DIY mini T2 Build Thread

[jamesmking]

22 hours ago, micon21 said:

Sorry,  but how does the signal get to pin8. I see that the cathode feeds transistor Q8, no signal comes to Base Q8 through R11.

Simplified explanation:

There are multiple methods for changing how a transistor or valve conducts. You can vary the voltage on valve grid or current on the transistor base. Another way is to vary the current supply to the transistor emitter, collector or valve anode or cathode, another way is to vary the voltage at the valve cathode. 

.. Just because the valve grid or transistor base does not have audio directly applied does not mean that there is no audio passing through the device....

• Q1 is constant current. Since its emitter has a constant load and voltage applied and so has a constant current available and the base also has a fixed resistor and fixed voltage. Rv1 can adjust slightly the overall emitter resistance and therefore available current to each half of Q1 so some current balancing can be done to compensate for mismatches in the halves of V1 and V2 etc.
• The Q1 current is shared between the second 6922 (V2) (the one which does not have its grids connected to the audio input) and Q4/Q5.
• If second 6922 (V2) pulls more current then less current must be available to Q4 and Q5 since the total current available is constant and visa versa is true if the 6922 pulls less current.
• So the current through Q4/Q5 is effected by the current pull from second 6922 (V2) whos current draw is effected by the audio on the grid 1 of the input 6922 (V1) . Since the second 6922 (V2) supplies the anode current to the input 6922 (V1).
• if Q4/Q5 are effected by input 6922, Q11 and Q12 must also be since they get their current directly from Q4/Q5 and the base of Q11 and Q12 is held constant by the resistor string .. R36, R49,50,60
• If Q11/Q12 have current draw effected by the audio input 6922 then the voltage drop across R11 and R12 and current through R11/12 must also be based on the audio input... since the current through R11 and R12 is directly controlled by the current through Q11 and Q12
• The current through R11 and R12 control the current at the base of Q8 and Q9 and therefore control the current through Q8 and Q9. Q8 and Q9 control the voltage at the EL34 cathode and that's how the audio gets into the EL34s and that's the AC audio voltage than can be measured with an multimeter at the cathode of the EL34s.

Edited January 6, 2023 by jamesmking

[boinger]

Anyone have the links to the bom and schematic for the group by? 

 

The first post seems to have the links unavailable.

[Pars]

Links in the first post work for me...

[boinger]

I got it to work. Had to open in new tab for some reason. 

[audiostar]

On 3/11/2019 at 2:36 PM, JoaMat said:

Well, I rather not make recommendations but I would go for +400V/-460V and add one 20K/3W resistor on each board between +220V and +400V pads.

I think I want something like this…

...for power supply.

Now, going to order some transformers 🙂

Is there anything speaking against using the following secondaries with this PSU? 

320Vac @ 0.18A -> 400Vdc
365Vac @ 0.18A -> -460Vdc

For the +220Vdc I'll use a separate GRHV, like the kgsshvpssicfetsinglenewright but without the Bias portion on it. 

 

Edited June 14, 2023 by audiostar

[jamesmking]

4 minutes ago, audiostar said:

Going to order transformers. Is there anything speaking against using the following secondaries with this PSU? 

320Vac @ 0.18A -> 400Vdc
365Vac @ 0.18A -> -460Vdc

 

 

thats close to what I used and should work fine.

I used 190Vac 0.18A for the 220VDC rail

[JoaMat]

 

mini T2 one channel lsk389 replacing 6DJ8/6922. Still on drawing table.

[yabba235]

On 9/21/2023 at 9:20 AM, JoaMat said:

 

mini T2 one channel lsk389 replacing 6DJ8/6922. Still on drawing table.

Did you run it with lsa389? What did you change in the schematic?

Have you considered using the SOIC version instead of the TO-71 transistors?

[Helium]

On 6/14/2023 at 2:30 PM, audiostar said:

320Vac @ 0.18A -> 400Vdc
365Vac @ 0.18A -> -460Vdc

Could someone explain the rule for choosing VAC secondaries of HV transformer?

Rule of 0.707 (400VDC as target + ~10V of voltage drop = 290VAC÷0.707) doesn't apply here? Should we add 30-50VAC  (or how many?) for voltage drop and regulation? Thx.

