1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.

    Dismiss Notice

Little Dot MK8SE / MK6 Super Mods (All verified mods are on first page)

Discussion in 'Headphone Amps (full-size)' started by redge78, Sep 26, 2015.
184 185 186 187 188 189 190 191 192 193
195 196 197 198 199 200 201 202 203 204
  1. baronbeehive
    Really! I don't find that about being analytical. I think that may be down to the choice of tube. The TS 6SL7, for example, is very magical sounding, no way is it analytical except that it does have great detail as well as musicality.

    Interesting to hear your comparisons.
  2. baronbeehive
    I don't know if this makes any sense:


    Edit: Booboo 1 - I think the binding post should go down between the 2 nodes above the transistor.
    Last edited: Feb 26, 2018
  3. MrCurwen
    baronbeehive all good with following notes;

    1) Why do you need a binding post between R19 and the Darlington pair / FET gate circuit? Just use R19 as the binding post?

    2) If instability occurs, put a 10R or more resistor between each CCS bottom and M1 source. Don't add these resistors "just in case", only if instability occurs. Probably won't, the active filter doesn't have that much gm.

    3) You might need a binding post between M3 and the cathode node. As Sonic mentioned, preferably use a small value like 10R. If instability occurs, increase to 47 or 100 R.
  4. baronbeehive
    I don't, that was supposed to be the one that Sonic mentioned which I think I got correct in the first draft but deleted it between M3 and the cathode node, it looked wrong when I looked at Sonic's pics.
  5. gug42
    Hello Guys,

    And thx to reply to me :)

    Here, it be call "transversal capacitor" Noted Ci. Don't know if it help.

    I use a mkp capacitor. 35 uF because I have thoses in my box ... so more simple than order a another 30uF :wink: The schematics is correct about the wiring. The objectif is to provide a little power boost in the bass (if I have understand everything)

    Yep SE amp. Sorry If I misunderstand something ...
    cascode constant current source ? and not used to provide some balance in a push-pull but for define a constent current in place of Rk.

    Please can you develop ? How to check this ? Can I see it with an oscilloscope ? Really interested indeed :)

    Well at the end, I'm pretty happy with it, and yes, it's a fabulous learning tool. I have learn lots of thing for sure !

    Last edited: Feb 26, 2018
  6. MrCurwen
    Kind of. The objective of the Ultrapath topology is to trade cathode bypass cap (of big capacitance value, electrolytic by necessity) to the Ultrapath cap between cathode node and B+. This cap can be of much smaller value to provide adequate bypass, therefore a good quality film cap can be used instead of electrolytics.

    Think of it like this; the cathode node needs to be at AC ground. This is achieved by letting all AC pass thru from cathode node via a capacitor. Now usually it passes to ground; the amp does have one more DC rail, the B+ rail. There is no reason not to 'ground' the other end of the cap to B+ instead of ground.

    The only meaningful thing for the cap is that the other end is at an steady DC rail. Both B+ and ground (in theory) are just that.

    Now why can the cap be of smaller capacitance if it is tied to B+ instead of ground? Well, the capacitance is determined by the R (or Z) component of the highpass filter. If the cap is tied to ground, the R is the cathode resistor, usually not of very large value. Therefore the cap must be of large value (Google RC highpass calculator and try out some values to see what frequency response you get).

    If the cap is tied to B+, the R component is made up of the impedance (Z) of the tube and the OT. These easily come up much larger than a typical cathode resistor, therefore less C is required to achieve the same frequency response.

    There's also some current loop stuff going on but it's less consequential. This is the meaningful point.

    No cascode, only one CCS. I'm not sure what you are saying there, please elaborate?

    In a long tailed pair (LTP), which is what the LD has, the tail is supposed to provide balance for the pair by acting as a current sink. The larger the impedance (compared to the tubes' anode resistances and plate loads), the better the balance.

    Now a strong CCS such as I and Sonic propose provides so many megaohms of impedance that it's practically infinite, therefore balance is practically perfect.

    Please elaborate on the harmful effects you mentioned.

    If we are indeed talking about the Ultrapath type retro SE amp, there is absolutely nothing in there that provides superclean transients. If by 'exaggerate' you mean slightly distorted transients then that's another thing.

    The simplest way to improve transient response in that amp would be

    1) Fixed grid bias. Trade big cathode bypass electrolytics to small good quality grid coupling caps.

    2) Regulate B+ with simple 2 FET regulator. This plays out most of the PSU electrolytics and most of all the tube rectifier. You do realise guitar amp guys actually deliberately put in tube rectifiers to get a compressed sound? Tube recs are ANTI TRANSIENT RESPONSE, big time.

    That's great! Do you plan to keep it as is, or experiment and mod?
  7. baronbeehive
    I was only speculating, I suspected some kind of bass/treble boost such as you find in ipods and such like which is there to make the sound acceptable for mini headphones. It can sound great but for me the frequency spectrum is too unbalanced.

    Glad you are liking your amp, well done for achieving this.
  8. gug42
    For sure I will do some modification, but not before the end of march ... no time before.
    Well I have some test in mind :
    - tests 6SL7
    - tests some interstage capacitors (I have some russian caps) And I can I would like to refactor the amp.
    - DC regulated heaters (LM317 or LT1084)

    And well in a galaxy far away :
    - tests some lundahl opt,
    - tests DHT tubes like 45 or 2A3 : Really I would prefer 45 but hard to find ... And in new production only emission labs ... costly ....
    - tests a full mkp dc link power supply line

    And I must read and understant more stuff about tailed stuff.

