Welcome! Need support, you got it. Or share your ideas and experiences.
Return to VK Music - Elekit & Triode Japan Forum
87.115.0.99
First may I say thanks to Victor for being a pleasure to deal with.
I'm about to build this amp. I have a fair bit of experience designing and building valve hifi but I've never done a headphone amp and the TU-882 seems a good place to start.
I plan to carry out some mods as I build; I was thinking of some component upgrades as follows:
. a better volume control
. omit C1 and C2; and having done that, R1 and R2 can be omitted
. use Takman resistors in all important positions
. coupling caps C5 and C6 to be either Jupiter copper foil or Mundorf silver/gold/oil; I use the latter and like them a lot (after *months* of running in), but an expert friend suggests the Jupiters.
But I'm now considering some mods to the circuit.
1. R3 / R6 grid stoppers ... 3.3K looks excessive; I generally use about 100 ohms, and some folks say that much more than that degrades the sound. The 2C51 is not an extreme valve and I have not seen any mention of a tendency to oscillate that might need a large grid stopper. So I plan to reduce the value to a few hundred ohms.
2. Stage 2 does not have grid stoppers; I'll see if it's possible to add them even though this may mean cutting the PCB strip.
3. Stage 2 grid resistors R13 /R14 are 1 megohm; this seems large, and a spec for one of the 2C51 types suggest 500K max grid resistor so I'm considering using this value.
However, the low output 'power' of stage 1 might make 1 meg work better so I'm undecided.
4. Elna Silmic caps for cathode bypass, with appropriate values. I use these and like them.
5. There is some global feedback via R9 and R7. The arguments run and run as to benefits of feedback. I'm considering putting in a removeable link so as to try it with and without.
6. A more fundamental change; V1 runs at only about half a mA; way too low in my view. (V2 runs at about 9mA at a good looking operating point).
So I'm considering increasing the current in V1 a bit. The supply voltage means the operating point can't be changed that much, but about 1.3 mA looks possible at about 80V on the anode and about 1.7V bias. R11 anode load resistor changes from 220k to about 80k, and R5 bias resistor changes to about 1300 ohms.
R11 will then still be 'large' compared to the valve internal resistance Ri.
Increasing the current should reduce the noise floor and give better drive to stage 2, so improving dynamics and subjective 'power'. But basically I'd expect quite simply an improved sound. Perhaps this is the reason for the slight lack of ultimate resolution that I've seen mentioned.
The best solution would be to provide a much higher voltage to stage 1 but of course that means a separate power supply; I may well try that later.
7. C7 caps, 220pF; these give a bit of high frequency rolloff. Possibly to remove a bit of noise, perhaps resulting from stage 1 running at such a low current? Or perhaps to remove HF resonance from the grid of V2 which doesn't have grid stoppers? I just don't know!
If stage 1 can be improved, C7 may not be needed; I'm tempted to reduce its value a bit and/or put in a removeable link to try with/without it.
In conclusion, I think there's a lot of potential for improvement, but even so I'm a touch confused as I don't know why some things were designed as they were, and evidently the unit works OK as standard.
I have enough confidence to try my mods, but any comments are welcomed.
Follow Ups:
Hi Mike, Well my 882 flew off the table as a result of a nightmare and will soon go to the generous Mr. Victor to repair a noise problem. If I wasn't so lazy these days I would have plucked the first 2 coupling caps. They probably should have never been there. I'm a tweaker not an electronics designer but I have a few ideas. Let me know if you have already built it. If not I'll have more to say. I've rolled quit a few tubes and the GE 5 star 5670, circa 1970,2, not 3 mica with a halo getter is stunning. I'm a big fan of conductive grease on contacts, pins, rca's... Used carefully they are a huge improvement. I use carbon in the signal path and silver grease in the PS. Silver's too bright for me in the signal path. Well I could go on and on. It's not all about the electronics. You made an excellent choice. Tweaker
Thanks for the recommendation for GE 5*; I've just got round to trying these and I'm very impressed Stunning, as you say!
