The Headcase Stax thread

[spritzer]

With no inside pics it is hard to tell but I had the Mk1 and it was utter trash and dangerous to the end user.  There are pics here somewhere of that but yeah, utter garbage

[K3cT]

23 hours ago, spritzer said:

With no inside pics it is hard to tell but I had the Mk1 and it was utter trash and dangerous to the end user.  There are pics here somewhere of that but yeah, utter garbage

Thanks, I would like to get into stats (particularly the CRBN) but it seems nicer if I can use my existing setup. 

[spritzer]

Good transformers are a great way to do that and are definitely getting more popular.  

[K3cT]

37 minutes ago, spritzer said:

Good transformers are a great way to do that and are definitely getting more popular.  

Yeah a couple of months ago I posted about the BHSE but unfortunately the seller went MIA so that didn’t work out lol.

I am a bit concerned my main amp, a SET 2A3 amp is a bit on the lower power side. In my experience it can drive the 32 Ohm transformer for RAAL Immanis fine but I don’t have much experience with e-stat interfaces. 

[spritzer]

2-3W or so?  That is very much on the low side but it always depends on overall system gain and how loud you listen.  A Stax SRD-7 Mk2 should work as the ratio on those is on the high side.  

[K3cT]

2 hours ago, spritzer said:

2-3W or so?  That is very much on the low side but it always depends on overall system gain and how loud you listen.  A Stax SRD-7 Mk2 should work as the ratio on those is on the high side.  

Yeah, around that range, if I'm not wrong it's about 2.5W into 32 Ohm load. I noticed that you have a Sansui variant of the SRD-7, will that work nicely as well?

[spritzer]

It should work, the ratio is a bit higher than the Stax ones.  Ultimately it's not about power but voltage in the system or voltage swing rather.  

[psyduck]

A closed back estat https://www.kaldasresearch.com/pages/inox

[Omega_ELS]

What do you think about these drivers?

can they have good sound?)

[chinsettawong]

They look nice!  I'm sure they can produce good sound.

[Milosz]

Hi all.  I've been trying to find measurements of electrostatic amplifiers - other sites typically don't know how to measure them- so looking for HD & IM distortion measurements, max voltage swing, SINAD, and also maybe slew rate.  I've spent some time looking around on this site but haven't stumbled across any yet... if someone could point me to a page here or maybe a site like Mjolnir or some other place that might list some measurements.  I just want to see what's considered good for electrostatic amps....

 

Thanks! 

[spritzer]

On 6/3/2026 at 7:26 AM, Milosz said:

Hi all.  I've been trying to find measurements of electrostatic amplifiers - other sites typically don't know how to measure them- so looking for HD & IM distortion measurements, max voltage swing, SINAD, and also maybe slew rate.  I've spent some time looking around on this site but haven't stumbled across any yet... if someone could point me to a page here or maybe a site like Mjolnir or some other place that might list some measurements.  I just want to see what's considered good for electrostatic amps....

 

Thanks! 

What the actual fuck is this?  How is posting in an open forum "reaching out to me"?  Copied from the Audioretards forum:

Quote

I've tried to contact Dr. Gilmore and his acolyte Spritzer of Molnir Audio to share with us their methods for measuring electrostatic amp performance, which is much different from the resistor-loaded method Amir used - but they are a clubby bunch and don't share much with mere hobbyists.

I know the testing requires a differential probe like the Tektronix THDP0100

I have a Carbon amp for my SR-007s, SR Lambdas, ESP950s and Dan Clark Corinas. Made a **HUGE** difference over the SRM-007 I used to use.

Did you send me a PM or an email?  No... so how did you reach out to me?  Go ask the retard in chief over there on how to test electrostatic amps... as he did such a good job last time.  Wasn't the quote something like, how can anybody listen to this as it measures so badly?  Yeah... stellar work there.  

Edited Thursday at 08:53 PM by spritzer

[Milosz]

I wrote an email to you using [email protected]

[spritzer]

Which has probably been bounced back as that is not my domain name.  

[kevin gilmore]

miloz never contacted me. i would likely not respond if he did.

get one of each of these from ebay.

keithley 2015 6.5 digit thd multimeter and a hp 82357b (or later) usb to gpib adapter.

yes, the 82357b is a fake coming out the wrong end of the agilent factory (no shielding on the plastic) in china. but work absolutely identical to the real one which is about $600 new. unlikely to pass fcc radiated noise tests

and a windows 11 laptop.

then contact james king for the software package.

ebay price about $600 total for the 2 pieces.  (about $7k brand new)

have at it.  rated for 2000 volts peak to peak. no attenuators or other hacks needed. 

many of the used 2015 will require replacement of the display for about $75, and unsoldering about 100 connections. but the software does not require a well lit display except to put it into the right mode after you turn it on.

accurate to .002% Thd+N. built in balanced signal generator that matches the same specs. single box solution.

better than any stax amp with 50 to 60db of gain. the Noise component overwhelming the signal.

no way i would pay $35k for an audio precision and then pray that the required attenuator does not cause trouble. I would much rather pay that much for a set of CCB brakes and wheels.

retired 9 years now. i only do things that are fun.  like 30 to 130 in the Z06.  over and over and over. and yet no tickets (yet).

