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.
First
 
Back
185 186 187 188 189 190 191 192 193 194
196 197 198 199 200 201 202 203 204 205
Next
 
Last
  1. gug42
    Hello,

    MrCurwen, When I will get more time, I will realy be interested ! My first amp is what it is : a first amp, a starting point, a learning machine :wink:
    Now I'm confident in myself to be able to build an amp :)

    Sorry I don't understant what you want to say / explain :frowning2:

    I'm totally agree. I have a good stock of 6SL7 and only two paris of 6SN7 .... I start with 6SN7 for timing reason .... Shipping delay for some resistor ....

    Yep I know. But it's to have exactly 6,3V whatever my local voltage could be :wink:

    I know that DHT tubes needs a careful filament supply. Planning one : filtered and regulated.

    I really want to heard thoses golden age tubes. I'm open to other stuff. 47 could be an option, in triode strapped modes. Indeed that's not the "good" or "best" approche, but ... well .... it's a pleasure, and, at the end pleasure is the objectif (if you see my point ?)

    Did you have some triode strapped curves for the 47 ?


    Ok I will please to talk about this with you :)

    Thank you again :)
     
    Last edited: Feb 28, 2018
  2. gug42
  3. baronbeehive
    Yes I know, sorry I put this post together rather hastily after searching out the parts all afternoon and looking for the best deal I just didn't know if the formula applied to a measurement from the board or to the type of part that was installed which MrCurwen answered. And yes I knew I needed the sense resistor 1% so just got the others the same

    Right thanks for the information, beginning to understand the functionality of the CCS.

    Ah, interesting. In that case gug42 should notice the missing frequencies. Strange that he said that the treble was almost too much.

    OK about the resistors.

    No I was intending to get 5 or 10 of the transistors as you suggested, but I was just saying what I needed for this mod.

    Thanks for the help!
     
    Last edited: Feb 28, 2018
  4. baronbeehive
    Just having a listen since burn in and I think the bass articulation problem seems to be much improved surprisingly, even with the TS 6SL7 in. I'm glad about that, it's my favourite tube, it has such great definition on drums for example. There is a very slight treble roll off with this tube on high hat type cymbal for example which I only just noticed when comparing it to the APPJ but on the whole the treble definition is superb as well.

    Also the soundstage problem, not a problem if you've got the right headphones, seems to be much improved, it seems much more spatial, even on my HE-500's.
     
    Last edited: Feb 28, 2018
  5. bloodhawk
    Is this with CCS? or just general amp burn in since the mods?
     
  6. baronbeehive
    bloodhawk likes this.
  7. baronbeehive
    OMG.... the sound of Gibson Les Paul, and the Strat on this amp, sounds equally amazing on the other amp, to die for!
     
  8. MrCurwen
    That is exactly as it should be.

    Do you plan to mod your existing amp or build a completely new one?

    OT means output transformer. Series feed OT means a topology where the OT primary is in series with the output tube. So it is between B+, the tube, and ground, having the same DC current pass thru it that passes thru the tube.

    Loudness filter means a frequency response with some degree of added highs and lows. Anti-loudness filter is the opposite of that, some degree of less highs and lows.

    Series feed OT amps almost always have some anti-loudness filter going on, that is because of the properties of the OT. With DC bias the OT has difficulty reproducing low frequencies. The DC "drowns out" any nearby (like 50 or 100 Hz) action. Also the windings have a lot of parasitic capacitance; think about it, lots and lots of wire stacked up right next to each other. This capacitance - if not satisfied with current - forms a low pass filter, depressing high frequency response.

    These problems can be satisfied by

    1) not having DC current thru the OT (fixes the bass issue completely)

    2) driving the OT primary from a low output impedance source (fixes the high frequency issue).

    Grid current is the umbrella term for parasitic properties of tube control grids. They generally require some current to satisfy. Imagine there is a capacitor inside the tube between control grid and cathode, and also control grid and anode. These capacitors in conjuction with other circuitry around the control grid act as low pass filters, dampening transients and often dampening high frequency response.

    The most common comments people give upon fixing this grid current insufficiency problem are that the sound is "more airy", "more 3D" and "faster". So, high frequency response and transient response are improved in technical terms.

    RC cathode bias does a similar thing but on the bass side of the frequency response. Point of the cathode C is that it is supposed to keep the cathode node at a steady DC and drown out all AC by charging and discharging accordingly.

    Now once the C has discharged some, it can only be recharged by current that arrives thru the tube from B+. This doesn't happen instantly; think about Ohm's law. There is a tube with impedance of X ohms between B+ and the cathode node. This impedance limits the amount of current that can be transported.