Edited August 27, 2024 by Helium

[jamesmking]

1 hour ago, Helium said:

Could someone explain the rule for choosing VAC secondaries of HV transformer?

Rule of 0.707 (400VDC as target + ~10V of voltage drop = 0.707*290VAC) doesn't apply here? Should we add 30-50VAC  (or how many?) for voltage drop and regulation? Thx.

The calculation are based on some assumptions and tradeoffs. Ideally you need to know some things about the high voltage power supply the secondaries will be powering.

1. you need to know the minimum voltage input to the power supply to get the required regulated output at the required current. (the minimum input will usually increase with higher current draw). typically this will be a fairly small additional voltage above the output voltage for low voltage power supplies e.g. a few extra volts for the golden reference low voltage and typically in the few 10s of volts extra for the golden reference high voltage power supply. Since this is the absolute minimum you probably want a little more input voltage than that to be safe. 

2. the voltage out of the wall also varies. In many countries during peak demand the power companies can reduce the voltage - typically by up to about 5 to 10% so you need about 10% more voltage on the output side than you might think to compensate the electrical grid giving you less than you expect. So we multiply the total secondary output voltage by 1.1 i.e. add 10%

3. Competent transformer design is also assumed. The more current you draw from a transformer the more inefficient it becomes. Which is why you specify the output voltage of the secondaries at the current draw you expect. If the manufacturers are crap and only design the transformer secondary output based on no load being applied you can have issues where the psu loads the secondaries and the output of the secondaries drops so much you don't get enough voltage for the psu to maintain regulation.  Conversely if the manufacturers design the transformer properly, specifying a transformer with a much higher secondary current draw than it will actually experience will result in the secondary voltage being higher than expected.

Any voltage going into the psu above the minimum amount required to regulate will be converted into additional heat. Any voltage below the minimum to regulate will result in loss of regulation and the output voltage dropping with greatly increased levels of ripple and noise.

you also need to make sure the input capacitors in the power supply can handle the peak voltage going into them or their life expectancy will be reduced and in more extreme overload conditions they can fail completely. (the output caps are mostly isolated from the varying input voltage and should only see the DC voltage output the power supply is setup to output assuming there is no failure in the psu which causes the regulation to be bypassed/short out and pass all the available voltage to the output e.g. the pass transistor to be fully switched on or become shorted)

 

[Since minimum dc voltage input for regulation - DC output voltage] is the additional voltage needed to regulate at the required current draw.

secondary VAC =  (minimum dc voltage input for regulation *1.1 )*0.707

Or

secondary VAC =  ((DC output voltage + minimum additional voltage for regulation)*1.1 )*0.707

 

so if we assume 10V more input is needed to regulate than output for 400V output we get:

secondary VAC = ((400+10)*1.1)*0.707 = 319VAC 

Edited August 27, 2024 by jamesmking

[JoaMat]

18 hours ago, yabba235 said:

Did you run it with lsa389? What did you change in the schematic?

Have you considered using the SOIC version instead of the TO-71 transistors?

Not much (nothing) has happened since my post 11 months ago.

Here is a schematic, original to the left and lsk389 version to the right.

My idea was to replace 6922 tubes with lsk389 and just make necessary changes and else keep the board unchanged. This is a draft - more work is needed.

Of course you can use SOIC lsk389, but I prefer TO-71 since I think it fits better in the layout.

[yabba235]

OK, thanks for explaining - best regards!

[Shawn]

Here is a long post I wrote, but may have a little help to someone who tries to build the T2 mini by JoaMat.

 

Transformer Specifications:

There are two transformers in total. 

Primary:

• 120V(USA or other countries like 230V)

Secondary:

• 320V 0.25A
• 365V 0.25A
• 190V 0.25A
•  
• 16-0-16V 0.5A
• 6.3V 4A × 2
• 6.3V 2A × 2

I highly recommend adding a soft-start circuit because the inrush current generated by the transformers is significant. High inrush current can easily blow the fuse and cause long-term stress on the filter capacitors. If a soft-start circuit is not added, a high-current fuse will be required, but that would compromise the protective function of the fuse.

