[jekjek]

I used 0.47uf WIMA MKP4 caps per Kryl0071's suggestion. I have some Jantzen Z-Superior 0.1uf caps on hand for my Eico HF81 project. I may try some cap rolling to see if its the caps or the capacitance that have the largest effect on the SQ.

On a side note, I'm glad the designers used actual Rubycon and Nichicon caps in this amp, although they aren't audio-rated.



Those four-number tube codes (1633, 7308, 8416, etc.) often mean the tube is industrial-rated. Not always though - like you said the tube designations don't make sense sometimes! However, just because the 1633 works in the APPJ doesn't mean it's working properly. When you run a tube's filaments (heaters) below their rated voltage, it reduces the tube's life by causing cathode poisoning.



Tube heaters have a operational tolerance of ±7%-±10% (depending on manufacturer suggestions). The APPJ's preamp section supplies 12.6V for the heater, so you're running the 1633 at 50% lower than what is is rated for. It probably won't last too long.

My bad. Sorry
I'm now using Raytheon 6414 as driver with a pair of EL3N
Result is good!

 

[CZ4A]

My bad. Sorry
I'm now using Raytheon 6414 as driver with a pair of EL3N
Result is good!

No need to apologize, it's your amp. I just wanted to warn you that the 1633's lifespan will be drastically reduced in the APPJ. If it was cheap enough and sounds good enough to you, use it.

 

[Pratt99]

Heads up. This amp is now available at Massdrop.(3 Days Left)
https://www.massdrop.com/buy/appj-1502a-tube-headphone-amplifier

 

[Minbeo]

Just want to ask how good this amp with HD800, how is it compared to Schiit Valhalla 2 or Ember???
Thanks alot.

 

[benjisun]

It sounds very good with my HD800. I would like the soundstage even wider but regarding tonality and details everything is ok.

However my amp have some mods : tube replaced : JJ 12A7x and Tungsol 6l6g + Capacitors replaced (audyn 0.47 mkp and 1000 uf electrolytic mundorf cap.

I think the 1000 uf electrolytic as mentionned earlier in the thread doesnt work for me: Imho it somehow bring more control but make the amp loose some dynamic.
I'm thinking to change again to a value close to original cap (220uf) Maybe 330 or something just a little bit higher than original.

 

[pvico]

For those interested, here is the schematic of the amplifier.
https://www.dropbox.com/s/ggm7a0la2era3n3/PA1502A APPJ Schematic.png?dl=0
Here is the LtSpice simulation package:
https://www.dropbox.com/s/rk5oukakn2bzwzg/PA1502A APPJ.zip?dl=0
Note:

• I did not dig into the switched power supply and only the left amplifier section is included
• I could not measure the value of capacitor E05, I just set it to 100nF. For E03, I measured 20uF with my LCR meter but this is affected by C09 through R27 and E03 is certainly much smaller than that. Anyway, it's just there to prevent Q2/IC3 from oscillating.
• I did not find any data for the "Japan Z11-EI48*24" output transformer, I measured the serial resistance of both loops and set reasonable values for the inductances and turn ratio
  
• I used 80 ohm for the headphone impedance (I use a Beyerdynamic DT770 PRO 80 ohm headset)
• Only the high impedance switch position is simulated. Setting the switch to "L" connects the output to an intermediate tap on the output transformer
• In the simulation, I used a 12AX7 model for the 6N4, a 6V6GT model for the 6P6P and a BUZ90 model for the 4N60 MOSFET

Whatever power tube you use, its quiescent current is set to 25mA by the combination of Q2/IC3/R29/R30

EDIT: I measured HT (179V), the heater voltage (11.7V, a bit low), the cathode voltage (9V) and the output transformer inductances. The schematic and LtSpice packages have been updated accordingly

 

[Maxx134]

Heads up. This amp is now available at Massdrop.(3 Days Left)
https://www.massdrop.com/buy/appj-1502a-tube-headphone-amplifier

Wow dam thats a killer pricetag!

Just want to ask how good this amp with HD800, how is it compared to Schiit Valhalla 2 or Ember???
Thanks alot.

Ember is more bassy compared to the APPJ anemic stock unit.
Changing output tubes to 6V6 and 6F6 solve that.
Upgrading caps gives massive upgrade in resolving nature to trounce most other amps.
Cannot say about Valhalla2.

Imho it somehow bring more control but make the amp loose some dynamic

Check the voltage rating of those caps.
I paralleled mine with an "Audyn" cap of smaller capacitance, but higher voltage to not have any transient problem.

BUZ90 model for the 4N60 MOSFE

Sk the BUX90 is better spec than the 4N60?
If so, I would assume better current capability?

Wow, how did you get the schematic thats great!
I could not even find link to the designer, much less any info...
This is great info...

 

[pvico]

Sk the BUX90 is better spec than the 4N60?
If so, I would assume better current capability?

Wow, how did you get the schematic thats great!
I could not even find link to the designer, much less any info...
This is great info...


No, but 4N60 is a chinese MOSFET for which no SPICE model could be found (but there is a datasheet somewhere). BUZ90 is close enough. Any MOSFET with similar ratings would fit the job. It has no impact on sound. The only important fact is that the quiescent current is fixed at 25mA.

As for the schematic, I used my iPhone camera, my USB microscope and my continuity tester...

 

[pvico]

I must say that I was surprised to see that the cathode capacitor was rated 200V. I just received my new capacitors: Mundorf .47uF EVO silver gold oil for the coupling capacitor and Nichicon 1000uF 50V for the power tube cathode capacitor. I measured the voltage at the power tube cathode at 9V. So the 50V electrolytic should be fine, but why did they put a 200V capacitor there?

 

[baronbeehive]

It sounds very good with my HD800. I would like the soundstage even wider but regarding tonality and details everything is ok.

However my amp have some mods : tube replaced : JJ 12A7x and Tungsol 6l6g + Capacitors replaced (audyn 0.47 mkp and 1000 uf electrolytic mundorf cap.

I think the 1000 uf electrolytic as mentionned earlier in the thread doesnt work for me: Imho it somehow bring more control but make the amp loose some dynamic.
I'm thinking to change again to a value close to original cap (220uf) Maybe 330 or something just a little bit higher than original.

I'm using Kaisei 1000uF, 25v caps and for me the dynamics have improved enormously. The only thing concerning me was whether I have lost any higher frequency response due to not using bypass caps but I don't think I have and I won't be able to check properly until I can compare it with my headphone amp which is currently out of order.

 

[Maxx134]

but why did they put a 200V capacitor there?

I have found and realized that manufacturers try to get away with whatever they have available to fit the job.
Most likely either the shipment of that spec cap was cheaper, or they are using cap for more than one location, so it work out cheaper for ordering parts if yoh can get away with whatever you already have...

The only thing concerning me was whether I have lost any higher frequency response due to not using bypass caps but I don't think I have and I won't be able to check properly until I can compare it with my headphone amp which is currently out of order.

The increase in Cathode cap is to stabilize that voltage drop for bass sustain and bass punch,
Which is optimized with larger Cathode Electrolytic value of 1000uf.

This is due to the "larger" cathode cap value causing the effect of holding the cathode voltage constant, in relation to current demands,
So the tube is kept at optimal gain.


Also, at such large capacitance of 1000uf, it should compensate for any performance or noise issues of the combo of Q2&IC3 wich, as "pvico" noted, are china parts which set the tube bias at 25ma.


Most good electrolytic caps perform similarly,
but the one we used is the "Audio Note KAISEI",
Which is supposed to be the best as it closest to the "blackgate" caps,
which were highly regarded but not made anymore.
So Audio Note Kasie caps are recommended first, then Panasonic or Elna silmic2.


Anyways, adding the second paralleled cathode cap(bypass) is for:
Quoted from member "Ridge78" :

"Bypass : in order to help the Lytic cap in the areas it is weakest (treble, microdynamics, ...), we add in parallel an another cap of a different techno, smaller and faster."