Edited Thursday at 09:24 PM by kevin gilmore

[Milosz]

I had some interaction with Dr. Gilmore a few years back discussing what kind of tests to perform on a DIY T2 that I had started, 7 years ago was it? Don't recall.

Thanks for giving the info on methods to test electrostatic amps, I'll post those on Audio Science Review there are actually people there curious about this, and yes I know Amir (the main guy) has a hard time thinking beyond his accepted paradigm of using a klippel robot to test speakers (doesn't work with Magneplanars but he can't see that) and testing amplifiers into load resistors.  However he's not the only one on the site with an interest in testing things, and there isn't much info out there about PROPER testing of amps for electrostatics.

I would LOVE to see some actual published test results on the "better" designs you have come up with as compared to the venerable Stax amps, but I can't find any such. I would have thought that Spritzer's Mjolnir Audio site would have published some test results, but I haven't found any.

[jamesmking]

In my opinion the current prices on ebay for a  Keithley 2015thd are rather high.

Some of its more interesting features are only available via gpib and the programming guide has multiple errors and omissions and does not in some cases give vital information which means figuring out simple things like how wide in Hz a FFT bin is, is a real pain in the **** - and the width varies with the fundamental frequency in non-linear way. Keithley provide no software or programming libraries and so you are left on your own. You can't even get the frequency of a bin or ask for the bin at a certain frequency on the 2025/6thd (you can on the (2015/6P) which seems crazy. The user interface on the keithley is not that intuitive. Due to some (I assume) design limitations it only has as a maximum of 1023 fft bins which limits its resolution and  for fundamental frequencies less than 60hz it can't differentiate noise from distortion since the bin width is the exactly the same as the fundamental, i.e. EVERY bin contains distortion and there are no noise bins. Despite this it will give you a thd+n result for a fundamental of 50hz with no warning that its actually a thd without n calculation! If you set the keithley up with filters or limit the number of harmonics the keithley can not provide unfiltered results. i.e. it only shows on screen and passes to a program the filtered results. There is no way to get at the raw readings. So from a programming point of view its often better to do a full fft and the calculate the thd, thd+n, sinad, noise and apply filters in software than rely on the keithley to do it. On the thd version the fft automatically locks onto the highest bin as the fundamental and if there are two highest bins this causes issues. The fft on the thd has no reliable way for you to manually specify the fundamental frequency, there is a hack but you can only specify a sub multiple of the fundamental frequency in the range 20-68hz and even then the fft is not as clean as automatically locking. The harmonic calculation is limited to 64 harmonics regardless of how many harmonics are present in the fft. there are also some firmware bugs (at least in the firmware I have) and there seems to be no reasonable way to get new firmware. The built in signal generator only goes from 10hz to 20K and has a surprising amount of jitter and at 20 hz it can output 19.Xhz which causes issues. The thd/thd+n and sinad only accept fundamentals from 20hz to 20K and at 20Khz there is only the fundamental and 1 harmonic since the fft is hard bandwidth limited to 50khz. Noise calculations are limited to fundamentals in the 60hz-20K range. In short the Keithley shows its age and requires a lot of programming and reverse engineering to understand what's actually going on under the hood. 

The 2015P and 2016P remove many of the programming limitations of the 2015/6thd and it would be interesting to know if a thd could be upgraded to a P model - it might only be a firmware change since the thd programmatically seems to be a gimped P with certainly very useful programming features missing. Unfortunately the P models are double the price of the thd...

Here are a couple of measurements of the Keithley's signal generator feed back directly into its inputs. At least on my unit, the generator has around 0.02% THD at 1V rms output (using single ended - not balanced). Using a good PC sound card instead the thd can be cut by around a factor of around 5-10. So the measurement of the keithley is more capable than the built in signal generator..

Signal generator harmonics at 20Hz (10 averages minimum and maximum values also shown)

 

Signal generator noise and harmonics at 1Khz (10 averages)

signal generator THD vs frequency of its output (averages of 10 measurements, minimum and maximum shown note log scale for x axis frequency)

Notice how there is considerably more THD at 20hz... this is because when 20Hz is requested the output of the signal generator is actually 19.X and this causes issues with the fft lock on the Keithley.

Noise of the keithley signal generator vs frequency of its output (120 averages min and max shown log scale for frequency)

 

notice the rising noise as frequency increases but there is a huge noise spike at 9.95Khz and smaller spikes at 85hz, and 19.86Khz. I have not put any effort into determining if this is caused by the measurement side of the Keithley, external noise or the Keithley signal generator.

THD+N of course shows the issues with 19.zHz locking and the noise spikes:

 

Edited 11 minutes ago by jamesmking