Rectifiers
You're correction is correct.
Even the smaller amps use a center tapped full wave bridge circuit, in effect the same circuit as used in the 'big' amps, but running from a common secondary winding.
There's something fundamental about it, but at the end of the day I'm happy to accept that if sounds better it is better and that's it.
My areas for deeper analysis have been the regulators themselves - Naim seem to test them for very low noise levels. I'm certain one could build a discrete regulator offering much lower noise - it's on my 'to do' list.
The transformers get inordinately bigger as the PSU's improve. The benefits this brings are lower impedances and hence better regulation of the raw supply (i.e. more consistent performance). There may well be filtering effects too, the newer products all use an electrostatic shield within the transformer, many earlier PSU's did not.
The larger smoothing capacitors used to reduce ripple increase transfromer heating (hence the need for bigger transformers again) coupled with the inefficiency of the circuit when used in a single rail PSU.
I'm currently listening to a modified entry level CD player though that sounds as good as my high spec LP12 in most areas so that's occupying my 'tweak time' at present!
Andy.
You're correction is correct.
Even the smaller amps use a center tapped full wave bridge circuit, in effect the same circuit as used in the 'big' amps, but running from a common secondary winding.
There's something fundamental about it, but at the end of the day I'm happy to accept that if sounds better it is better and that's it.
My areas for deeper analysis have been the regulators themselves - Naim seem to test them for very low noise levels. I'm certain one could build a discrete regulator offering much lower noise - it's on my 'to do' list.
The transformers get inordinately bigger as the PSU's improve. The benefits this brings are lower impedances and hence better regulation of the raw supply (i.e. more consistent performance). There may well be filtering effects too, the newer products all use an electrostatic shield within the transformer, many earlier PSU's did not.
The larger smoothing capacitors used to reduce ripple increase transfromer heating (hence the need for bigger transformers again) coupled with the inefficiency of the circuit when used in a single rail PSU.
I'm currently listening to a modified entry level CD player though that sounds as good as my high spec LP12 in most areas so that's occupying my 'tweak time' at present!
Andy.
Andy, you are right. At the end of the day it is not really necessary to why something works - if it works good! But I am a curious person, some facts do not stop me thinking about until I am going to find it out...
Regulators: I changed the regulator ICs (LM317T from ST) in the custom made preamp PSU I am currently using (the only one which is left, but I have some more 42 boards lying around) with LT317AT (from Linear Technologies). Although they should be in all aspects superior to the original ones, I can only detect a slight difference. With a temporarily installed HICAP instead my preamp sounded much better :-(
I am also going to try different approaches in the future (also parallel regulators etc.), when time prevents. I have - and I mentioned that before - good experiences with accus, but this solution is not really convenient for me, because I like to leave the preamp switched on all the time without any "loading breaks".
Klaus
ps: Do you have something on the net ?
And: Can you tell me/us more about your CD tweaks?
I am also thinking about that, I have a Arcam Alpha player around, which does not sound so very good, and which I occasionally use as a CD drive with an outboard DAC. This machine is quite well build and surely could be tweaked.
Regulators: I changed the regulator ICs (LM317T from ST) in the custom made preamp PSU I am currently using (the only one which is left, but I have some more 42 boards lying around) with LT317AT (from Linear Technologies). Although they should be in all aspects superior to the original ones, I can only detect a slight difference. With a temporarily installed HICAP instead my preamp sounded much better :-(
I am also going to try different approaches in the future (also parallel regulators etc.), when time prevents. I have - and I mentioned that before - good experiences with accus, but this solution is not really convenient for me, because I like to leave the preamp switched on all the time without any "loading breaks".
Klaus
ps: Do you have something on the net ?
And: Can you tell me/us more about your CD tweaks?
I am also thinking about that, I have a Arcam Alpha player around, which does not sound so very good, and which I occasionally use as a CD drive with an outboard DAC. This machine is quite well build and surely could be tweaked.
Tweaks and stuff
Regulators: I changed the regulator ICs (LM317T from ST) in the custom made preamp PSU I am currently using (the only one which is left, but I have some more 42 boards lying around) with LT317AT (from Linear Technologies). Although they should be in all aspects superior to the original ones, I can only detect a slight difference. With a temporarily installed HICAP instead my preamp sounded much better :-(
My own home brew PSU (in a test I conducted with a friend recently) wipes the floor with a Hicap and comes within 75-80% of a Supercap (when only using two rails a per a Hicap).
It also uses double-regulated 317AT's, a 530VA transformer and almost 100,000 uF of smoothing.
Most DIY supplies,IMHO, are let down by construction / wiring problems, particularly with regard to earthing.
For the CD player trust me on this one (!).
Go to http://www.lcaudio.com, buy an LCLock XO2 on the correct frequency for your CD player, fit it and be prepared to not want to sleep for weeks.
Clock jitter is the biggest factor in good CD replay and most CD players are rubbish in this respect. Even the entry level Naim players, whilst outperforming their rivals (IMO) by a significant margin are fundamentally limited in this area.
