JeffYoung's DIY Torpedo 3 Build

Discussion in 'DIY' started by JeffYoung, Jun 17, 2017.

  1. JeffYoung

    JeffYoung Friend

    Pyrate
    Joined:
    Apr 11, 2017
    Likes Received:
    714
    Trophy Points:
    93
    Location:
    Ireland
    @TomB & @dsavitsk, I'm building a DIY T3, but since the PCB boards are no longer available I've no automatic way to pay my portion of the development costs. I could just PayPal $50 to each of you, but that presumes I know the business arrangements between you.

    It might be cleaner to add a Blood, Sweat and Tears "product" to Beezar.com (with say, a price of $100) that I could "buy". Thoughts?
     
  2. dsavitsk

    dsavitsk Friend

    Pyrate
    Joined:
    Oct 2, 2015
    Likes Received:
    1,616
    Trophy Points:
    93
    Home Page:
    You're just building a point to point amp based on our design? If that's the case, it's awfully nice of you to want to pay us, but I don't think it's necessary. I do hope you'll experiment a bit.
     
  3. JeffYoung

    JeffYoung Friend

    Pyrate
    Joined:
    Apr 11, 2017
    Likes Received:
    714
    Trophy Points:
    93
    Location:
    Ireland
    Actually, I'm going to try a DIY-PCB version. I've made some changes to the board layout since a home-etched PCB won't have thru-hole plating -- I've basically gone to a single-sided version. I had to lose a little flexibility (such as the 110/220v jumpers), and go to panel-mounted hardware at the front (but that will have the added benefit of allowing for more tolerance in case design). And there are four jumper wires on the top going from the output CCSes to the parafeed caps.

    Speaking of which, I also took the opportunity to stretch the board 1/2" in length and lay out the parafeed cap section for Mundorf EVOs.

    I also wasn't sure toner-masked etching would do the finer details well, so I'm planning on point-to-point wiring the input jacks to the volume pot and on to the tube section. If that is done with twisted-pair shielded cable, do you think I could use a 100K pot? (I've got an Audio Note going spare.)
     
  4. JeffYoung

    JeffYoung Friend

    Pyrate
    Joined:
    Apr 11, 2017
    Likes Received:
    714
    Trophy Points:
    93
    Location:
    Ireland
  5. TomB

    TomB MOT: Beezar

    Joined:
    Dec 14, 2015
    Likes Received:
    1,753
    Trophy Points:
    93
    Location:
    Atlanta
    Home Page:
    It's just a power LED (unlike the LEDs in the T1). Note that Doug spec'd a fairly low current with the resistor, though. The milliamps add up, especially with the CCS boards and if you start rolling higher-current heater tubes in there (6829, 6414, etc).
     
  6. JeffYoung

    JeffYoung Friend

    Pyrate
    Joined:
    Apr 11, 2017
    Likes Received:
    714
    Trophy Points:
    93
    Location:
    Ireland
    The LED I mentioned will produce on the order of 100mcd with the resistor / current draw Doug has spec'd (360K, 0.6mA), so I'll just go with that.

    (The forward voltage of the LED turns out not to make a material difference when we're talking a couple of volts on the LED side vs. 200 volts on the supply side.)
     
  7. JeffYoung

    JeffYoung Friend

    Pyrate
    Joined:
    Apr 11, 2017
    Likes Received:
    714
    Trophy Points:
    93
    Location:
    Ireland
  8. TomB

    TomB MOT: Beezar

    Joined:
    Dec 14, 2015
    Likes Received:
    1,753
    Trophy Points:
    93
    Location:
    Atlanta
    Home Page:
  9. JeffYoung

    JeffYoung Friend

    Pyrate
    Joined:
    Apr 11, 2017
    Likes Received:
    714
    Trophy Points:
    93
    Location:
    Ireland
    Thanks, Tom. BDenterprises had them, so I'm sorted.
     
