That should be good.
The star quad is an improved version of the twisted pair.
The screen then adds on some further attenuation of internally generated interfernce. Connecting the screen to the chassis is correct.
I too used a handmade star quad, but unscreened.
I was asked for numbers.
Would you care to measure and post:
shorted input, output mVDC, output mVac
open input, output mVdc, output mVac
For me I find that a DMM that has a lowest scale setting of 199.9mVdc and 199.9mVac gives sufficient resolution of mV for inaudibility with normal <90dB/W @ 1m speakers.
Now back to internally and externally generated interference.
Internally generated interference will be at a minimum when when the amp is quiescent (no input and low Rs).
Externally generated interference will be transferred in via all the connected cables: mains power, speaker and signal input.
Attenuating this interference at the chassis entry points helps. Screening the internal cables helps.
Big Loop Areas as a result of separating the Flow and Return wires makes both internally generated and externally generated interference worse.
Hi AndrewT,
I think I did a good job twisting and separating the cabling. Only the sense cabling is on it's own in the amp. And speaker return, but it is impossible to position things in a way that lets you combine those.
I will post those measurements you asked for tonight.
Could you take a really close look at my cabeling and see if anything could be done Better? I have learned a lot from you guys over here
and I am not done learning
I think I did a good job twisting and separating the cabling. Only the sense cabling is on it's own in the amp. And speaker return, but it is impossible to position things in a way that lets you combine those.
I will post those measurements you asked for tonight.
Could you take a really close look at my cabeling and see if anything could be done Better? I have learned a lot from you guys over here
and I am not done learning 
Could be Sjoerd that on your other amps HF is a bit rolled off. The finals represent a very high capacitive load. On an amp with a higher output Z, this would tend to soften the HF.
One 'trick' I read about was to insert a low value resistor - say 0.5 Ohms - in series with the speaker. This more emulates a 'softer' tube sound wrt HF if that's what you are after.
I am using B&W703's - quite a benign load and 89 dB efficiency - I am very happy with the sound. Some reviewers noted that they were a bit 'bright', but I am not getting a harsh sound out of them. They have fantastic imaging I must say.
Anyway, hope things settle down and you get to enjoy your new amp.
One 'trick' I read about was to insert a low value resistor - say 0.5 Ohms - in series with the speaker. This more emulates a 'softer' tube sound wrt HF if that's what you are after.
I am using B&W703's - quite a benign load and 89 dB efficiency - I am very happy with the sound. Some reviewers noted that they were a bit 'bright', but I am not getting a harsh sound out of them. They have fantastic imaging I must say.
Anyway, hope things settle down and you get to enjoy your new amp.
If I am reading your pics correctly, it looks like you have a twisted triplet for the power cabling into the amp PCB. That's good. You are stuck with the three separated leads going to the respective connectors. I could not see any smple way to avoid that.
The speaker leads seem to be separated.
I think there is a note in the build guide that the speaker Flow and Return should be twisted.
i.e. from the amp PCB to the protection PCB and from the protection PCB to the terminals.
The thin red sense wire carries a tiny current. There may be a very small advantage to run this along with the speaker pair back to the protection PCB. Then it will separate to it's own terminal. I tapped into the protection PCB and used that to connect to the sense terminal. It works.
The speaker leads seem to be separated.
I think there is a note in the build guide that the speaker Flow and Return should be twisted.
i.e. from the amp PCB to the protection PCB and from the protection PCB to the terminals.
The thin red sense wire carries a tiny current. There may be a very small advantage to run this along with the speaker pair back to the protection PCB. Then it will separate to it's own terminal. I tapped into the protection PCB and used that to connect to the sense terminal. It works.
I am driving the metal tweeter & metal bass/mid of the AE1 with this NX. I don't find any treble exaggeration or grittiness.
