My first valve amp - Luxman A3600 running KT88s

[xlr8or]

Hi Greg. If you have the equivalent digitalised version of the original A3600 book please send it to me. I have the original A3300 book if you're interested in it. Thanks!

[couchy]

10 hours ago, Luckiestmanalive said:

 

Wow - what a great catalogue, thanks, Kirk! It's such a shame it is in such a low resolution - I'm going to find a copy and buy it! And then I will be able to digitise it at high resolution. I can see the A3700 TCR-A7 has an ultralinear circuit with KT88s, which is how my A3600 is wired. The difference in NFB components is interesting...

 

Also, Craig, I found this copy, sans power supply, which might be a little easier to read?

 

Thanks Greg.

[Luckiestmanalive]

14 hours ago, xlr8or said:

Hi Greg. If you have the equivalent digitalised version of the original A3600 book please send it to me. I have the original A3300 book if you're interested in it. Thanks!

 

I haven't yet, but I must get onto that...

[Luckiestmanalive]

I decided to learn more about how the power supply of my A3600 works. I downloaded PSUDII, and modelled the S-2007 Lux transformer by inputting the specs and the measured resistance between the primary and secondary windings. I modelled the draw from the output and phase inverter stages as current taps. The dropping resistors are as per the spec (and my amp) and the C-2006 choke is modelled as per spec. The capacitance values are as measured on my DMM. I tweaked the final resistor (R3) until the resting voltages at each stage (C2-C4) are near enough to bang on.

 

So here is the model and how the voltage on C2 (ie B+) comes up on turn on. There is both overshoot and a little ringing I could tweak. I also found this interesting piece on tuning a tube amp's power supply. Has anyone tried tuning a power supply using PSUD and have any advice? 

 

 

 

 

 

Edited March 1, 2024 by Luckiestmanalive

[Luckiestmanalive]

For instance, I could rearrange the configuration of the existing twin-can caps (parallelling the 2x 108uf can for C2) and install a much smaller 10uf  film cap at the end to get this (noting I've also increased the R1 1Kohm resistor to 2.2Kohm as per KT88 conversion instructions):

 

The model suggests a 2.2Kohm resistor would reduce the voltage for the driver/phase inverter and input tubes from 480V to 461V and from 260V to 249V, respectively. This would have little effect on the operation of these tubes but it (and the much smaller 10uf cap) reduces ripple significantly (see table below for results after allowing for a delay of 11 seconds):

 

Edited March 7, 2024 by Luckiestmanalive

[Luckiestmanalive]

Update on changes to the amp in the last few months:

1. The C2006 choke (0.35H 400mA 11ohm) has overheated at some stage and the black goop (pitch?) that is packed in the container to dissipate heat has leaked and been cursorily cleaned up. I unscrewed the two bolts holding it and looked underneath and it was a bit of a mess so I cleaned it thoroughly. I then found a second-hand replacement C2006 choke on the Yahoo Japan auction site I bid for and won and have replaced the old one. Using my multifunction T7 tester I measured the inductance and resistance (no load) of the old choke (~1H 11ohm) and the replacement (~1.4H 11ohm). The seller of the replacement choke advertised the inductance at 0.66H with 1kHz signal (presumably with load of 400mA).
2. I removed the 100uf electrolytic bypass caps for the 6AQ8 input/preamp tube (C103-104), as recommended in at least one of the recommended set of mods for KT88 conversion. This appeared to reduce "smearing" and produce a slightly more well-defined (holographic) image.
3. I used PSUDII to optimise the power supply and ended up using a Kemet 600V 75uf cap (measures 82uf) with 970mA ripple current capacity at C1 (PSUDII indicates ripple current of ~708mA is needed). I connected it to a tag strip I added to one of the power transformer mounting bolts close to the first can cap (C2), which I changed to a Kemet 600V 390uf capacitor (measures 352uf). The second can cap I changed to a F&T 550V 3x 15uf (measures 18uf) multi section cap to provide filtering for C3 and C4. I also increased the resistance of R301 (R1 in PSUD model) from 1Kohm to 2Kohm. Mods for the conversion to KT88 suggest using a 2.2Kohm resistor but my PSUD model indicated 2Kohm would be enough to bring the B2 down closer to the voltage indicated in the schematic and it is near enough to 470V to be bang on. Next time I get the bottom off it I will take a pic and post it here. This took a few hours to complete as I wanted to make sure I got it right.