    So it doesn't charge instantly. While the C is not charged, the DC bias point of the tube (cathode node voltage) is different. What happens is the cathode node voltage floats up and down with the signal, a little bit. Now since the tube only amplifies the VOLTAGE DIFFERENTIAL between cathode and the control grid, this means that if the cathode node voltage is floating in sync with the signal, there is less amplification. This effect happens most on the biggest amplitude frequencies, because they resemble DC the most.

    If the cathode C has to discharge a long long time, it takes a long long time to recharge it, allowing the cathode node voltage lag.

    Why does the cathode node voltage fluctuate in the first place? Well, same reason as the anode node voltage fluctuates. The tube doesn't know which resistor - anode resistor or cathode resistor - is supposed to be the output.


    That's understandable. I'd go for 6.1V instead; doesn't hurt operation in any way and doubles lifespan of the tube.

    So what I'm hearing is you like the way the ST envelope tubes look? That's fine. Go for the 47 then, it's every bit as linear as any 45 or similar.

    But I'd suggest testing new circuit topologies with ugly TV sweep tubes first. After you've got them working, then bring in the pretty tubes with difficult PSUs.

    RCA47T01.JPG

    Please note 0V is not at the origo, but rather at the point where the first curve begins. Individual tubes will differ +- 15 % on anode resistance.

    Yes it's basically the "Ultrapath solution" applied to the grid. You can trade a large (and in those times very poor quality) electrolytic for a nice high quality film cap (or paper in oil in those days).

    The FET needs about 4.5V positive on the gate to start conducting. The higher the gate voltage, the more it conducts.

    The Darlington pair needs a certain positive base voltage to start conducting.

    Remember Ohm's law. Look at the resistor above the Darlington pair, let's call it the R. At rest the FET gate is at B+; since there is no current thru the R (the Darlington pair is basically 'open', not conducting) there is no voltage drop across the R.

    Once there is current thru the pair, and thus, thru the R, there is a voltage drop across the R.

    So, the gate voltage lowers. FET conducts less current. There is a smaller voltage drop across the sense R underneath the FET.

    This means the base voltage for the pair is lower, meaning it conducts less current. Less current thru the R means FET gate voltage rises -> FET conducts more -> more current and thus voltage across the sense R -> higher Darlington pair base voltage -> Darlington pair conducts more -> more current thru the R -> FET gate voltage drops -> FET conducts less current etc etc ad infinitum.

    It doesn't actually oscillate like this (best case scenario...), but rather it stabilizes so that any change in the feedback loop is very strongly resisted.

    Every feedback loop contains a time paradox, so it's rather difficult to fully imagine. Don't think about how the loop is first set up, at power on. You'll hurt your brain; I know I cannot imagine it. Which component needs to conduct first?

    The Darlington pair and the FET work in a see-saw. If one begins to conduct more, the other conducts less. But each has their input at the output of the other (the pair is set up like a normal gain stage, with 'anode' output, the FET is a source follower).

    This is inside the loop; looking in from outside all you can see is the FET changing the voltage across itself. When the FET gate voltage changes, this also changes the voltage across the FET as well as how much current it is conducting. This effect is what makes the CCS useful in a LTP circuit. It floats the cathode node voltage to achieve constant current. Cathode node voltage changes tubes' biases. Tube biases change the voltage conditions inside the tube.

    Possibly some high frequency distortion.
     
  9. baronbeehive
    I'm trying not to think about it lol, not hard for me.

    Yes, I thought there must be some sort of dynamic interplay going on inside the CCS to achieve balance, but it's interesting having the filter part of it coming into play to clean up anything bad at the source of the driver stage as well.

    Yes, way back in the beginning.... ah 10 years ago now, when I got the LD I used it to amplify my ipod classic and was impressed, but it in no way compares with a proper clean source that I got some time after that. I realised I was amplifying garbage, noise and distortion! Anyway all that's prehistory now.

    I will go on putting my order together today with the proper parts for when I'm ready to implement it. Nothing much else to do atm we're snowed under, something that rarely happens here!
     
  10. gug42
    Hello,

    Hum, If I can, economically speaking, I'd like to start a new one.

    So a Push-Pull in class A ?
    But loosing some microdetails when the signal cross the tubes ?

    I have read that a requirement is a x3 ratio between tube impedance (Ri) and Z presented at the tube by the transformer ?
    Thank you, I will need to re-read this multiple time :D

    I'm using a 35 uF mkp (so PP film capacitor)

    Thank you for the graph.
    Well I will think but after some reflexion ... the 4P1L could be a very good option for my need !
    Evryone says that's a damm good tube. What is the internal impedance of the 4P1L in triode ?

    Thank you I will take some times and find the times to work on the next one. I will surely post a thread :)

    Regards :)
     
    Last edited: Mar 1, 2018
  11. MrCurwen
    I'm cheap, so my specialty is getting maximum sound quality at the lowest possible price. Best results can be had much cheaper than you probably think at the moment!