Chassis:

I housed the power supply and amplifier in two separate chassis. For the PSU section, I used Kerry-designed GRHV and GRLV. Pay attention to the chassis height clearance to ensure it can accommodate the transformers. The specific drilling hole specifications can be found in the current thread—I recall JoaMat mentioned them somewhere.

All panel holes were made using a desktop CNC router. I believe it's a good investment to have a desktop CNC router, as it ensures precise hole dimensions. The text and graphics on the panels were also laser-engraved using the same machine. Watching the CNC mill parts is quite satisfying.

Power Supply:

Three different high-voltage power sources are required:

• -460V, +400V, +220V
• ±15V for the servo

For details on Kerry Design’s GRHV/GRLV, refer to this link:
GoldenReference Low Voltage Power Supply

Two mainboards are required—one for -460V and +400V, and another for +220V, leaving the rest of the board empty. To save space, I designed a new PCB specifically for the +220V rail.

Notes:
If using Kerry Design’s GRHV/GRLV, I suggest disabling the high-voltage delay function. In my tests, the 220V rail initially outputs only 20V, while the 460V and 400V rails start around 200-300V before reaching their nominal voltages after approximately 37 seconds (time varies based on capacitor size). Although this delay is short, it prevents the servo from functioning correctly. During this period, the balanced voltage remains around 200V, which could be harmful to headphones over time. A better solution is to add a dedicated tube warm-up circuit.

Output voltages may have minor deviations. For instance, the actual 220V rail might measure around 217V. Even with 0.1% tolerance resistors, small variations can be amplified. A good approach is to use a trimmer resistor—for example, replacing a 20kΩ resistor with an 18kΩ resistor in series with a 2kΩ trimmer.

 

Amplifier Assembly:

SMD components should be installed first, followed by through-hole components and tube sockets. To simplify the process, I used a paste stencil and a heating plate, which significantly saves time. For the other side, I used a syringe and a heat gun.

A more efficient method would be using a heating plate for both sides, with two different solder pastes. For example, using 183°C solder paste for one side and 138°C solder paste for the other.

Notes:

• Don’t forget to connect the +15V jumper and the servo jumper. Otherwise, the balanced voltage may stay at 400V, potentially damaging the servo. This damage may not be immediately obvious, but you might notice the servo stabilizing more slowly, tiny background noise, or excessive sensitivity of the EL34 filament power supply to external interference, leading to noticeable microphonic effects.

• The T2 Mini’s heat dissipation is moderate, somewhere between the Grounded Grid and KGSSHV. If your enclosure is large enough (not wooden), power transistors like KSA1156 can be directly mounted to the chassis. However, a better approach is to use an L-bracket mounted to a heatsink. My enclosure measures 300mm × 297mm × 62mm, resulting in an internal temperature of around 45°C. Do not use this approach for KGSSHV, as SiCFETs generate a significant amount of heat.

• 01N100D must be properly insulated.

• Use single-point grounding as much as possible to avoid ground loops.

• If using aviation connectors, ensure they are rated for at least 500V.

• HN4C51J and HN4A51J look very similar—don’t mix them up.

• Ensure correct polarity for LEDs and 1N914 diodes.

• Test all components before powering on to verify continuity and check for shorts.

• For EL34 filament wiring, use at least 18AWG wire.

 

One Last thing: Enjoy the build🙃

 

 

 

 

 

[G600]

Thank you Shawn.

Many words of wisdom and common sense.

And quite badass to have CNC and laser etching machines at home 😀

[JoaMat]

 

Yep, thanks a lot Shawn.

It seems that KSA1156 has vanished from Mouser and other places. Mini T2 has a bunch of them. Today I’ve replaced one channel’s KSA1156 with STN9360.

Surface mounted device, 90th between pins same as A1156. I soldered them in standing and power ON. Worked great – after two hour I measured something like 60+ degrees Celsius. Perhaps one can put the STN9360 on a small PCB board for cooling purpose. How about a board with aluminum substrate?

[Shawn]

You may need this for stn9360 plus the aluminum or copper board.

https://www.fischerelektronik.com/web_fischer/en_GB/heatsinks/C04/Heatsinks for D PAK and others/$catalogue/fischerData/PG/FK256/search.xhtml

The KSA1156 is abandoned by Mouser. But other suppliers still have some in stock.