  10. JeffYoung

    JeffYoung Friend

    Pyrate
    Joined:
    Apr 11, 2017
    Likes Received:
    714
    Trophy Points:
    93
    Location:
    Ireland
    Here's my design for case ventilation:

    [​IMG]

    The three rows of holes will also continue down the sides of the case.

    The design is predicated on the tube CCSs being the major heat-source, with the voltage regulators and rectifiers being next, and the power transformer and output CCSs not being significant sources of heat.

    Note that my board is stretched 1/2" around the output CCSs so they have more room and will be mounted vertically on the top of the board. (Everything needs to be on the top as a DIY PCB won't have plated-through holes.)

    Premises sound correct? Hole pattern look reasonable?
     
  11. peef

    peef Friend

    Pyrate
    Joined:
    Oct 5, 2015
    Likes Received:
    311
    Trophy Points:
    63
    Location:
    Montreal
    Having never built one, I believe that in order of hottest to coolest, you're looking at tube > output CCS > follower >> regulator / tube CCS. The latter only sees what, 12 V x 2mA or so? You may want to put the output CCSs on the main board instead of creating daughter boards.

    Not sure if it's within the scope of your build, or if the power transformer can keep up, but you might want to bump up the idle current on your output stage. Something like this could be nice and will take the extra heat a bit better. Watch out since the package isn't isolated.
     
  12. JeffYoung

    JeffYoung Friend

    Pyrate
    Joined:
    Apr 11, 2017
    Likes Received:
    714
    Trophy Points:
    93
    Location:
    Ireland
    @peef Pardon the ignorance, but I'm a digital guy and stumble around analog: the followers are Q2/Q3/Q4/Q5? (Or at least the ones that get hot? Would Q6/Q7/Q8/Q9 also be considered followers?)

    While there's physical room on the board for the output CCSs, my one-layer traces get a bit congested there. (Plus, I've already got the daughter boards.)

    Out of curiosity, one would adjust the idle current of the output stage by decreasing the values of R19/R20/R21/R22? Or do I have the wrong end of the stick?
     
  13. JeffYoung

    JeffYoung Friend

    Pyrate
    Joined:
    Apr 11, 2017
    Likes Received:
    714
    Trophy Points:
    93
    Location:
    Ireland
    @bazelio Do you have any thermal images from back a bit farther showing the whole amp? Your images of the tubes/tube CCSs were very helpful, but I can't quite see what's going on back at the voltage regulators / rectifiers / power transformer.
     
  14. JeffYoung

    JeffYoung Friend

    Pyrate
    Joined:
    Apr 11, 2017
    Likes Received:
    714
    Trophy Points:
    93
    Location:
    Ireland
    Oh wait, those are the resistors that get replaced with the output CCSs. So it's the output CCSs that control the idle current?
     
  15. JeffYoung

    JeffYoung Friend

    Pyrate
    Joined:
    Apr 11, 2017
    Likes Received:
    714
    Trophy Points:
    93
    Location:
    Ireland
    Based on @bazelio 's earlier thermal images, I've altered my ventilation design:

    [​IMG]

    I was trying to avoid slots as I have only the one case to feck up, but all the heat appears to be concentrated in the middle, so slots it is. (I do have a mill, it's just from the 1950s with no numerical readout or control, so one has to pay attention to what they're doing.)

    I hope I don't owe Cisco any royalties over this. ;)
     
  16. JeffYoung

    JeffYoung Friend

    Pyrate
    Joined:
    Apr 11, 2017
    Likes Received:
    714
    Trophy Points:
    93
    Location:
    Ireland
    PCB layout.

    Changes from Torpedo III:
    1. Mundorf EVO cap layout (including length increase of 1/2")
    2. point-to-point shielded twisted-pair wiring from RCA jacks through pot to amp input
    3. changes to accommodate DIY PCB without plated-through holes:
      • addition of 4 jumpers from amp output to parafeed caps
      • 220V-specific
      • removal of unused diodes
      • tube-CCS-specific
    4. floating front panel controls (to allow lower-accuracy DIY casework)
    5. additional PCB support near PSU transformer
    6. addition of PCB chimneys to supply air from case-bottom vents
    [​IMG]
     
    Last edited: Jun 30, 2017
  17. Mr.Sneis

    Mr.Sneis Friend

    Pyrate
    Joined:
    Oct 2, 2015
    Likes Received:
    1,320
    Trophy Points:
    93
    Location:
    Phoenix, AZ
    Can't catch a break. Despite the picture, that switchcraft is red on top :(

    Studying datasheet PJRAS1X2S02AUX might be the proper number?
     
    Last edited: Jun 29, 2017
  18. peef

    peef Friend

    Pyrate
    Joined:
    Oct 5, 2015
    Likes Received:
    311
    Trophy Points:
    63
    Location:
    Montreal
    Yep, they're all followers. Q6/Q2 and others form Darlington pairs. Usually, the gain of power transistors like Q2-5 is fairly low, so using them in a Darlington configuration is a good way to bump up the input impedance (which avoids loading the tubes, improving distortion performance) and reduce the output impedance.

    Yes, but it's a tradeoff. The follower needs to put current into both the resistor (R19-22) and into the primary of the transformer; and it's only the latter current that will couple into the headphones.

    So, if the transistor is idling at 100V (the tube's plate voltage), and the resistor has a value of 10k, you can calculate the follower's quiescent current.

    V = RI
    100V / 10k = 10mA

    The follower sees a load of 1/2 of the transformer primary impedance (Zprim) in parallel with the resistor (Re).

    Re // Zprim = Zload
    10k // 5k = 3.33k

    Now, supoose you have a 10V signal...

    V = RI
    10V / 3.33k = 3mA

    That means for an idle current of 10mA, 3mA serves to drive the load. 2mA for the transformer, 1mA for the resistor. Reducing the value of Re has a few effects.
    • The transistor and resistor idle current is higher, so both will dissipate more heat.
    • The transistor's transconductance goes up, and its output impedance goes down-- but not by much, and not in a way that significantly impacts the output impedance of the whole amplifier.
    • The resistor requires more current to drive, so the extra current is not necessarily available to drive the load.
    • The transistor works into a lower equivalent load, which probably isn't great for distortion.
    Yep, and those eliminate the tradeoff above as the CCS's impedance is enormous-- the current that passes through it is constant regardless of the voltage applied across it, meaning all the current is available to drive the transformer primary.
     
  19. TomB

    TomB MOT: Beezar

    Joined:
    Dec 14, 2015
    Likes Received:
    1,753
    Trophy Points:
    93
    Location:
    Atlanta
    Home Page:
    Looks good, but you may want to re-evaluate your position on the ptp wiring from the RCA jacks to the volume pot. Maybe using coax will fix it, but Dsavitsk did extensive testing when we first made the original Torpedo (T1). He found that everything he tried, direct wiring above or below the PCB to the volume pot, resulted in picking up hum from the PT. The only thing that worked was running the traces on the bottom of the PCB, so that they were shielded from the ground plane.

    Further, we almost thought the T3 was a failure in the beginning. Turns out, Doug had flipped the signal traces and put them on the same side as the 110/220 option pads. Those pads were close enough to the signal traces to introduce hum again. He flipped it back, the hum disappeared, and we went to production with the PCBs.

    As I said, maybe the coax will fix it, but it might be a good idea add the traces anyway. That will give you an option in case the coax is noisy.
     
  20. JeffYoung

    JeffYoung Friend

    Pyrate
    Joined:
    Apr 11, 2017
    Likes Received:
    714
    Trophy Points:
    93
    Location:
    Ireland
    Thanks for the input, Tom. I can try and squeeze them back in, but the reason I was going to point-to-point wasn't really for the hum but because the traces were really thin and long and I wasn't sure the ironed-on-toner-mask for a DIY PCB would be up to the task.

    (The "coaxial" is also a typo; I'll fix that. The wire I have in mind is shielded twisted-pair: http://www.lakecable.com/index.php/...udio/single-and-dual-channel/item/965-avb221a.)
     

Share This Page