It's not worth doing any other tweaks until the clock is right. I've modded several players now and they've all improved beyond recognition, the better engineered players benefit more, but even a Marantz CD63 KI Sig sounds very close to a standard entry level Naim player when modded.
Try it - you'll like it I promise!
Once you've done this consider their inexpensive AD825 opamp modules - they're great as well (used them in a CD63).
Andy.
Regulators: I changed the regulator ICs (LM317T from ST) in the custom made preamp PSU I am currently using (the only one which is left, but I have some more 42 boards lying around) with LT317AT (from Linear Technologies). Although they should be in all aspects superior to the original ones, I can only detect a slight difference. With a temporarily installed HICAP instead my preamp sounded much better :-(
My own home brew PSU (in a test I conducted with a friend recently) wipes the floor with a Hicap and comes within 75-80% of a Supercap (when only using two rails a per a Hicap).
It also uses double-regulated 317AT's, a 530VA transformer and almost 100,000 uF of smoothing.
Most DIY supplies,IMHO, are let down by construction / wiring problems, particularly with regard to earthing.
For the CD player trust me on this one (!).
Go to http://www.lcaudio.com, buy an LCLock XO2 on the correct frequency for your CD player, fit it and be prepared to not want to sleep for weeks.
Clock jitter is the biggest factor in good CD replay and most CD players are rubbish in this respect. Even the entry level Naim players, whilst outperforming their rivals (IMO) by a significant margin are fundamentally limited in this area.
It's not worth doing any other tweaks until the clock is right. I've modded several players now and they've all improved beyond recognition, the better engineered players benefit more, but even a Marantz CD63 KI Sig sounds very close to a standard entry level Naim player when modded.
Try it - you'll like it I promise!
Once you've done this consider their inexpensive AD825 opamp modules - they're great as well (used them in a CD63).
Andy.
Transformer query
Andy, out of interest do you use a proper centertapped transormer for your psus or just a conventional off the shelf with the dual secondaries joined to form one centertapped secondary?
Andy, out of interest do you use a proper centertapped transormer for your psus or just a conventional off the shelf with the dual secondaries joined to form one centertapped secondary?
Off the peg
ced,
Currently using off the shelf dual secondaries, joined in the middle.
Also it has no electrostatic shield as per newer Naim transformers.
Seems to work well though 😉
Andy.
ced,
Currently using off the shelf dual secondaries, joined in the middle.
Also it has no electrostatic shield as per newer Naim transformers.
Seems to work well though 😉
Andy.
Hmmm
Andy,
Thanks for your reply. It was kinda what I expected but leaves me puzzled. I gave up banging my head against a wall a while back with my homebrew psu- try as I would I could not get it to really change the sound of my preamp- quality was indentical to the poweramp internal supply regardless of whether I used single or double regulation, heatsinking, full or half wave rectification, 10k or 60k uf of caps etc etc etc.
But you've sparked off my curiosity again as you do seem to know what you're doing. You say most homebrews are let down by wiring/earthing or construction iyho. What are for you the key points in wiring & construction if I may be so bold? I know Jason Hector said he got the most improvement in his trials by using really thick hookup wire (will try this with some 16A stuff some time) which was surprising.
cheers
Ced
Andy,
Thanks for your reply. It was kinda what I expected but leaves me puzzled. I gave up banging my head against a wall a while back with my homebrew psu- try as I would I could not get it to really change the sound of my preamp- quality was indentical to the poweramp internal supply regardless of whether I used single or double regulation, heatsinking, full or half wave rectification, 10k or 60k uf of caps etc etc etc.
But you've sparked off my curiosity again as you do seem to know what you're doing. You say most homebrews are let down by wiring/earthing or construction iyho. What are for you the key points in wiring & construction if I may be so bold? I know Jason Hector said he got the most improvement in his trials by using really thick hookup wire (will try this with some 16A stuff some time) which was surprising.
cheers
Ced
Doin' it right
ced,
I'm not being big-headed or boastful here, but most of the DIY-cap's I've seen previously are let down by poor standards of construction, or poor / incorrect earthing arrangements.
To understand where most designs go wrong you need to understand some fundamental principles of the Naim design. As you have no doubt noted the Naim PSU sits in the signal path between pre-amp and power amp. It seems counter-intuitive at first glance to feed a delicate audio signal through the PSU, but it is fundamental to correct earthing of the pre and power amps.
The idea is that the signal 0V within the PSU is a reference point (signal star earth point). Both the pre-amp and the power amp are referenced to this point. Many people 'hang' a PSU on the preamp, whilst routing signals direct from the preamp to power amp. A simple circuit diagram will enable you to understand why this is wrong.
Any current flow to the pre will modulate the 0V point within it, the power amp is referenced to this point and consequently the power amp is 'measuring' the signal relative to a noisy (in relative terms) ground.
So golden rule no.1 - Observe correct reference signal earthing to preamp and power amp.
