That's is what I would call a discussion.
We present our experiences and we reach conclusions.
Your design decision to locate the Amplifier Output Zobel at the PSU is based on your experience that it is the better of a variety of compromises.
I just don't happen to agree on the compromises. But I don't have any technical data/measurements to show which is better.
We present our experiences and we reach conclusions.
Your design decision to locate the Amplifier Output Zobel at the PSU is based on your experience that it is the better of a variety of compromises.
I just don't happen to agree on the compromises. But I don't have any technical data/measurements to show which is better.
Aksa and Bonsai seem to agree that Pmax of the low value NFB is not a serious issue.
I disagree.
I would change the 5 parallel resistors to 10 series connected resistors giving 6W of dissipation capability. Or as I did on the PCB, changed to 5 series connected paralleled pairs.
I used 220r in parallel to give 110r and 5 groups in series to give 550r.
This needed changes to the copper pads/traces on the PCB.
I did relocate the tempco sensing BJT to the collector lead of the output device. I glued on a sot23 to directly sense the temperature of the Tc as close to Tj as I am able to access.
The standard "gap" did not work for my build.
Access to the VR adjustments on the vertically installed boards is not easy, even in my very large enclosure.
I would mount the PCB horizontally to allow access either through the bottom panel, or with the lid off.
If I were to attempt a redesign (but I won't) I would seriously look at incorporating a DC servo to allow the DC coupled amplifier to operate. The protections for this DC coupled amp are already in the PSU+prot board.
I disagree.
I would change the 5 parallel resistors to 10 series connected resistors giving 6W of dissipation capability. Or as I did on the PCB, changed to 5 series connected paralleled pairs.
I used 220r in parallel to give 110r and 5 groups in series to give 550r.
This needed changes to the copper pads/traces on the PCB.
I did relocate the tempco sensing BJT to the collector lead of the output device. I glued on a sot23 to directly sense the temperature of the Tc as close to Tj as I am able to access.
The standard "gap" did not work for my build.
Access to the VR adjustments on the vertically installed boards is not easy, even in my very large enclosure.
I would mount the PCB horizontally to allow access either through the bottom panel, or with the lid off.
If I were to attempt a redesign (but I won't) I would seriously look at incorporating a DC servo to allow the DC coupled amplifier to operate. The protections for this DC coupled amp are already in the PSU+prot board.
Last edited:
Why don't you just design something yourself Andrew and publish it on the forum?
I think I've asked this of you two or maybe three times in the last 5 or 6 years - you've shown nothing. Zip. Nada.
In the meantime I've published 4 power amp designs and 1 preamp design plus other stuff. Ditto loads of other active designers here. I make design choices and tradeoff's (not always the best) but at least I do them based on the practical, physical situation I face on the ground.
Stop floating around like the proverbial gadfly. We don't need you acting as 'Mr. Corrector' on everything we put up. If you look around, you'll notive very little direct criticism of other peoples designs on the forum. Its called courtesy. If someone feels they need YOUR help, I'm sure they will ask for it. Quite a few others have also gotten the hell in with your approach and that includes lecturing newbies about the size of their picture uploads. So I'm not the first one to point this out.
Get a life man.
I think I've asked this of you two or maybe three times in the last 5 or 6 years - you've shown nothing. Zip. Nada.
In the meantime I've published 4 power amp designs and 1 preamp design plus other stuff. Ditto loads of other active designers here. I make design choices and tradeoff's (not always the best) but at least I do them based on the practical, physical situation I face on the ground.
Stop floating around like the proverbial gadfly. We don't need you acting as 'Mr. Corrector' on everything we put up. If you look around, you'll notive very little direct criticism of other peoples designs on the forum. Its called courtesy. If someone feels they need YOUR help, I'm sure they will ask for it. Quite a few others have also gotten the hell in with your approach and that includes lecturing newbies about the size of their picture uploads. So I'm not the first one to point this out.
Get a life man.
Last edited:
That depends, actually. You can read my thoughts on the topic on my Taming the LM3886 Chip Amp - Grounding page.
It sounds like you're using a star ground scheme. That's also what I gather from the build pictures on your website. That is actually not the way to go if you want high performance. Specifically, the THD will rise quite dramatically versus frequency due to the ground inductance (that 0.3-0.4 uH you mention). You can see some of my initial investigation on the topic in the LM3886 P2P vs PCB (with data) thread.
If you are using a star ground scheme, the ground star is your reference. Then, as you say, the Zobel return current will introduce an error voltage in the ground path. In that case, as you say, the Zobel should return to the ground star. The unfortunate thing with that connection is the 300-400 nH you introduce in series with the Zobel by doing so. This limits the effectiveness of the Zobel as its impedance will rise with frequency.
