Music listening test
Inspired by recent 'music listening' reviews of another amp that have its root in Russia (from another thread), I decided to do little music listening with my half-baked amp too.
Keep in mind that this is just a board, without speaker protection circuit,
so being hesitant to risk damaging my JBL speakers, I connected small speaker I harvested from the old 80s boombox my neighbor put out in the garbage the other day..
Also - this is just one channel being tested. Imagine how much better it would sound, with music playing from both channels!
The sound of this amplifier can be described as balanced, detailed, rich and very energetic, and it also features stunning transparency.
Plays cheerful, energetic and organized.
The famous song "Dick in the box" by Justin Timberlake delighted with its swiftness, clear rhythm, a very large and perfectly structured stage, as well as fun and drive.
"Happy Birthday" - something that everyone here knows and can relate to - performed by the folk orchestra, left a very good impression
with the excellent elaboration of the musical scene, the authenticity of the voices of the instruments and their groups, the large-scale dynamics and special energy inherent in this interpretation of the popular work.
The song "Da Da Da" of the Trio group pleased with a biting, elastic rhythm, articulated with a natural and perfectly focused vocals,
depth, concentration and weight of the bass, detail and mobility of the treble.
Inspired by recent 'music listening' reviews of another amp that have its root in Russia (from another thread), I decided to do little music listening with my half-baked amp too.
Keep in mind that this is just a board, without speaker protection circuit,
so being hesitant to risk damaging my JBL speakers, I connected small speaker I harvested from the old 80s boombox my neighbor put out in the garbage the other day..
Also - this is just one channel being tested. Imagine how much better it would sound, with music playing from both channels!
The sound of this amplifier can be described as balanced, detailed, rich and very energetic, and it also features stunning transparency.
Plays cheerful, energetic and organized.
The famous song "Dick in the box" by Justin Timberlake delighted with its swiftness, clear rhythm, a very large and perfectly structured stage, as well as fun and drive.
"Happy Birthday" - something that everyone here knows and can relate to - performed by the folk orchestra, left a very good impression
with the excellent elaboration of the musical scene, the authenticity of the voices of the instruments and their groups, the large-scale dynamics and special energy inherent in this interpretation of the popular work.
The song "Da Da Da" of the Trio group pleased with a biting, elastic rhythm, articulated with a natural and perfectly focused vocals,
depth, concentration and weight of the bass, detail and mobility of the treble.
Last edited:
Minek, you win the internet today, because you used ‘Dick in a box’ to test a newly built amp. Well done.
One of the best SNL skits of all time. 😀 JT was always fabulous on that show.
Love the rest of your impressions, as well.
One of the best SNL skits of all time. 😀 JT was always fabulous on that show.
Love the rest of your impressions, as well.
Last edited:
My post was related to this thread - post #20 and subsequent posts.
Check it out.
Amazing how these 'evaluation' statements from the review match the songs I selected 🙂
Check it out.
Amazing how these 'evaluation' statements from the review match the songs I selected 🙂
Yes, I reviewed that thread after I posted. 😀 Good stuff.
Now if you don’t mind, I have to move to another room to keep up with the mobility of the treble.
Now if you don’t mind, I have to move to another room to keep up with the mobility of the treble.
I managed to find spice model for TLE2071 - it seems like very nice and fast, new op-amp, so I decided to use it in my sim, instead of usual LT1056.
It doesn't hurt that TLE is much cheaper than LT.
I tested actual amp earlier - it works fine with TLE2071 just fine.
Sim attached - but new op-amp model needs to be added to LtSpice first, and 'Solver' needs to be changed to 'Alternate' in Tools->Control_Panel->Spice->Engine->Solver
Default solver doesn't like TLE2071's model...
Anyway - results indicate that:
1) Thd and FFT profiles are pretty much the same
2) Rise time on square waves is 15% faster
3) Compensation needs to be slightly changed to make this amp more stable:
Feedback cap C6 needs to be lower (e.g. 5pF), and C12/C13 shunts for the bases of the drivers need to be higher (e.g. 68pF).
Phase margin is slightly lower than before, but still solid. Maybe this could be improved further...
I suspect the same rules will apply to any other faster op-amp.
I run out of LT1056 op-amps, and I guess I'm going to move to TLE2071 from now on, for all new builds.
It doesn't hurt that TLE is much cheaper than LT.
I tested actual amp earlier - it works fine with TLE2071 just fine.
Sim attached - but new op-amp model needs to be added to LtSpice first, and 'Solver' needs to be changed to 'Alternate' in Tools->Control_Panel->Spice->Engine->Solver
Default solver doesn't like TLE2071's model...
