Craig Sawyers

Careful of the feedback paths, particularly since it goes over two boards with I would guess wires to connect. The T2 is prone to oscillation which can be quite tricky to sort out. The thing you have going for you in this regard is the compactness with SM. Why go to the trouble of mirroring? If you have a layout that works, you might well run into problems with a mirrored version with different tracking (since active devices themselves do not mirror).

Well the thermal resistance seems to be pretty constant from lowish torque (at least from the graphs in the apps note). The importance is that it stays put as temperature changes and polymer materials (like the AAVID insulation washer) creep with time under pressure and temperature.

Oh oh oh - I cannot cope with such a headphone overload. Phzzt phzzt grkkk

Just do a search on Google. Lots of suppliers Stateside. And the OnSemi note is from the US and says Belleville. And no JoaMat - I do not think that any compromise is advisable as being "good enough" for a T2. Any break from best practice is not a good idea given the voltages and power dissipation.

Have a look at the OnSemi note I linked to, pages 7 and 8, for why split ring washers are not the best idea.

Ha ha! Such a serious movie it absolutely needed that treatment. And apart from dollars and Holyoke all the references are just as applicable to the UK.

You are right to use 4/40 UNC rather than metric, because the interface pressure depends on the thread pitch (it acts like a lever in this). 4/40 has a thread pitch of 0.635mm and M3 is 0.5mm. The onsemi apps note torques all apply to UNC threads.

I use http://www.onsemi.com/pub_link/Collateral/AN1040-D.PDF as my reference for power device mounting. Table 1 shows a test torque of 8in-lb = 0.9Nm. But figure 6 suggests 6lb-ft = 0.67Nm although the thermal resistance is rather flat from 3in-lb = 0.34Nm. However, if you tighten to 0.34Nm it seems just too slack. So I recompensate a tad and up to 1.1Nm for TO220 (9.7lb-in). I haven't had any problems with my T2 since I built it - but there is nothing absolute about the torque I chose. If you stick to the 0.5-0.7Nm it will be absolutely fine. But follow the apps note's recommendation and use a Belville washer (a dished spring washer) to maintain tightness as temperature changes. If you use plastic screws of any kind (I bought some PEEK ones from a list member in Japan) you can't get close to those torques - the threads strip out. Nylon would be a disaster, polycarbonate strips as does PEEK. As you can see I tried hard to not use something conducting! But the long AAVID shoulder washers, steel hardware and spring washer works fine, and you can get the torque with confidence. The TO126 transistors definitely need a lower torque - the surface area is smaller, and the important thing is contact pressure. About 75% of the TO220 torque is about right. Springs, either this http://www.pts-uk.com/Products/Washers_Disc_Springs or this http://www.pts-uk.com/Products/Washers_Locking_Washer_S_Type or this http://www.pts-uk.com/Products/Washers_Schnorr_Locking_Washer_S_Type would do the job. I stocked up with these http://www.bellevillesprings.com/belleville-washers.html . But any of those will do the job.

Aha - I thought you had struck lucky! The one I'd like to get my paws on is the Vendetta phono stage.....hey, maybe it is cloneable.....

Oh for fuck's sake. Prince??????? How much more of this - 2016 is a shit storm of untimely exits.

The JC-2 is a nice box for sure to put something in. Has a nice two gang Penny & Giles pot in there too.

Have you ever tried to export to China? I have. They refuse to accept CE and UL approval, and insist on putting a product through their own process (protocol 60) - the company is 9 months in and counting, driven by a Chinese speaking employee. Oh - and you have to supply full engineering drawings, full BOM and documentation pack translated into Chinese, and a product for "testing" - no ifs and buts. The process takes an eternity, and you have to down your pants and ask them to kick you right in the ass in order to sell into China. So do I believe that there is a policy of obstruction to exporting to China? Yes, absolutely. Where does that policy originate? The Chinese government in whatever shape you want define it. That is leaving aside that they have had a product and manufacturing data for 9 months......

I had a pair of 57's on stands - and my wife knocked one over. It was saved from damage by the speaker cable, but the super expensive silver connectors got trashed. But those 63's were responsible for my obsession with high end audio - visited a room at the Heathrow show in the early days of Meridian when they were importers of Mark Levinson. Meridian CD player into dual mono ML pre, class A power into a pair of 63's on stands. Brothers in Arms was playing - and it sounded so completely different to what I was used to hearing. I left that room determined to aspire to that sort of sound quality. Little did I know where that go....

^But delicious.

RIP the amazing Victoria Wood, cancer age 62. CBE, dual BAFTA winner, comedian, musician, scriptwriter, actor, director. http://www.bbc.co.uk/news/entertainment-arts-36094827 Out of a clear blue sky. Bugger.

IIRC my original T2 clone build gets to about that.

In fairness to the 2 transistor circuit, the original 1965 circuit used germanium transistors, Mullard OC44. Price in 1965? 8s3d from the Mullard price list, or in today's money 41p. Correct that for price inflation and you get £7.61 or $11, times 2 or $22, times 2 for stereo - $44. - for which you can now buy 4 AD797 or 16 LM4562 or 68 NE5532. The 1967 Quad 33 used the same circuit, but moved across to silicon BC109 - which were equally expensive back then. The 12 transistors in the 33 would have been the determining factor in the unit's cost. You can make that 1965/7 performance much better by adding a transistor (to up the open loop gain, and give enough drive current for the RIAA), something they probably knew - but that would have added a lot to the semiconductor cost. Which is all horribly irrelevant, since it is not going to be used!

The mag PU design was in a Mullard apps note from the mid 1960's, published by Dinsdale in Wireless world Jan 1965. You're right to leave it alone - the performance was marginal at best. In fairness to Stax they are running it from 30V, which improves linearity as compared with the usual 12V, and the emitter bypass on the second transistor of 100uF kills of capacitor nonlinearity - so they thought about squeezing every ounce of performance out of mediocrity. It's noise performance is either dominated by the 2.2k resistor in series with the input, or the poor noise performance of the first 2SC458 transistor (16dB noise figure at 30Hz). The red line is the correct treatment!

I take my hat off - that is a thing of beauty.

^This

From a purely thermal management point of view - absolutely, 100%. But from the product safety perspective to CE or UL, the general rule is 25C above ambient (to max ambient of 40C and external part temperature of 65C) for touchable anodised casework - and an external heatsink counts as casework. So most manufacturers of power hungry audio gear now put the heatsinks inboard. Like Krell and Emotiva - but not interestingly Mark Levinson who still put their heatsinks on the outside and don't bother to certify for external temperatures.

I'd do it somewhat differently - release final Gerbers, BOM etc but only to those on the list who are (a) trustworthy (by some criterion like long term, regular contributors, no newbies etc (b) agree formally not to release to a third party. Something like that might be workable and prevent blatant commercial design duplication. Probably achieve nothing other than satisfaction, but why not send a letter to the Russians who are selling the BHSE. Also I know Brian Sowter - I am sure he would be devastated to know that he is supplying his attenuator transformers to someone ripping off the hard work of someone else.

If it isn't the Chinese plagiarizing products it is the Russians. In a business meeting a year ago or so, the visiting guy had been shown the inside of a unit in China where they had a team of 200 whose job was to decompile embedded code, FPGA code, ROM contents etc of Western products. We don't stand a chance in hell against that level of industrial scale ripping off.

I sure hope they are paying Kevin a royalty on sales. The Sowter (UK, Ipswich) attenuation transformers used are really good though, and very well thought of.