• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Tear down of Massdrop THX AAA 789 Headphone Amp

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,298
Location
China
You can gain up to10dB of s/n, depending on the input opamp (FET vs. bipolar) by going to a lower value pot for the volume control (10k to 1k in this case). However, as has been stated, 1k ohm is generally considered too low by the industry as a whole for an input impedance. If you use a buffer with high input impedance to drive the 1k pot, then it works great. So, no, I don't think it's a good idea to just swap the pot for a much lower value. Some kind of low noise, low distortion buffer has to be added to drive the low Z pot. Now the issue becomes one of designing an awesome buffer that can handle the full output from a DAC or CD player, which is another story.
Old standard is not the limitation of modern good performing equipment. If you want the best performance you have you pair it with suitable front end. In this case, 1kohm isn't really asking too much. Sure I get it, versatility is something to care about. But in some cases, majority of modern/or very new dacs are very capable of doing 2v to 1kohm.

On the input buffer, not necessarily the best idea because interference can definitely go in and then be buffered, 50hz main hum being pucked up by cable is definitely a thing. So imo it's still better to have low input impedance. Maybe not 1kohm, 2kohm is good choice too, if input buffer is used/ or often time combined with gain stage.

It's somewhat a trade off but not necessarily. Because most mid tier dacs only has -110db something thd+n. The top tier ones have -125db. So relatively speaking the output buffer still produce lower distortion than whole system.

Also I guess that's not the main problem with 1k pot. Many dac has 100ohm output impedance meaning we lose 1/10 of the signal level. I don't know why they have it. But to me that's the only thing I consider whether to use 1k pot or not.

In the end, from my measurements and testing, 1kohm is an acceptable option. For most 10k is fine too. But 50k or 100k? Not acceptable in my book.
 

Dirk Wright

Member
Joined
Apr 28, 2018
Messages
23
Likes
13
Well, you can do whatever you want with your system. No one cares about that. You're just expressing your personal opinion. I'm speaking about general industrial practice, and I can tell you that the standard input impedance is 10k or higher. There are many source devices out there that cannot drive 1k at low distortion, and it would be unwise to manufacture a device that had a low input impedance like that.
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,298
Location
China
Well, you can do whatever you want with your system. No one cares about that. You're just expressing your personal opinion. I'm speaking about general industrial practice, and I can tell you that the standard input impedance is 10k or higher. There are many source devices out there that cannot drive 1k at low distortion, and it would be unwise to manufacture a device that had a low input impedance like that.
I am not talking about industry standard practice but actually no bullshit conclusion supported by my measurements. Most (more like all) dacs I measured can drive 1k load perfectly.
And the benefits are reduced interference and low noise. All measurable. All tested.

And again the original question is asking to change the 10k pot to 100k pot. And that's 0 benefit.
 
Last edited:

maxxevv

Major Contributor
Joined
Apr 12, 2018
Messages
1,872
Likes
1,964
I am not talking about industry standard practice but actually no bullshit conclusion supported by my measurements. Most (more like all) dacs I measured can drive 1k load perfectly.
And the benefits are reduced interference and low noise. All measurable. All tested.

Have you shown that here or somewhere ?
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,298
Location
China
Have you shown that here or somewhere ?
Ahh there wasn't a comparison test. But there are measurements measured with 1k ohm pot. On mu2p thread the graphs are actually going through my amp but without load.
Also in neurochrome hp-1 thread the measurements are done with 1k ohm pot.
I did some test while I was testing the su-8. There isn't much difference with through the amp and directly going to the measurement input.
https://audiosciencereview.com/foru...ents-of-musiland-mu2-plus-dac-amp.6964/page-3
https://audiosciencereview.com/foru...ochrome-hp-1-high-performance-amp.5324/page-9
 

maxxevv

Major Contributor
Joined
Apr 12, 2018
Messages
1,872
Likes
1,964
Ahh there wasn't a comparison test. But there are measurements measured with 1k ohm pot. On mu2p thread the graphs are actually going through my amp but without load.
Also in neurochrome hp-1 thread the measurements are done with 1k ohm pot.
I did some test while I was testing the su-8. There isn't much difference with through the amp and directly going to the measurement input.
https://audiosciencereview.com/foru...ents-of-musiland-mu2-plus-dac-amp.6964/page-3
https://audiosciencereview.com/foru...ochrome-hp-1-high-performance-amp.5324/page-9

I seriously don't see the correlation with all you said here and the ones you linked other than what you said you measured.

I'll leave it as that, electronics isn't my domain, shall leave it to the experts here on these forums.
 

trl

Major Contributor
King of Mods
Joined
Feb 28, 2018
Messages
1,978
Likes
2,542
Location
Iasi, RO
The potentiometer will connect amplifier's input directly to the ground when is maxed out (10 KOhms in this case). So no additional noise will get into our ears, unless we're using extremely sensitive cans and an amp with a very high gain (also, pot to the middle...that means around 2-3 o 'clock). However, the Johnson thermal noise will lower the dynamic of the input signal, although...given the extremely low volume when pot is around minimum, I don't really think it matters.

Choosing the best gain (if amp has adjustable gain) for a specific pair of headphones is probably more important than lowering pot's value. Worth mentioning that if a DAC has DC-blocking capacitors on its outputs, then having a very low pot value will most likely cause a low-end frequency roll-off, because the output caps with the potentiometer's resistance from the amplifier will create a high-pass filter. Basically, the DAC will always "see" amp's potentiometer value.
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,298
Location
China
The potentiometer will connect amplifier's input directly to the ground when is maxed out (10 KOhms in this case). So no additional noise will get into our ears, unless we're using extremely sensitive cans and an amp with a very high gain (also, pot to the middle...that means around 2-3 o 'clock). However, the Johnson thermal noise will lower the dynamic of the input signal, although...given the extremely low volume when pot is around minimum, I don't really think it matters.

Choosing the best gain (if amp has adjustable gain) for a specific pair of headphones is probably more important than lowering pot's value. Worth mentioning that if a DAC has DC-blocking capacitors on its outputs, then having a very low pot value will most likely cause a low-end frequency roll-off, because the output caps with the potentiometer's resistance from the amplifier will create a high-pass filter. Basically, the DAC will always "see" amp's potentiometer value.
The maximum noise is at 50% position. You meed to take all the use cases in considerations. 250ohm equivalent noise resistance for 1k pot, 2.5k for 10k pot.
Also the benefit is not only the thermal noise but immunity to interference. It's very overlooked and absolutely crucial. 50hz/60hz main noise will be very easily picked up by cable.
The gain is always 0db in my case.
Sure this is like the first world problem in audio. I never said it's crucial to do this i just rather to do that. The original problem was to change to a 100k pot.
About the dc blocking capacitor, yes you are right that's something to consider.
edit: most dac with blocking capacitor use 22uf or higher and most likely there will be a series resistor after that. So still like what I said, the voltage drop due to output impedance of the dac is more a consideration than the unwanted high pass filtering.
 
Last edited:

trl

Major Contributor
King of Mods
Joined
Feb 28, 2018
Messages
1,978
Likes
2,542
Location
Iasi, RO
We're so off-topic here. :)

http://www.sengpielaudio.com/calculator-noise.htm: 1 KOhm means about 0.6uV RMS @25C, while 10K Ohm means about 1.8uV RMS @2V RMS (similar with a very low-noise audio opamp), calculated up to 20KHz badwidth, and that's the total noise injected if we remove the grounding pin of the pot and pot's cursor is at the minimum (max. carbon resistance). In real life, we don't hear background noise above 12KHz (predominant noise if at the midrange and low-end freqs), so if we stretch the bandwidth we get even lower noise values.

Pot. to the middle means about 3 o'clock which in most circumstances is insanely loud.

About the DC-blocking caps from DAC's outputs: I've see so many audiophile DIY-ers replacing 100uF polarised or 10-20uF non-polarised caps (usually Nichicon or WIMA) with 1uf Mundorf and...funny thing, soundstage increased and musicality got seriously added on top of every sound. Well, no test to prove that, of course. :)

In-house test with sensitive 121dB/V SPL 16-Ohms IEMs: I got a +20dB headamp (I guess that makes it 10X of internal gain) with a 50K pot in series with a 20K resistor per channel. When I shorted the 20K resistor (volume was maxed out) in my headphones the background noise decreased a little bit (really tiny difference to my ears), so I had those resistors replaces with much lower value (220 Ohms actually); indeed, thermal noise in front of a high gain amplifier could be a problem. I also tried to see what's the background noise diff. between the 50K pot. to the min. vs. the same 50K pot to the max. and it was not a big difference (but it was a bit higher than the 20K resistors above)....and I 'm speaking about 50K potentiometer (ALPS Blue). Also, on +10dB of gain I was almost unable to hear any background noise when volume to the max., so for unity gain (0dB) doesn't even worth trying. :) I did this test just to see if my ears will benefit of a background noise decrease in case of swapping the expensive and hard to find 4-gangs 50K ALPS with a 10K ALPS, but given the fact that the background noise was only heard with 121dB/V SPL IEM's and only when using the internal +20dB high gain of the amplifier (20V RMS theoretically, so pretty close to 150dB SPL with my sensitive IEMs when 2V on inputs) I considered that this upgrade makes no sense to me.
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,298
Location
China
We're so off-topic here. :)

http://www.sengpielaudio.com/calculator-noise.htm: 1 KOhm means about 0.6uV RMS @25C, while 10K Ohm means about 1.8uV RMS @2V RMS (similar with a very low-noise audio opamp), calculated up to 20KHz badwidth, and that's the total noise injected if we remove the grounding pin of the pot and pot's cursor is at the minimum (max. carbon resistance). In real life, we don't hear background noise above 12KHz (predominant noise if at the midrange and low-end freqs), so if we stretch the bandwidth we get even lower noise values.

Pot. to the middle means about 3 o'clock which in most circumstances is insanely loud.

About the DC-blocking caps from DAC's outputs: I've see so many audiophile DIY-ers replacing 100uF polarised or 10-20uF non-polarised caps (usually Nichicon or WIMA) with 1uf Mundorf and...funny thing, soundstage increased and musicality got seriously added on top of every sound. Well, no test to prove that, of course. :)

In-house test with sensitive 121dB/V SPL 16-Ohms IEMs: I got a +20dB headamp (I guess that makes it 10X of internal gain) with a 50K pot in series with a 20K resistor per channel. When I shorted the 20K resistor (volume was maxed out) in my headphones the background noise decreased a little bit (really tiny difference to my ears), so I had those resistors replaces with much lower value (220 Ohms actually); indeed, thermal noise in front of a high gain amplifier could be a problem. I also tried to see what's the background noise diff. between the 50K pot. to the min. vs. the same 50K pot to the max. and it was not a big difference (but it was a bit higher than the 20K resistors above)....and I 'm speaking about 50K potentiometer (ALPS Blue). Also, on +10dB of gain I was almost unable to hear any background noise when volume to the max., so for unity gain (0dB) doesn't even worth trying. :) I did this test just to see if my ears will benefit of a background noise decrease in case of swapping the expensive and hard to find 4-gangs 50K ALPS with a 10K ALPS, but given the fact that the background noise was only heard with 121dB/V SPL IEM's and only when using the internal +20dB high gain of the amplifier (20V RMS theoretically, so pretty close to 150dB SPL with my sensitive IEMs when 2V on inputs) I considered that this upgrade makes no sense to me.
Distortion doesn't make sense too. You can't hear 0.1% below. Why bother low distortion?
The thd+n is a sum of cumulated(integral) of noise + the distortion. If we care about thd+n a tiny bit. At normal listening level any distortion will be drown in the noise. If you want to achieve good thd+n you have to reduce noise. Other wise by your standard, 0.1% distortion is good enough. Any more performance is useless, only higher power will benefit. Don't tell me noise is not important but distortion is important, this is double standard. My standard is to achieve best thd+n at any level at any volume pot position.

Also 10k input impedance will pick up a lot of 50hz noise. Like shown in the jds atom graph, there is a high 50/60hz mains hum caused by cable. It's easily eliminated by 1k input pull down resistor. This is not the power supply design issue. If you like to see mains hum then it's fine.

Also the volume at max equals no potentiometer noise, it's essentially shorted. Remember that as well.

In the end, if you care only about "audible transparency" go ahead, I won't argue things in electronics. Almost every product here is essentially audibly transparent. You may only need more power in some cases.

If you care about absolute performance, then your argument is contradictory. Quit double standard and pursue best measurable performance.
 

Dirk Wright

Member
Joined
Apr 28, 2018
Messages
23
Likes
13
The thermal noise advantage of a 1k pot only occurs when the input devices are bipolar. For FET inputs, it is the opposite, lower noise is achieved with a higher value pot. The optimum source impedance varies a lot based on the input device. So, no one can make a blanket statement that lower value pots are always better. It is not true. See figure 1, for example: https://www.onsemi.com/pub/Collateral/AN-6601.pdf.pdf
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,298
Location
China
The thermal noise advantage of a 1k pot only occurs when the input devices are bipolar. For FET inputs, it is the opposite, lower noise is achieved with a higher value pot. The optimum source impedance varies a lot based on the input device. So, no one can make a blanket statement that lower value pots are always better. It is not true. See figure 1, for example: https://www.onsemi.com/pub/Collateral/AN-6601.pdf.pdf
I guess you can't read right? Sorry to be so harsh.
Resistance thermal noise will always be there. Lower noise isn't achieved with high value pot.
 

Attachments

  • Screenshot_20190422-015157.jpg
    Screenshot_20190422-015157.jpg
    796.3 KB · Views: 164
Last edited:

Dirk Wright

Member
Joined
Apr 28, 2018
Messages
23
Likes
13
I guess you can't read right? Sorry to be so harsh.
Resistance thermal noise will always be their. Lower noise isn't achieved with high value pot.

You're not sorry for "being harsh". You've been nothing but that all along. Stop lying about it.

Facts are facts. You don't understand my point, which is that if you have an amplifier with FET inputs, then changing the 10k pot for a 1k pot will not make much difference. The reduction in noise is much more for an amplifier with bipolar inputs. Sorry you do not understand that simple fact.
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,298
Location
China
You're not sorry for "being harsh". You've been nothing but that all along. Stop lying about it.

Facts are facts. You don't understand my point, which is that if you have an amplifier with FET inputs, then changing the 10k pot for a 1k pot will not make much difference. The reduction in noise is much more for an amplifier with bipolar inputs. Sorry you do not understand that simple fact.
Does not make much difference =/= "lower noise is achieved with a higher value pot"
Stop quibbling.

Also you can always parallel fet device to lower the voltage noise. Not even mentioning, there was no context of using fet in the first place.
 

Dirk Wright

Member
Joined
Apr 28, 2018
Messages
23
Likes
13
Does not make much difference =/= "lower noise is achieved with a higher value pot"
Stop quibbling.

Also you can always parallel fet device to lower the voltage noise. Not even mentioning, there was no context of using fet in the first place.

You started off making a blanket statement that a 1k pot will always result in lower total circuit noise than a 10k. I'm telling you that you are wrong. It's as simple as that. It depends on the amplifiers used, as well as other circuit characteristics. For vacuum tube inputs, there is no difference, for example. Likewise, a 10k pot on an FET input op amp will usually have lower noise than the same pot on a bipolar input op amp. So, you can't say that lowering the value of the pot makes a difference in all circuits.
 
Last edited:

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,298
Location
China
You started off making a blanket statement that a 1k pot will always result in lower total circuit noise than a 10k. I'm telling you that you are wrong. It's as simple as that. It depends on the amplifiers used, as well as other circuit characteristics. For vacuum tube inputs, there is no difference, for example. Likewise, a 10k pot on an FET input op amp will usually have lower noise than the same pot on a bipolar input op amp. So, you can't say that lowering the value of the pot makes a difference in all circuits.
When did I say that?
Origin post was: Do not use high resistance pot(100k alps blue to replace the 10k in thx 789).
I would rather to replace to 1k pot.
Any amplifier with big pots(100k) are shit in my book.

The context is thx 789. It uses opa1602. Also when I'm talking about 1k pot it's immediately the last bit of performance.

If you want lowest thd+n at all volume position, 1k pot is important, unless you use a lot of gain then use inverting amplifier to step down. If have a noisy circuit, then it's out of context in the first place, it's noisy. It's not the other way around. It's not "1k pot will always improve" but "if highest performance is needed, 1k pot is very helpful and important".

Also the interference noise picked up through cable will be reduced by 1k pot. This is always true. This is a different topic than thermal noise.
 
Last edited:

304290

Member
Joined
Dec 27, 2018
Messages
72
Likes
81
I have one of these that will not power back on after being left on overnight. In my search I found a few who have had issues with the power on circuit. Is it possible to bypass this circuit to power it directly?
 

elira

Member
Joined
Feb 8, 2018
Messages
73
Likes
46
I have one of these that will not power back on after being left on overnight. In my search I found a few who have had issues with the power on circuit. Is it possible to bypass this circuit to power it directly?
You can bypass the auto-off that some people have issues with. It has a switch in the back. That way it won't turn of by itself.
 

304290

Member
Joined
Dec 27, 2018
Messages
72
Likes
81
I have one of these that will not power back on after being left on overnight. In my search I found a few who have had issues with the power on circuit. Is it possible to bypass this circuit to power it directly?


Maybe you didn't read the post correctly. I can't turn it off if it won't power on to begin with. But thanks for the comment.
 
Top Bottom