When I decided last year to explore some Stax electrostatic headphones, I discovered that even their older-model headphone amps are jaw-droppingly expensive. So I got an SRD-7 Adaptor and converted it to power high- and normal-bias Stax 'phones. Unfortunately, the sound quality left much to be desired.
The good news is, it's most glaring problems are fairly easily addressed. I just completed these detailed mods for the SRD-7 adaptor:
Transforming the Stax SRD-7
Comments/questions welcome. (This is my first crack at creating web content of this ilk outside of a forum, so be gentle... 🙂 )
The good news is, it's most glaring problems are fairly easily addressed. I just completed these detailed mods for the SRD-7 adaptor:
Transforming the Stax SRD-7
Comments/questions welcome. (This is my first crack at creating web content of this ilk outside of a forum, so be gentle... 🙂 )
Ha! You have a mixed-vintage rig. The SR-X MkIII were concurrent with your Lambdas, no? But the SRD-7 replaced the 6, supposedly with better transformers. So these mods won't directly apply. But if your 6 has a PTH, definitely get rid of it!
Oh dear... that stinks. It looks like the 27R terminating R's are not installed. Odd.
That's an "SB" model... Self-Bias... it gradually builds up the bias voltage from the rectified xfmr output. Clever.
That's an "SB" model... Self-Bias... it gradually builds up the bias voltage from the rectified xfmr output. Clever.
"One example is the L300, which have a fairly “polite” sound with soft, pillowy bottom and smooth-but-recessed highs. The slight (0.7dB) rising response above 1kHz compliments the L300 nicely. But for natually-bright cans like the SR-X, we’ll want to tame the upper-octaves rise as well."
I have a Stax SRD7 Pro as well as a custom transformer unit (mjolnir-audio), but as you have observed, Stax 'stats can have a very different tonalities, so modifying these transformer units can sound better with one 'stat but actually sound "worse" with others. For example, my Stax Omega II and Stax SR009 represent both extremes of tonal/frequency response, so a mod like this will require a toggle switch to present different options.
We're thinking the same, drjlo. What you describe factored strongly into making the compensation a "two-stage affair". The output comp circuit flattens the overall slope. This is beneficial in all cases, so it can be permanently installed in the enclosure.
The input RL can then be tailored "to suit" (or bypassed). It does not have to be permanently installed, so a switch is not necessarily needed. One approach is installing them in the power amp with separate output terminals, as I chose to do. Another would be to install them in a plastic junction box that is plugged into the amp output. It doesn't need to be shielded. That would actually be the most flexible. Different ones could be optimized for different model 'phones.
Few are going to expand $399 to $11,000 for a "head-phone" microphone system, but it begs the question as to whether the electrical transfer function replicates the acoustic.
But the impedance balancing you describe is indeed worthwhile, and I am going to try it out with the Stax Lambda!
Well it doesn't, any more than a power amp replicates the TF of the speakers... nor should it be expected to, yes?
Keep us posted, of course!
Update: Recent communication with someone doing the mods to their SRD-7 made me take a closer look at the situation regarding the parallel LR circuit at the input. I had perhaps not made it clear enough that this input filter is ONLY needed with the SR-X headphones. This was based on using it in my systems, which do not have any tone controls.
Well, after using it with an integrated amp that has even the simplest bass and treble shelving controls, I can now say that the tone controls do a better job of correcting the SR-X tonal imbalance than the input LR filter does. So the input filter is ONLY needed if there are no tone controls in the system.
This simplifies the mods even more, and removes the objections some might have to a) having an inductor in series with the amp output, and b) making mods to the adaptor that narrow its use to one model of headphone only.
The instructions have been updated to correct for this.
Well, after using it with an integrated amp that has even the simplest bass and treble shelving controls, I can now say that the tone controls do a better job of correcting the SR-X tonal imbalance than the input LR filter does. So the input filter is ONLY needed if there are no tone controls in the system.
This simplifies the mods even more, and removes the objections some might have to a) having an inductor in series with the amp output, and b) making mods to the adaptor that narrow its use to one model of headphone only.
The instructions have been updated to correct for this.
Modding SRD will only provide incremental improvement. The glaring deficiency is in the transformer.
On my STAX journey, I was expecting SRD transformers to be of better quality. But on close examination, transformers in SRD-5 and SRD-7 had cheap steel cores and simple windings, like those in power transformers. Unsurprisingly, these transformers had self-resonance at 8-10 kHz. This not only screws frequency response, but causes high frequency distortion. So, my plan to utilize SRD transformers as output chokes in directly-coupled tube amplifier had to be scrapped.
SRDs are sorely inadequate for the STAX phones. They cannot be radically improved. The concept of using an amplifier designed for standard speakers to be used with STAX is also flawed. It has to be a dedicated STAX amplifier.
On my STAX journey, I was expecting SRD transformers to be of better quality. But on close examination, transformers in SRD-5 and SRD-7 had cheap steel cores and simple windings, like those in power transformers. Unsurprisingly, these transformers had self-resonance at 8-10 kHz. This not only screws frequency response, but causes high frequency distortion. So, my plan to utilize SRD transformers as output chokes in directly-coupled tube amplifier had to be scrapped.
SRDs are sorely inadequate for the STAX phones. They cannot be radically improved. The concept of using an amplifier designed for standard speakers to be used with STAX is also flawed. It has to be a dedicated STAX amplifier.
"Don't compare me to the almighty; compare me to the alternative." (Joe Biden)
No offense, but methinks you're taking this way out of context. The point of this project was not to create the "ultimate" Stax adaptor. For that, I would not choose a transformer output. It was to see if the affordable SRD-7 could become a good, worthwhile entry-level adaptor. Stock, it most certainly is not.
This sounds like total BS, deep Stax-KoolAid-drinking to me. The SRD-7 is not difficult at all to drive, the impedance is above 20 Ohms from about 15Hz to 9kHz.
I am listening to this setup (SR-X MkIII) as I write this. It is much better than you give it credit for. Is it the best, most accurate HP system I have? No. But I really enjoy it. And it is HUGELY better than stock.
I stand by my assertion that the modded SRD-7 is easily the best affordable adaptor/amplifier for Stax cans, especially the older low-bias ones.
No offense, but methinks you're taking this way out of context. The point of this project was not to create the "ultimate" Stax adaptor. For that, I would not choose a transformer output. It was to see if the affordable SRD-7 could become a good, worthwhile entry-level adaptor. Stock, it most certainly is not.
I totally disagree. These mods provide significant incremental improvement.
Actually, the most glaring deficiency is the PTH thermistor. It's distortion is almost two orders of magnitude worse than the xfmr.
That I agree with, and said so in the article. They're nothing to write home about, that's for sure, especially the core material. But I was very pleased with improved results after compensation.
The SRD-5 and -7 xfmrs are not the same; the -7's are larger than the -5's.
Not sure how you measured that. Actual resonance is around 50kHz. The dip below that is from the headphone and cable capacitance.
Again, I disagree. What is "radical"? Taking something from borderline unlistenable to quite enjoyable is not radical enough? And again, let's keep things in context. Show me a better $100 alternative.
This sounds like total BS, deep Stax-KoolAid-drinking to me. The SRD-7 is not difficult at all to drive, the impedance is above 20 Ohms from about 15Hz to 9kHz.
I am listening to this setup (SR-X MkIII) as I write this. It is much better than you give it credit for. Is it the best, most accurate HP system I have? No. But I really enjoy it. And it is HUGELY better than stock.
I stand by my assertion that the modded SRD-7 is easily the best affordable adaptor/amplifier for Stax cans, especially the older low-bias ones.
Last edited:
No argument, if you like it, you like it.
I measured SRD transformer resonances by observing 1,000 Hz square wave. 1 Ohm source, secondary terminated with 300 pF capacitor.
I measured SRD transformer resonances by observing 1,000 Hz square wave. 1 Ohm source, secondary terminated with 300 pF capacitor.
Well no wonder. Why 300pF? Stax cans with cable are typically less than 100pF. The SR-X are around 90pF. L300 with flat cable lower than that.
Also, your source impedance was too low, something in the 4-5 Ω range is optimal.
Those factors would definitely bring out an 8-10k resonance.
Last edited:
300 pF was worst case assumption.
Just measured my SR-3 and SR-X on capacitor bridge. SR-3 is 170 pF. Surprisingly, SR-X is only 44 pF! I guess they improved the SR-X cable big way.
1 Ohm is a typical output impedance of a no-feedback triode amplifier. Solid state amplifiers are at least an order of magnitude lower.
Lower source impedance provides better damping of transformer self-resonance. With 4 Ohm source, ringing is worse than with 1 Ohm source.
That is not the case for any xfmr I've worked with. Going below the optimum source impedance creates an underdamped response, i.e. ringing at the upper bounds of its bandwdth. There also has to be an input termination R to tame the inductance out-of-band.
There was a typo in my previous post that I didn't catch in time to correct. Optimum source impedance for the SRD-7 xfmrs is in the 4-6 Ω range, with tradeoffs of linear bandwidth vs ringing determining the choice. I'll post scope plots later if I can get to it.
Here are 1kHz square waves through the SRD-7 with 100pF on the 2ndary, along with the 1nF+619kΩ comp that my mod recommends. The results would be even more exaggerated without the comp network.
This one is with 2.3Ω source R. Lots of overshoot - nearly 30%..
This one is with the standard 4.7Ω source R. Much better.
And this one is with 5.7Ω source R. A little cleaner, but now it is starting to lose overall level.
I can't recall any transformer I've optimized that didn't behave this way. Lowering the source R below optimum always resulted in more overshoot/resonance.
This one is with 2.3Ω source R. Lots of overshoot - nearly 30%..
This one is with the standard 4.7Ω source R. Much better.
And this one is with 5.7Ω source R. A little cleaner, but now it is starting to lose overall level.
I can't recall any transformer I've optimized that didn't behave this way. Lowering the source R below optimum always resulted in more overshoot/resonance.
That leaves much power behind. The classic prescription is load half of Rp. Or 4r Zout on the 8r tap.