TU-8600S

I managed to get the Elekit installed in my main system, the handyman is fixing the damage to the barn caused by a falling tree this winter, he is much younger and stronger than I.

The TU-8600S is definitely a rather good amp.  A bit laid back in comparison to the XA25, but good bass, luscious mid-range, and plenty of air at the top.  Expanding on my opinion after listening to the BluSound - Elekit - P3ESR setup.  There was somehow a slightly fuller bodied sound than from the XA25.

One issue I noticed was 60Hz hum when the volume was at 100%.  This hum must be picked up by the PCB traces that run from the back right to the front-center where the volume control is, as the hum volume is directly affected by the volume control position.  I was initially running with the amp volume at max thinking that that would essentially remove the pot from the signal path, but things are better when the amp potentiometer is at 50%, no audible hum when you listen at 1 ft. from the speakers.  The amp is shipped with a "cheater" mains cord, no ground pin but a lead that could be screwed to the outlet.  I did not use that,  as running an ungrounded metal box is not acceptable.  I am not sure why so often signal ground is connected to power ground. 

Then I took delivery of the Cardas Clear Reflection cables.  XLR from the K-01XD DAC to the LA4 pre, RCA from pre to the amp.  WOW!  Cables DO matter, I had been using the Clear Cygnus, a pretty good cable one level down from the Clear Reflection.  There was more detail and, at the same time, a greater sense of space and a warmer presentation. Not subtle.

I am now one happy customer and can just sit back and listen (until the next urge strikes, though I am not sure where).

I borrowed the Harbeth P3ESRs and a BluOs box from my "kitchen" system, inserted the three NOS input & driver tubes, and a pair of WE 300Bs, hooked it up with Kimber PBJs and 4TCs and turned it on.

Given its location streaming did not work so I had to run a Cat 5 so I could access the library on the Aurender.

First impression: it's a very good amplifier.  Clearly a step up from the TU-8000R; tighter bass, cleaner treble, better soundstage and the instruments are resolved better, each in its own space.  The resolution of lower level passages is outstanding (and delicious), even though the P3ESRs, which sound lovely, are not the most revealing speakers.

As expected the mid-range is gorgeous.  I played a number or choral pieces, from the Fauré requiem to the (inevitable) Spanish Harlem Chesky demo track.  Rebecca Pidgeon floated a couple of feet behind the speakers, as did the soloists in the requiem; I was listening very near field, the speakers and I in a 5 or 6 ft equilateral triangle.  On good recordings there was a clear sense of the recording space.

With the Lundahl upgrade and premium (Jupiter Copper-Wax) coupling caps it is not an inexpensive amp, a bit over $4,500 with the installed tubes.  That is more than I paid for my XA25 which was a floor demo unit.  So it is in a price range of some excellent performers.

I really want to move it over to the "reference" (please forgive) system but I am currently limited to 15 lbs lifting and the XA25 is in the way and is over 50 lbs!  It will be interesting to see how good it is between the K-01XD and the Atrias, with Cardas interconnects and filtered power.  There will be a decision whether to run it direct or through the LA4.  It is nice to have remote volume and not to have to rewire stuff to listen to Vinyl.

As an aside, looking at the circuit diagram I found TWO NFB loops for the input.  The NFB winding on the OPT feeds back to part of the input cathode resistor chain, and the input cathode resistors are not bypassed!  There is a total of 4 NFB links.  So much for the simplicity of SET topology.

I'll update in a few weeks when my surgery has healed and the amp has more hours on it.

Congrats on your new build!  Let us know your thoughts after it has settled into your system.  Lots of helpful info in your posts.

I finished the build, I think it took about 15 hours overall.  This kit is significantly more challenging than the TU-8600R.  There are many more components (about 300 solder points) and there are a number of components whose leads must be bent rather precisely to make them fit.

Fitting the sockets for the 300Bs was a game - the sockets had L shaped solder tags that must be almost straightened to fit the slots in the PCB.  The 5 pin SCRs required care to get the pins in the holes.

I made my task longer as, after measuring the provided resistors, I purchased a set of Vishay/Dale 1% metal film parts, with a few spares, and carefully selected those closest to nominal, achieving a worst case of 0.2% and usually better that 0.1%.  Also I matched the l-r pairs, with only a couple of instances where I did not a have enough spares to get an absolute match, and always less that 0.05% (a difference of 1 in the LS digit of my meter, with an MS digit of 2 or more) between the channels.  I don't know is this will improve SQ but it cannot hurt!

They had done a good job with the OPT PCBs, they were tailored specifically to the TU-8600S so no jumpers required.

