Small amplifier for Stax

[JimL]

9 hours ago, wink said:

But, but, - I don't want to.....   

"Go do something you hate.  Being miserable builds character!"

- Calvin and Hobbes 

[JimL]

9 hours ago, joehpj said:

Interesting thing is, Stax uses CCS loads in their SS amps and resistors in tube amps. I could imagine 007t would be a nice amp with CCS.

Also, the KGST is similar to an SR007T but with a much better output tube (6S4A vs two halves of 6CG7) and CCS load, plus a regulated power supply for that last extra bit of sophistication and oomph.

Edited March 25, 2016 by JimL

[wink]

11 hours ago, kevin gilmore said:

Let's not forget the total pile of crap known as the Eddie current Electra.

center tapped output inductors that saturate due to the DC problems at the input stage.

 

But, that wuz their big selling point.......    Exclusivity.......  Who else would build something like that...?     Cavalli, perhaps..??  

 

His amps are getting rave reviews at the SoCal meet thread over on the dark side,,,,,,,   

9 hours ago, JimL said:

"Go do something you hate.  Being miserable builds character!"

- Calvin and Hobbes 

“That propaganda is good which leads to success, and that is bad which fails to achieve the desired result. It is not propaganda’s task to be intelligent, its task is to lead to success.”
― Joseph Goebbels

 

If it weren't for electricity, we'd all be watching television by candlelight.

George Gobel
 

Edited March 25, 2016 by wink
Four matting

[Pirx]

JimL, Spritzer - I can't agree with you about chokes and electrostatic headphones.

I build an amp with 6C4C as a power stage with big 100 H chokes. It worked very well.

I couldn't measure any resonance in audio spectrum.

I didn't finish it only because it was too big and heavy.

It was build only as a prototype, so I can't prove now anything else.

[spritzer]

...and this proves nothing at all. 

[Pirx]

2 hours ago, spritzer said:

...and this proves nothing at all. 

You don't have to be so cruel

It is possible to build something good even if you don't think so.

Did you build an amplifier for electrostatic headphones with power stage with choke?

[kevin gilmore]

Nope, chokes are too big.

current regulated supply far superior.

[spritzer]

Chokes in the PSU vs. proper regulation?  Well then chokes are just stupid. 

Reminds me of an email I was sent where Lampazitor was defending his crappy power supplies because people liked them at some meet.  No actual testing, just "we liked the sound".  Goes to show what shitty crap that stuff is. 

[JimL]

The heart of the matter is this, it's not about what sounds "good" or "OK".  It's about how to get the best sound.  An electrostatic headphone is inherently a high impedance device.  The best resistor or choke is always going to limit how good the sound can be, because it is always going to suck the majority of current from the output device to itself, diverting it from the headphone, because it is lower impedance than the headphone.  That is simple Ohm's law.  That means the output device is always going to more distorted than necessary.  A really good current source is MUCH higher impedance than a electrostatic headphone.  That means most of the current from the output device actually goes to drive the headphone rather than being wasted generating heat in the resistor or choke load.  Better efficiency, lower distortion, better sound.  Simple.

 

So, Pirx, build an amp with choke output and the same amp with a good MOSFET cascode constant current source, listen to both and then get back to us.

Edited March 26, 2016 by JimL

[PretentiousFood]

17 hours ago, kevin gilmore said:

Nope, chokes are too big.

current regulated supply far superior.

These aren't mutually exclusive. I've used a choke input power supply into a shunt regulator for a dynamic HP amp to good effect. Excessive, but hopefully not inferior. Jim, aren't you doing the same for your SR-X? With a properly sized choke, you can get away with smaller power transformer and b+ caps. 

[Pirx]

JimL, do you suggest that electrostatic headphones should be driven from high impedance output?
 Can  you explain why?