The better solution would be to use the output ground connection as your reference (you can see my "Taming" site for my justification for doing so). In that case, as Andrew pointed out, the best return path for the Zobel is to the local decoupling by the output stage.
While I do agree that Andrew's communication style is often counterproductive to his goals of providing help (at least I assume that's his goal), he does actually have a valid point here. It just appears he assumed you were using a grounding scheme that is different from a star ground. I am not trying to defend him or pile onto the criticism of you. My intent here is to point out that you're both right given your respective assumptions. The only ulterior motive I might have is to nudge you a bit further by encouraging you to improve your grounding scheme. I think you would also benefit from dropping a few bucks on some test & measurement gear so you can validate your design decisions. Alternatively, I will encourage you to develop a simulation methodology to determine the optimal grounding.
I do like your website. It's nice and cleanly laid out. Kudos to you for bringing your designs to market. I do know what that takes.
Should you be curious about my background, design capabilities, and measurement capabilities, feel free to poke around on my website. You can find my (incredibly outdated) resume/CV here.
Tom
Tom,
As I pointed out earlier, the ground reference on the amplifier module is good but not perfect. The amplifier is very wideband (design choice). The Zobel is returned to the PSU ground and tightly coupled to the amplifier module ground and also bundled where possible to the speaker return.
So the return inductance estimated is .3 to .4 uH worst case based on a probable wiring layout. The question then becomes is this worse than the ESL, ESR of the local decoupling network on the amplifier board since we don't want HF garbage, should it arise,
I have a picture of a100 kHz square wave at 70 V Pk to Pk output on the nx-Amp. If you look at the top of the waveform, you can see ringing at about 500 kHz. It's caused by the local trade inductance on the PCB, the cap ESL and the decoupling caps (220uF //0.1 uF). The layout on the amp board is very compact. The distortion on either the sx or nx amp is not going to dramatically increase because of .3 to .4 uH of Zobel inductance I can assure you.
I am quite happy I made a satisfactory tradeoff's in my design. Hopefully you did the same with yours.
As I pointed out earlier, the ground reference on the amplifier module is good but not perfect. The amplifier is very wideband (design choice). The Zobel is returned to the PSU ground and tightly coupled to the amplifier module ground and also bundled where possible to the speaker return.
So the return inductance estimated is .3 to .4 uH worst case based on a probable wiring layout. The question then becomes is this worse than the ESL, ESR of the local decoupling network on the amplifier board since we don't want HF garbage, should it arise,
I have a picture of a100 kHz square wave at 70 V Pk to Pk output on the nx-Amp. If you look at the top of the waveform, you can see ringing at about 500 kHz. It's caused by the local trade inductance on the PCB, the cap ESL and the decoupling caps (220uF //0.1 uF). The layout on the amp board is very compact. The distortion on either the sx or nx amp is not going to dramatically increase because of .3 to .4 uH of Zobel inductance I can assure you.
I am quite happy I made a satisfactory tradeoff's in my design. Hopefully you did the same with yours.
Last edited:
Hi by seeing the posts above I like to add some words here.I am not an expert in technical aspect of design or like that but as a practical builder and a listener of music.I have used many commercial hi end amps in my audio journey now doing some diy builds.
I have build some amps including Hiraga Lemonstre,P3A,DX amp and some chip amps my last build was NX..I am still using the NX amp as main amp with akido tube preamp my other two friends already build SX amp for there full range speakers they all are very happy with the result.Coming back to NX build this is the amp I have build with full confident and it worked first time itself when I powered on as it was not the case with my previous build always make issues becuse of less experience from my side.
The documentation part is excellent . we can see every detailed explanation on the single pdf file that helped me to complete this amp as I am a beginner also his support in the forum.The sound is as I said earlier excellent my amp dc offset is stable and in two channels and I am getting less than 3mv without any audiable hum and noise to my ears(put very close to tweeter and mid bass unit).The great advantage of the design is the integration of muting and speaker protection. Everything working perfect as he described in his documentation.I like to thank bonsai for his contribution a million times.I like to point out no design in this world is perfect we cannot say a 100% but keeping close to it is the perfection what I think as an end user I am very happy with his design and is very stable even soo many builders are happy with it.
Now Dear "AndrewT "
I can see you in all places whenever I follow a thread you are there no you are everywhere in the forum.So I can see you are very experienced diy person who knows exact deflect in all designs.So I challenge you to build an AMP for us "a perfect amp".Not only me many others from this forum are very eager to see your design a perfect one but should sound good too.
So hope to see your design soon.........🙂
I have build some amps including Hiraga Lemonstre,P3A,DX amp and some chip amps my last build was NX..I am still using the NX amp as main amp with akido tube preamp my other two friends already build SX amp for there full range speakers they all are very happy with the result.Coming back to NX build this is the amp I have build with full confident and it worked first time itself when I powered on as it was not the case with my previous build always make issues becuse of less experience from my side.
