Building A Tube Preamp: The Aikido

Discussion in 'Audio Hardware' started by fully_articulated, Jun 13, 2019.

  1. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Continuing on with the power section.

    [​IMG]

    I screwed the steel shield to the chassis floor and threaded the PSU wires through...

    [​IMG]

    ...and then undid it all again because I realised I needed a couple more holes!

    [​IMG]

    Everything back into place, the wires were (hopefully) cut to the correct length ready for tinning before being soldered to the PS-21 PSU board. The green wire carrying 6.3VAC for the volume control PSU wasn't long enough, so was extended. Also, the circuit ground wire was attached to the chassis near the rear panel. This would also be connecting to the PS-21 board.
     
  2. fully_articulated

    fully_articulated Forum Resident Thread Starter

    [​IMG]

    With the PS-21 board positioned on it's 20mm supports, the prepared wires were inserted up through the board for soldering.

    [​IMG]

    Complete and ready for testing.

    At first I fired it up connected to a dim-bulb tester to check for shorts. The bulb quickly illuminates as the capacitors charge, and then goes out signifying all is well in the circuit. If the bulb stays brightly lit however, it means there is a problem.

    [​IMG]

    Removing the dim-bulb tester I began to test the power supply boards input and output voltages. Here is the B+ being measured at 212VDC - close enough to where it needed to be. The other voltages were also in line with what was hoped for, except the 268VAC the HV transformer produced was higher than I would have liked. The volume control PSU was then soldered into place and tested.

    A note about testing and working with high voltages (or any voltages really): If you're going to attempt this kind of thing, you must familiarise yourself fully with best safety practices. Do something foolish here, and you can quite simply become dead.
     
    Echo likes this.
  3. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Next it was time to attach the Aikido board to the power supply.

    [​IMG]

    The complimentary design of the two boards enabled the wiring between them to be very short.

    [​IMG]

    I did make the wires longer than necessary just in case I needed to lift the board again for anything. Also, at that point I wasn't quite sure where I was going to position the signal wiring, so keeping the power wires out of the way made some sense.

    There was nothing to do now but insert some tubes and begin measuring to see if the previously theoretical would be reflected in the newly created reality.
     
    struttincool, ScottRiqui and blc__ like this.
  4. fully_articulated

    fully_articulated Forum Resident Thread Starter

    TESTING
    Poking around the Aikido board revealed the idle currents to be relatively reflective of what was calculated. But, being tubes and therefore not perfect, there were some frustrating readings across different areas. The input tubes affected the output tubes readings and vice versa, if there was a mismatch between the two triodes within the same tube it would alter the characteristics of the circuit, etc.

    It was disappointing to find a couple of the expensive NOS tubes I owned had massively unmatched triode sections. The closest tube pairing was some new production Gold Lion 6922s matched very nicely it seems by Jim McShane. Too bad I don't really like their sound... An unmatched pair of bog-standard Electro Harmonix were also very close.

    I spent a few days inserting tubes, letting them warm up, and taking readings (helped along by a supportive Canadian diyaudio forums member). If there's one thing that came out of all that work, it's that I've certainly learned which tube pins are for what and where!

    Thankfully, a few days before this I had learned (by chance) of Kirchhoff's voltage and current laws. This new knowledge benefited my understanding of circuits and the way electricity interacts within them, allowing me to make more sense out the readings I was getting from the Aikido board.
     
    ScottRiqui likes this.
  5. bluemooze

    bluemooze Forum Resident

    Location:
    Frenchtown NJ USA
    Images are great as they are, please don't downsize them. :tiphat:
     
    fully_articulated likes this.
  6. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Absolutely bored of prodding at the Aikido, I was keen to get started on the signal section.

    [​IMG]

    I purchased some very nice (and genuine) CMC RCA sockets for this build. Here are the right channel connectors attached to the rear panel. You can no doubt tell there are actually two different types - the 'standard' CMC-805-2.5 version with the red insulator, and the CMC-805-CU-G uber-expensive version made of pure copper and thickly plated in gold.

    I bought those specifically for the primary input and output, but they turned out to be more trouble than they were worth.

    Apparently there is a tool available to aid in tightening these sockets. I should have bought one, as I ended up marking some of the jacks trying to secure them with a pair of pliers.

    [​IMG]

    The Seiden selector switch is positioned at the rear of the case to keep the signal wiring short. Unfortunately the same couldn't been done for the volume attenuator due to layout issues. With some scrap cable I started to plan the wiring to the connectors.