    And By the way I wait my usb scope to mesure the amp :)

    You think the bass/trebble boost can be in the recorded tracks ?? I mainly use tidal in hifi quality.
    I use a schiit modi multibit dac.

    Yeah I'm pretty happy to acheive this, a challange for me :) And of course, now, I want more !

    Thanks guy !

  9. baronbeehive

    Parts List for CCS mod:

    1 - 2 x transistor (Q3) 2N2222 - NPN
    2 - 2 x transistor (Q4) 2N2222 - NPN
    3 - 2 x mosfets (M3) IRF830 - Vishay IRF830APBF N-channel 5 A 500V 3-Pin TO-220AB
    4 - 1 x mosfet (M1) IRF9640 - Vishay IRF9640PBF P-channel 11 A 200V 3-Pin TO-220AB
    5 - 1 x cap (C3) Nichicon Aluminium Electrolytic 22uF 160V +105C
    6 - 1 x cap (C4) Nichicon Aluminium Electrolytic 10uF 160V +105C
    7 - 2 x resistor (R19) Takman Metal Film 220K 0.5W 1%
    8 - 2 x resistor (R18) Takman Metal Film 470R 0.5W 1%
    9 - 2 x resistor (R20) Takman Metal Film 100R 0.5W 1%
    10 - 2 x resistor (R21) Takman Metal Film 10R 0.5W 1%
    11 - 1 x resistor (R6) Takman Metal Film 1K 0.5W 1%
    12 - 1 x resistor (R7) Takman Metal Film 1K 0.5W 1%
    13 - 1 x resistor (R8) Takman Metal Film 10K 0.5W 1%
    14 - 1 x resistor (R9) Takman Metal Film 220K 0.5W 1%

    I haven't decided the value of the 470R sense resistor yet, I'm not sure if it is taken from board measurement of the B+ or calculated from the B+

    I'm not sure how to put 2 resistors between each CCS bottom and M1 source if there is instability, given that both the CCS's are joined, surely only 1 resistor will be possible here?
    Last edited: Feb 28, 2018
  10. bloodhawk

    Does any one have images of the CCS modules installed in the MK VI+?
  11. SonicTrance
    I answered this on the previous page. There’s a formula to calculate the current. The CCS will keep that constant current at all times. Has nothing to do with B+.
    Only R18 needs to be metal film 1%. The rest can be 5% and 1w. Especially R19 should not be 0.5w, at least 1w.
    Otherwise it looks good. Binding post also not necessary.
    Last edited: Feb 28, 2018
  12. SonicTrance
    I posted some very clear and detailed pics a few days ago.
  13. bloodhawk
    Interesting, i only saw the ones of the modules themselves, but not wired up in the amp. I might have totally missed the post though, will go back and double check. Thanks!
  14. MrCurwen
    Series feed OT amps pretty much always have an anti-loudness frequency response, meaning at least some depression in bass and high treble. It's the "sweet mids tube sound" thing.

    Add grid current and RC cathode bias and you've got a pinch more loudness filter.

    I'll comment very shortly here. If you're interested in some suggestions and further comments, please PM me or make a new thread and link it to me. I'm very happy to provide suggestions and info. I started out with SE amps, I've gone thru quite a lot with them.

    Tube types aren't the important thing (up to a point, they need to have nice curves). Simple changes in circuitry bring back so much more in sound gains than any tube type change.

    Same here. I'd suggest trying to go around the problems with the caps.

    But do use russian caps. They're very rugged, real NOS, and if nothing else they can be used as binding posts in P2P layouts. That's a big plus. Cheap also.

    Unless you've messed up with the wiring, this will do nothing for the sound quality. No need for it at all. Badly filtered DC WILL sound worse than clean AC, because it has spikes in the voltage (triangle wave or similar).

    Why pay for iron? I'd suggest engineering around it. A cheap Hammond will sound better in the right circuit than your Lundahl.

    DH tubes will sound horrible unless filament supply is actively filtered. This is a real hassle once you go up in current. If you want to taste the DH magick, I suggest you use 4P1L.

    It's the most linear and best output tube there is anyway. Why go for second rate tubes like 2A3? 45 is crazy expensive; if you want Golden Age Americana, go for 47 or 1619 or similar.

    Crazy crazy expense. A simple 2 FET reg will solve these issues AND sound better.

    All good, except you need to use 1W for all except the sense resistor. Also just use carbon films.

    Don't buy 4 FETs, buy at least 10. Same for 2N2222. Trust me.

    Sonic gave the formula for that value; 1.16 / R = desired I

    I use 1.13 as the factor. It varies, it's determined by the base voltage of the 2N2222(A) you happen to get. Use whatever, it's not that sensitive. Basically the formula comes from keeping the Darlington pair at the exact threshold of conductance. That's where the CCS stabilizes.

    In that scenario the CCS bottoms would not be joined, except via the two resistors. That's the point. Don't worry about it, I don't expect there to be instability, so just don't use them unless there is.
  15. SonicTrance
    Right. I misread your post. I did post pics on them installed in the LD also some time ago when I did the testing and measurements.
184 185 186 187 188 189 190 191 192 193
195 196 197 198 199 200 201 202 203 204

Share This Page