iFi Pro iESL Electrostatic Headphone Energizer Review

I've been holding on to the Drop x Koss ESP95X hoping one day that I would eventually get an amp to my liking that would do them justice. For backstory: I once owned a T2 DIY along with SR-009, SR-007, SR-507, and HE Jade. The HE Jade I modified and it became my preferred electrostatic headphone. My favorite was always the Orpheus, but I didn't want to plunk down $10k or more to own one back then. I also had an ESP950 handy and this actually was my second favorite electrostatic headphone. The T2 DIY did wonders for the Koss. Looking back, I guess you can say that I had developed a preference toward any electrostatic headphone which wasn't STAX.

Eventually I returned the T2 DIY early (it was on a 99-year lease) to go back to SET (single ended triode) amps and regular headphones. My foray into electrostatics was interesting, but I found myself searching for good sound on merely a technical level, and missed the engagement factor of SET amps. It's not that SET amps and electrostatics were mutually exclusive. Perhaps the greatest electrostatic amp I had ever heard was the one that Frank Cooter brought to a Bay Area meet 10 years ago. This was the first meet that @CEE TEE put together and the first large Head-FI meet which I attended. Unfortunately this amp was never seen again as Frank got spooked that some idiot would sit on the mercury vapor rectifiers and cause a major toxic (literally, not figuratively toxic) incident at a headphone meet. The secret to that amp was those rectifiers.

Anyway, what I wanted wasn't a T2 DIY for the ESP950. I simply wanted an amp which was at least semi-competent with these headphones. There was a BHST (not BHSE) on Head-FI for cheap a few years ago, but I missed out on the opportunity. No way was I going to bother with KGSS or its variants such as HV: macrodynamics and control all right, but talk about flat and boring on the microdynamics plus analytical and sterile - the antithesis of SET. What about STAX? You kidding? I dislike their low-end amps, which are expensive enough already; and kind of dislike their higher end amps. They all sound too soft. Oh, the SRM-T8000? Wow, what a piece of shit for $6k+. I don't know how STAX gets away with making stuff like this. BTW, this review here I totally pulled my punches: https://www.superbestaudiofriends.org/index.php?threads/stax-srm-t8000-first-impressions.4761/. What about those STAX energizers? Those made the SR-009 sound like low fidelity headphones, syrupy, non-resolving, goo - and this from the Ragnarok 1, which was otherwise a great small speaker amp if you knew how to get it up right.

So when iFi came out with their Pro iESL, I was rather intrigued about the possibilities. iFi would be my only hope. I trusted the brand enough that they would actually use good transformers which I would be crucial. And to remind readers again, I simply wanted something which was competent, not necessarily fantastic, so that I could enjoy the ESP95X in some way. I would say that my expectations have been exceeded, rather significantly.

Let story time begin...

 

I've actually had the iFi Pro iESL since last year when I was in California. I had intended to write a full review. However, I relocated to Texas, hence the delay. I'll start the story from last year.

I had heard from others who had experience that the iESL didn't offer much gain, hence my first instinct was to pair it up with a cheap solid-state speaker power amp I had on hand: the Schiit Vidar. I also like tubes, and given my success with Saga as a preamp (light touch of tube sound), I stuck it in front of the iESL. The Saga also serves as a nice volume control, which is needed in this situation. Ideally, I would have liked to put a iFi based front-end, but other than the JBL Nano Patch, the Saga was all I had on hand. And really, I don't mind taking a best-of-breed approach, curating pieces from different brands, to get the result that I want.

In addition to the Vidar, I also tried another high-gain amp, the Crest CA200 and also the Schiit Aegir. What I was impressed by was how well the iESL was able to transmit the attributes of each power amp. The Vidar sounded warm and powerful; the Aegir did its Aegir thing, but with rounder transients with less heft; the CA200 was delicate with a endlessly deep stage, but a bit veiled. (In retrospect, I wish I had tried the Crown CDi1000, sharp attacks, massive control, third-order harmonics and all, to emulate the KG sound).

Holy cow I thought. This was pretty neat. As I write this now, after much more experience, I can say that the iESL does have a sound of its own. Everything does. But the bottom line is that the iESL is capable of transmitting a significant portion of the sound of the source amplifier without imposing a strong sound of its own. That's impressive because those STAX energizers... yuck.



Oh yeah: if anyone says they have a piece of gear which is fully transparent, they are either full of shit or haven't been exposed to enough gear. By fully transparent, I mean no sins of omission and no sins of commision. Most often, people cannot tell when there are sins of omission.

 

The iFi Pro iESL has several novel features going for it:

It utilizes a big capacitor bank for maintain the high voltage bias on the headphone diaphragm. The circuit charges up the caps and only tops it off every once in a while to maintain the voltage. I don't know if this has any effect upon the sound, but I would give iFi engineers the benefit of the doubt that they tried various solutions and in the end, came up with this very interesting solution.

The ability to select multiple gains with a middle tap on the primary and two sets of taps on the secondary of the transformer (needed to get the high voltages necessary to drive electrostatic headphones). The lower the gain, the easier the load on the source amplifier. At the highest gain with nominal 16-ohm load, 20Vrms at the input gets us 640Vrms. At the lowest gain, 20Vrms gets us 320Vrms. That's x32 and x16 gain respectively. I will explore this a bit later with measurements. Some very interesting stuff coming up!

A low and high-impedance setting for the shared node between the channels for bias. iFi calls this AC Termination. Yes, this darn thing works. When AC Termination is turned on, the headstage gets pushed deeper out. I think there's some expense to localization of images, some diffuseness. Overall I like the effect and prefer the AC Termination turned on. The only time I would have this turned off is if I wanted to do some stereo mixing work. However, I would probably be using speakers instead of headphones for this kind of work. There is a big difference between headstage and soundstage.

A selector for bias voltage! Koss ESP950 liked to be pushed to 620V, 640V even. Push it until the diaphragm starts to get small holes and melts! Conversely, the Nectar Hive seem to like lower bias. Higher bias sounds harder and lower bias sounds softer. I cannot stress enough what a handy feature this turned out to be!

The "no duh" one: can use a universe of different source amps to get almost any kind of sound to one's content.

 

No, and I have no intention of doing so because life is too short and also because I have no idea how it works. I would need to be convinced because I would much rather run headphones from dedicated headphone amps or low-efficiency planars directly from the taps of speaker power amps. Why put stuff in the way of the signal path if it's not necessary? I can see this as a handy feature for traditional pure two-channel guys, but this not one that fits my usage needs because I have dedicated systems for two-channel and headphones.

P.S. Super cool stuff coming up...

 

So you guys know like how no one the universe had any real fricking idea how electrostatic headphones measure with respect to impedance vs. frequency? I believe there was talk in the ASRR thread about how Amir may have improperly measured the performance of an electrostatic amp because he used the 100k-ohm load of this AP-555 and that 'stats may have impedances in the megaohms? Or that the STAX site mentioned something about the load being the equivalent of a small cap in their amplifier specs?

Well now I've got some solid data, of which we can infer the impedance curve of electrostatic headphones. The best way of course is to measure directly, with the 580V bias turned on. The second best is to do so with a transformer, which of course would provide some level of isolation of protection of DC in case of a high-voltage arc from the biased diaphragm.

So here we go:

Koss ESP950
Impedance at iESL speaker inputs 96-ohm setting


Koss ESP950
Impedance at iESL speaker inputs 64-ohm setting (x16 gain per spec)


Koss ESP950
Impedance at iESL speaker inputs 32-ohm setting


Koss ESP950
Impedance at iESL speaker inputs 16-ohm setting (x32 gain per spec)


Does anyone know how the transformer gain will translate into impedance? It would be interesting if someone who knows the formulas at the top of their head can convert the above to drive the native impedance curve of the ESP950. Perhaps this data would allow ASR to develop a more realistic curve to test amps for electrostatic headphones. Science is about discovery, not assumptions or half-assery!

Note that none of these loads are especially different for any amplifier, including headphone amplifiers, with the exception of the higher-frequencies at the higher gains. The key really is sufficient voltage output.

 

Yeah. Quite frankly, the sciency person in me is a lot more intrigued at being able to finally set an idea on the impedance curve of various electrostatic headphones.

Calling @Biodegraded. It appears so, at least at a glance, from the x16 x32 gain data. We should be able to derive the impedance for x1. My brain is too fried today.

 

I forgot to post this. This is what happens when there is no high-voltage bias applied to the diaphragm. It just don't work. Electrostatic headphones are not like regular headphones. They need a bias voltage to work.

ESP950
(No bias voltage applied)
Impedance at iESL speaker tap (any gain)

 

So much for the "megaohms" and perhaps the small cap (I forgot the numbers) assumptions.

 

Nectar Hive headphone impedance measurements below. Not too different from the ESP950. A bit higher and broader after 1kHz.

Nectar Hive
Impedance at iESL speaker inputs 96-ohm setting (x16 gain per spec)


Nectar Hive
Impedance at iESL speaker inputs 16-ohm setting (x32 gain per spec)


One thing I didn't notice before is that the impedance is a little bit more peaky in the mids at the higher gain lower impedance 16-ohms setting compared to the 64-ohms on the iESL.