Anyway - results indicate that:
1) Thd and FFT profiles are pretty much the same
2) Rise time on square waves is 15% faster
3) Compensation needs to be slightly changed to make this amp more stable:
Feedback cap C6 needs to be lower (e.g. 5pF), and C12/C13 shunts for the bases of the drivers need to be higher (e.g. 68pF).
Phase margin is slightly lower than before, but still solid. Maybe this could be improved further...
I suspect the same rules will apply to any other faster op-amp.
I run out of LT1056 op-amps, and I guess I'm going to move to TLE2071 from now on, for all new builds.
Last edited:
Thank you for sharing your experience. Good news that TLE2071 works well in this circuit. I am confident that it will do in "LMK & PHILIPS Beast" circuit as well. What an epic nickname: "Excalibur". How TLE2071 sounds compared to OPA134?
From my tests it's clear that they all sound the same 🙂
The most economical op-amp is TL071.
Cheap and sounds as good as OPA.
The most economical op-amp is TL071.
Cheap and sounds as good as OPA.
Maybe because of the low signal mode. That's why I'm interested in LMK & PHILIPS, as that amplifier up to the base of output drivers works in low-signal mode.
I would be surprised if any mortal can distinguish OPA134 from TL071 in blind audio test...
Regardless of the mode.
Try it yourself. Get an amp, handful of chips with their name/model masked by the tape (just number them underneath), and put them in the socket one by one, randomly, and try to listen and rate the music. Repeat several times (the more the better) with randomly selected chips.
Or even better - ask someone to help you with changing chips, and you just focus on the listening.
I guarantee there should be no preference pattern in you ratings.
If I'm wrong, I'll agree that the proverbial mobility of the treble is a real thing.
I'm not saying that there is no difference between these chips.
I'm just saying this difference can not be detected in an audio hearing test by a human. Maybe if you are a bat you can do it 🙂 ?
Regardless of the mode.
Try it yourself. Get an amp, handful of chips with their name/model masked by the tape (just number them underneath), and put them in the socket one by one, randomly, and try to listen and rate the music. Repeat several times (the more the better) with randomly selected chips.
Or even better - ask someone to help you with changing chips, and you just focus on the listening.
I guarantee there should be no preference pattern in you ratings.
If I'm wrong, I'll agree that the proverbial mobility of the treble is a real thing.
I'm not saying that there is no difference between these chips.
I'm just saying this difference can not be detected in an audio hearing test by a human. Maybe if you are a bat you can do it 🙂 ?
Last edited:
It needs to be double blind - otherwise the brain cheats (confirmation bias). That means no humans in the room know which device is which during the test. An automated ABX blind tester is a realistic approach, where the test schedule is kept secret until after the listening test.
And of course to be statistically significant you need to run a tediously large number of tests - its time consuming and hard work, but that's how you get real data about human perception abilities.
Mark, so according to your knowledge - human should distinguish TL071 from OPA134 in such test?
What's your opinion?
I did this test with my wife switch between two amps - one supposedly good one, and another one built with 2N3055 outputs.
The both sounded the same to me.
What's your opinion?
I did this test with my wife switch between two amps - one supposedly good one, and another one built with 2N3055 outputs.
The both sounded the same to me.
Last edited:
I have only heard a very small difference between the sounds of opamps once in my life. When I replaced the NE5532 with LME49720NA in my FX502S amplifier. Probably the NE was fake. The difference was barely noticeable. The replacement of the power filter elko (from noname to BC Components) brought a much greater improvement in sound. In AH578 I compared TL071 and LT1056A, I didn't hear any difference in their sounds.
The AH578 has long been the first amplifier to have an opamp at input. The first was QUAD405 in the early eighties that I built. Since the QUAD405 didn’t really deliver the expected sound quality, there is a widespread view that opamps cannot be used to build a really good quality amplifier. Now, however, I find that this is not true. Moreover, it is much easier to build a good sounding amplifier with IC input than with discrete components.
The AH578 has long been the first amplifier to have an opamp at input. The first was QUAD405 in the early eighties that I built. Since the QUAD405 didn’t really deliver the expected sound quality, there is a widespread view that opamps cannot be used to build a really good quality amplifier. Now, however, I find that this is not true. Moreover, it is much easier to build a good sounding amplifier with IC input than with discrete components.
Perhaps some improvement in supply rail filtration of opamps, or even feeding them from a battery, makes a bigger difference
>there is a widespread view that opamps cannot be used to build a really good quality amplifier. Now, however, I find that this is not true. Moreover, it is much easier to build a good sounding amplifier with IC input than with discrete components.
Exactly. It's much EASIER to build good amplifier with an op-amp, rather than with discrete components.