    [​IMG]

    After considering using some exotic OCC copper wire for the signal path, I decided to be a bit more sensible and utilise Van Damme cabling throughout. The selector switch to RCA socket wire chosen was Van Damme console cable. Small in diameter, with a foil shield and drain wire, it seemed ideal for this part of the project.

    Unlike the many preamplifier builds I had taken inspiration from, I decided to forgo the use of a bus wire connecting all of RCA grounds together. The Seiden switch, being double-pole / double-throw, was wired so that each ground wire would be independent of the others - only selected with it's companion signal wire - if you understand my meaning. This was done mostly in the hope of preventing internal ground loops, but also to protect the more important inputs (for example, my DAC) from being polluted by the perhaps noisy grounds of devices like a television.

    That may be airy-fairy audio magic, but it made me feel better nevertheless. As long as there are no ill effects, why not?

    [​IMG]

    All the wires measured, cut to length, and tinned, ready for soldering to the RCA jacks.
     
    blc__, 33na3rd and bluemooze like this.
  7. Good tube gear does produce larger imaging with better depth, and hopefully with that elusive sonic holography. So larger quality images of the tube gear getting made is appropriate and appreciated.
     
    bluemooze likes this.
  8. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Well, they certainly are larger. Not sure about the quality though :oops:
     
  9. fully_articulated

    fully_articulated Forum Resident Thread Starter

    The console cable turned out to be a bit of a disaster.

    [​IMG]

    It wasn't nearly as easy to work with as its size suggested, being quite stiff. The drain wire was a pain, and the insulation melted against the heat I needed to use on the RCA jacks. The soldering of RCAs has been my worst work in all three of the builds I've done!

    The CMCs, though nice, are quite thick - especially the ground tabs - and I found them difficult to heat adequately. The more expensive ones were particularly terrible for this.

    [​IMG]

    In the end I just cut it all away, deciding to try again even though there's nothing I hate worse than redoing things (perhaps other than fighting a losing battle).

    [​IMG]

    For the second attempt I stripped the jacket, shield, and insulation from some Van Damme lavalier mic cable and removed the conductors. I soldered them to the selector, then twisted them tightly for noise rejection. Each twisted pair was then heat-shrinked together.
     
  10. The pictures are very good. The important parts are in proper focus. Very good close-up photography.
     
    fully_articulated likes this.
  11. fully_articulated

    fully_articulated Forum Resident Thread Starter

    The second try worked out much more pleasingly, with the new cables XLPE insulation being much more tolerant of high heat, and for longer periods.

    [​IMG]

    You may be able to see that I actually twisted the cables the wrong way as they exited the heat-shrink, but I'd done too many before I noticed and wasn't going to go back for a third attempt. I just continued the mistake for the whole lot, for uniformity's sake. :shh:

    [​IMG]

    [​IMG]

    Selector switch wiring completed. I wasn't, and am still not, overly happy with it. I also wish I'd chosen differently coloured heat-shrink.

    Continuity testing with a multimeter proved all was well signal-wise, so that would have to be good enough.
     
    blc__ likes this.
  12. fully_articulated

    fully_articulated Forum Resident Thread Starter

    [​IMG]

    The leads for the output coupling capacitors were not long enough to reach the board connections, so I cut them and soldered longer wires on each end. This time I used two lengths of fancy solid core Neotech OCC wire twisted together. The biggest hazard this process posed was the possible overheating of the delicate capacitor internals, especially with these ham-hands operating the iron.

    [​IMG]

    I wanted to keep the area clear for a second set of board-mounted capacitors in the future, so aimed to terminate the leads at the rear output connections.

    [​IMG]

    The biggest danger with this arrangement, I suspect, is the risk of these long leads becoming antennas for any kind of airborne interference. I do like the look though...
     
    blc__ likes this.
  13. fully_articulated

    fully_articulated Forum Resident Thread Starter

    The output cabling here is another type produced by Van Damme and I like it very much.

    [​IMG]

    A twisted pair with a 4.5mm diameter jacket filled with what looks to be cotton, inside a high coverage braided shield, it has the same XLPE insulation around a combination of OFC and silver-plated OFC strands.

    It would make a very nice quality, flexible interconnect I reckon. The support stand was one of the reasons I had to drill those last minute holes.

    [​IMG]

    The major downside of using shielded cable, is that its shield has to be connected at one end so it doesn't also function as an antenna. Luckily I had left a stalk of wire sticking up from the main ground on the board for the attachment of test clips while taking measurements. I'm not sure what would have done otherwise!

    Whether it's a correct assumption or not, all of the cable used follows the 'interconnect rule' in regards to its shield: connected at the source (upstream) end and not at the other.
     