BTW, to avoid linear distortion (avoiding deviations in frequency response), the output impedance of the source amp should be as close to zero as possible. Impedance ZMA files are attached below. Format is frequency, impedance, electrical phase.

 

So given the performance of the solid-state speaker power amps, how would a flea-powered headphone amp do?



It turns out, very very well. Now this is it! However the following conditions needed to be met:

• The max gain (x32) at 16-ohms was needed to obtain sufficient volume. The gain on this amp isn't that high. Given the impedances seen, there shouldn't be a huge problem driving this load- and there wasn't. The volume knob needed to be cranked all the way to max on the 64-ohms setting. That's not a good idea with this amp. It simply won't put out that much voltage without starting to go into clipping.
• Listening needs to be done at moderate volumes. Distortion (not crackling) but softening and a kind of warbling, could be heard at higher SPLs. But really, it's not that bad. I'd say that I can crank it up quite a bit to moderately loud volume levels even!

--

For reference, and also a sanity check, let's compare measurements to the included freebie Koss amp:

Koss freebie amp -> ESP95X
Distortion (1kHz ~104db SPL)


EC 45 --> iESL --> ESP95X
Distortion (1kHz ~104db SPL)


Basically what we are seeing here is that the EC 45 amp is going into clipping. This is a SET amp which will naturally have higher second order than third order distortion, unless it's clipping. It's one into clipping alright. The big question is how far into clipping. I don't think it's that far since third order isn't even 10% yet. So let's try backing this off a little bit, say 5db.

EC 45 --> iESL --> ESP95X
Distortion (1kHz ~99db SPL)


Ahh, so that's a bit more like it. However, it would seem that the amp may still be having its effect on the distortion with the higher second order harmonics throughout. The amp is being pushed hard, but it's not clipping anymore. Let's see what happens when we push down the volume just a bit more, say about 4db.

EC 45 --> iESL --> ESP95X
Distortion (1kHz ~95db SPL)


Looks pretty good. I don't think we can get any lower because we are running into the microphone / ambient level noise floors now. It's like what I said before I took the measurements, it sounds fine as long as you don't crank it up to insane levels. And no, I don't crank it up until I can hear distortion because that is a really stupid test (I prefer to keep my hearing intact).

The only thing I miss with this combo is the lack of slam. But then again, I'm comparing from memory to the DIY T2, and these are electrostatic headphones after all which inherently do not slam. And the lack of slam isn't any worse compared to STAX amps (I haven't heard their latest revamped models) or most other DIY electrostatic amps. The most important part is that the seductive qualities of SET are largely intact, and as such, I prefer this to any of the solid-state speaker amps I have around, even though I can get more volume and a bit more slam (particularly with the Nectar Hive).

 

Note: Bias cranked up to 640V for the ESP95X.

The iFi Pro iESL is somewhat pricey, but in my estimation well worth it for those who want flexibility and choice with respect to amplifiers for electrostatic headphones. The iESL allows one to leverage almost the entire universe of speaker power amplifiers and a good portion of the more powerful headphone amplifiers. Personally, I much prefer this approach given the lack of choice, poor value, or generally either antiseptic or soft presentation of existing dedicated electrostatic headphone amplifiers. The iESL exceeded my expectations for sound quality, committing no sins of commision and very minor sins of omission.

I would like to thank @Victoria Pickles at iFi for helping to make this happen, Darin at DarinFongAudio for the ESP950 to STAX adapter, @zerodeefex for the camera and lenses so my photos looks slightly better (or at least have nice bokeh), and @CEE TEE for the pyrate figurine. Finally, it's members like you who make this possible. Your support makes a difference at NPR SBAF.

 

Well if we think about it, the "capacitance" is still within 10% until we get below 60Hz. The extremes are tough for any transformer. I had initially worried about distortion at the extremes, but the measurements don't show this to be an issue. Under good and proper conditions, the distortion of the drivers will dominate over that of the transformer (and amp).

There is a switch on my EC 45 which selects between the 4-ohm and 8-ohm taps which affects both the speaker and headphone outs. TBH, I'm not even sure if there's a significant difference in the output impedance. You are correct that the "higher-Z" tap outputs more voltage. Not only that, but less distortion for a given voltage. I only use the 4-ohm taps for the big JBL speakers with the 4-ohm woofer where the impedance drops to as low as 2.8 ohms somewhere in the lows.

It's very cool to be able to listen to a electrostatic headphones from a SET amp. I cannot emphasize this enough. I believe the EC Studio B would do even better than my custom EC 45 with the iESL.

Of course a good classic push-pull such this this could be very interesting...