IXTP01N100D is out of stock as well.☹️

2SC3324 can be replaced by 2SC2713 with higher noise as a compromise.

[JoaMat]

There is about 1/4 W each KSA1156.

Below STN9360 (in red) replaced A1156. Bare copper (no solder resist) under the bracket. Put thin Sil-Pad to isolate bracket from the copper. Time will tell… if this is a good idea.

And 2SC3324 is now 2SC2713 on silk screen, Thank you.

[Shawn]

How about something like this with the aluminum oxide isolation pad + screw insulator.

It can be mounted just like the KSA1156 to the L-bracket or straight to the enclosure.

 

[JoaMat]

Sil-Pad 477-4862 under bracket at stn9360

Edited March 15 by JoaMat

[Shawn]

One of my OPA197 had a sparking firework show and died today.

It happened when I switched on and off, then on the amp quickly.

I guess it is the ripple current or something which turns out the mandatory of a soft starter.

Fortunately, I have a bunch of OPA197 in my hand. Swapped and got it back.

The other thing I realized is that I have trouble with the tiny humming in my left channel.

The grounding is probably bad since when my hand gets close to the left EL34, I do not even touch it, and the humming reduces and disappears.

Any ideas?

 

[JoaMat]

I’ve had more than a couple of OPA197 failing. Original I used OP27 in offset servo position and they failed on a regular basis. OPA197 seems a bit more robust, but some casualties there too. When offset servo fails output voltages go high, almost all the way up to +HV (in my case anyway) and it has, so far, always been Op-amp problem.

I’m working on a new version of a mini T2 board. STN9360 replacing KSA1156 and also moving offset  servo up on tube side of the PCB – making it easier to replace a faulty OPA197. No problem with balance servo… knock on wood.

Any ideas regarding a more reliable offset servo are more than welcome.

Edit. I mixed up the balance and offset servos in the text above – now  it’s correct.

Edited March 18 by JoaMat

[Shawn]

4 hours ago, JoaMat said:

Any ideas regarding a more reliable balance servo are more than welcome.

I have the same issues with OP27. The OPA197 seems to be more reliable but still has a chance of going wrong.

According to my experience, those circumstances may be relative to falling:

1. Kerry Design PSU with HV delay On. The +220V will keep at ~+20V and get back to +220V after 40 secs. In this case, it may somehow overload the offset servo for a while.

2. If the three high-volt sources are not engaged at the same time. I forgot to plug in +220V, after plugging it in, it burned my OPA197.

3. Switch off the amp and turn it on immediately. It happened to me once.

I lack the knowledge to analyze the falling but those are my experiences after burning a couple of the OP27s and OPA197. The OPA197 has a higher tolerance of supply voltage(36V) with protection.

 

After a few hrs troubleshooting the left channel humming issue, it seems like it is relative to the left heater wire of EL34, When I touched and wiggled it, the humming tone changed. I guess the parasitic capacitance changed when I touched it but not sure since the other channel is totally fine without the humming issue. The two pairs of heater wires came from the same space connector with different windings. Don`t know if it is bad grounding or not.☹️

In this way, I made this(attached below), and it somehow fixed my problem.

LT1083( 7.5A Max output at 5V) since I have it on hand. The good thing is the voltage drop is 1V which means the 6.3VAC can be accepted as input without any problems.

The bad news is the heatsink is super hot and I have to redo the layout and get the larger heatsink instead.

Still don`t know how the heater AC Coupled to Signal circuit.😑

Joatmat, Can you take a screenshot or photos of the empty left channel board(Front and Back) by any chance? There may be something wrong with my board layout.

 

 

[G600]

Install that reg remotely on the side of the case?

Edited March 18 by G600

[JoaMat]

Glad you found a solution to the hum problem.

I don’t have any empty board left – last one sent to Finland a week ago

Below print screen from my PCB software, silk screen disabled, version .24.

 

[Shawn]

Thank you so much for sharing the pictures. After checking everything, I dont have any mismatches. BTW there is a wire cut in your back here but I believe it's the software problem.

6 hours ago, G600 said:

Install that reg remotely on the side of the case?

I still have space for two regs in the PSU case. Mounting the LT1083 to the case maybe a good idea to shrink the board even smaller. I noticed the time for balance to be stable became longer when you used DC heaters.