I have dropped the RF input filter down a bit to ~240kHz and incorporated the LF input filter @ ~1.6Hz
Last edited:
Hi AndrewT,
Thanks for taking the time to check and give me feedback. I was wondering how to go about those speaker wires. So since twisting for 3 centimeters did not seem like a winner to me, I selected a speaker wire that has teflon insulation and for better insulation. Same goes for the thin red wire. That is no ordinary red wire But it has no screening. I don't quite get what you mean by tapping in to the psu.... Could you try and be more clear? A picture would help
Thanks for taking the time to check and give me feedback. I was wondering how to go about those speaker wires. So since twisting for 3 centimeters did not seem like a winner to me, I selected a speaker wire that has teflon insulation and for better insulation. Same goes for the thin red wire. That is no ordinary red wire
But it has no screening. I don't quite get what you mean by tapping in to the psu.... Could you try and be more clear? A picture would help Bonsai and AndrewT,
It's not that the high's sound grainy or bad. Believe it or not, I do have very good ears.... And that is also a challenge I describe the high's as slightly metallic like. Not bad, grainy or gritty. This is also what I had in my first Hiraga build with metal film resistors. The resistors used in the NX are good ones. But I can't help thinking they are the problem here. Shame you guys cannot come over and have a listen
So it is not a really big problem, more a detail
It's not that the high's sound grainy or bad. Believe it or not, I do have very good ears.... And that is also a challenge
I describe the high's as slightly metallic like. Not bad, grainy or gritty. This is also what I had in my first Hiraga build with metal film resistors. The resistors used in the NX are good ones. But I can't help thinking they are the problem here. Shame you guys cannot come over and have a listen So it is not a really big problem, more a detail
Thanks AndrewT, this is what I was thinking when I see single wires run like that, it has been discussed greatly by the likes of Scott Wurser and others on the Blowtorch thread the reason to use twisted pairs and more to limit problems of egress of signals into a large loop. But it looks pretty to have everything look so neat, just not a great design decision, I was surprised when Bonsai didn't say something about it.
ps. SjoerdSmits, there are more than a few reasons that teflon insulation is not a great solution, also discussed ad nauseaum over on the Blowtorch thread.
ps. SjoerdSmits, there are more than a few reasons that teflon insulation is not a great solution, also discussed ad nauseaum over on the Blowtorch thread.
Last edited:
No change for the mVac tells me the stage is not too noise sensitive to seeing an open input and the high impedance that presents. That is down to using the 10k input impedance.
The noise levels are high.
0.2mVac in the quieter channel is about 12dB more than I would expect. I got slightly worse than this in my initial build on the bench, but it kept varying. It had a bit to do with cabling. The NX amplifier seems to be very sensitive to cabling layout. I eventually got it down to showing 0.0mVac (meaning less than 0.05mVac) with careful routing.
0.5mVac is about 20dB more. This noisier channel is a failure in my book.
That could be hiss, or hum, or buzz, or all three.
What can you see on the scope? What does it sound like using headphones connected via a 2200uF capacitor?
The change in output offset when shorted, or open, is typical of DC coupling.
I never do that.
The output offset is affected by temperature. as temeprature varies the offsets will vary. In an LTP style stage a setting with poor offset gets much worse as temperature drifts, whereas a low initial setting due to very well matched LTP sides (currents and volts) ends up with a low offset tempco, i.e. it does not drift as much when temperatures change. I don't have any expertise in this complementary single ended input stage.
You and others have not metioned that output offset varies with the volume contol setting. Are you all using a buffered vol pot?
m = milli multiplier = 10^-3
M = Mega multiplier = 10^6
The noise levels are high.
0.2mVac in the quieter channel is about 12dB more than I would expect. I got slightly worse than this in my initial build on the bench, but it kept varying. It had a bit to do with cabling. The NX amplifier seems to be very sensitive to cabling layout. I eventually got it down to showing 0.0mVac (meaning less than 0.05mVac) with careful routing.
0.5mVac is about 20dB more. This noisier channel is a failure in my book.
That could be hiss, or hum, or buzz, or all three.
What can you see on the scope? What does it sound like using headphones connected via a 2200uF capacitor?
The change in output offset when shorted, or open, is typical of DC coupling.
I never do that.
The output offset is affected by temperature. as temeprature varies the offsets will vary. In an LTP style stage a setting with poor offset gets much worse as temperature drifts, whereas a low initial setting due to very well matched LTP sides (currents and volts) ends up with a low offset tempco, i.e. it does not drift as much when temperatures change. I don't have any expertise in this complementary single ended input stage.
You and others have not metioned that output offset varies with the volume contol setting. Are you all using a buffered vol pot?
m = milli multiplier = 10^-3
M = Mega multiplier = 10^6
Last edited:
Noise
You should short your meter leads together on AC mV to make sure you are reading zero. I would also only consider any OP mV readings taken with a meter as valid with a load in place - i.e. 8 Ohms. Remember on the nx-Amp, when the power is OFF, the output is open circuit due to the SSR.
