Now for something different...

[justin]

Wow, $3.50 is insane. I'd sell 100 in original packaging for that

really? YGPM then!

[deepak]

Wow, $3.50 is insane. I'd sell 100 in original packaging for that

It is, I paid $1.82 each. But it's cheaper than the 2sa1968

edit: just looked at Craig Sawyers/KG T2 BOM which I think was way early in the T2 builds, $1.57 each back then

[ujamerstand]

You can still get them for about that price.

[luvdunhill]

Wow, $3.50 is insane. I'd sell 100 in original packaging for that

I see what you did there Justin... Smart guy you are

[spritzer]

The 2SA1831 is 1/3 the price and should also work. It's not isolated but no heatsinks needed here.

Since this is the "something different" thread I've stumbled over an even simpler and crappier way of doing the Darkstar and it's all on one chip. Stay tuned for what will probably be the worst amp I'll ever build...

[luvdunhill]

What about STN9260 or 2SA1807, both are at Mouser.

[kevin gilmore]

surface mount. But sooner or later we are going to have to make adapter boards

to use these parts.

[Beefy]

surface mount.

Is that a problem? Most people will be going balls-deep for a build like this, so surface mount shouldn't be too much in additional difficulty.

I would certainly prefer surface mount devices from Mouser than the increasing risk of fake parts from obscure sellers.

[kevin gilmore]

If you ever have to remove a smt device for repairs you risk damaging

The board. On the preamp board I have turned the opa1632 into a dip

On a smt adapter because some of the chips don't meet dc specs.

Nothing wrong with smt to to220 adapters these are the future for stuff like this

An all smt stax amp will never happen because the resistors can't handle the voltages.

[luvdunhill]

Just make a TH pattern for the sot223 I'd bet it would work, just stand the part up vertical. The Vishay SMD SMM resistor line looks like what Stax uses. Anyways to each his own, seems like reasonable decisions...

[jezz]

I designed these for a similar need a while ago but ended up not using them (SOT-223 and DPAK PNP transistors are somewhat layout-compatible). Probably wouldn't cost much to make in great quantity--also probably has some layout defects, because at the time I was thinking "oh, what if I slapped a heatsink on these?"

[justin]

Just make a TH pattern for the sot223 I'd bet it would work, just stand the part up vertical. The Vishay SMD SMM resistor line looks like what Stax uses. Anyways to each his own, seems like reasonable decisions...

i'd like to know if there's any advantage to using the Vishay SMM over regular 1206 or 1210 thin-film resistors. The round ones are more annoying to solder.

[justin]

i'd like to know if there's any advantage to using the Vishay SMM over regular 1206 or 1210 thin-film resistors. The round ones are more annoying to solder.

i went with the Vishay SMM0204. way better than thick film chip resistors, but price isnt too crazy.

Thick Film 1206 1% - around $12/5000pcs

Vishay SMM0204 1% thin film (the round MELF package) - I managed to find these for about $35-50/3000pcs

Vishay TNPW 0.1% thin film - about $900/reel...ehh no thanks, i need around 15 values...

[kevin gilmore]

i figured i would toss these two out for your entertainment

http://gilmore.chem....edu/electra.pdf

http://gilmore.chem....du/electra2.pdf

obviously you have to add adjustable bias for the output tubes.

Edited August 27, 2012 by kevin gilmore

[livewire]

Will be interesting to see what lurks under the hood...

[spritzer]

It should be noted that we are well aware of the Electra only having one cap in the signal path but this setup yields better distortion numbers. Personally I'll take an extra cap and lower distortion.

[Remolon]

I am working on an amplifier with 2A3 triodes for the output. This is the schematic:

V1 12AU7; V2, V3 12AX7; V4, V5 2A3

The values used for the simulation are:

Rload 10K

R1 4K75

Rk1 1K74

Rg1, Rg2, Rg3, Rb4, Rb5 274K

Rp1 475R

Rk2, Rk3 100R

Rp2, Rp3 100K

Rg4, Rg5 1K

Rk4, Rk5, R45 10R

Rp4, Rp5 2K49

C2, C3, Cp2, Cp3 1uF

Cs4, Cs5 0.1uF

B+ 300VDC and Bias -50VDC

It is my first aproach to the Stax system, and any comment is welcome.