Since the KAISEI caps are already good enough,
my reason for adding bypass was to offset any tonality changes of the Cathode cap (KAISEI, have a slight darker tonality) compared to the coupling cap.

So my coupling cap and bypass cap are same exact brand & model.
But just using the Audio Note KAISEI is just fine as it is "supposed" to be the best..



I editing this post a year later because I have tried various caps since in different amps and in these types of circuits where there's a cathode cap, it will play a factor to the sound, and Kaisei caps in tbis situation will provide a darker richer tonality, while a Nichion KZ will be more neutral, and I prefer them over any Silmic2.

Remember the caps are in a situation to affect tonality only because they are in a position of being a "cathode cap".

 

[baronbeehive]

.............

The increase in Cathode cap is to stabilize that voltage drop for bass sustain and bass punch,
Which is optimized with larger Cathode Electrolytic value of 1000uf.

This is due to the "larger" cathode cap value causing the effect of holding the cathode voltage constant, in relation to current demands,
So the tube is kept at optimal gain.

...........



.........

Yes, and that is what I have found, the gain is now very strong, it almost seems higher, but I realise it must be more stable, and recovery from transient peaks and troughs is at a very high level now. As you say the transients are very sharp now and this is what is so shocking, as I said it knocks me off my chair. The volume doesn't have to be high but these leading edges are now very prominent, and they have so much more textural detail, so I have no doubt that the cathode caps are doing their job. It all goes to add realism to the music and I'm pretty sure that applies to the high frequencies as well.

 

[pvico]

Additional considerations, as a reply to some private messages:

R35 sets the feedback level and E05 increases the feedback at higher frequencies. I have no idea of the value of E05, .1uF is just an arbitrary value used to run the simulation.
Removing E05 would reduce the feedback at higher frequencies and increase the treble output. It would modify the phase shift of the feedback and this could cause some stability issues (oscillations).

As there is no audio signal through them, I don't think that IC3 & Q2 contribute significantly to the noise of the amp.

If you want to increase the bias current, you can add a resistor in parallel to R29/R30. Let's say you want to increase the bias current to 50mA, IC3 controls Q2 current to have 2.5V at its ref input, so it needs 50 ohm between ref input and GND. You need a 100 ohm resistor in parallel to R29/R30.

The 4N60 is rated at 4A & 106W, well above the default parameters of the amp. Out of the box, Q2 dissipates about 240mW according to the LtSpice simulation.

Running the simulation with a 6L6 instead of the 6V6 and with 100 ohm in parallel with R29/R30, the power dissipation of Q2 is now 450mW. You should probably put it on a heatsink.

Another factor may be the power limitation of the switching power supply. Out of the box, it provides 16W:

• Heaters (the power tubes heaters are connected in series): 450mA + 150mA at 11.7V = 7W
• HT: 2 x 25mA + 2 x 0.5mA at 179V = 9.1W

According to other users of this forum, it can accept tubes requiring 0.9A of heater current. Heater power is then 12.3W (11.7 x 1050mA) for a total power of 21.4W.

Increasing the bias current is most likely wanted when using bigger tubes like 6L6's. Increasing the bias current to 50 mA will increase the HT power to:
(2 x 50mA + 2 x 0.5mA) x 179V = 18.1W
For a total of 12.3 + 18.1 = 30.4W
This probably exceeds the power supply capability

EDIT: the power calculations have been updated with the measured voltages

 

[Maxx134]

Yeah,
I am very tempted to increase PSU capacitance (by C07&C08),
but do not know what is the capability of the PSU to handle increased initial surge current from such increased capacitance.
Oh well..

 

[pvico]

Yeah,
I am very tempted to increase PSU capacitance (by C07&C08),
but do not know what is the capability of the PSU to handle increased initial surge current from such increased capacitance.
Oh well..

There is not much space to put bigger capacitors there. Moreover, my schematic is incomplete on the power supply side. There is another 4N60 MOSFET between those 2 caps adding either some kind of regulation or just an enable/disable function. I don't know what drives the gate of that MOSFET