Insisting on using ONLY discrete components is like insisting to only use vacuum tubes.
We all make irrational decisions when it comes to building amps, but it's good to know the reality 🙂
Just BUILDING amps by yourself is irrational enough.
Exactly. It's much EASIER to build good amplifier with an op-amp, rather than with discrete components.
Insisting on using ONLY discrete components is like insisting to only use vacuum tubes.
We all make irrational decisions when it comes to building amps, but it's good to know the reality 🙂
Just BUILDING amps by yourself is irrational enough.
I tested 2nd channel of the amp.
Surprise - it started to oscillate somewhere under 10MGHz range, because I could see it on my oscilloscope.
I considered 2 different actions:
1) play with compensation - most likely driver's shunts 56pF need to higher, or bigger gate stoppers, etc...
2) 'downgrade' op-amp to TL071, as a brute-force quick solution
As you can guess, I went for 2) first.
I hate re-soldering this dense PCB (already did it several times) - I tested TL071, and of course everything works perfect.
Gonna leave it as is, as I spent already too much time on this amp, and I want to get this amp out of the door.
There is nothing wrong with using TL071.
All usual tests went fine.
2nd channel is built exactly the same as the 1st one with one diff:
220uF caps are different (both versions are Nichicons, but different),
and of course mosfets are differently matched - most likely this affects stability...
So I guess there is still a lot of to learn in the general subject "How to build fast mosfet amps with multiple pairs", and how to correctly match them.
Even as I spent 2 full days matching them using different methods, currents, voltages, temperatures, they are still not perfect. Acceptable, but not perfect.
I wonder perhaps these problems are the reason why we don't see Russians building LMK-type amps with mosfets. From all the amps I've seen on Russian sites, haven't seen even one with mosfets.
There must be a reason for it. People build amps even with voltage regulators, so surely someone wold try mosfet LMK in the last 35 years..
So this build is waiting for chassis/heatsink now, once it's finished, will proceed with next one.
With next amp (Philips LMK mongrel) - which will have the same OS, will spend more time with simulations using different op-amps models, to understand better what's going on.
Surprise - it started to oscillate somewhere under 10MGHz range, because I could see it on my oscilloscope.
I considered 2 different actions:
1) play with compensation - most likely driver's shunts 56pF need to higher, or bigger gate stoppers, etc...
2) 'downgrade' op-amp to TL071, as a brute-force quick solution
As you can guess, I went for 2) first.
I hate re-soldering this dense PCB (already did it several times) - I tested TL071, and of course everything works perfect.
Gonna leave it as is, as I spent already too much time on this amp, and I want to get this amp out of the door.
There is nothing wrong with using TL071.
All usual tests went fine.
2nd channel is built exactly the same as the 1st one with one diff:
220uF caps are different (both versions are Nichicons, but different),
and of course mosfets are differently matched - most likely this affects stability...
So I guess there is still a lot of to learn in the general subject "How to build fast mosfet amps with multiple pairs", and how to correctly match them.
Even as I spent 2 full days matching them using different methods, currents, voltages, temperatures, they are still not perfect. Acceptable, but not perfect.
I wonder perhaps these problems are the reason why we don't see Russians building LMK-type amps with mosfets. From all the amps I've seen on Russian sites, haven't seen even one with mosfets.
There must be a reason for it. People build amps even with voltage regulators, so surely someone wold try mosfet LMK in the last 35 years..
So this build is waiting for chassis/heatsink now, once it's finished, will proceed with next one.
With next amp (Philips LMK mongrel) - which will have the same OS, will spend more time with simulations using different op-amps models, to understand better what's going on.
Last edited:
My Philips LMK baby mongrel PCB is slowly being completed, with a single FQA pair, of course, without source resistors, but for safety reasons I designed the GD snubber on the board. This will help reduce the high frequency cross conduction so it may be stable. Tvicol uses the same FQAs in his amp, without a GD zobel, it seems stable, though he uses 330 Ohm gate resistors.
Is there an .asc file for Tvicol's amp?
>with a single FQA pair
Hey, that's cheating 🙂 You are doing it the easy way....
12 big exposed mosfets looks much more cool than only 4 of them 🙂
>he uses 330 Ohm gate resistors.
Bigger stoppers will slow mosfets down. I wanted to try to build fast amp..
>with a single FQA pair
Hey, that's cheating 🙂 You are doing it the easy way....
12 big exposed mosfets looks much more cool than only 4 of them 🙂
>he uses 330 Ohm gate resistors.
Bigger stoppers will slow mosfets down. I wanted to try to build fast amp..
Last edited:
🙂 Surely. Although in my 20 square meters room it's enough for my 91 dB sensitive speakers.