The circuit is very complex, with many components in the power supplies. There are totally separate supplies for the left and right output tube filaments. The rectification and filtering for the left and right B+ are separate and there is a separate stabilized supply for the fixed bias.

Interestingly the heaters of the driver tubes are AC, the input tube has heavily filtered DC.

There are two NFB loops, from an extra winding on the OPT, one to the grid of the output tube, the other to the cathode of the drivers.  In both cases the NFB in isolated using opto-couplers.

Now to try running it on the bench hooked up to an old pair of small speakers and a BluOs node.  Unfortunately it will be a month before I can put it in the "big rig" to get an idea of how good it is as I have just had surgery and cannot lift the stuff around!

I would say 6 out of 10.  Nothing too difficult but it does require care to not get solder bridges between soldering points as they are often very close, and there are a few large joints to be made, there are large solder only (no mechanical connection) joints between the back panel and the main PCB.

There are a lot of components!  40 capacitors and 81 resistors - that is 242 solder points.

And probably another 50 or 60 connections to other  components, diodes, headers etc. so there are over 300 solder joints to be made.

I think I would not recommend it as a first project.  There is a company Jameco that has some fun, interesting, beginner projects (this is a great store, along with DigiKey and Mouser for components.)  With a kit such as Jameco has you should quite quickly get the rhythm of how long to hold the iron on the joint and how and when when to apply the solder to the joint - NOT the iron.

The Tube Store https://www.thetubestore.com/elekit-tu-8600s-tube-amp has kits available at a good discount ($‎1,359.96, though I don't know if they have the Lundahl Upgrade available.  I previously built the basic level TU-8600R and it has excellent sound quality.

If you want to do it I would suggest you get the TU-8600S kit before they sell out and a couple of learning kits for fun from Jameco.  Go to Jameco.com and search for learn to solder.

You will of course need a good soldering iron, I have the Weller WE1010NA Digital Soldering Station from Amazon (about $120), it holds an accurate temperature and has an auto standby mode (Jameco sells this but for about $70 more).  You will also need solder - 63/37 is much better than 60/40 as it is a "eutectic" alloy - it immediately goes from liquid to solid which helps to prevent dry joints, 60/40 goes through a semi-liquid stage and ANY movement at that time ruins the connection. Jameco's Kester 24-6337-0027 Solder is just the right diameter (.031 - thinner than stuff from the hardware store), and is the 63/37 alloy.

A lead bending jig - a few $$ from Amazon will help to bend the leads of the resistors to go through the holes accurately - though I planed pieces of wood to the right thickness in my wood shop.

Also soldering wick to suck up solder if you have to undo something and a desoldering pump may also be useful.  Figure on about $200 for tools, as you will also need a good diagonal cutter and needle nose pliers.

I use a small fan to provide airflow so I don't breath fumes from the rosin flux.

I have also found that a single ended 300B will drive my 87dB (in)sensitive Atrias to any volume that I need.

I believe that you will enjoy the overall experience, though there may be some initial frustration with learning to solder cleanly - hence the beginners' kits.  And you will end up with a great sounding amp (with a volume control) for about $2k with tubes.

My apologies if I carried on for too long, this did turn into a treatise on soldering, but I hope it was encouraging in a measured way.  Good luck!

What an interesting little amp project. On a scale of 1-10, how difficult would you think this build would be? Assuming 1 is a complete novice who has never touched a soldering iron, and 10 is a proficient DIY builder who can handle most anything out there. 

The variance in resistor capacitance does not seem promising though. Did you ask the seller about that?

The kit arrived - but was missing the PCBs for the Lundahls.  An Email to Jon at Tubestore and the parts were shipped immediately.

This kit is VERY different from the 8600R, it appears to be a completely new design, with many more parts and a very complex bias system.  The NFB is taken from a separate OPT winding and there are opto couplers involved.

The amp no longer supports 2A3 tubes.

There are now 3 small signal tubes.  Each half of the input 12AX7 feeds both halves of a 12AU7, connected in parallel, to drive the 300Bs.  

The 300B sockets are now ceramic.

I was labelling the resistor packaging to simplify finding the correct value when building, and measured the actual resistance.  The Rs are marked with a gold band => should be 1%.  However I found a variance of almost 20% between different samples in some packages!  I am very surprised.  That feels as though it would affect SQ negatively.

Logging on to Mouser I have ordered a kit of Vishay Metal Film 1% units (with a bunch of spares in case I drop one through the cracks in the tongue in groove flooring) for about $30.