The documentation part is excellent . we can see every detailed explanation on the single pdf file that helped me to complete this amp as I am a beginner also his support in the forum.The sound is as I said earlier excellent my amp dc offset is stable and in two channels and I am getting less than 3mv without any audiable hum and noise to my ears(put very close to tweeter and mid bass unit).The great advantage of the design is the integration of muting and speaker protection. Everything working perfect as he described in his documentation.I like to thank bonsai for his contribution a million times.I like to point out no design in this world is perfect we cannot say a 100% but keeping close to it is the perfection what I think as an end user I am very happy with his design and is very stable even soo many builders are happy with it.
Now Dear "AndrewT "
I can see you in all places whenever I follow a thread you are there no you are everywhere in the forum.So I can see you are very experienced diy person who knows exact deflect in all designs.So I challenge you to build an AMP for us "a perfect amp".Not only me many others from this forum are very eager to see your design a perfect one but should sound good too.
So hope to see your design soon.........🙂
Perhaps he is very capable of design and build a good amp, but his obsession of a perfect solution is stopping him from doing it.
Why not perfect it? All you need to do is to connect the input reference ground to the output ground and everything else will follow. That may require you to hack the PCB, which is not fun. Perhaps you could consider the change to the ground scheme for a future design.
If you measure THD vs frequency, or THD at 10-20 kHz, you'll see a significant improvement from the grounding scheme I suggest.
Fair enough. That just means you have to be extra careful with those nH. What do you consider "very wide band" by the way?
It absolutely is worse. The HF energy comes from the local decoupling and will need to return to the local decoupling. By forcing it to return through a longer path, you reduce the effectiveness of the Zobel network.
That's good data to have.
I agree with you there. The Zobel network has nothing to do with THD in the audio band. It deals with frequencies around 1 MHz and beyond.
I'm quite happy with the design tradeoffs I made. The 0.000067 % THD (40 W, 8 Ω) of my Modulus-86 should speak for itself. 😉
Tom
Both the sx and nx-Amps are minimalist CFA designs with low to moderate loop gains (= low to moderate feedback). Because they use a classic CFA architecture they are inherently very fast. The sx Amplifier without the front end filter for example has a -3 dB bandwidth of c. 1.5 MHz. The nx is also fast - see the write-ups for the details on both designs and especially the section on the comp design for the sx-Amp.
They do not specifically target low distortion because of the low feedback (wiring/layout considerations aside). There is plenty of anecdotal evidence to suggest that you don't need ultra low distortion to have good sonics. You simply have to address the correct aspects - distortion below hearing threshold and/or benign wrt harmonic make-up, suitable dynamic range, ability to drive any reasonable load to realistic sound levels and flat frequency response. The result is you can get very good sound from quite simple circuits without the complexity that comes with high feedback or the additional challenges wrt compensation. It seems to me Nelson Pass, Jean Hiraga and JLH have/had similar thoughts. Their designs are very simple, the distortion not particularly low and yet people keep going back to them. In the case of JLH, his class A amp is still getting built 45 years later and all are noted for their 'good sound'.
I've designed a high feedback, very low distortion amp - the e-Amp - you can read about it on my website. But, it is a lot more complex and required a lot of effort to build it.
Wrt the input grounding on the sx and nx Amps: you link J2 and the input is referenced to the common ground. Please see the schematics where this is detailed.
I am very happy that you achieved low distortion on your amp. That is indeed a good result.
They do not specifically target low distortion because of the low feedback (wiring/layout considerations aside). There is plenty of anecdotal evidence to suggest that you don't need ultra low distortion to have good sonics. You simply have to address the correct aspects - distortion below hearing threshold and/or benign wrt harmonic make-up, suitable dynamic range, ability to drive any reasonable load to realistic sound levels and flat frequency response. The result is you can get very good sound from quite simple circuits without the complexity that comes with high feedback or the additional challenges wrt compensation. It seems to me Nelson Pass, Jean Hiraga and JLH have/had similar thoughts. Their designs are very simple, the distortion not particularly low and yet people keep going back to them. In the case of JLH, his class A amp is still getting built 45 years later and all are noted for their 'good sound'.
I've designed a high feedback, very low distortion amp - the e-Amp - you can read about it on my website. But, it is a lot more complex and required a lot of effort to build it.
Wrt the input grounding on the sx and nx Amps: you link J2 and the input is referenced to the common ground. Please see the schematics where this is detailed.
I am very happy that you achieved low distortion on your amp. That is indeed a good result.
