Welcome! Need support, you got it. Or share your ideas and experiences.
I'll start by giving the results of my substantially modded build.
A good friend Richard is very fussy about sound quality; he has a TW Raven turntable and superb electrostatic speakers.
I am fussy about sound quality; I have a Teres / Moerch /Zyx turntable, excellent valve electronics, and a superb Gryphon Scorpio CD player.
We both have AKG K812 headphones. We both know *good* sound.
Richard currently uses a MF X-Can 2 headphone amp, but is always on the lookout for something better and has tried several.
When Richard heard my TU-882, after only ten seconds came the firm exclamation 'Best yet!'. And that's it really.
A week later we had a shootout at his place, using AKG K812. A Graham Slee amp was easily outclassed by the TU and by my Cayin HA-1A (modded and with seriously good valves), both of which were not even warmed up.
After some warming up, things were very interesting. Both the TU and the Cayin were sounding superb, with a preference by all for the TU. The MF X-can was good, but not quite as good.
Richard was again highly impressed and has ordered a kit.
I'm using the TU882 with the Gryphon Scorpio (£6000 and worth every penny) and AKG K812s, giving *seriously* good sound; Tracy Grammer's fine 'Flower of Avalon' playing now, gorgeous.
Fine tonality in every way, giving character to instruments, not over 'lush' but naturally warm and full bodied.
Superb transparency, detail and air.
Dynamics, transients, fast leading edges, giving a lively, rhythmic, engaging sound - tapping feet etc; very musical.
An even frequency response, nothing exaggerated; natural, and all there.
In other words, just about everything; a fine, highly enjoyable listen.
Now a bit about myself; I love music!
I've been designing and building valve gear for several years and my best efforts compare well with a top UK brand. I like the tonality of valves but I don't like the slow, rolled off sound of some valve amps.
I have also modded a number of units and have a good idea of the sound of various components, and (being an engineer) I can often identify where a unit can be improved, including my own of course! When I have a good unit, I typically build another to try component changes and circuit variations in a controlled way. I have learned a lot - relevant to my mods to the TU.
Getting late now, more tomorrow.
Mike, I am told that the zener protected fet STF20N95K5 which is in an FP insulated package and has a gate threshold of 3v instead of the 2v GT for the 2SK2750 looks OK to work in the TU-882. Think about it. Larger devices sound a lot better from my experience. Tweaker
Mike, Would you say that for C12 and 13 that one would want lowest ESR and highest ripple in the Pano or Nichicon series that have the 22uf and 350v? There is a huge number of non audio grade caps out there, Nichicon PW,Cy,CA, PJ, VZ,TX,PS... Pano,EEu, Rubycon TX. Some like the PW some hate it. Not as much opinion out there to go on as say the Silmics, Nichicon Muse... Which Panosonic you thinking of and why. Tweaker
Here are details of my mods.
First I must say I appreciate what a great job the designer has done to produce a very good sounding unit at such a reasonable price. OK, I've changed it and I know I've improved it but that does not imply any disrespect for the designer - quite the opposite!
I didn't build it as standard for some good reasons. It's not that easy to change components on a pcb (I know, I've done it several times) and there is the danger of damaging the pcb. Also I have confidence in my changes. While it would be nice to hear each change one by one, it would be a real pain to do this taking many months (allowing for running in every time). You'll have to trust me ... I know what I'm doing with valve circuits. I did intend to try varying two things but results are so good I have not yet tried this; when I get a unit this good I don't mess about with it.
The main mod is to change the operating point of the 1st stage valve. As standard, it runs at only 0.5mA which is low for any valve, but especially low for the 5670 which looks to run well at about 5mA. This low a current will struggle to drive the 2nd stage well; 'slew rate' will be limited. Transients, leading edges and dynamics may be compromised. Also, the low current will likely result in a degree of noise.
The voltage of about 180V does not allow much change to the op point, but I increased the current to 1mA by changing R5 to 1800 ohms and R11 to 100kohm (in fact I got 0.9mA but that's OK; the 1800 could be reduced by one value to get closer to 1mA).
The volume control pot was changed to Tocos 10kohm stereo (yes, 10k) in conjunction with a 47kohm series resistor. This is a good sounding pot, and the way I use it improves it further and reduces the gain usefully giving a reasonable volume control setting.
To do this: omit the pcb with volume control. The Tocos pot fits the front panel OK. Identify the earth and wiper terminals on the pot (the 3rd terminal is not used). Fix the 47k resistor to the wiper. Connect a wire from Vol-in G to both channel earth terminals on the pot, and from there to the Vol-out G.
Connect a wire from Vol-in L to the free end of the 47k R; connect a wire from the wiper to Vol-out L. Do similar for R channel.
The result is that the attached source will see a load of about 50kohm, and only the lower half of the pot track will be in circuit. I use this method a lot and with a good pot, I find it is about as good as a stepped attenuator; yes, that good (in fact I prefer it!). As it fixes the issue of the over sensitive volume control, this mod is most highly recommended.
Coupling caps: omit C1, replace it with a link; also remove R1 as it's not needed now.
C5 must be changed to something good. Victor's suggestion of a Teflon cap is good, but I have my favourites (after trying *many*, including Teflon); Mundorf silver/gold/oil, and Jupiter copper/beeswax. Both are superb, very transparent, fine tonality and natural sounding; I use Jupiter here, perhaps it's a touch more natural.
Rectifiers: standard, two ready made bridges are used. A useful improvement can be had by replacing these with schottky diodes. I have not found suitable ready made units so I made up a bridge, on the pcb, using in each case 4 individual diodes. For the heater (the big unit) I suggest 4 off, 2amp 40V schottky diodes (try ebay). For the main circuit (the small one), I use 4 off 1 amp 150V schottkys. In theory 100V ones will do but go a bit over for safety.
Look up on the web how to wire these up.
This mod will give a bit more 'purity' and tonality to the sound and for the small cost involved, it is definitely recommended.
More to follow .....
I've just looked up slew rate calculations, posted by Gordon Rankin on SET asylum.
I made a silly slip in my original calculations; slew rate current is very much OK in the original configuration.
But the mod to increase the current in stage 1 is still recommended.
Having changed the op point of the first valve, some other changes were appropriate as follows.
The circuit uses some global negative feedback via R9 and R7. This will increase the subjective 'drive' of the 1st stage and also tighten the bass; however, global NFB is widely disliked as having an adverse effect on some aspects of sound quality e.g. tonality. I avoid it, but I'm not totally against it, I accept there are cases where it works. Now I have increased the 'drive' capability of stage 1 so less NFB should be needed, but I was reluctant to remove it totally. I have kept R9 the same but reduced R7 to 30 ohms to reduce the amount of feedback. My intention was to try removing R9 to see the result, but results to date are so good that I have left it as it is. I may try this some time but I feel no need to mess with it. So this mod is recommended along with the higher current in the 1st valve.
I looked into C7, 220pF to ground giving some rolloff of high frequencies. I am not sure why it is used but speculate as follows.
Perhaps some noise was generated in the 1st stage and C7 was put in to remove some of this? But I have increased the current in stage 1 which should reduce any such noise, in which case less of this filtering should be needed.
Perhaps C7 is there to remove RFI? However I added a grid stopper to do this, so again a lower value cap should be OK.
Perhaps C7 is there for general filtering, allowing say a less perfect CD player to sound good by removing any 'edge' from hf distortion? Well I have good quality CD players (several ... too many) so this tailoring is not needed.
As a result I reduced C7 to 150pF, raising the hf cutoff by about 20kHz IIRC. Again I intended to remove or adjust this cap but I feel no need to do so; the high treble is all there, bright shimmering but without harshness, so it's pretty good as it is (i.e. excellent!).
R13, the grid resistor of stage 2; 1 mohm as standard. I generally use 500kohm as max, and I've seen this value quoted for the 5670 and considered going for this. But stage 1 is still marginal re current and drive, and a high value of R13 will avoid loading it down. So I compromised and went for 750kohm.
Stage 1 has R3 as a grid stopper but I considered it could be improved (again, in conjuction with higher current in stage 1 making it less critical). The value 3.3kohm is high for this function, 100 ohms is more normal. In case the cct needs more I chose 360 ohms. I use carbon composition as that's normal here.
The fitting was improved to put the resistor body close to the valve base pin. The pcb track from R3 output to the valve base was cut near the valve base (chisel or craft knife does it) and the resistor was soldered to the R3 input position and to the valve base. I'll take photos of my next build.
A 100 ohm grid stopper was added to stage 2, again carbon composition. You have to identify the pcb track, from R13 non earthy side to valve base pin 7; cut the pcb near the valve base, and solder the resistor from that end of R13 to the valve base.
The grid stoppers were a bit tricky, it's not nice messing with PCBs but the mod is recommended as removing rfi gives a bit of purity to the sound; not major, but worth having!
Resistors: all the important resistors (i.e. in the circuit path) were changed to Takman carbon which I find give good results at a very reasonable cost.
Mains lead: it is not easy to add an iec socket to allow different mains leads to be tried. I used high quality Belden mains cable (see Cable Asylum), easily available on ebay; soldered in position much as standard. OK, it's a bit tricky; the hole in the chassis needs enlarging and protection MUST be added to prevent the cable being chafed through, but it's possible and easier than adding an iec. Next time I'll use something even better such as Furutech FP-S022N, highly recommended.
I think that's it.
I believe these mods have had a significant effect, adding some purity and tonality; increasing transparency and detail; increasing transients, speed of leading edges and dynamics. OK, as I didn't build it as standard I can't swear to all this, but my experience with valve gear allows me to have much confidence in what I say and I try to give reasons why and how the mods work. I hope people will trust me and try these changes. Anyone in the UK can come and listen for themselves (I'm Essex).
I'd like to try a separate power supply for stage 1 for several reasons. Mainly, a higher voltage will enable the stage 1 valve to run at much higher current, say 3 to 5 mA. Also, having separate PSUs for each stage has significant benefits; and I'd try a low dcr unreg PSU. But I feel no real need for such a radical change; it's really making very good music indeed.
Post a Followup:
Post a Message!
This post is made possible by the generous support of people like you and our sponsors: