Geshelli Lab Erish coming soon

[solderdude]

How would an amplifier affect 'microdetails' ?

 

[Bluess]

How would an amplifier affect 'microdetails' ?

I have no idea, I'm not an electrician. Maybe you can help enlighten me why I didn't noticed them before and proper way to test this out so I can try again and comeback to you.

 

[minmsee]

nice! first I thought it looked ugly like a diy project, but now I think it looks cool something very different on your desktop getting tired of all black stuff. I got myself a red M500 and I love the color and now I am leaning towards that purple box to go with it. Is there any pictures that show how it looks like with different led color settings?

Here you are. It looks very cool to my eyes because it's a true giant killer!

 

[solderdude]

I have no idea, I'm not an electrician.

I have no idea either, that's why I asked.

 

[MalinYamato]

The T1 = 600 Ohm
The Erish is around 0.4W in 600 Ohm and 0.75W in 300 Ohm.
However, when the design is really fully balanced as advertised and given it has a 4-gang potmeter changes are when connecting it to an RCA out of the source via an RCA to XLR cable the output power may well be 1/4 of the above mentioned output power in that case.

When a headphone amp can deliver 6W in 32Ohm it can deliver around 0.3W in 600 Ohm (probably a bit more like 0.4 or 0.5W in 600 Ohm).

If I were Geshelli I would also list output powers in various impedances.

I expected those numbers to be much much lower. Indeed fantastic numbers and I do the same reflection as why Gashelli doesnt adverice those great numbers. I have XLR outs on my M500 so I suppose I am fine.

 

[MalinYamato]

Here you are. It looks very cool to my eyes because it's a true giant killer!

lovely,,, this box is such a fancy toy.

 

[JustAnandaDourEyedDude]

if thats so, then this little box despite its 1w@32 should be enough for all my future needs. My knowledgeable in audio tech is poor so my questions my appear odd> Considering that Eirhs is enough for 600ohm phones, I am a little confused why you need wopping 6w@32 or 3w@32 such as the performance of those AAA offerings.

SMSL M500 HPA (6.35mm SE output) : max power per channel 570 mW into 32 ohms; 63 mW into 300 ohms
Geshelli Erish Balanced HPA (4-pin XLR BAL output) : max power per channel 1000 mW into 32 ohms load

Denon AH-D7200 will be louder than Sundara
HiFiMan Sundara : 37 ohm impedance; 94 dB/(1)mW sensitivity
Argon MkIII / Fostex T50rp Mk3 : 50 ohm impedance; 92 dB/(1)mW sensitivity
Beyerdynamic T1 (2. Gen) : 600 ohm impedance; 102 dB/(1)mW sensitivity

Max SPL calculation for M500 driving Sundara : power at knob max volume is 570 x 32/37 = 493 mW. Corresponding max SPL is 94 + 10 x log(493mW/1mW) = 121 dB.
Erish driving T50: max power is 1000 x 32/50 = 640 mW. max SPL is 92 + 10 x log(640mW/1mW) = 120 dB.

[Edit: updated table below with Topping L30 HPA just released. It is more powerful than Erish and gets four times as loud as M500, so +6dB relative to M500.]
[Edit 2: corrected T1's SPL driven by Erish, based on @solderdude's estimate.]
Proceeding similarly, table of max vol knob SPL (rounding to nearest 1dB):
[HP] Sund T50 T1
M500 121 118 117
Erish_ 123 120 119 128
TpL30 127 124 123

So what are all your future needs? From the table above, it seems you have no need for a new amp. You might be better off waiting until you get the next set of headphones, and get a new HPA only if the M500 is unable to drive them to satisfactorily high levels without noticeable distortion.

If you use an amp that delivers max power of 6W into 32 ohms, then it will give about an extra 8 dB as compared with Erish for each headphone (so 131 dB SPL from Sundara). Basically, Erish is almost twice as loud (+2.5dB) as M500, and a 6W-32ohm amp would be about ten times as loud (+10dB) as M500. Please try out your favorite music through Sundara driven by M500 at max volume knob position for a few seconds, at your own risk, to help you judge what the numbers in the table mean. Or else please consult Harmful Noise Levels to see what dB levels you want out of your headphones.

Amir likes a HPA to max out above 100 mW into 300 ohms, which the Erish does (106 mW into 300 ohms if it maintains the voltage), and which helps a little with less sensitive headphones.

 

[Jimbob54]

SMSL M500 HPA (6.35mm SE output) : max power per channel 570 mW into 32 ohms; 63 mW into 300 ohms
Geshelli Erish Balanced HPA (4-pin XLR BAL output) : max power per channel 1000 mW into 32 ohms load

Denon AH-D7200 will be louder than Sundara
HiFiMan Sundara : 37 ohm impedance; 94 dB/(1)mW sensitivity
Argon MkIII / Fostex T50rp Mk3 : 50 ohm impedance; 92 dB/(1)mW sensitivity
Beyerdynamic T1 (2. Gen) : 600 ohm impedance; 102 dB/(1)mW sensitivity

Max SPL calculation for M500 driving Sundara : power at knob max volume is 570 x 32/37 = 493 mW. Corresponding max SPL is 94 + 10 x log(493mW/1mW) = 121 dB.
Erish driving T1: max power is 1000 x 32/600 = 53 mW. max SPL is 102 + 10 x log(53mW/1mW) = 119 dB.

[Edit: updated table below with Topping L30 HPA just released. It is more powerful than Erish and gets four times as loud as M500, so +6dB relative to M500.]
Proceeding similarly, table of max vol knob SPL (rounding to nearest 1dB):
[HP] Sund T50 T1
M500 121 118 117
Erish_ 123 120 119
TpL30 127 124 123

So what are all your future needs? From the table above, it seems you have no need for a new amp. You might be better off waiting until you get the next set of headphones, and get a new HPA only if the M500 is unable to drive them to satisfactorily high levels without noticeable distortion.

If you use an amp that delivers max power of 6W into 32 ohms, then it will give about an extra 8 dB as compared with Erish for each headphone (so 131 dB SPL from Sundara). Basically, Erish is almost twice as loud (+2.5dB) as M500, and a 6W-32ohm amp would be about ten times as loud (+10dB) as M500. Please try out your favorite music through Sundara driven by M500 at max volume knob position for a few seconds, at your own risk, to help you judge what the numbers in the table mean. Or else please consult Harmful Noise Levels to see what dB levels you want out of your headphones.

Amir likes a HPA to max out above 100 mW into 300 ohms, which the Erish does (106 mW into 300 ohms if it maintains the voltage), and which helps a little with less sensitive headphones.

@MalinYamato mentioned wanting Argons - I tested my T60 (same driver) with Erish, no problems at all driving. Sounded good. Would need to get balanced cable to use with Erish obviously. But yes, the L30 would seem to be an even better bang for buck.

My quick and dirty trial of M500 HPA with T60 didnt sound good to my ears- but not a proper test at all. You may be right about sticking with M500 , getting new cans then buying new amp to suit if unsatisfied . Should always be the default option in fact, unless your trigger finger is itchy like mine.

 

[solderdude]

Erish driving T1: max power is 1000 x 32/600 = 53 mW. max SPL is 102

I suspect the Erish is around 0.4W in 600 Ohm (balanced) = 128dB SPL peak and will go louder than the L30.
In your calculations you forgot to take into account the power rating of the internal DCDC converter which is the limiting factor for low impedance headphones.

 

[Jimbob54]

I suspect the Erish is around 0.4W in 600 Ohm (balanced) = 128dB SPL peak and will go louder than the L30.
In your calculations you forgot to take into account the power rating of the internal DCDC converter which is the limiting factor for low impedance headphones.

Lets put it this way, I dont need to go to high gain to go plenty loud with my T1. But thats the same with them and my 789 (also running balanced) - low gain plenty. I wouldnt want to run them off a smartphone, but they are not that hard to drive

 

[JustAnandaDourEyedDude]

@MalinYamato mentioned wanting Argons - I tested my T60 (same driver) with Erish, no problems at all driving. Sounded good. Would need to get balanced cable to use with Erish obviously. But yes, the L30 would seem to be an even better bang for buck.

My quick and dirty trial of M500 HPA with T60 didnt sound good to my ears- but not a proper test at all. You may be right about sticking with M500 , getting new cans then buying new amp to suit if unsatisfied . Should always be the default option in fact, unless your trigger finger is itchy like mine.

I saw you tested Erish with both T60 and T1.2 and found it had plenty of power, which are the best checks that should reassure her that the Erish should be a great option. She still seemed anxious, so I put together that table to show that all the headphones she has or is considering are not super hard to drive to satisfying levels. I've suffered from the itchy trigger finger syndrome too , but I am in recovery now, or so I claim!

I suspect the Erish is around 0.4W in 600 Ohm (balanced) = 128dB SPL peak and will go louder than the L30.
In your calculations you forgot to take into account the power rating of the internal DCDC converter which is the limiting factor for low impedance headphones.

I did not forget, rather I simply did not know about DCDC converters and a resulting big droop here in the I-V curve of the Erish. For lack of knowing any better, I was assuming constant amp voltage across the load range. I will edit the post to correct that estimate. Thank you for explaining the complicating factor at play in the Erish. May I say that your posts here on ASR and the content of your website (which I was reading well before I came across ASR) all have a SINAD well over 200 dB , and I always find them enlightening and enjoyable to read!

 

[solderdude]

The Erish has a +/- 12V DCDC converter inside. This limits the max output power.
The amplifier in the amp thus has a +/-12V rails. Generally this means the maximum available voltage swing is just below +/-12V as the used opamps are probably not the so-called rail-to-rail types.
My estimation is +/-11V might be available at each driver output. This means it can put out 7.8Vrms.
But this is a balanced amplifier so there is another driver that outputs -7.8Vrms. Combined (balanced) this is about 15Vrms.
In 600 Ohm this is about 0.4W. As this is lower than 1W the DCDC converters will provide the full 12V.
This is how I came to my estimation.

The L30 has a higher voltage rails (I assume +/-15V) and can reach +/-13V = 9.2Vrms = 0.14W in 600 Ohm (so 4.5dB lower than the balanced out of the Erish). This is still impressive as it is not a balanced output and still is capable of 100mW+ in 600 Ohm.

Above 1W the +/-12V power rails of the Erish will drop below 12V where the L30 does not so this one can provide more power in low impedance headphones (like the 37 Ohm Sundara)
That dropping of the +/-12V voltage rails will drop the output power of the Erish.
However, as music consists of peaks and these will be drawn from internal capacitors the short peak power delivery of the Erish may be above 1W for short (music) moments.

The problem here is that output power is measured as continuous power and not also the 'old' music power output numbers which used to be given for power amps.

Perhaps it would be fun to measure peak output power capabilities with the Erish in several impedances as well as continuous.
@Geshelli doesn't provide these numbers alas (they easily could).
Power levels for 16, 32, 50, 120, 300 and 600 Ohm would look nice in their specifications.

 

[JustAnandaDourEyedDude]

Thank you for the detailed and insightful analysis. It lets me make sense of the characteristics of the Erish. The L30 certainly has impressive power for a SE design. The Erish would be excellent for high impedance high efficiency headphones like the T1 2nd. Yes, Geshelli are too modest, listing only the power into 32 ohms in the specs; they do themselves a disservice. Interesting how the caps provide higher power transients giving more headroom even into 32 ohms.

The Erish has a +/- 12V DCDC converter inside. This limits the max output power.
The amplifier in the amp thus has a +/-12V rails. Generally this means the maximum available voltage swing is just below +/-12V as the used opamps are probably not the so-called rail-to-rail types.
My estimation is +/-11V might be available at each driver output. This means it can put out 7.8Vrms.
But this is a balanced amplifier so there is another driver that outputs -7.8Vrms. Combined (balanced) this is about 15Vrms.
In 600 Ohm this is about 0.4W. As this is lower than 1W the DCDC converters will provide the full 12V.
This is how I came to my estimation.

The L30 has a higher voltage rails (I assume +/-15V) and can reach +/-13V = 9.2Vrms = 0.14W in 600 Ohm (so 4.5dB lower than the balanced out of the Erish). This is still impressive as it is not a balanced output and still is capable of 100mW+ in 600 Ohm.

Above 1W the +/-12V power rails of the Erish will drop below 12V where the L30 does not so this one can provide more power in low impedance headphones (like the 37 Ohm Sundara)
That dropping of the +/-12V voltage rails will drop the output power of the Erish.
However, as music consists of peaks and these will be drawn from internal capacitors the short peak power delivery of the Erish may be above 1W for short (music) moments.

The problem here is that output power is measured as continuous power and not also the 'old' music power output numbers which used to be given for power amps.

Perhaps it would be fun to measure peak output power capabilities with the Erish in several impedances as well as continuous.
@Geshelli doesn't provide these numbers alas (they easily could).
Power levels for 16, 32, 50, 120, 300 and 600 Ohm would look nice in their specifications.

 

[Jimbob54]

Thank you for the detailed and insightful analysis. It lets me make sense of the characteristics of the Erish. The L30 certainly has impressive power for a SE design. The Erish would be excellent for high impedance high efficiency headphones like the T1 2nd. Yes, Geshelli are too modest, listing only the power into 32 ohms in the specs; they do themselves a disservice. Interesting how the caps provide higher power transients giving more headroom even into 32 ohms.

A rumour I read somewhere is that is exactly what it was designed for . 600 ohm Beyers.

 

[solderdude]

Interesting how the caps provide higher power transients giving more headroom even into 32 ohms.

This kind of depends on how big the reservoir caps behind the DCDC converters are. When these are 2200uF peak power could be higher than if it were 100uF or so. This all would depend on what the max. capacitance is after the DCDC converter but they can easily increase this by adding inductors after the DCDC converter.

 

[JohnYang1997]

A rumour I read somewhere is that is exactly what it was designed for . 600 ohm Beyers.

Hmmm. Not quite. There are many 600 Ohm beyers being single ended only. How about making a single ended amp with 400mW into 600 Ohm?

 

[solderdude]

Already designed one

of course this isn't a SINAD driven design but 'borked in certain ways' on purpose.

 

[JohnYang1997]

Already designed one

of course this isn't a SINAD driven design but 'borked' on purpose.

Almost almost
https://www.diyaudio.com/forums/ven...erformance-cmos-audio-op-amp-post6263956.html

 

[JustAnandaDourEyedDude]

This kind of depends on how big the reservoir caps behind the DCDC converters are. When these are 2200uF peak power could be higher than if it were 100uF or so. This all would depend on what the max. capacitance is after the DCDC converter but they can easily increase this by adding inductors after the DCDC converter.

Hmmm. Not quite. There are many 600 Ohm beyers being single ended only. How about making a single ended amp with 400mW into 600 Ohm?

This causes a question to pop up in my mind that I have had from time to time but never asked for fear of sounding ignorant. When an amp is driving a reactive headphone or speaker, does any energy get returned to caps in the output circuit of the amp at any time, or is it all stored in the load so that the power output from the amp is always positive or zero? Does this depend on whether the reactance of the load is due to inductance in a crossover network or just due to mechanical inertia in the drivers? I have a confused picture in my mind, because I am not familiar with the electro-mechanics of such a situation. Basically, I am asking whether the interface between amplifier and load is designed to be one-way only for transfer of electrical energy, or can energy be transferred from the load to the amp during transients, though of course the average power is transferred from the amp to the load? Perhaps the amp itself has negligible electrical inertia, and it responds practically instantaneously to changing demands in the load and so energy is never absorbed by the amp even at high audio frequencies? I guess this is off-topic for the thread, but on the other hand, two experts in amp design are posting here.

 

[Jimbob54]

Hmmm. Not quite. There are many 600 Ohm beyers being single ended only. How about making a single ended amp with 400mW into 600 Ohm?

Good point.