[kevin gilmore]

2.5K for the plate resistors, what is the B++

No way that is going to work.

The B++ has to be a minimum of 600v to be

able to get enough linear voltage swing out of

the output section. 800v or 1kv much better.

I think that the 2a3 is not going to like 600v.

But you will have to try it. A plate resistor

that is a minimum of 25k is reasonable.

211's or 845's would be much more fun.

[nattonrice]

A 211 stat amp would be a complete pita to sit in front of from a glare point of view

[Victor Chew]

What about 300B Kevin? The 2A3 outputs about 3w and the 300B at 9w. Would this make it a better candidate?

211 are going to costs if the number of tubes are the same as in the Megatron. And yes, I agree with nattonrice about the glare and also that it will give us a "free suntan" each time it is turned on.

Edited September 3, 2012 by Victor Chew

[spritzer]

The 300B is very voltage limited and can't handle anything like the 211/845 or even a triode strapped EL34. What we want here is voltage swing which is for instance why the 6S4A is such a perfect tube for the job, far better than most of the DHT's. No use anywhere else though which is good for us.

If you want to use the 300B/2A3 then you have to use an output transformer for some voltage "gain". B+ of 300V will not get you a whole lot so at least double that but that means the 2a3 won't work. Something like the EML20B with a 600V B+ and anode resistors would work but it's not cheap.

[kevin gilmore]

The lowest plate resistance of all electrostatic amps out there is the hev90 which has a plate

resistor (actually 3 in series) with a total resistance of 18.3K and runs on +550 volts.

This is the maximum rating for an ecl86.

Most of the stax amps have plate resistors in the range of 47k to 60k.

birgir pointed me to this

http://www.bartola.co.uk/valves/category/valves/dht-valves/

6P21S would handle at least 600 volts.

The emission lab tubes would fit the bill, but with good sockets are over $250 each and you need 4

[spritzer]

66K is the highest I've seen in a Stax amp (T1 and 006t) but that HEV90 implementation made no sense what so ever.

The 6P21S looks interesting but I'd really like to try something with my Bendix 7403's. IDHT's but they are some serious tubes....

[FrankCooter]

Interesting project Remolon.

Unfortunately I'll have to second all Kevin said about the resistively loaded 2A3 outputs. There are also factors of low gain, high plate current for decent linearity, and high Miller Capacitance at the input. Although some people do it, the 12AX7,with its high rp and low current cabability is a poor choice to drive a DHT. Cascaded 6SN7's would be the traditional way to go. Better yet would be a single high mu/high current pentode like a 7788 or D3A.

The only reasonably sized traditional class A DHT that I'm aware of that can handle 600V on the plate with an appropriate current and gain level for a project like this is the 801A. It's a great tube and would be my first choice for a project like yours.Unfortunately it's not in current production and an NOS quad, when available, will probably set you back $400.00.

If you're stuck with the 2A3 outputs and 300V B+, you can abandon the plate resistors in favor of chokes or output transformers.There's a lot to be said about not burning up half your B+ in waste heat. Any pp speaker transformer will work as long as the primary is suitable for your output tubes. The easiest thing to do would be to build a classical pp design like the Williamson and take your headphone output off the output transformer primaries via coupling capacitors.

The 6P21S does indeed look interesting. I'm going to hop over to Ebay and pick some up.

Don't get discouraged, you're not alone, For me at least, rarely does something work right the first time (and sometimes not the 10th).

[Remolon]

When you say that 2K5 for the plate resistor is too low, is because of the requirements of the amplifier or the headphones?. Even, I am considering to use 1K5 because the RCA 2A3 datasheet specifies for a push pull amplifier 3K plate to plate, and I am curious if the valve makes a distinction between the 3K image trough the output transformer of the loudspeakers impedance and the 1K5 resistor between each plate and ground. In normal operation, each 2K5 resistor would waste less than 2wats.

With B++= B+=300VDC and Rp4= Rp5= 1K5 Spice calculates a current of 38.6mA for each valve, and a output of Vpp= 266v with a source of 2v(CD player), and, if is true, I hope that it is enough to move the SR-007 at an useful level. Facts will say.

I have already used the input stage to modify, some time ago, an old Dynaco ST70 that works properly.