I've been using the stock valves (not bad at all), and Sylvanias (better, very good sound).
The GE's add to the strengths of the unit, taking it to an even higher level. Transients (always good) are even more explosive, bass is tighter, top end a touch more detailed, and sweet and civilised with it; rhythmic and musical, gorgeous vocals.
I'm sure you'll have seen my posts giving full details of the mods and the results; but I'm interested to hear your other ideas as I'm going to build another for a friend. Also, another friend is building one now.
Thanks for the recommendation for GE 5*; I've just got round to trying these and I'm very impressed Stunning, as you say!
I've been using the stock valves (not bad at all), and Sylvanias (better, very good sound).
The GE's add to the strengths of the unit, taking it to an even higher level. Transients (always good) are even more explosive, bass is tighter, top end a touch more detailed, and sweet and civilised with it; rhythmic and musical, gorgeous vocals.
I'm sure you'll have seen my posts giving full details of the mods and the results; but I'm interested to hear your other ideas as I'm going to build another for a friend. Also, another friend is building one now.
Woops, Sorry for the double post. Usually on other forums I can delete it but it wouldn't let me here. Mike, I saw in your original post on what you were planning and thinking that you already know about the Silmic's. Seems we are on the same page. Tweaker
I'm thinking Elna Silmic II if they come in hi voltage for main PS cap and 10v cathode bypass caps. They can be a bit big. Pretty sure the 10v 47uf are about if not even smaller in diameter than the os-cons. If you haven't heard the Silmic they must be listen to. It could be that the SimicII is the GE 5 star of the electrolytic cap world, just sayin'. YMMV Tweaker
I'm thinking Elna Silmic II if they come in hi voltage for main PS cap and 10v cathode bypass caps. They can be a bit big. Pretty sure the 10v 47uf are about if not even smaller in diameter than the os-cons. If you haven't heard the Silmic they must be listen to. It could be that the SimicII is the GE 5 star of the electrolytic cap world, just sayin'. YMMV Tweaker
Just don't tell anyone else about the 5 stars. I thought I was set and found a used pair at a reasonable cost and said what the heck. The designer says not to remove R1 and R2 to protect the amp from oscillating if the vol. control disconnects for some reason?? I was thinking of using .22uf for the one coupling cap but the designer says this will or can cause the amp to oscillate. I'm not an electronic trained person, mainly a parts upgrader and a tweaker. Once again carefull use of carbon grease on the tube pins and any connector in the signal path. Silver conductive grease on PS connectors. Very messy stuff, like shoe polish and needs to be used sparingly with care not to get it on clothing... Soon I plan to finally venture into the amp. How do you like the Takman (REX?) Tweaker
Hi,
thanks for the comments. I like Takman carbon film resistors a lot; I'm not sure I can tell much difference if any to Shinkoh which were my resistors of choice until Takman. There are many choices of high end resistors nowadays and I'm sure there are better than Takman; but in general I've rarely found much difference between decent resistors.
Removing R1 and R2; I can't see this as a risk; it's just like the first stage of a preamp and I've never put a resistor in parallel with the pot in this way, never had any problem.
Elna Silmic are indeed fine as cathode bypasses but I can't find them in a suitable voltage rating for PSU use. In my own DIY gear I always use multiple chokes and film/oil caps, by far the best way IMO (if well designed, PSUD is fabulous); I may get round to trying this on the TU one day but the standard PSU is very good.
I can't see any benefit in going above 0.1uF for the coupling cap. I'm using 0.047uF and it seems fine. If you're going to try changing this cap I do very much suggest the Jupiter copper (top of the range).
Thanks for the interest!
Mike , When you talk about removing R1 and R2 along with the two input coupling caps do you have to jump those? It's not obvious what to do to me. Also, did you give C12 and C13 any attention?? I calculated the -3db point at around 60hz with a .047. If correct I would think that would be to high. Tweaker
Hi,
Simply remove R1 and R2.
