Jumping around a bit, back to...
Parts
Ok the idea is to use top quality parts where possible, but I also want to allow for a bit of parts experimentation if options present themselves. So, looking at the amp boards first.
The Amp boards:
Looking back at this layout,
Ouput transistors.
You can see Q10, Q14, Q11 and Q12 allow options on the layout for TO3 and TO-264 devices. I've not played around madly with output devices in the past. Not sure I'm that inclined to much. Now orientation and pitch of the TO3 footprints is as per Naim boards so there are no compatibility issues slotting straight into my Nap140 (or any other olive naim amp by extension). I think I will use the NA001 parts from my Nap140 initially. However, my current 'big amp' uses the TO264 footprint MJL3281s that Neil mcbride recommended many years ago. They are nice and speedy parts with ft of 30Mhz though a lower-than-some SOA capability. I've also been sitting on a dozen On Semi MJ15003s for a couple of years which Les Worstenhold has recommended in the past.
Basically the immediate options are
NA001 ;TO3 Naim standard part (house badged)
MJ15003 ;TO3 (Avondale recommended- cheap)
BUV20 ;TO3(Avondale recommended- expensive, discontinued, high current capability)
MJL3281 ;TO264 (Neil McBride recommended)
http://www.onsemi.com/pub/Collateral/MJ15003-D.PDF
http://www.semelab.com/pdf/bipolar/BUV20.pdf
http://www.onsemi.com/pub/Collateral/MJL3281A-D.PDF
I think Les Worstenholm described the sound of the NA001s as an allrounder, MJ15003s as lithe and articulate and the BUV20s as 'power and glory'.
The parts I am most interested in trying some time however, are the BUV 20s which are pretty high current and the Semelab BUR50 (350W, 70A, 125V, hfe min 15) or BUR52 (350W, 60A, 250V, hfe min 20) which as well as being high current items have a very nippy ft of 16mhz.
I personally believe that the BUR50s are the rebadged items that Naim use for the Nap500 though can't be sure- Browsing the Semelab power transistor stable, picking up some specifications that Naim have dropped over the years eg Vce ratings, Max Ic, would indicate so anyway.
Also shockingly expensive at £20-£30 a pop
if you can find them but that is really something for another day.
http://www.semelab.com/pdf/bipolar/shortform/SF_BUR50.pdf
How differences pan out sound wise in the grand scheme of things? I don't know, not done the AB comparisons yet. I don't
expect there to be that much between them sonically in truth (probably more a certain 'polish' or 'mood' to the sound than significant leaps in quality) and more a 'tuning' thing.
The technical theory however is that faster transistors (higher ft generally) will less likely cause closed loop phase shifts and lead to more overall amp stability. However, faster parts are more likely to oscillate, so it’s a balancing act. Check out the datasheets if you're interested. More critically, the power rating, SOA curves and hfe determine how the amp functions and what sort of difficult speakers it can drive and what sort of currents it can deliver without melting/going volcanic (along with other factors such as heatsinks etc). More the nuts and bolts stuff.
driver transistors
I'm going to use MJE15032/3. These are pretty ubiquitous in a lot of amp designs. I'm not inclined to care to experiment. They have good gain, linearity, voltage rating. No problems. A bit of heatsinking is all that’s required.
http://www.onsemi.com/pub/Collateral/MJE15032-D.PDF
Small signal transistors
Inputs BC546c (same as the preamp BC550s but higher voltage rating).
VAS and CCS cascodes: 2sc2705/2sa1145 There are good reasons for using these (low Cob, medium ft- critical parameters for a good VAS transistor or cascode device. also good to 150V) and they do sound better in practice.
CCS BC546c. Not particularly critical IMO. Could play with this. Reason being this part is well protected by the cascoding transistor.
VBE mulitiplier (Q6): Hmm, probably a ztx part or similar. Reasonably high hfe is desirable. Not really bothered. The decoupling cap (C9)will be most important in maintaining a constant voltage across the circuit IMO and keeping high frequency noise at bay. I will go for a film cap in the 10-47uf range. Randy Sloane says 10uf upwards is good for this element of a circuit which is good enough for me. Definitely a 47uf Evox MMK or similar works wonderfully though I will dabble in an upgrade- I've certainly planned on it
.
Capacitors
Over time my view on caps has settled somewhat and I now have 2 golden coloured rules-of-thumb: a)they must be film if at all humanly possible b)must be non inductive construction if at all possible.
Heirachy of film caps is roughly Teflon (best), polystyrene, polypropylene, PPS and polyester.
Metail foil is better than metallised, but I don't worry too much about that these days. The differences are often a matter of taste in the grand scheme of things. Usually. Generally. As a rule, sort of. The MAIN thing, which does matter, is that they aren't electrolytics (inc tants).
My other thing is I try not to touch anything apart from Evox or Wima box capacitors if possible. They are reasonable value and most importantly great sounding in general; and my guess is the reason being they are true non-inductive designs and inductance can play havoc with sound. Basically, although they are wound film foil construction, each butt of the wound roll is effectively 'smothered in solder' (actually vacuum metal vapour deposition IIRC) so shorting out any inductance- Current flows across the wound foil not around it- between the electrodes, ergo no inductance (well as much as a short piece of wire only). Have a look at the Wima website some time or this explains it well:
http://www.illinoiscapacitor.com/uploads/papers_application/51D69991C5554E6B95CB9BE30B82B0BB.pdf
A lot of audio boutique caps are actually just plain wound (the electrodes attach to the foils at one point only) so although they may use fancy materials, they have whopping inductances which in my experience leads to often 'weird sounding caps'. Basically inductance is completely the wrong sort parasitic to have in a cap as it 'fights' ac when we generally want caps to respond fast to ac changes; In coupling it will distort a signal voltage through it, and will also make a cap slower in decoupling applications.
Even the Mundorfs and Audyn cap plus use an inductance cancelling winding topology where opposite inductances cancel out. In theory. I tried the Audyn cap plus years ago and they sounded truly bizarre both in coupling and decoupling. Just wrong. Ron who tried some at the same time agreed, so I think the theory and practice don’t quite marry; my opinion based on limited experience only.
In general however, caps are the most important
type of part (eg inductors, transistors, resistors etc) to worry about in terms of material construction and quality IMO, but follow the above guides and you can't go far wrong. I won't say it’s a truism as that is a bit dogmatic; I'm sure there are exceptions, but its expensive testing, ceased to hold much interest as an area of experimentation a while back, and its incidental to the focus of this project. I have a pretty good idea of capacitor bang for buck now and the 2 above guides get 95% of any possible sound gains from caps IME.
Its really the difference between film and electrolytics that is critical. Walt Jung and Richard Marsh did some seminal articles on this about 30 years ago and it still seems to be truer with time.
http://www.reliablecapacitors.com/pickcap.htm
Now if you read, Walt does give the polyesters a bit of a kicking. However, they can be useful, even essential sometimes. In signal coupling the sound is actually quite forgiving- not the last word in detail but well balanced. Personally I don’t use em, but they are quite popular with several pink fishers. I use polyesters as lot in decoupling though. Why? Because actually I have had polypropylenes cause low level oscillation several times which, while doesn't initially do much physical harm, sounds terrible. Swapped with a polyester, problem gone. You have to be a bit careful where you use polyprops, so I am. Sometimes it can be like sticking a rocket booster to a mini; you are going to get a mini adventure. And its going to end at the top of a tree on fire. Also sometimes size does play a part- You can get roughly 10x as much capacitance in a polyester as a polyprop of the same size. Also the difference in sound is not that great. Usually. As always there are no rules just optimum solutions.
So for the Redbox amp I'm going for polystyrenes, polypropylenes and polyesters pretty well throughout unless I really can't get away from using an electrolytic. This will be only for space (need v. large uf in v. small space) or market availability (i.e isn't manufactured) reasons, not cost. If its electrolytic, then most likely rubycon ZLs.- top quality, very low impedance with voltage ratings up to the 100V mark. Sorry, the ZAs only go to 35V and forget Oscons which also fry at 25 or 30V i think is the max. Panasonic also do a lovely line in electroytics; FMs i think.
The polystyrenes I'm going to be using also have a non-inductive construction from my inspection though they don't seem to be promoted as such. Can't remember the exact type and manufacturer off the top of my head (maybe the LCR FSCEX???- sort it out at the BOM stage.
www.evoxrifa.com/cap_catalog/generalp/mmk.pdf
www.evoxrifa.com/cap_catalog/pulsecap/phe426.pdf
www.rubycon.co.jp/en/catalog/e_pdfs/aluminum/e_zl.pdf
input coupling- I'm planning on polyprops, probably wima mkp4 or evox phe426s. They're sweet, detailed and well balanced generally in this role and 1uf is absolutely fine. I may eventually swap to polyesters or SMRs to do a bit of tuning to taste so I've allowed varied pitches to fit most Evox and Wima components. Board space is tight so they will have to be metallised not film and foil. One day I may go mad, and try a Teflon Vcap just because I've had one in mind for so long. However its big dollars, lots of space required and I doubt there's much more to be gained. Strictly out of curiosity only. Also Vcaps wouldn't fit in my 140, so this would only be for my big amp and it would be flying leads lash up. In context I've played with polystyrene film/foils (RTXs) on the preamp and they're only a hair better than a decent metallised polyprop in my book so the Vcap is really just an itch that has been around a long while.
Polyprop for the feedback cap too. If you know what to look for, it can be better than a polyester though there are some caveats, both in sound and in practical execution; What I'm planning may simply not work and sound horrible if it doesn't, in which case back to Evox MMKs for the NAP. More shortly.
Power resistors. (e.g R29,R30)
Naim use some ghastly inductive wirewound horrors for this. Rated at 3W I think? Maybe 5 as the wirewounds are quite tolerant of heat. I think they've used various manufacturers in the past, including Meggits.
Im going to use Caddock MP930s- TO220, non-inductive, 50ppm temp co (edit: oops, its 200ppm actually), metal film resistive elements. Very nice if a tad pricey.
www.caddock.com/Online_catalog/Mrktg_Lit/MP9000_Series.pdf
Now the TO220s are good to 2 watts unheatsinked. I have been running my protoype amp boards for a couple of years now with MP930s and no problems- they don't even get warm. If your speakers are current hungry a small heatsink might be in order. However do the maths, for .22R current sharing resistors they can take 10A dc at max power rating. Your amp can't so its not a worry. There's space to fit a small heatsink if worried or using TO264 output transistors.
Otherwise resistors will be the good old Vishay Dale RN60s, 0.25W mil spec rated (actually 0.5W industrial) except around the drivers where I'm shoving in RN65s, 0.5W mil spec (1 W industrial) as the dissipation is a bit larger here.
www.vishay.com/docs/31027/31027.pdf
I did consider going for RN65s all through but its not really warranted and I needed the space. One thing I might do is actually move to PRPs which are the new Vishays and becoming pretty common in high end audio, reason being the Vishays sometimes come in industrial i.e non copper leads, which are not so great sounding. Not always easy to know when your buying. I will also consider some vishay dale PTFs which are 0.1% tolerance, 10ppm temp co parts (yum) or caddock MKVs for the feedback loop (specs not exceptional but very positive sound reviews) which I judge most sensistive position. Depends on availability. Something top quality anyway. Tantalums? I'll think about it. A few select good resistors are worthwhile but when some of Vishays zero tempco top foil offerings can be £50-100/pop you have to draw the line. Boutique resistors are relatively bad value in bang-for-buck stakes as a general rule- difference between super costly and cheap is small sound wise.
The zoble
Polypropylene cap (0.22uf), Caddock MP930 resistor, home made inductor 20 turns of 1.25mm insulated copper on an 8mm diameter dowel IIRC (I have this worked out somewhere but its something like that- comes to 4uH IIRC).
So all parts are in the top 95-98% of what it is possible to buy sound quality wise IMO, but paying only the upper end of the economy price range. Those are figures I like a lot and actually, I'm not sure you can do much better whatever money you spend on parts. The most critical single component is, as ever the feedback cap, and that is covered by the best on the market available IMO. There is a little trick I'm trying to pull off with this which I hope works though; If it doesn't, I'm screwed as far as the Nap140 goes. But if it goes well that alone will make this amp exceptional and why I don't believe a Nap500 will stand a chance soundwise (touch wood, not tempting fate, proof in the eating.) when the regulation and a few other tweaks are thrown into the mix.
OK, regulator boards parts breakdown to come shortly.