Secondly, its likely if you are not hearing anything out of you speakers with your ear right against the cone when there is no input, that you are just seeing noise pickup around the output wiring and your meter leads. Cable dressing will help you to get a lower reading on your meter, but it is doubtful that it will have any practical impact on your amplifier performance.
I don't believe the nx or sx amps are anymore sensitive to wiring layout than other amplifiers. To get low noise in any amp, you need to adhere to good wiring practices - minimize loop areas, keep inputs away from outputs bunch wires appropriately - like the PSU and speaker wires for example.
I good trick that I use is to connect a pair of headphones directly to the output (with no input of course). Since headphones are 30 to 40 dB more sensitive than typical speakers, you can really get a handle on how good your cabling layout is – any hum will be clearly audible.
Move the cables around, experiment with bunching, and once you have a 'null' then fix your wiring layout. I used this technique on all my nx and sx builds and they are extremely quiet.
Keep in mind, if you are using a mV AC meter on the 200mV scale and trying to discern 0.5mV readings you are at 0.25% of FS, and at 0.05mV you certainly will have serious problems taking any sensible readings.
Offsets
I have not had offsets as high as you report. With no input signal, the current that flows through the 10k input bias resistor is primarily difference between the base bias currents of the input transistor pairs. If these are matched, along with the 2nd stage buffer pairs and you are using high beta devices like the BC5x7C types, then the input base current required is about 2 uA which flows mainly from one base to the other base . With zero input, its mainly the difference current that flows in the 10k resistor and that should be very low i.e. <<<2 uA.
The second point is that it is not normal in an amplifier to have the DC bias resistance changing drastically - e.g. going from 50 or 100 Ohms which is a typical preamp output - to 10k. So, you will get a change in output offset if you do this, although as I have said, 50/70 mV looks high to me. I would also expect your source output to be AC coupled or to specify very low offsets - I think I mention this latter point in the write up and that the offset should be well under 1mV). If offsets are a concern, then use a cap on the input or, if DC coupling, connect your preamp and adjust the offset – but make sure you preamp offset is zero if you do this.
If you are going to use a pot on the front of either of these two amps, then you should use a DC blocking capacitor - how to do this is shown in the back of the construction write up. If you don’t, you will get pot scratch noise.
You should short your meter leads together on AC mV to make sure you are reading zero. I would also only consider any OP mV readings taken with a meter as valid with a load in place - i.e. 8 Ohms. Remember on the nx-Amp, when the power is OFF, the output is open circuit due to the SSR.
Secondly, its likely if you are not hearing anything out of you speakers with your ear right against the cone when there is no input, that you are just seeing noise pickup around the output wiring and your meter leads. Cable dressing will help you to get a lower reading on your meter, but it is doubtful that it will have any practical impact on your amplifier performance.
I don't believe the nx or sx amps are anymore sensitive to wiring layout than other amplifiers. To get low noise in any amp, you need to adhere to good wiring practices - minimize loop areas, keep inputs away from outputs bunch wires appropriately - like the PSU and speaker wires for example.
I good trick that I use is to connect a pair of headphones directly to the output (with no input of course). Since headphones are 30 to 40 dB more sensitive than typical speakers, you can really get a handle on how good your cabling layout is – any hum will be clearly audible.
Move the cables around, experiment with bunching, and once you have a 'null' then fix your wiring layout. I used this technique on all my nx and sx builds and they are extremely quiet.
Keep in mind, if you are using a mV AC meter on the 200mV scale and trying to discern 0.5mV readings you are at 0.25% of FS, and at 0.05mV you certainly will have serious problems taking any sensible readings.
Offsets
I have not had offsets as high as you report. With no input signal, the current that flows through the 10k input bias resistor is primarily difference between the base bias currents of the input transistor pairs. If these are matched, along with the 2nd stage buffer pairs and you are using high beta devices like the BC5x7C types, then the input base current required is about 2 uA which flows mainly from one base to the other base . With zero input, its mainly the difference current that flows in the 10k resistor and that should be very low i.e. <<<2 uA.