Edited September 24, 2024 by Luckiestmanalive

[Luckiestmanalive]

Wow. The third change has a had a significant effect for the good and thanks to the nice folks on DIYaudio.com (trobbins and Tubesaregreat) who provided valuable feedback.

 

Changing the configuration (choice of caps and their values) of the power supply and the dropping resistor to reduce by 10V the plate voltages of the driver/phaser inverter and input/preamp tubes has made a very noticeable improvement in the focus, PRaT and transparency of music. It was very good before - the PRaT is also a function of the room interaction with my loudspeakers and volume - but there is no longer any harshness as I turn the volume up in the upper range I like to listen at and I can see every instrument and voice in space. There is a more realistic 'being there' vibe.

 

I'm very pleased with the outcome from using the combination of the PSUDII tool to model the performance of the power supply and the vtadiy.com site to model the performance of the tubes. I've checked the way the voltage rises as the amp powers up and there is a graceful rise in B3&4 voltages to near 400V before they decrease to their nominal level (260V), which all happens over a 15-30 second period (it used to take longer). The B2 voltage jumps to 540V before increasing to 550V and then decreasing to 500V for a pause, before decreasing slowly to 470V. The first bit up and down to 500V takes about 30 seconds and the last decrease to 470V takes about another minute.

 

You'll see from the pics below that I have made only minor mods to the internal layout. C1 (75uf 600V Kemet cap) is now inside the amp on a tag board attached to one of the power transformer bolts. C2 is the tall and thin Kemet 390uf 600V can cap (KT120 output tubes) on top of the chassis in the back left corner (see bottom pic), behind the power transformer (the bottom of which you can see in the top left of the first pic). C3 (driver/phase inverter 6GU7 tubes) and C4 (input/preamp 6AQ8 tube) are in the small blue F&T 3 x 16uf can cap next to the C1 cap (the bottom of which you can see in the first pic directly below the C1 cap). The green resistor is the 2Kohm dropping resistor (R301), the blue one is the 100Kohm resistor (R302) and the brown one is the 330Kohm bleeding resistor (R303).

 

After a couple of listening sessions, I did a re-bias check of the KT120s and balanced them at 70mA each (down from 75mA) to reduce the total quiescent current and power draw. 
 

 

 

 

Edited September 28, 2024 by Luckiestmanalive

[Luckiestmanalive]

This is the new configuration and wiring for the power supply (note the capacitance for C1 and C2 on the diagram are the measured values).

 

 

 

Here's my updated PS schematic:

 

Edited October 5, 2024 by Luckiestmanalive

[Luckiestmanalive]

This week I altered the ground wiring for the speaker outputs - making a separate ground wire for each speaker common back to the negative terminal of the can cap for the bias circuit's power supply, rather than daisy chaining as per the manual.

 

Here is the rest of my list of things to do/try:

1. Make separate ground wiring back to negative terminal of the C3 and C4 can cap for each of the push-pull sides (the ground wires for the output tubes are daisy-chained as per manual)
2. Add large resistor to ground for the other side of 20Kohm bias pots for each push-pull side (channel) - just in case wiper lifts...
3. Replace R125-129 from 68Kohm to 39Kohm to lower the SG1 resistance (KT120s like <52Kohm resistance)
4. Replace R133-134 2.7Kohm resistors in NFB circuit to 2.2Kohm to bring NFB back to stock (16dB) level
5. Add 10-15uf bypass caps (poly) for C103-104 (currently without and stock value is 100uf). Such a low value might reduce bass as roll off increases as capacitance is reduced
6. Create separate supplies for left and right channel by splitting B3 and B4 using spare 16uf section.

