I guess you are the minority in the 0.000001% who got lucky
iFi audio iDSD Diablo - A portable reference done our way!
- Thread starter iFi audio
- Start date
Yup, 2 black label iMicro IDSD and both have imbalance at low volumes. The iCan Pro is okay so is the Gryphon and the xCan.
Thorsten Loesch
Member of the Trade: Studio RaumklanG, Studio Cocktail
Hi,
I suspect had you instead gotten an iDSD micro Black or Signature (largely the same except 4.4mm jack) you would have been fine.
That was because the "Diablo" deleted the iEMatch and S&M insisted making the output balanced at any cost, which made the 4.4mm balanced out much worse for headphone use than it would have been with a single-ended amp and a 4.4mm jack connected in what I called "Single Ended Compatible Balanced" (aka S-Balanced), which is ALWAYS a better choice for headphones.
And there are good reasons why in addition to using an SE Amp you want iEMatch build in.
To cover the whole volume control thing (again), you get a number of options.
A classic "Pot(entiometer)" is usually sonically the best, but it has limits. Typically a Pot is intended to cover a 40dB range between all the way up and near the lower travel limits. A typical specification of channel error for a high quality product (ALPS RK-168) reads like this: "−60dB to −40dB 5dB max. −40dB to 0dB 3dB max."
A 3dB error at -40dB is significant, 5dB beyond this quite high.
So realistically you get a reliable range of 0dB to -40dB with most potentiometers. Also, the "12 O'Clock" setting commonly corresponds to -20dB. In many applications this is adequate. Until the advent of very high sensitivity IEM's that was even the case for Headphone Amplifiers.
Let's take the iDSD micro as example.
We have a 10k volume control (to lower noise), fed from a DAC with 2V output @ 0dBFS for PCM and 1V @ 0dBDSD for DSD. The official acceptance specification for the volume control was < 2dB Channel Imbalance at -40dB and up. Most of the Black label's should have a Volume Pot that passed this test.
If we set the volume control to 12 O'Clock we get 200mV @ 0dBFS or 100mV @ 0dBDSD at the pot output and the pot has a noise resistance of around 900 Ohm, giving 0.54uV Noise and a SNR of 111dB for PCM and 105dB for DSD.
The Amplifier in the iDSD micro has a gain of 0dB (no gain) in ECO mode and a self noise of around 1uV. Given how random noise adds, we get around 1.2uV noise or 98dB unweighted SNR @ 100mV out.
If we have (say) a Headphone with 118dB/1V sensitivity this setting gets us a noise level of 0dB and a peak SPL of 98dB (DSD) or 104dB (PCM) which is a good loud listening volume level with a typical average SPL of 89dB (PCM) or 83dB (DSD).
For reference, an Etymotic 4 XR is 118dB/1V and Focal Utopia ~ 116dB/1V. This is a "normal sensitivity" IEM. This combination is noise free and allows a good reserve of volume for quiet DSD or PCM recordings or to "Rock Out" with a +/-20dB adjustment range and range left to compensate modest variations in headphone sensitivity.
The two additional gain steps (which also adjust the power supply voltage) are +12dB and +24dB, with the same "12 O'Clock" setting this is respectively an output at the Headphone Jack of 800/400mV (PCM/DSD) and 3.2V/1.6V respectively.
It means the "design center" setting of the 12dB gain step targets a Headphone with around 106dB/1V sensitivity, which are Headphones like the Sennheiser HD-800/820 and Audeze LCD-4 (2/3 as well) etc. within a few decibel. Again, there will be no audible noise and a good adjustment range. In other words, a perfect match.
For the 24dB gain step, the target headphone is 94dB/1V, which are headphones like the AKG K240 Monitor 600 Ohm, Hifiman Susvara etc.
Now, if we use the same system with a balanced output and 4.4mm Jack, first all the gain is 6dB higher on the balanced output (as is the output voltage into higher impedance's) and noise from the amplifier is ~9dB higher. So the same headphone would need a volume control setting lower than before and would have much more noise, both of which are "bad things" in my books.
But what if we use some super sensitive IEM's, say Campfire Audio Andromeda's at a whopping 140dB/1V?
On the iDSD micro with the volume control at 12 O'Clock and PCM playing that is a whopping 126dB Peak SPL. Even using the FULL reliable -20 dB adjustment from our Volume Control leave us much louder (106dB peak vs 98dB) than the Etymotic Example. And we will have a background noise (from a generally rather low noise Headphone Amplifier) of around 28dB absolute which is a quiet but very noticeable hiss.
So the volume control at the best usable setting DESPITE using 0dB Gain setting leaves us TOO LOUD with nowhere to go and there is noticeable noise. If we use a balanced Amp like in the Diablo, noise would be 37dB absolute and minimum reliable volume put's us at 112dB SPL peak, which is VERY LOUD. I would argue that the original iDSD micro case is borderline unusable, but Diablo, forget it.
Let's set iEMatch to Ultra, which is -24dB. This means on the 12 O'Clock volume control setting we get 102dB SPL peak and around 4dB absolute noise, with good adjustment range into both directions. So that is why, to really drive very high sensitivity IEM's well, you need very low noise.
The noise on the iDSD micro, of 1uV is 120dB SNR vs 1V or 94dB at 50mV (which is about the maximum in the 50V SNR chart over at ASR), doing much better (not just 1 or 2 dB) is not easy. BUT, with iEMatch at -24dB the noise is dropped to 62.5nV which on the 50mV SNR Chart would be 118dB!!! Now of course, this IS CHEATING, this is precisely why I designed iEMatch. To get much less noise than 1uV in a Headphone Amp is very hard, but using a "cheater" plug (iEMatch as stand alone device) or a cheater switch (iEMatch build in).
Of course, TANSTAAFL applies, the SNR and volume control range improvement of iEMatch doesn't come for free. The key price is that iEMatch raises output impedance.
The < 1 Ohm of the original iEMatch in "Ultra" is "mostly harmless". Now I did not design the 4.4mm iEMatch, whoever did never understood the design principles behind the iEMatch. A 4.4mm iEMatch to work with Diablo should have offered 30dB attenuation with 1 Ohm output impedance. That would have put the whole system on an even footing with iDSD micro and iEMatch at "Ultra".
As is, with, 3.6 Ohm rated (and if I read right over 5 Ohm actal) output impedance the changes of frequency response with something like the CA Andro's will be unacceptable. I'd say the 4.4mm is "broken by design" and should not be used at all. And yes, given that iFi build in the same "faulty by design" iEMatch into the Diablo 2 which retains the amplifier of the Diablo, the same comments apply.
So, whereas the iDSD micro Black Label or Signature can indeed handle ANY IEM and Headphone under the sun well, out of the box. The Diablo never will, not even once you add 4.4mm iEMatch. And iFi's S&M should have known that and written their add copy accordingly as they had specified the Diablo features and were warned (by me) about the consequences).
Why claim something patently untrue? If you simply copy/paste text from previous products without thinking and nobody who understand the product proof-reads the copy, that is what you get. It is not even intentionally misleading, it's just not wanting to put in the minimum effort into the job to get the correct result.
So, anone with the same application as CH23, buy a 2nd hand iDSD micro, use 6.3mm Jacks or get the iDSD Signature if you need 4.4mm jack.
I feel greatly sorry for the overall poor experience some of iFi's customers, such as CH23, have been having. He is absolutely correct in being upset with the waste of money and and time he experienced. For those customers that are sufficiently "mainstream" this may not matter.
Morale of all this? If you buy an audio product, make sure to test yourself before you buy or buy with a right to return and if you do not get the result you want within the timeframe to return, ruthlessly return the product and get something else.
Thor
PS, if there had been an "xDSD micro" in 2018/2019 it would have had an iEMatch with electronic instead of mechanical switches and transparently tied into the Volume control and most likely "Headphone Fingerprinting" to measure the headphone impedance at multiple frequencies and thus be able to identify headphones and either (if the headphone was used before) use the last settings from this headphone (including Bass Boost and Crossfeed on top of gain etc.) or set sensible defaults for new headphones.
Seeing what iFi delivers instead five years later...
I suspect had you instead gotten an iDSD micro Black or Signature (largely the same except 4.4mm jack) you would have been fine.
That was because the "Diablo" deleted the iEMatch and S&M insisted making the output balanced at any cost, which made the 4.4mm balanced out much worse for headphone use than it would have been with a single-ended amp and a 4.4mm jack connected in what I called "Single Ended Compatible Balanced" (aka S-Balanced), which is ALWAYS a better choice for headphones.
And there are good reasons why in addition to using an SE Amp you want iEMatch build in.
To cover the whole volume control thing (again), you get a number of options.
A classic "Pot(entiometer)" is usually sonically the best, but it has limits. Typically a Pot is intended to cover a 40dB range between all the way up and near the lower travel limits. A typical specification of channel error for a high quality product (ALPS RK-168) reads like this: "−60dB to −40dB 5dB max. −40dB to 0dB 3dB max."
A 3dB error at -40dB is significant, 5dB beyond this quite high.
So realistically you get a reliable range of 0dB to -40dB with most potentiometers. Also, the "12 O'Clock" setting commonly corresponds to -20dB. In many applications this is adequate. Until the advent of very high sensitivity IEM's that was even the case for Headphone Amplifiers.
Let's take the iDSD micro as example.
We have a 10k volume control (to lower noise), fed from a DAC with 2V output @ 0dBFS for PCM and 1V @ 0dBDSD for DSD. The official acceptance specification for the volume control was < 2dB Channel Imbalance at -40dB and up. Most of the Black label's should have a Volume Pot that passed this test.
If we set the volume control to 12 O'Clock we get 200mV @ 0dBFS or 100mV @ 0dBDSD at the pot output and the pot has a noise resistance of around 900 Ohm, giving 0.54uV Noise and a SNR of 111dB for PCM and 105dB for DSD.
The Amplifier in the iDSD micro has a gain of 0dB (no gain) in ECO mode and a self noise of around 1uV. Given how random noise adds, we get around 1.2uV noise or 98dB unweighted SNR @ 100mV out.
If we have (say) a Headphone with 118dB/1V sensitivity this setting gets us a noise level of 0dB and a peak SPL of 98dB (DSD) or 104dB (PCM) which is a good loud listening volume level with a typical average SPL of 89dB (PCM) or 83dB (DSD).
For reference, an Etymotic 4 XR is 118dB/1V and Focal Utopia ~ 116dB/1V. This is a "normal sensitivity" IEM. This combination is noise free and allows a good reserve of volume for quiet DSD or PCM recordings or to "Rock Out" with a +/-20dB adjustment range and range left to compensate modest variations in headphone sensitivity.
The two additional gain steps (which also adjust the power supply voltage) are +12dB and +24dB, with the same "12 O'Clock" setting this is respectively an output at the Headphone Jack of 800/400mV (PCM/DSD) and 3.2V/1.6V respectively.
It means the "design center" setting of the 12dB gain step targets a Headphone with around 106dB/1V sensitivity, which are Headphones like the Sennheiser HD-800/820 and Audeze LCD-4 (2/3 as well) etc. within a few decibel. Again, there will be no audible noise and a good adjustment range. In other words, a perfect match.
For the 24dB gain step, the target headphone is 94dB/1V, which are headphones like the AKG K240 Monitor 600 Ohm, Hifiman Susvara etc.
Now, if we use the same system with a balanced output and 4.4mm Jack, first all the gain is 6dB higher on the balanced output (as is the output voltage into higher impedance's) and noise from the amplifier is ~9dB higher. So the same headphone would need a volume control setting lower than before and would have much more noise, both of which are "bad things" in my books.
But what if we use some super sensitive IEM's, say Campfire Audio Andromeda's at a whopping 140dB/1V?
On the iDSD micro with the volume control at 12 O'Clock and PCM playing that is a whopping 126dB Peak SPL. Even using the FULL reliable -20 dB adjustment from our Volume Control leave us much louder (106dB peak vs 98dB) than the Etymotic Example. And we will have a background noise (from a generally rather low noise Headphone Amplifier) of around 28dB absolute which is a quiet but very noticeable hiss.
So the volume control at the best usable setting DESPITE using 0dB Gain setting leaves us TOO LOUD with nowhere to go and there is noticeable noise. If we use a balanced Amp like in the Diablo, noise would be 37dB absolute and minimum reliable volume put's us at 112dB SPL peak, which is VERY LOUD. I would argue that the original iDSD micro case is borderline unusable, but Diablo, forget it.
Let's set iEMatch to Ultra, which is -24dB. This means on the 12 O'Clock volume control setting we get 102dB SPL peak and around 4dB absolute noise, with good adjustment range into both directions. So that is why, to really drive very high sensitivity IEM's well, you need very low noise.
The noise on the iDSD micro, of 1uV is 120dB SNR vs 1V or 94dB at 50mV (which is about the maximum in the 50V SNR chart over at ASR), doing much better (not just 1 or 2 dB) is not easy. BUT, with iEMatch at -24dB the noise is dropped to 62.5nV which on the 50mV SNR Chart would be 118dB!!! Now of course, this IS CHEATING, this is precisely why I designed iEMatch. To get much less noise than 1uV in a Headphone Amp is very hard, but using a "cheater" plug (iEMatch as stand alone device) or a cheater switch (iEMatch build in).
Of course, TANSTAAFL applies, the SNR and volume control range improvement of iEMatch doesn't come for free. The key price is that iEMatch raises output impedance.
The < 1 Ohm of the original iEMatch in "Ultra" is "mostly harmless". Now I did not design the 4.4mm iEMatch, whoever did never understood the design principles behind the iEMatch. A 4.4mm iEMatch to work with Diablo should have offered 30dB attenuation with 1 Ohm output impedance. That would have put the whole system on an even footing with iDSD micro and iEMatch at "Ultra".