    Retro push-pull class A still has DC current thru the primary, only in a kind of sort of balanced way so as to cancel the magnetic bias. What I mean is no DC current at all.

    Yes that's the old time rule. What I prefer is about 20 to 50 ohms output impedance driving a 10k primary. Sounds like there's no OT there at all!

    Basically it's this: grids eat current, especially at transients. If sufficient current is not available, transients sound lazy.

    RC cathode bias (with a resistor and a cap bypassing it) always takes some definition and force (transient impacts) out of bass register.

    Solutions: drive hungry grids with a low output impedance; use fixed grid bias.

    Yes. I'd still suggest going with 6P31S. It is a very very good sounding linear tube. You can upgrade to 4P1L later if you wish.

    Anode resistance is about 1k, maybe 1.2k or something like that.
     
  12. baronbeehive
    Just trying to get a handle on this CCS.

    I've looked at the FET specs for the IRF830APBF and I'm assuming they're the same for that type of FET. They all seem to list the specs slightly differently but from what I can gather the max gate source voltage is 30V, min is 2V, and the threshold voltage is 4.5V, as MrCurwen said.

    For the Transistor, specs for the 2N2222 are, collector-emitter voltage: 40V, collector-base voltage: 75V, and emitter-base voltage: 6V. I'm assuming the transistor pair start conducting at 6V then.

    The formula for the resistance value is 1.16 / R = desired I, or 1.13/R = desired l. I don't know how this is related to the base voltage of the 2N2222. I see that the formula comes from keeping the transistor pair at the exact threshold of conductance to stabilize the CCS, where it stops oscillating I assume. This is the point I was making about whether to take the desired CCS current value from a measurement from the board or from the parts spec ie from the CCS requirements. Anyway MrCurwen said that the current at the CCS is below 2mA, which is at each CCS I assume. This doesn't agree with what Sonic said about his I value being 2.37mA. So do I take the Sonic value, or if not where do I get the value from on my amp?

    I'm a lot happier about the interaction between the transistor pair and the FET now after MrCurwen's explanation!
     
    Last edited: Mar 2, 2018
  13. MrCurwen
    These are breakdown voltages.

    Look at page 2: http://www.farnell.com/datasheets/296640.pdf (don't worry, it's the same part, just the old metal envelope)

    Saturation voltage 0.6V at 15mA, and it decreases when current decreases (we're looking at 1mA or so here). Also Darlington is configured so that the saturation voltages are stacked (both transistors need to be turned on), so that adds up to about 1.13 or so. It does depend a bit on what production batch and manufacturer you get.

    Once again, it's not so sensitive. Just use 1.14 or 1.15 or whatever. Even if you use 0.1% resistors, the FET and the transistors have temperature coefficients, so the current is never going to be 100.00% where you plan it to be. Close enough is good enough.

    I don't know what the context is for my statement. But you can just pick a current you like, and use the formula. It'll be ok. 0.3mA here or there is not going to matter much.

    I'm very happy to hear that!

    Have you thought about the external functioning of the CCS? As in, imagine it's a module, a box dropped in the circuit; what does it do, what is it's function in the circuit?
     
  14. baronbeehive
    I don't want to hog this thread too much so I'll keep it simple.

    Not sure what saturation voltage is, is it max voltage handled, is it optimum voltage for the stabilization of current, I don't know. The ImA or 1.13mA, is that the amount of current the CCS can use to keep everything stable, I don't think so because the test condition specified up to 500mA, so I will stop babbling, until I know more I'm at my limits atm, sorry.

    I've tried several options for calculating R:

    1.15/3.0=.38
    1.15/2.0=.58
    1.13/3.0=.38
    1.13/2.0=.57
    1.16/2.0=.58
    1.16/3.0=.39

    You're right they don't diverge that much, at least it doesn't look like it to me so I will probably pick a value of R somewhere between .38 and .58.

    I have only thought superficially about the CCS in a circuit, obviously it needs to iron out differences when the balanced signals are combined, and it does this within the limits defined by the sense resistor. When the current changes the voltage also changes in the same direction after it passes through the CCS, and when this change is noted in the circuit it adjusts again to keep the current the same.
     
  15. SonicTrance
    I don’t think MrCurwen gives the CCS enough credit. In my experience it’s very accurate and sensitive. I use 1.16 when calculating and it’s always very close IRL. Also, baron, you need to calculate in amps, not mA. So move your comma three places to the right and you have your sense R.
     
First
 
Back
185 186 187 188 189 190 191 192 193 194
196 197 198 199 200 201 202 203 204 205
Next
 
Last

Share This Page