  14. fully_articulated

    fully_articulated Forum Resident Thread Starter

    It was then time to wire up the Khozmo attenuator.

    [​IMG]

    As the incoming and outgoing ground wires both share the same connection, and also due to needing space to connect the cable shields in the same spot, I increased the surface area by soldering in some clipped resistor leads. Once again, to avoid any ground loop issues the two channels grounds remained independent.

    [​IMG]

    The input wires were soldered to the front faces...

    [​IMG]

    ...while the output wires and shields are soldered behind. The extra surface area for the grounds certainly came in handy.
     
    JMAC, 33na3rd, JMT and 1 other person like this.
  15. fully_articulated

    fully_articulated Forum Resident Thread Starter

    [​IMG]

    I screwed the volume control to some temporary supports and began preparing the wires at the other end. What I didn't take any photos of was me cutting the wrong wires (making them too short) and having to re-solder new cables to the Khozmo. The second time I was a bit more careful.

    [​IMG]

    Left and right channel inputs were soldered to the Aikido board first, at the same time ensuring there would be enough clearance between them and the selector switches extension bar.

    [​IMG]

    The signal, ground and shield ready to be soldered to the selector switch output.

    [​IMG]

    A power cable from the 5V PSU was then connected to the Khozmo attenuator. Up until now that power supply had been ungrounded, and I was expecting to have to connect it to a ground point somewhere, however it appears the connection to the Khozmo solved that problem.

    As I said to @waterclocker earlier in this thread, I assume connecting it somewhere else now would risk a ground loop?
     
    waterclocker likes this.
  16. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Now the build was largely complete, save for the addition of a proper front panel and the connections to it.

    [​IMG]

    At this point it looked much better than I had been anticipating, especially during the last half of the build when it felt like it had left the rails.

    [​IMG]

    Time to spark it up to see if I could get any actual music out of the thing!
     
    shadowlord, JMAC, blc__ and 3 others like this.
  17. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Feeding my First Watt F6 clone amp hooked to some test speakers, I was very pleased to see the Aikido come to life with the Khozmo display illuminating as hoped. I slowly turned the attenuator up all 63 steps to full volume and the speakers were almost silent. They are not particularly sensitive I don't think.

    This proved not to be the case when the first source was connected though. Plugging in a 3.5mm to RCA cable for my phone, it buzzed increasingly as the volume was once again increased to full. Interestingly, this ceased when a signal was put through, even a silent one. When playback was paused it remained silent for a couple of seconds, then the buzz starts again. Changing the source to an Oppo disc player dispensed with that noise problem, and testing continued.

    The only other gremlin I encountered was a light pop through the speakers when the volume was adjusted at some steps, most noticeable between steps 31 & 32, and toward higher volume levels.

    I suspect the source of both issues may be with my grounding scheme.

    [​IMG]

    Above is a diagram of how I understand my current grounding pattern. Any comments?

    The two pairs of uber-expensive CMC RCA sockets again proved troublesome - for some reason the central signal barrel will twist in its insulation, so care had to be taken inserting/removing the interconnects without rotating the plug. Turned too much and the internal wiring could twist and break. Could this be due to excessive heating during solder I wondered.

    The volume control relays click internally every step as well, which I know many people hate, but I quite enjoy it. Overall I was happy and relieved with the performance of the Aikido so far. It make sound!
     
    33na3rd and waterclocker like this.
  18. fully_articulated

    fully_articulated Forum Resident Thread Starter

    I cut the piece of MDF for a front panel test before finalising the file to be sent to Front Panel Express. I had been designing it with their program during the whole build process, and although it wasn't to be cheap, I hoped the final product would be both simple and beautiful (and have its holes in the right place).

    The next day I placed my order, and like Hifi2000 a few months earlier, Front Panel Express were great to deal with. They spotted a small error in my perspex window design and corrected it for me, and helped me through the rest of the process quickly and pleasantly. It's nice to get such service even when you're only a hobbyist, especially with what I am sure is a relatively small job.

    They have a good info page that tracks your job at the different stages of production too. Thumbs up!

    We've now caught up, and everything from now on will be in real-time. The front panel should clear customs soon, so hopefully will arrive by the end of the week!
     
    JMAC and waterclocker like this.
  19. JMAC

    JMAC Forum Resident

    Location:
    PDX, OR, USA
    Neat and clean! Nice work!
     
    fully_articulated likes this.
  20. That Khozmo attenuator is neat. Who knew an attenuator that gets hidden inside could look so good.
    You need to put a display window in the front panel to show off the sexy red beast inside.
     