[JimL]

Nope.  The active device, we'll use tube for simplicity, should be relatively low output, but because the headphones are so high in impedance, even an output of 5-10k is adequate - in fact the Stax headphones all have a 5.1k resistor in line between the output device and the headphone to protect the latter.  Being high in impedance, the headphone needs large voltages but relatively low currents.  Hence, a 6SN7GTA with its DC voltage limit of 450volt and running 7-8 mA makes a good output device - IF all its output current goes to driving the headphones.  But, if you waste a good proportion of that current in the amp load, it isn't nearly as good.

 

Remember that the output tube drives the headphone in parallel with its load, whether that be a resistor, a choke, or a constant current load.  This means that the music signal current is distributed between the headphone and the load depending on their relative impedances.  

 

On the other hand, the output impedance is the impedance of the tube and the load in parallel.  With a tube plate resistance of, say, 8k, even a perfect constant current load with an infinite impedance results in a combined output impedance of 8k.  With a resistor load of, say 50k, the output impedance is about 6.9k, a negligible difference. With a choke load, the output impedance varies with frequency, being higher as the frequency rises, but mostly dominated by the plate resistance.  So the output impedance is pretty similar regardless.

 

In order for the tube to have a reasonable standing current, the resistor has to be around 50 kilohms unless you have a very high voltage power supply.  Since the impedance of the headphone is much higher than 50 kilohms, this means that most of the music signal current is wasted in the resistor.  With a choke of 100H, the impedance is only 12.5 kilohms at 20 Hz, and only gets to 50k and above at 80 Hz and above, and doesn't exceed the impedance of the headphones until you get above 1 kHz.  With music signals, about 1/2 the power is in the range below 300-400 Hz.  Again, this means that the majority of the music signal current is used to drive the choke, and the leftovers go to the headphone.  This is not only inefficient, it causes more distortion because the tube has to supply much more current.  With a really good current load of very high impedance, there is practically NO current going to drive the current source, so nearly all the current goes to drive the headphones.  Do a search for the thread "Output Stage Current Requirements for Electrostatic Headphone Amps" for my full technical discussion.

 

The bottom line is, the output load is the plate resistor/choke/constant current load in parallel with the headphone.  A perfect constant current source needs no current drive so it is invisible to the output tube, which sees only the headphone load.  This is the best possible situation for the output tube.  

 

On the other hand, the output impedance is the impedance of the plate resistor/choke/constant current load in parallel with the plate resistance of the tube.  Since the tube plate resistance is much lower than that of the plate resistor, etc., the output resistance is primarily the plate resistance.  So a constant current load has no significant disadvantage in terms of output impedance compared to the other two options.  

 

Finally, a constant current load does require a heat sink, so it is  more expensive than a resistor, but still significantly less expensive than a good plate choke.  So, the constant current load is by far the most efficient in directing the signal current to the headphone, resulting in the least distortion and most output, has no significant disadvantages, and reasonable expense.

Edited March 26, 2016 by JimL

[JimL]

As an example of the difference in sound between a resistor load and a constant current load here is a post by Congo5 on the Output Stage Current Requirements thread from April 24, 2015:

 

After building the SRX Plus I decided to try the CCS on an Egmont.

Copying Kevin's layout I drew up this:

 

Having two Egmont's I can A/B them to compare.

It is a huge improvement, more volume and much less distortion.

Much more of a usable amp now. Noticeably cleaner bass. 

like Night and Day.

[Pirx]

JimL, thank you for your detailed explanations. I promise to search and read the thread you proposed.

I know how CCS works, and I agree with you in fundamental rules you described.

I build others amplifiers with CCS or resistive and I don't  want to build another the same topology's amp.

When you decide to build tube amplifiers there is a space for many imperfections.

In your eyes my decisions of using chokes is technically wrong. And you are right.

For me it is just another way.

Is Megatron with it's CS on EL34 technically better than KGSS? I bet it is not, and that proves nothing, as Spritzer would say

[dsavitsk]

17 hours ago, JimL said:

The heart of the matter is this, it's not about what sounds "good" or "OK".  It's about how to get the best sound.