Edited October 5, 2024 by Luckiestmanalive

[Luckiestmanalive]

Update - a few months ago, I started a project to copy the PCB digitally and have a new one made. Today, I finished that project and installed the new PCB. I populated a new PCB with new parts and returned the various feedback components to their original values:

1. C113-114 from 330pf back to 560pf
2. R133-134 from 2,200ohm back to 2,700ohm
3. R135-136 from 11ohm back to 22ohm.

The cathode resistors for the LTP/driver stage remain at 10Kohm (up from 8.2Kohm) and I left the input stage's cathode bypass cap (100uf) out.

 

 

It goes and the voltages and temperatures are pretty much the same as the old board but this one sure looks a lot newer! I have four more boards coz I the minimum order was five. PM me if you are interested. My emulation wasn't perfect, mind, as my mounting holes were a little off and I don't know how to make the test points with sticky-out eyelets (vias).

 

 

Edited November 3, 2024 by Luckiestmanalive

[couchy]

 Amazing work.

[Luckiestmanalive]

Thanks, Craig! It has been a fun project - I learned a lot! Once I pulled the old board off I noticed the a couple of traces have been compromised (lifting and cracking) so I'm glad I've changed it over. I've made a list of six minor edits to the PCB gerber file. Once that work is done the file will better match the original board.

 

I may then move the components on the old board onto a new one so I can swap them out if I prefer the way the old configuration sounded... 

Edited November 3, 2024 by Luckiestmanalive

[couchy]

On 04/11/2024 at 7:57 AM, Luckiestmanalive said:

Thanks, Craig! It has been a fun project - I learned a lot! Once I pulled the old board off I noticed the a couple of traces have been compromised (lifting and cracking) so I'm glad I've changed it over. I've made a list of six minor edits to the PCB gerber file. Once that work is done the file will better match the original board.

 

I may then move the components on the old board onto a new one so I can swap them out if I prefer the way the old configuration sounded... 

I wish I had your skill.

 

[Luckiestmanalive]

I wouldn't call it skill! I just seem to possess an entirely misplaced confidence in my ability to learn new stuff! Once I started taking the cover off my A3600 amplifier I started to demystify it and now I'm hooked on altering the circuit myself to improve the sound, rather than trading up to do it. I'm happy I bought a Luxkit and not a Luxman model of my amplifer as it kinda gives me permission to make mods (not to mention the need to do so by changing the output and driver tubes). And I haven't had a hankering to buy new components for a few years now!

[Luckiestmanalive]

I'm reporting back that my new PCB works just fine (of course it does - I had complete faith!) and returning the feedback components back to stock values (reducing negative feedback level) has made a very noticeable improvement in fidelity! I like to think that my board has contributed some (ha ha!) but this combo is creating the best reproduction of my reference tracks I've ever heard. The soundstage is wider (much wider than the speakers) and deeper (well behind the speakers) and instrument/voice placement is more distinct. PRaT and presence has improved, too. And yet I can still hear there is a smidge room for improvement on a couple of tracks...

 

Here is the remainder of my list of things to do/try:

• Create separate power supplies for left and right channel by splitting B3 and B4 using spare 16uf section of the 3x 16uf multi-section cap being used for input signal and driver stages. There is potential for it to provide a noticeable improvement in channel separation. 
• Make separate ground wiring back to negative terminal of the C3 and C4 can cap for each of the push-pull sides (the ground wires for the output tubes are daisy-chained as per manual). This one might improve channel separation but it is a pain to do because the black wires are in a bundled loom...
• Add large resistor to ground on the other side of 20Kohm bias pots for each push-pull side (channel), just in case wiper lifts...
• Replace R125-129 from 68Kohm to 39Kohm to lower the SG1 resistance (KT120s like <52Kohm resistance). Isn't likely to have any effect on sound quality - more just peace of mind.
• Experiment by adding 15uf bypass caps (poly) for C103-104 (currently missing stock 100uf cap). I'm concerned such a low value might increase bass roll off so I'm in no hurry to try this one.
• Take the components off the old PCB (NFB as per KT88 conversion instructions) and reinstall them on another of my new PCBs. Then maybe use it to replace the one I have installed now, for comparison.