The two input caps have to be replaced by a link.
I left all PSU caps as standard including C12 and C13 but I may replace these with good quality Panasonic.
You must have made an error in the calculation - we've all done it! :-)
Z = 1/ (2 x pi x F x C) where F is the frequency and C is capacitance.
Rearrange to:
F = 1 / (2 x pi x C x Z) where Z is the impedance.
At low frequencies we can totally ignore the small filter cap C7 and the miller capacitance of the output valve, so the load is just R13.
When the impedance of the coupling cap equals the load R13 (1 megohm), half the signal will be lost across the cap and half will appear across R13 (look on it as a potential divider). This gives - 3 dB.
So use C = 0.047 / 1 000 000
Z = 1 000 000
and you'll get F = 3 Hz near as dammit and that's ideal.
RC coupling give phase shifts, roughly speaking these affect only up to about 10 x the -3db frequency, so there will be no appreciable effect above 30Hz (and nothing much to worry about above 15Hz).
Hope this helps; do ask about anything you are not sure about.
"The only silly question is the one that should have been asked but wasn't". :-)
Mike, C12 and C13 replacement is a no brainer. Just what to use? The Panasonics get huge respect, one person thinking that the ED is an equal to the BG. You, Nelson Pass, Michael Samra and others all like the Panasonics for the 350v cap. In Panasonic I see two different alternatives, the ED and the EE. I guess I'll try them both. Any notion of how big they can safely go up to from 22uf. Tweaker
Hi,
I know very little about electrolytic caps as I rarely use them.
I've tried Black Gates in the past and I'm not convinced even though many people rave about them. From what people say, Panasonic are worth trying. As for ED or EE, check the data; 105 deg C is good, and you want minimum series resistance and maximum rated ripple current (the two tend to go together); also look them up on Tube DIY Asylum.
As for capacitance value ... I can't help as I have no experience of regulated supplies as used here, or of RC dropper stages. I do know that too much capacitance can be harmful; it can sometimes give a 'slow' sound and mess up leading edges. My unit has a fast, clean, engaging sound with fine clean leading edges so myself, I'm leaving the values as is.
My unit is very quiet so I see no need to increase the capacitance.
It's worth seeing what other people have done and try things out yourself carefully, just be aware more capacitance may not be an improvement.
Actually I think that the power in this circuit is not regulated. I believe that the fet is used as a ripple filter. Not exactly regulation, I'm pretty sure. I'm no expert but there are no regulators. Tweaker
Mike, My experience with SS power supplies is that more better. Improvement in bass speed, impact... I have a note to the designer as to how high it can go. The Panos are well loved, as stated before. Being in the more is often better camp a higher rated FET would be an all but certain improvement ,IMO. A little like a higher gauge power cord. I would replace the 2SK2750 with the highest current, wattage rated device that would operate the exact same as the 2750, in the same or similar TO-220 package. Can't pick this out myself, it's out of my league. Tweaker
The phase shift thing. Read your post to the end. The V-cap person in a conversation said you can probably get away with 2x. Printed the 10x suggestion on his site. There have been blind tests as to the perception of phase shift that people have failed. It may be a bigger issue in headphones than in speakers? I went from a overboard 4.7 junk cap to a Muse .47 on my dac output , going into a 47k input,(got that one right) and only noticed a big improvement. Tweaker
"What is the input impedance of the component you'll be feeding" This Mike, is the question asked on the V-Cap coupling cap calculator so I don't have to go back to school. The input impedance of the amp is stated at 50k, I measure 45kish. The headphones I use are 75 ohm and of course this is irrelevant. So then, the component I'll be feeding is the second stage of the tube at 1M. This is what I didn't get. The V-cap calculator works fine, gets around 3hz with 1 meg plugged in. A minus 3db at like 15hz should be ok shouldn't it?? This would mean much smaller than a .047 could be used?? Is it just a phase shift issue or would it be something else?? The next problem is how to get silk fibers into Pano caps? Great info, Thanks, Tweaker
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: