Jan 29, 2019 at 11:00 AM
The CCS amplifikation is quite high and the bandwith also. I have not calculated on the poles so I don´t know the phase margin. When i got time I want to do a spice simulation. Anyway the decoupling worked well.

Reagarding the 150H choke. A choke supply will generate an opposite current and voltage, that in turn will feed the headphone. Giving 3dB higher voltage swing or 6dB more power than just a CCS. The voltage out can actually swing higher and lower than the power supply.... Thats way the headroom gets bigger. And as I said the secon adwantage is that at high frequencies where the impedance of the CCS is increasing, thats where the choke comes in... at 10kHz the 150 H choke will have an impedance of 10Meg ohm.

Um, no. A choke by itself can have the voltage swing higher and lower than the power supply voltage, but if you put a choke in series with a CCS, the CCS forces the choke to have a constant current through it, so there is absolutely no voltage variation across the choke whatsoever. The whole idea of the CCS is that it forces the current through it to be constant regardless of the voltage across it. When you use a CCS as the load for an output device, that means that any voltage variation pushes all of the current variation through the transducer.

Second, the CCS impedance does decrease somewhat at the highest frequencies, but since it is likely over 100 megohm anyway, the extra 10 megohm doesn't make that much difference (and remember that is assuming a perfect choke, real chokes generally have leakage capacitances that will drop that value significantly).

Choke loading can be useful on its own, and actually I use chokes in my power supplies all the time. It's just not particularly helpful in conjunction with a CCS - the CCS just turns it into a very expensive resistor. Another way to think of a CCS which may be useful is to look at it as an extremely high value resistor. If you have a bazillion ohm resistor in series with a choke, the choke really doesn't make any difference - it's the resistor that determines the behavior of the combination.

Feb 3, 2019 at 10:21 AM
I'm looking for a cost effective way to power my 007, what stax amps are recommended besides the expensive kgss's ?

Feb 3, 2019 at 10:48 AM
I'm looking for a cost effective way to power my 007, what stax amps are recommended besides the expensive kgss's ?

Depends, If you are not opposed to buying used, a used Stax SRM-717 or SRM-727 with global feedback mod, or a used SRM-323 if you're really hard up. If you are willing to do a little DIY or have it done for you, an SRM-T1 or T1S with constant current load mod will do the job and is IMHO somewhat more refined sounding than the solid state amps.

If you're willing to do quite a bit of DIY an SRX-Plus (tube) or KG amp can be done for significantly less money than a built amp.

Feb 3, 2019 at 11:00 AM
All those have the same power ?

Feb 3, 2019 at 11:32 AM
Pretty close to the same power. Note that the T1 (or any other Stax tube output amps) use plate resistor loads in stock form, so without the constant current load mod they have less power. With the constant current load mod, the T1 has a bit less voltage output, but it's only about 0.7 dB less, which is negligible. Any of them will put out 110 dB at 1 kHz max output, which on a continuous basis is enough to cause permanent hearing loss within a few minutes. The KG amps with +/-400V power supplies will put out another dB or so, and the DIY T2 with +/-500V power supplies will put out another 3 dB.

Feb 3, 2019 at 4:56 PM
Pretty close to the same power. Note that the T1 (or any other Stax tube output amps) use plate resistor loads in stock form, so without the constant current load mod they have less power. With the constant current load mod, the T1 has a bit less voltage output, but it's only about 0.7 dB less, which is negligible. Any of them will put out 110 dB at 1 kHz max output, which on a continuous basis is enough to cause permanent hearing loss within a few minutes. The KG amps with +/-400V power supplies will put out another dB or so, and the DIY T2 with +/-500V power supplies will put out another 3 dB.

So in terms of max sustained output, the KG amps don't have much on the Stax amps, but what about max transient outputs? Can upgrading caps alone help in this regard?

Feb 3, 2019 at 7:03 PM
So in terms of max sustained output, the KG amps don't have much on the Stax amps, but what about max transient outputs? Can upgrading caps alone help in this regard?

Are you referring to high frequency transients or are you referring to higher voltage on a transient basis. In terms of the former, the ability to deliver an undistorted high frequency transient at a high voltage is dependent on an adequate slew rate. That said, most studies of the slew rate of musical signals suggest that the ability to deliver an undistorted sine wave at 6 kHz is sufficient to reproduce almost any musical signal, although the late Peter Baxandall reported that he found one musical excerpt that required enough slew rate to reproduce a full power 15 kHz sine wave.

Since electrostatic headphones resemble capacitors, this means that amps with a higher current output stage have the potential to have a higher slew rate, because the equation describing current required into a capacitor load is:

current (mA) = 2000*pi*voltage*frequency*capacitance (in farads).

So, for example, into a typical headphone resembling a 100 pf capacitor, 400VRMS at 20 kHz requires a max current of about 10 mA RMS, or about 14 mA peak. A T1S with constant current load mod can provide about 400VRMS but the output stage only runs about 9 mA, and some of that is needed to drive the feedback resistors, so only about 5.8 mA of the total output stage current is available to drive the headphones, which means it will go into hard slew limiting driving the headphones around 8 kHz, and it won't sound too good doing it as amps never sound their best when driven to their limits.

The Stax amps run between 9 and 14 mA/channel quiescent current output stages. The KGSS ran around 16 mA,/channel, the DIY T2 around 28 mA, the BHSE around 36 mA and the Carbon up to 40 mA/channel, so the KG amps definitely have higher slew rate. However, note that in most cases this potential is not needed because 1) most people don't listen at these high levels, and 2) most music has less slew rate requirements. In other words, most people will not approach the slew rate limitations of even the Stax amps.

The other option that you may be talking about is having a higher voltage limit for transients. A prime example of this is the original NAD3020 which was rated at 20 WPC continuous but had a "soft" power supply which allowed it to reproduce transients at a higher power, somewhere around 40 watts IIRC. However, this is because the power supply had a higher quiescent voltage which dropped under load to a lower voltage, and this occurred because the NAD was a class A/B amplifier, meaning it drew more current from its PS under continuous conditions (i.e. putting out a high power sine wave) than under music conditions.

However, all the Stax amps and the KG amps are class A, which means that they draw the same amount of power from the PS whether they are reproducing music or not. Moreover the KG amps also have regulated power supplies, so the voltage is further stabilized. So adding extra capacitance won't really change the transient power of either the Stax or KG amps. All of these amps are differential circuits, so current draw is pretty constant anyway (if the amplification devices were perfectly matched there should be NO current variation at all up to clipping).

Last edited:
Feb 3, 2019 at 8:24 PM
Are you referring to high frequency transients or are you referring to higher voltage on a transient basis. In terms of the former, the ability to deliver an undistorted high frequency transient at a high voltage is dependent on an adequate slew rate. That said, most studies of the slew rate of musical signals suggest that the ability to deliver an undistorted sine wave at 6 kHz is sufficient to reproduce almost any musical signal, although the late Peter Baxandall reported that he found one musical excerpt that required enough slew rate to reproduce a full power 15 kHz sine wave.

Since electrostatic headphones resemble capacitors, this means that amps with a higher current output stage have the potential to have a higher slew rate, because the equation describing current required into a capacitor load is:

current (mA) = 2000*pi*voltage*frequency*capacitance (in farads).

So, for example, into a typical headphone resembling a 100 pf capacitor, 400VRMS at 20 kHz requires a max current of about 10 mA RMS, or about 14 mA peak. A T1S with constant current load mod can provide about 400VRMS but the output stage only runs about 9 mA, and some of that is needed to drive the feedback resistors, so only about 5.8 mA of the total output stage current is available to drive the headphones, which means it will go into hard slew limiting driving the headphones around 8 kHz, and it won't sound too good doing it as amps never sound their best when driven to their limits.

The Stax amps run between 9 and 14 mA/channel quiescent current output stages. The KGSS ran around 16 mA,/channel, the DIY T2 around 28 mA, the BHSE around 36 mA and the Carbon up to 40 mA/channel, so the KG amps definitely have higher slew rate. However, note that in most cases this potential is not needed because 1) most people don't listen at these high levels, and 2) most music has less slew rate requirements. In other words, most people will not approach the slew rate limitations of even the Stax amps.

The other option that you may be talking about is having a higher voltage limit for transients. A prime example of this is the original NAD3020 which was rated at 20 WPC continuous but had a "soft" power supply which allowed it to reproduce transients at a higher power, somewhere around 40 watts IIRC. However, this is because the power supply had a higher quiescent voltage which dropped under load to a lower voltage, an this occurred because the NAD was a class A/B amplifier, meaning it drew more current from its PS under continuous conditions (i.e. putting out a high power sine wave) than under music conditions.

However, all the Stax amps and the KG amps are class A, which means that they draw the same amount of power from the PS whether they are reproducing music or not. Moreover the KG amps also have regulated power supplies, so the voltage is further stabilized. So adding extra capacitance won't really change the transient power of either the Stax or KG amps. All of these amps are differential circuits, so current draw is pretty constant anyway (if the amplification devices were perfectly matched there should be NO current variation at all up to clipping).

Thank you. What I was really wondering is that if the KG amps only offer a mere 1 dB more power than the Stax amps, then what really distinguishes them from Stax amps besides the supply voltage regulation? I'm not inclined to think that voltage regulation and a mere 1 dB higher output potential alone can account for that big of a sonic improvement.

As a separate inquiry. If the srm-252s only offers 1.2 mA, would that mean that moving up to something like the srm-323s would add ~6 dB (x4) more max output, or is the calculation not so linear and straightforward?

Feb 3, 2019 at 11:57 PM
Thank you. What I was really wondering is that if the KG amps only offer a mere 1 dB more power than the Stax amps, then what really distinguishes them from Stax amps besides the supply voltage regulation? I'm not inclined to think that voltage regulation and a mere 1 dB higher output potential alone can account for that big of a sonic improvement.

As a separate inquiry. If the srm-252s only offers 1.2 mA, would that mean that moving up to something like the srm-323s would add ~6 dB (x4) more max output, or is the calculation not so linear and straightforward?

Well, what I've written about are some of the most obvious technical differences. Spritzer, for example, says that the Stax biased the 717 lower making it less linear than the KGSS that it was reportedly based on. The BHSE, GG and Carbon use a different output stage (in tube terms, a grounded grid topology) which is a more linear method of driving the output devices, but requires a low impedance driver. Also, as the KG amps tend to have significantly more quiescent current available, they are running well within their current limits when the Stax amps are approaching their current limits. That could make a difference when listening at high levels. I can discuss circuit differences, but I can't tell you how much of a difference in sound it will make for you - you'll have to listen and judge for yourself.

The question is, do you really think that these basic specifications tell the whole story?

As far as regulated supplies, I have heard a good regulated supply significantly increase the soundstage and dynamics of a LTA MicroZOTL, but that is a different beast altogether. AFAIK, nobody has built a KG amp with passive PS and then added a regulated supply and compared them directly. OTOH most commercial amps including Stax have passive power supplies, not because they are better, but because they are much cheaper and take up much less space and less weight.

Also keep in mind that Stax and KG have two different objectives. KG does the best design he can regardless of cost. Stax designs amps to sell at a reasonable cost and to make a profit, so they have to judge cost-effectiveness and size (a larger and heavier amp incurs higher shipping cost). Also they have to consider profits for distributors, retailers, etc. in pricing their amps. Until the T8000 they had limited themselves to a certain size for their amps. For example, the KGST is essentially the same circuit as the Stax T1/006/007 series, but with more powerful output tubes, constant current output loads, and a regulated supply.

The Stax SRM-252 is listed as putting out 280VRMS at 1 kHz vs. the SRM-323 which is specified as putting out 400 VRMS at 1 kHz, which is a 3dB difference. The 252 is CURRENT limited in how much output it has, whereas the 323 is VOLTAGE limited in its output. However, if you compared the 252 to 323 side by side at say, 280 VRMS, the 323 should sound significantly better as the 252 would be at its limit whereas the 323 would be well within its limits.

Last edited:
Feb 4, 2019 at 3:55 AM
At first, somebody need to start DIY a KG Amplifier first and when he/she achieved this, he/she will know and understand from paper to personal experiences as of why (many things happens, technological and sound performances...etc..)

There are a huge different from theory VS real practices. I admire @JimL11 for his ability to be patiently explaining all this

Feb 4, 2019 at 9:46 AM
Peter Baxandall reported that he found one musical excerpt that required enough slew rate to reproduce a full power 15 kHz sine wav
JimL11, Nice post! Can you cite a reference for Baxandall's report?

Feb 4, 2019 at 12:36 PM
The Wireless World has his earlier research (Nelson Pass also found similar data that he wrote about in an article about his A40 (?) class A power amp in The Audio Amateur. Baxandall wrote about the 15 kHz full power signal in a chapter he wrote about electrostatic loudspeakers for High Performance Loudspeakers, by Martin Collums, I think it was the 2nd or 3rd edition. I'm not at home this week so all this is from memory.

Feb 4, 2019 at 4:54 PM
... High Performance Loudspeakers, by Martin Collums,..

Nice. Thanks. I haven't seen a copy of that in ages. I'll look around for one.