  21. fully_articulated

    fully_articulated Forum Resident Thread Starter

    It's a heavy thing too. The casing is aluminium.

    What I didn't photograph was the accompanying LED display / IR receiver board. There'll be a window in the front panel to show that through at least. :agree:
     
    Ham Sandwich likes this.
  22. JMT

    JMT Forum Resident

    Location:
    Rocklin, CA
    Your build skills are amazing. The wiring is so clean and your solder joints look perfect. Well done!!
     
  23. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Thanks, I really appreciate that. I think I got a bit lucky, soldering seemed to come pretty naturally once I tried it.
     
  24. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Since there's no more photos for the moment, I guess I can bore everyone with the last theory post. With these two things I am coming close to the limit of my understanding, and the info that follows is the least likely to be correct. Still, gonna try it anyway.

    POWER SWITCH ARC SUPPRESSION
    Many times a capacitor will be placed across the tabs of a power switch in order to suppress electricity arcing between the contacts at switch-off. As I understand it this protects the switch contacts from deterioration, or even welding themselves together. The VTA preamps looked to incorporate this feature, although their power sections were wired differently to the way I had planned mine. I needed to research the most useful way of implementing this arc suppression, which brought me to RC snubbers.

    A resistor and capacitor in series, the snubber seemed to be a better solution than a capacitor by itself. Although a capacitor will absorb the excess energy at the switch contacts when turned off, it may do the opposite at turn-on, releasing the stored energy quickly back into the circuit. The addition of the resistor would seem to prevent this.

    [​IMG]

    Most of the snubber examples I found were of the first variety in this picture, with the RC placed across the switch. This arrangement was easily dismissed as I did not want any kind of leakage through to the circuit when the switch was off, and other texts warned against using it in AC circuits for that very reason.

    In the end I didn't spend much time working out the resistor and capacitor values for the snubber - not as much as I probably should have. Using the general guidelines offered in that image I selected a conservatively valued all-in-one RC network device by Cornell Dubilier called a Quencharc. From the info I was able to find, the chosen Quencharc's 150 ohm and 0.1uF capacitance rating appeared to be relatively standard for this kind of application.

    INRUSH CURRENT PROTECTION
    When a piece of equipment is turned on current draw through the transformers spikes at many times the normal operating value, which may weaken or damage the power supply components. For this particular preamp the risk should be relatively small considering it's rather low power rating. However, the simple inrush protection utilised in the First Watt amplifier designs got me thinking I may be able to satisfy some other nagging issues by mimicking that setup.

    The thermistor is a resistor whose resistance varies according to its temperature. When used as an inrush protection device it presents a high resistance into the transformer when the unit is switched on, shielding the power supply from the current spike. Then, as it heats up, the resistance drops significantly to normal operating level. You can also use a timer/relay to remove the thermistor from the circuit after the soft-start has been acheived.

    Mine was going to stay in place though, located before the B+ transformer specifically, not only to calm the current spike at switch-on but perhaps to add enough resistance in operation to bring down the 268V the transformer was producing. As long as it didn't also drop the 6.3V too much from the same transformer. Additionally, I hoped this would be a basic way to retard full B+ voltage to the plates at power-up, giving the heaters a chance to warm the cathodes a bit first.

    Leaving the thermistor active the entire time isn't destructive, but if the power is turned off and back on again before it has had the chance to cool it cannot provide any protection ability.

    Finding how to correctly spec the thermistor proved difficult until I came across this very cool video produced by Ametherm:



    Short, but extremely helpful, the video allowed me to work out the main parameters for my own use case.
     
  25. fully_articulated

    fully_articulated Forum Resident Thread Starter

    Oi, I'm back! The front panel arrived sooner than I'd expected and the urge to rush to the finish line overcame me. This will happen every time it would seem, making mistakes more of a likelihood.

    [​IMG]

    The power switch had a very particular cut-out shape that I got from the datasheet along with the dimensions. It was still tough to slide it in cleanly, and on the first attempt I broke those little hinge things on the side. The second time I still had to cut away a piece of one to get them to click in.

    DIY tip: Buy more than one of a thing when you order parts. Luckily I had 3 of these, because I wreck parts often.

    [​IMG]

    I attached the strengthening bars without sanding any of the beautiful powder-coat finish. It appeared I could work enough paint from the thread holes so that the screws themselves were enough to enable conductivity.

    [​IMG]

    The brass bushing for the extension bar to sit in. The bar itself required hacksawing to size.
     

Share This Page