Some final shots of the amplifier after I wired up the IEC inlet last night... despite having a blocked nose really wanting to go to bed...

I'm expecting the DN2540 DMOSFET to arrive today. Don't think I'll be up to building the CCS replacement for the Buffalo2 project tonight. The ones that I did based on MJE350 had difficulties getting the correct resistor settings even after following the instructions to a T. The DN2540 cascode with LM317 in theory will make it easier to set the current. ;)

It really took a long time. Cutting the IEC inlet and the hole for the switch, measuring the angle bracket to fit under existing holes so that I can use the screws to tighten the angle bracket, tapping the required holes (i used the power drill to do the taps this time round. M3 and M4, all done quite nicely), rounding off the edges of the aluminium panel, sanding it again, wiring up to the power transformer... All this for just one channel. I'm also working on my Buffalo 2 dac at the same time. Having some problems with the CCS. More pics of the 1626 here.

Used my jigsaw to cut up an aluminium top plate from a failed prototype. I needed some pieces of metal to secure the IEC inlet socket and power switch for my 1626 amps. Yup... I've not done this yet. Still very much enjoying the amp to do additional work on it. ;)

And also, I prefer to do this type of cutting when I'm home alone. No one to distract me. I'm thinking of doing a base for the amps as well. Need to look thru my stockpile of aluminium plates to see if I have something suitable.

Next up would be to cut out an opening for the IEC inlet and switch and to mount it onto the amps.

I could not find suitable 5k 10W resistors for cathode duties at the audio stores. Both RS and element14 (Farnell) did not have stocks for the Dale aluminium power resistors. I looked around for a while, and eventually ordered Caddock MP820 series from RS. They were extended availability items and would have to be shipped from elsewhere. Total cost for 2 of these babies is around SGD25. I sure hope they would do the job well.

The MP820 resistors dissipates a bit of heat. Mounting the cathode capacitor on the aluminium plate is not ideal. I then looked around for suitable smaller sized capacitors that can be directly soldered onto the resistor leads rather than mounting the large Nippon Chemi-con. Anyways, there is not that much headroom for the Chemi-cons. It's a tall order to fulfill since the voltage would need to be around 200v, and yet be of a manageable size.

John Broskie happens to have an old post about calculating cathode capacitor value and the corresponding capacitor value to inject the right amount of power supply noise to counteract output noise in the circuit. The formula to calculate cathode capacitor is Ck = 1 / (2pFRk) or Ck = 159155 / F / Rk. I think this formula is pretty much standard. For my circuit, with a cathode resistor of 5k, this works out to 2uF for a 16Hz cutoff. Standard capacitor values guided me to this cutoff frequency... not that I play pipe organs that much in the system. There surely must be some mistake I thought. Is a mere 2uF enough?

Well... I tested with a 1uF capacitor that I had on hand and it worked pretty well to replace the 220uF Nippon Chemi-con that I used in my tests so far. It turned out to be quite an advantage for a direct coupled circuit as this is a small enough value to consider a quality polypropene capacitor in this location. The same post also mentioned a formula to calculate a corresponding capacitor (ultrapath capacitor) to inject ac signal to the cathode, thereby canceling the power supply output noise. The 1626 has a gain of about 5. This results in 1/6 of the 2uF Ck, giving 0.33uF.

I got some GAD-viva capacitors from Koba since they had these values (and others too... their GAD 0.1uF Silver Gold is around SGD105! ). Managed to wire them up tonight, together with the MP820. Dialed in the operating points and finalized the design. 6N6P (gain of 20) is running at 110v, -2v bias (actually the yellow LED drops 1.95v). 1626 is running slightly hot, at 185v, -18v, 26mA.

The final schematic is also in the pictures above. The schematic for the CCS used is the HV PNP CCS from diyaudio.com (pdf can be found here). I used an additional red LED for the reference voltage string (LED2). This increases the reference voltage, causing the emitter resistor to be larger, and raising the CCS impedance. The pdf for the Janus Shunt Regulator would provide detailed information on how to build it. My implementation used a Ohmite rheostat for R5, allowing me to adjust the voltage, and a JJ ECC832. C1 used is 22uF, C3 & C4 are 220uF. Bypass capacitors are GAD-viva 2.2uF ones.R4 and R7 are Kiwames.

I just need to finish up the IEC inlet and switch, plus the top aluminium plate. I guess these can be done at a more leisurely pace since music is flowing now. ;)

Was wiring up some portion of the amplifier this afternoon, and proceeded to hook up the CCS, cathode resistors, and used my spare speakers as a test load. After some false starts, it managed to sing with my iPhone as a source. As expected, the gain was low. Strange thing is that voltages on the 6n6p is more or less in the calculated ballpark rather than the 85v previously observed when testing the CCS with the 6n6p alone. Together with the output transformer and 1626 tube in the circuit, it had different operating points. B+ was also lower. This was expected due to the extra current the 1626 is pulling. I just have to adjust the Broskie shunt regulator to provide for more B+. I ran it for a short while before switching everything off and going for my reunion dinner.

