this arrangement at diyaudio forums, probably quite rightly, for being unnecessarily complex. Now planning to use a standard power switch with a single turn-on movement, my main concerns were twofold: switch protection and inrush current protection. But before that, I'd need a fuse.
I’m impressed by the quantum leap from building an ANK DAC and designing your own Aikido project! I wish I had taken electronics classes in school.
I know what you mean. I wish I'd started doing this 25 years ago. The Aikido isn't that far removed from a full kit. You still get a map to the top of the mountain - it's just that you're responsible for packing your own camping gear. I'm always impressed by those that can fashion something from just a schematic, or even more, create their own!
You're probably right. I suppose the great thing about DIY is that something like that can be added in later on. I just wasn't in the mood to learn one more thing!
As the chassis preparation was now at an end it was time to start final construction ...finally. IEC socket mounted, and the all-important safety ground lug attached and lock-nutted. The live and neutral wires are on their way to the front of the case where they'll eventually be hooked up to the front power switch. The HV transformer appears to be designed for top-mounting to a case, so there needed to be a bit of clearance created. Because of this I took the opportunity to mount the transformer on rubber grommets to help isolate it from the rest of the components in case it produced any vibration. Isolating the vibration from the case in this way also meant that the transformer was isolated electrically as well - with no path to ground. So should the transformer malfunction and the outside become live, it would stay live and perhaps be a potentially lethal hazard if the case was open. I attached a ground wire from it's exterior to one of the sanded holes to connect it to the chassis. Can anyone confirm for me whether this was the right thing to do? The heater transformer was also mounted on a rubber pad for vibration control, and so it too was connected to the case with a ground wire. I had to sand away a small area of the varnish covering the transformer to make this work. This photo should show clearly how the transformer primaries are connected in series using the terminal block.
Thanks to everyone that has commented, or 'liked' a post. It's been appreciated. A FUSE My project currently only has a single fuse located at the IEC inlet, but more should probably have been considered. The guideline I used for rating the fuse was 125% of operating current using the following formula: Fuse = (Watts/Voltage)1.25 The wattage was assumed to be the cumulative sum of the two transformers VA ratings, and with NZ wall voltage at 240V the equation was: ([28+40]/240)1.25 = 0.354 Amps (or 354mA). At this time a slightly underrated 315mA time-delay (slo-blo) fuse seems to be working fine. Slo-blo fuses are generally used so they can withstand the high inrush current experienced at turn-on.