 

Edited November 29, 2024 by Luckiestmanalive

[xlr8or]

1 hour ago, Luckiestmanalive said:

I'm reporting back that my new PCB works just fine (of course it does - I had complete faith!) and returning the feedback components back to stock values (reducing negative feedback level) has made a very noticeable improvement in fidelity! I like to think that my board has contributed some (ha ha!) but this combo is creating the best reproduction of my reference tracks I've ever heard. The soundstage is wider (much wider than the speakers) and deeper (well behind the speakers) and instrument/voice placement is more distinct. PRaT and presence has improved, too. And yet I can still hear there is a smidge room for improvement on a couple of tracks...

 

Here is the remainder of my list of things to do/try:

• Create separate power supplies for left and right channel by splitting B3 and B4 using spare 16uf section of the 3x 16uf multi-section cap being used for input signal and driver stages. There is potential for it to provide a noticeable improvement in channel separation. 
• Make separate ground wiring back to negative terminal of the C3 and C4 can cap for each of the push-pull sides (the ground wires for the output tubes are daisy-chained as per manual). This one might improve channel separation but it is a pain to do because the black wires are in a bundled loom...
• Add large resistor to ground on the other side of 20Kohm bias pots for each push-pull side (channel), just in case wiper lifts...
• Replace R125-129 from 68Kohm to 39Kohm to lower the SG1 resistance (KT120s like <52Kohm resistance). Isn't likely to have any effect on sound quality - more just peace of mind.
• Experiment by adding 15uf bypass caps (poly) for C103-104 (currently missing stock 100uf cap). I'm concerned such a low value might increase bass roll off so I'm in no hurry to try this one.

 

 

I will send you my one to refurbish after you've fully restored/updated your one.

[couchy]

My mind is blown at what you are achieving  here Greg. As I said before I wish I had your skills to work on  my Luxman pieces or even my Chinese tube amps for improvements. Keep up the great work.

[Luckiestmanalive]

Thanks, Craig. I've certainly come a long way but there is still a lot to learn!

[Luckiestmanalive]

On 29/11/2024 at 5:04 PM, xlr8or said:

 

I will send you my one to refurbish after you've fully restored/updated your one.

 

Hmmm... how about I build another board, send it to you and you can swap yours out and test mine. If you prefer it, you buy it (component cost). If you don't want to keep it, you put your old board back in and send my one back?

[xlr8or]

What would be good to do as well as part of this restoration and modification process is to trace out the signal path if you have not done this already. From there you can look to use audio grade caps and resistors for some of the critical signal path areas.

 

Edit: I forgot to mention that changing the grid leakage resistor value for KT120 use (reducing in this instance) will have an effect on changing the tonal characteristics. This video from Stephe indicating what happens when changing this resistance value is quite useful.

 

 

Edited November 30, 2024 by xlr8or

[Luckiestmanalive]

Thanks, Kirk. I've watched nearly all of content on the Skunkie channel but I watched the video again. As the A3600 is a push-pull design, I thought the 2nd harmonics cancelled out? Of course, it might have other unknown effects, too. OTOH, in my research around the mods documented for the circuit for other output and driver tubes, I have seen mention of the replacement of R125-128 (68Kohm) resistors for 39Kohm values to lower distortion, if needed. I was wondering if the motorboating I was experiencing at times might be due to this higher grid resistance but it isn't much of a priority now as I've not experienced it for over a year.

 

The A3600 has a very clean signal path compared to many other amps. There is a 22Kohm resistor from the vol pots to the PCB and C101-102 is an 0.1uf cap blocking DC from any preamp connected. The next components in the path, apart from the input and driver tubes, are the four coupling capacitors between the board and the output tubes. So the only components on the PCB that are directly in the signal path are C101-102, which I've bought WIMA poly MKS4 box caps to fill.