Last edited:
Hello class-A fans! I am looking into building the SX 15W amp. I want to replace a pair of JLH 20W monoblocks which I built in 2003. These fell out of use about three years ago when I got into a massive Stax headphones and ES amplifiers binge. Recently I acquired some efficient speakers and found that the old JLHs were in need of a re-build, so looked for an alternative to the JLH and Hiraga designs.
I discovered the SX and thank you to Bonsai for his generous gift of time and skill for this circuit which looks well worth doing.
When I built the JLH it had a long history of successful builds. On first perusal of this thread it seems most of the discussion is about the NX amp without specifically naming it in posts. For the SX it is hard to determine whether any/many successful builds have been completed. The pcbs available from Jims appear to be original release and not correct and up to date. Could someone very kindly point me to or link me to up to date correct gerbers for the sx pcb.
I am reading the pdf from Bonsai’s website hifisonix, the changes seem small and the chance of success high if successful with transistor matching.
I discovered the SX and thank you to Bonsai for his generous gift of time and skill for this circuit which looks well worth doing.
When I built the JLH it had a long history of successful builds. On first perusal of this thread it seems most of the discussion is about the NX amp without specifically naming it in posts. For the SX it is hard to determine whether any/many successful builds have been completed. The pcbs available from Jims appear to be original release and not correct and up to date. Could someone very kindly point me to or link me to up to date correct gerbers for the sx pcb.
I am reading the pdf from Bonsai’s website hifisonix, the changes seem small and the chance of success high if successful with transistor matching.
Buy from Jims audio. Good quality. Only some small changes and it plays wonderful. Bonsai has already explained what to do.
Hi bambadoo, congratulations on your successful build. I am not keen to start a new project knowing that the available boards from Jims are not correct. Why doesn’t he bring them up to date? It is only a small job. I noticed a photo in the thread (perhaps it was your nice build) where a wire link was added to correct a circuit error and some component labels are wrong. His pcb is original vers 1.0 from way back.
Regards GCL
Regards GCL
I would expect that Jim's make a profit by selling thousands of boards and kits every week.
He probably did a run of many SX boards that he needs to sell before updating.
If the changes are minor, then just follow the advices in Bonsai's webpages and this Thread.
He probably did a run of many SX boards that he needs to sell before updating.
If the changes are minor, then just follow the advices in Bonsai's webpages and this Thread.
Indeed these changes are minor. I think I captured everything from Bambadoo's feedback plus the others in the latest doc update.
There are two versions of the nx-Amp. The latest version is V2.0.
Also, if you get any problems, there's a friendly forum that will help you through to successful completion.
I also just want to confirm that the boards Jim's Audio sells for the sx and nx Amps are indeed very high quality. Through hole plated, solder mask plus silk screen and the exposed solder pads are gold flashed. The boards are exactly per my Gerbers as well.
There are two versions of the nx-Amp. The latest version is V2.0.
Also, if you get any problems, there's a friendly forum that will help you through to successful completion.
I also just want to confirm that the boards Jim's Audio sells for the sx and nx Amps are indeed very high quality. Through hole plated, solder mask plus silk screen and the exposed solder pads are gold flashed. The boards are exactly per my Gerbers as well.
Thank you for the replies. No one has mentioned adding a flying lead correction to the pcb so I assume the picture I saw was that particular builder's experiment. I did quickly read through the entire thread. There is no mention of modifying the pcb in the build doc V2.10 so I assume I got the wrong impression. I have ordered the pcb set and look forward to building it.
If you look on page 56 of the write up on the nx-Amp, the history of the changes is listed.
V2.0 PCB includes all the changes.
As a general note: Jim's audio only supplies the PCB's based on the Gerbers he got from me - he is the supplier and not involved in the PCB design or anything like that.
Hi guys,
I'm a little worried about the SMD components (I've gone through most of the thread and still am a bit confused).
On the sx amp pcb there is a 10uF (marked) near Q2, so that's fine. But at the back there's another set of pads -- one end goes into the 15R and the other into the center leg of the variable resistor.
There's also, worryingly, a transistor -- the BC847 -- which is not only obsolete and out of stock, but supposed to cling on to the back somewhere, but I can't find any solder pads for it. What is the closest contemporary equivalent for it... and, er, where does it go?
Thanks in advance.
I'm a little worried about the SMD components (I've gone through most of the thread and still am a bit confused).
On the sx amp pcb there is a 10uF (marked) near Q2, so that's fine. But at the back there's another set of pads -- one end goes into the 15R and the other into the center leg of the variable resistor.
There's also, worryingly, a transistor -- the BC847 -- which is not only obsolete and out of stock, but supposed to cling on to the back somewhere, but I can't find any solder pads for it. What is the closest contemporary equivalent for it... and, er, where does it go?
Thanks in advance.