As is, with, 3.6 Ohm rated (and if I read right over 5 Ohm actal) output impedance the changes of frequency response with something like the CA Andro's will be unacceptable. I'd say the 4.4mm is "broken by design" and should not be used at all. And yes, given that iFi build in the same "faulty by design" iEMatch into the Diablo 2 which retains the amplifier of the Diablo, the same comments apply.
So, whereas the iDSD micro Black Label or Signature can indeed handle ANY IEM and Headphone under the sun well, out of the box. The Diablo never will, not even once you add 4.4mm iEMatch. And iFi's S&M should have known that and written their add copy accordingly as they had specified the Diablo features and were warned (by me) about the consequences).
Why claim something patently untrue? If you simply copy/paste text from previous products without thinking and nobody who understand the product proof-reads the copy, that is what you get. It is not even intentionally misleading, it's just not wanting to put in the minimum effort into the job to get the correct result.
So, anone with the same application as CH23, buy a 2nd hand iDSD micro, use 6.3mm Jacks or get the iDSD Signature if you need 4.4mm jack.
I feel greatly sorry for the overall poor experience some of iFi's customers, such as CH23, have been having. He is absolutely correct in being upset with the waste of money and and time he experienced. For those customers that are sufficiently "mainstream" this may not matter.
Morale of all this? If you buy an audio product, make sure to test yourself before you buy or buy with a right to return and if you do not get the result you want within the timeframe to return, ruthlessly return the product and get something else.
Thor
PS, if there had been an "xDSD micro" in 2018/2019 it would have had an iEMatch with electronic instead of mechanical switches and transparently tied into the Volume control and most likely "Headphone Fingerprinting" to measure the headphone impedance at multiple frequencies and thus be able to identify headphones and either (if the headphone was used before) use the last settings from this headphone (including Bass Boost and Crossfeed on top of gain etc.) or set sensible defaults for new headphones.
Seeing what iFi delivers instead five years later...
Last edited:
Thorsten Loesch
Member of the Trade: Studio RaumklanG, Studio Cocktail
Hi,
First, ANY Potentiometer will have imbalance at very low volume. There was an issue with iDSD micro in general due to poor adherence to formal specification by a supplier, However MOST Black Label should be fine. Was the imbalance above ~ 8 O'Clock?
With iCAN Pro, if you wind down the volume far enough you you will get imbalance, see ASR:
There is variability in production of the Pot's. iCAN Pro uses ALPS RK168 Series.
All products in the "X" series uses a digitally stepped attenuator instead of a potentiometer. So these have minimal channel imbalance and a much wider adjustment range.
Had I done the follow-up's on the iDSD micro "my way" the same volume control would have been used, just two in parallel to reduce impedance and noise and the next generation "Pro's" would have used them too, just more in parallel again.
Thor
First, ANY Potentiometer will have imbalance at very low volume. There was an issue with iDSD micro in general due to poor adherence to formal specification by a supplier, However MOST Black Label should be fine. Was the imbalance above ~ 8 O'Clock?
With iCAN Pro, if you wind down the volume far enough you you will get imbalance, see ASR:
There is variability in production of the Pot's. iCAN Pro uses ALPS RK168 Series.
All products in the "X" series uses a digitally stepped attenuator instead of a potentiometer. So these have minimal channel imbalance and a much wider adjustment range.
Had I done the follow-up's on the iDSD micro "my way" the same volume control would have been used, just two in parallel to reduce impedance and noise and the next generation "Pro's" would have used them too, just more in parallel again.
Thor
Haven't listened to my Diablo in a little while, but started again recently and still love the sound of this thing. Listening with the Atrium Closed currently and it's awesome.
Do you still have these issues?
Last edited:
I wish ifi uses 18650 batteries instead of batteries that can only be replaced by them and it is very costly
Enleum HPA23RM use 18650
- Joined
- Mar 20, 2013
- Posts
- 16,614
- Likes
- 12,419
I will certainly pass this suggestion along at my next meeting, Thanks for suggesting it!
Cheers!!
Too high for my budget, shanling h7 seems more like it but the tonality is not suited for me. Cayin C9 is probably the one. Unless an ifi with 18650 batteries come up off course.
I miss the Diablo.
Why should one consider the Version2 over the original?
What has been improved on the circtuit?
Why should one consider the Version2 over the original?
What has been improved on the circtuit?
Last edited:
Yup, USB-C (which I would have preferred vs that clunky male USB-A in the Diablo), and Bluetooth (which I personally do not like to see.. additional distortions for something I would not use). Can bluetooth be turned off definitively btw?Screenshot of the significant part of a response from Thor previously.
There is an on and off and pairing button on the rear with bluetooth on being displayed on the front indicator light.
Users who are viewing this thread