After dinner, I came back and hooked up my preamp, cd player and Coral Beta 8 speakers. Volume was much louder, but could do with more. I did some fine tuning of the operating points and B+. The CCS now outputs 112v. This results in 110v, -2 bias for the 6n6p. The yellowish LED is a constant 2.048v. 1626 now runs at 189v, -16v. Cathode resistor is a Dale 3k9 power resistor. With 128v across the cathode resistor, this should provide ~33mA of current. This did not change the volume output of the amplifier much, but is within the calculated values and should be a safe operating point for the circuit.

Adding a cathode capacitor to the 1626 resulted in more gain and I thought it sounded so much better with it in place. I used a 220uF 450v Nippon Chemi-con that I bought in Hong Kong years ago. I almost finished listening to a complete cd, but kids got to sleep, so further tweaking will be for another day when the stereo is built (add a balance pot for the 1626 filament and lift it nearer within the 128v cathode. I also want to try it with the 7k output tap, but this means lesser power output, but greater linearity - probably need help to A/B the difference).

Next up is to mount and wire the CCS, wire the ground, yellow LED and 3k9 cathode resistor, do up a panel to hold the IEC, switch and fuse. Then I will start with the other channel, followed by the top plate for both amplifiers. I would say this is a successful build. The next post would probably not be soon. Got lots of work on hand. ;)

Happy Chinese New Year to all!

The chassis is having more holes drilled into it. I tried to reuse holes where possible. For example, one of the hole that is meant to secure the Broskie high voltage regulator on the bottom is reused for a standoff mounting the 1626 octal socket on the top. I also had to do a small panel for the rca input and speaker outputs, and drill the holes for the standoffs as well. Had to hand sand these down as the piece is too small for the Makita sander. In total, 32 one inch standoffs were used for a mono block. I'll need to buy more when building the other mono block. I think the challenge would come when trying to fit in the top plate of the amplifier... especially when trying to mark out the holes for the tubes.

While rummaging in my parts bin, I found some nice Cardas rca(eutectic Brass, rhodium over silver plate) and speaker binding posts(pure copper billet) that will be used for this project. The rca is the short version... not exactly the best fit, but still manages to accept an interconnect quite nicely. Drilling the holes for the speaker binding post is a little tricky as the black mounting plate needs to be aligned nicely in order to fit into the two circle extrusions. Hmm... difficult to describe until you see how it looks like. Anyways, a 13mm drill bit size for these is fine. The rca will do with a 11mm drill bit size (most are around 10-11mm for rca female mounts).

I wired up the filaments to the tubes (except the 1626) yesterday so that I can hook up the CCS to re-confirm the operating point for the 6n6p. Turned out that at -2v bias (yellowish LED), i will get around 85vdc on the plates and 22-23mA of current for one section of the tube. This differs quite a bit from the datasheet. I did not put in another tube to check though. The label on the tube reads 6Н6П-И. Searching the net, I found an interesting post on diyAudio. So the tube is basically...

"H" for N, Double Triode.
"П" for P or Pi. Small Glass Envelope.
"И" for I, Pulsed operation

6n6p-i. Pulsed operation. Not sure when or where I got 4 pieces of these or what I had in mind for them when I ordered them. I'll just run it in the 1626 and see how they sound. Should be able to listen to one channel this weekend as the Chinese New Year holidays are coming up this week. Nice. ;)

Wishing everyone a properous year of the Rabbit!!!

It sure is taking some time since I posted about the tube purchase. The concept for the chassis is largely determined by the FT-10 output transformers. Nick suggested that I mount them outside as they are such beauties. So, it's gonna be mounted that way. However, doing it this way does not mean less chassis work. In fact, I think it is more... and doubled since this will be mono blocks.

I only managed to drill and cut some holes for the tubes last night. Voltages were tested more or less ok few nights ago in a rat nest of crocodile clips wiring. The Broskie regulated supply worked like a charm. The slow start up offered by the 6AX5GT rectifier complemented the regulated supply nicely.

The constant current plate load for the 6N6P drivers are working fine, however, will need to fine tune the operating points of the tube as it does not seem to match the stated curves as much. This will impact the cathode resistor under the 1626 tube since they are directly coupled.

Stay tuned for more updates.

Some fella was asking me if I had any updates on the FT-10 output transformer that I bought during my Japan trip. Well, unfortunately, not yet.

So I asked for suggestions on a output tube that can go with either a 5k or 7.5k loadline and the 1626 was one such suggestion. I did some reading up on the 1626 and decided to do something around this tube. The FT-10 would be mounted naked, to be seen as part of the amp. Perhaps this build should take some cue from Don Garber, but maybe not so elegant. ;)

I have some 6N6P tubes that would be great as drivers for the 1626. Possibly do something direct coupled, with a 25mA CCS on the driver. B+ would be around 280vdc. Wanted to stack my transformer or use a voltage doubler to achieve the required voltage, but then I decided to use one of my larger power transformer for this build since I'm gonna use a FT-10 output. It's got to be good from the power supply onwards.

Rectification would be via RCA6AX5GT that I have in my stash. This will then go to the John Broskie shunt voltage regulator that was meant for the 6T10. Not too sure if it will do 50mA though. Will have to try it out.

I just need to trim down 4 aluminium plates to a 230mm x 230mm size and get a 1uF capacitor to test the power supply voltage.

Keeping my fingers crossed for this build, but gonna have to travel next week and the week after for work.