The second point is that it is not normal in an amplifier to have the DC bias resistance changing drastically - e.g. going from 50 or 100 Ohms which is a typical preamp output - to 10k. So, you will get a change in output offset if you do this, although as I have said, 50/70 mV looks high to me. I would also expect your source output to be AC coupled or to specify very low offsets - I think I mention this latter point in the write up and that the offset should be well under 1mV). If offsets are a concern, then use a cap on the input or, if DC coupling, connect your preamp and adjust the offset – but make sure you preamp offset is zero if you do this.
If you are going to use a pot on the front of either of these two amps, then you should use a DC blocking capacitor - how to do this is shown in the back of the construction write up. If you don’t, you will get pot scratch noise.
Last edited:
Since your amplifier is claimed to be no worse than most normal power amplifiers with regard to noise+hum, you have not said as much, but is 0.5mVac as output noise+hum indicative of an assembly fault?
For comparison 0.5mVac is roughly equivalent to Ein = 95nV/rtHz, when the amplifier gain is 37times
0.2mVac ~ 38nV/rtHz
An amplifier that reads 0.0mVac meaning less than 0.05mVac (if the reading is believed) would have an indicative Ein<9.5nV/rtHz.
Some very quiet power amplifiers claim a noise + hum at the output of around 20uVac (-91.8dBu) and this for a lower gain amplifier of 28times (+28.9dB) would have Ein~5nV/rtHz
For comparison 0.5mVac is roughly equivalent to Ein = 95nV/rtHz, when the amplifier gain is 37times
0.2mVac ~ 38nV/rtHz
An amplifier that reads 0.0mVac meaning less than 0.05mVac (if the reading is believed) would have an indicative Ein<9.5nV/rtHz.
Some very quiet power amplifiers claim a noise + hum at the output of around 20uVac (-91.8dBu) and this for a lower gain amplifier of 28times (+28.9dB) would have Ein~5nV/rtHz
Thanks! Very important info!
Now, i checked all connections, and my fuse holders were actually NOT holding the fuses on the boards, changed them. I soldered all power lines directly on the amp boards, they were not very tight. The larger ones on the PSU are fine. Next I put in screened dual awg solid core speaker cables(that was not an easy job) and I re-routed some cabling according to the tips everyone has given me. I agree with Bonsai, if you do not hear anything, chances are there will be no impact in overall performance. So, Right channel is all done. When the left is also done, I will measure again and adjust the DC ofset with DCB1 preamp Connected. My DCB1 has 0.3mv DC ofset, so maybe I should use decoupling on the input of the amp. How, I do not know. If someone could post a schematic, that would be great!
Keep you posted!
Now, i checked all connections, and my fuse holders were actually NOT holding the fuses on the boards, changed them. I soldered all power lines directly on the amp boards, they were not very tight. The larger ones on the PSU are fine. Next I put in screened dual awg solid core speaker cables(that was not an easy job) and I re-routed some cabling according to the tips everyone has given me. I agree with Bonsai, if you do not hear anything, chances are there will be no impact in overall performance. So, Right channel is all done. When the left is also done, I will measure again and adjust the DC ofset with DCB1 preamp Connected. My DCB1 has 0.3mv DC ofset, so maybe I should use decoupling on the input of the amp. How, I do not know. If someone could post a schematic, that would be great!
Keep you posted!
I think if you can get to 200 uV of wide band noise and hum on the output of a power amp, you are doing very well. 0.5mV on the output is probably wire routing - especially if you are getting a much lower reading on the other channel. As I mentioned earlier, use a pair of headphones on the output to optimize the wiring layout.
Evidently the connections were to blame for some of the "problems". Ofset is now 0.6 mv with preamp connected. Also the better speaker cable makes a huge difference Very happy with the sound!
Thank you all for all the help!
I could use some gain on my preamp, but that is just not possible on a DCB1.... So I will have to live with the fact that the amp will not be able to go full power. But it is loud enough for me.
Very happy with the sound! Thank you all for all the help!
I could use some gain on my preamp, but that is just not possible on a DCB1.... So I will have to live with the fact that the amp will not be able to go full power. But it is loud enough for me.
The NX is already a higher gain Power Amplifier than most....................
I could use some gain on my preamp, but that is just not possible on a DCB1.... So I will have to live with the fact that the amp will not be able to go full power. But it is loud enough for me.
It has 2k7/5 for the upper leg and 15r for the lower leg giving a gain of 37times.
Mosy Power Ampliifirs have a gain between 20times and 30times.
Most domestic set ups have too much gain using that typical amp gain of 20 to 30 times.