Even with all your justifications, which I'll stipulate are true (a CCS will reduce resonances, divert more output to the load, and lower distortion), there are good reasons to use a choke load. Chokes store energy and can swing above the power supply. As a consequence, a CCS loaded tube needs a power supply more than double the voltage of one with a choke load. Besides the huge reduction in efficiency and the need to dissipate the heat, this also requires the use of a very different class of high voltage components and very different construction techniques. Building a 400V supply is trivial. Building an 800V one is not.

Additionally, while the choke load will have more distortion, almost all of it is at low frequencies where distortion is considerably less audible. Your 200H choke may only provide a 50K load at 40Hz, but at 1000Hz it is well above 1M2 Ohm which is on par with a CCS.

For reasons of safety, personal comfort in working with high voltages, or possibly even cost, the trade off of slightly increased low frequency distortion chokes may be justified in some cases. You don't have to use them, but it does not mean that they are a blanket bad idea.

And Lundahl makes a 270H choke, so 70K at 40Hz.

[JimL]

6 hours ago, PretentiousFood said:

These aren't mutually exclusive. I've used a choke input power supply into a shunt regulator for a dynamic HP amp to good effect. Excessive, but hopefully not inferior. Jim, aren't you doing the same for your SR-X? With a properly sized choke, you can get away with smaller power transformer and b+ caps. 

I AM using that topology for my SRX Plus.  But it is rather bulky and heavy.  I used surplus transformer and chokes so those tend to be bigger than absolutely needed.

[JimL]

2 hours ago, dsavitsk said:

For reasons of safety, personal comfort in working with high voltages, or possibly even cost, the trade off of slightly increased low frequency distortion chokes may be justified in some cases. You don't have to use them, but it does not mean that they are a blanket bad idea.

And Lundahl makes a 270H choke, so 70K at 40Hz.

I think you make some cogent arguments, and you are correct about the Lundahl choke - but it also costs over $90.  Electrostatic headphones are an extreme example of high impedance which requires the highest available impedance anode chokes for outputs.  These are more difficult to design and more expensive.  And the fact is that with music signal, which is power weighted toward bass and low mid-range frequencies, there is going to be significantly more current draw to drive the chokes than a constant current source, likely by 2-3 fold.  While choke distortion may be limited to the bass frequencies, one could argue whether the large voltage swings will increase choke distortion as a whole, and the extra current draw increases tube distortion as a whole.  

 

It is true that current sources are wasteful from a static point of view (although no more so than a resistor) compared to a choke, but they are more efficient from the point of view of providing power to the transducer because very little signal power is wasted in the current source, compared to what is wasted in a choke or power resistor.  And while choke outputs can use a lower voltage supply, most stat amps and power supplies are bipolar - constructing two 400 volt power supplies is no harder than constructing one, 

 

 While choke outputs are not necessarily a bad idea, I think that current source outputs are a better idea for stat phones.  Perhaps the trade-offs of choke outputs may make more sense for dynamic drivers which have a much lower impedance, but that's a wild-ass guess as I haven't really done any analysis.

Edited March 26, 2016 by JimL

[dsavitsk]

4 hours ago, JimL said:

While choke outputs are not necessarily a bad idea, I think that current source outputs are a better idea for stat phones.

And that's really the point. This is your opinion. It happens to be one I agree with -- I am, and have been, a proponent of CCSes in audio circuits for years. I think they are almost always the right decision -- but that does not mean that other solutions do not have merit, and that there aren't good reasons for someone to solve a problem in a way I might not. The world is not black and white. The above discussion seemed to suggest it might be.

[JimL]

Ah, OK, got it.  I agree with you.  Engineering decisions are rarely good vs evil, which is why I usually try to present advantages and drawbacks in a discussion.

Edited March 27, 2016 by JimL

[Pirx]

JimL - I hope now that you won't call me "heretic" when I publish my amplifier with tubes and chokes

I read your thread about using current sources and  driving electrostatic headphones. Very interesting, thank you.

[JimL]

Honestly, it doesn't matter to me what you build, that's what DIY is about.