The second problem is the earthing arrangements within the PSU itself. The best way to get the correct result is to define the reference 0V which is the centre tap of the transformer - usually terminated to the -ve of the smoothing cap(s).
Golden rule no.2 - All signal 0V returns from the DIN sockets must terminate to this reference point.
Earthing of the regualator circuits must also follow 'good practice'. My own PCB went for overkill and provides a 0V return for every single component back to the reference point. There are no shared current paths anywhere. This applies to the positive raw DC feeds to the regulators also, they must come directly from the smoothing / reservoir cap(s).
Golden rule no.3 - Do it properly(!)
The rest of the improvements come from achieving low noise through regulator selection or testing. Linear Technology have a good application note (AN83) detailing methods for testing low noise regulators. A good starting point is an LM317A series or better still an LT317(A) as the linear tech part has 1/3rd the noise of any competitors part.
DO NOT tie mains earth to signal earth within the PSU, it should only be used for safety earthing / electrostatic shielding. The mains earth should be tied to signal 0V at ONE point in the system only. Some integrated amps do this, but it's usually done by the turntable or the CD player. If you don't have this connection made then it's OK to do it in the PSU.
I measured the performancve of my own PSU using a cheap but quiet PC sound card (SBLive) and some spectrum analysis software. I could get a good FFT of the PSU output giving some useful information on noise against bandwidth. It's possible to get noise peaks with certain values / types of output capacitors.
Hope this helps, if there is a phrase to sum the above up it's 'Attention to Detail'.
Now it's possible to understand why the genuine article is relatively expensive 😉
Andy.
P.S. I should add I don't consider my PSU to be a 'clone' since it was designed without reference to any genuine Naim PSU's. I built it as an educational / lower cost alternative long before I owned any genuine supplies.
ced,
I'm not being big-headed or boastful here, but most of the DIY-cap's I've seen previously are let down by poor standards of construction, or poor / incorrect earthing arrangements.
To understand where most designs go wrong you need to understand some fundamental principles of the Naim design. As you have no doubt noted the Naim PSU sits in the signal path between pre-amp and power amp. It seems counter-intuitive at first glance to feed a delicate audio signal through the PSU, but it is fundamental to correct earthing of the pre and power amps.
The idea is that the signal 0V within the PSU is a reference point (signal star earth point). Both the pre-amp and the power amp are referenced to this point. Many people 'hang' a PSU on the preamp, whilst routing signals direct from the preamp to power amp. A simple circuit diagram will enable you to understand why this is wrong.
Any current flow to the pre will modulate the 0V point within it, the power amp is referenced to this point and consequently the power amp is 'measuring' the signal relative to a noisy (in relative terms) ground.
So golden rule no.1 - Observe correct reference signal earthing to preamp and power amp.
The second problem is the earthing arrangements within the PSU itself. The best way to get the correct result is to define the reference 0V which is the centre tap of the transformer - usually terminated to the -ve of the smoothing cap(s).
Golden rule no.2 - All signal 0V returns from the DIN sockets must terminate to this reference point.
Earthing of the regualator circuits must also follow 'good practice'. My own PCB went for overkill and provides a 0V return for every single component back to the reference point. There are no shared current paths anywhere. This applies to the positive raw DC feeds to the regulators also, they must come directly from the smoothing / reservoir cap(s).
Golden rule no.3 - Do it properly(!)
The rest of the improvements come from achieving low noise through regulator selection or testing. Linear Technology have a good application note (AN83) detailing methods for testing low noise regulators. A good starting point is an LM317A series or better still an LT317(A) as the linear tech part has 1/3rd the noise of any competitors part.
DO NOT tie mains earth to signal earth within the PSU, it should only be used for safety earthing / electrostatic shielding. The mains earth should be tied to signal 0V at ONE point in the system only. Some integrated amps do this, but it's usually done by the turntable or the CD player. If you don't have this connection made then it's OK to do it in the PSU.
I measured the performancve of my own PSU using a cheap but quiet PC sound card (SBLive) and some spectrum analysis software. I could get a good FFT of the PSU output giving some useful information on noise against bandwidth. It's possible to get noise peaks with certain values / types of output capacitors.
Hope this helps, if there is a phrase to sum the above up it's 'Attention to Detail'.
Now it's possible to understand why the genuine article is relatively expensive 😉
Andy.
P.S. I should add I don't consider my PSU to be a 'clone' since it was designed without reference to any genuine Naim PSU's. I built it as an educational / lower cost alternative long before I owned any genuine supplies.
cheers Andy
Hey Andy,
Great explanation and I'm with you all the way. The earthing arrangement is exactly how I've done it and the rationale behind it is logical so I don't reckon that's the problem. My bet is now on a something wrong with my transformer (of which I have my suspicions) or I've got a dud batch of LM317ATs. Think I'll try the LT317ATs like you suggest.
Out of interest what are your thoughts on heatsinking though? I find it interesting that Naim use hefty heatsinking and T03 packages in their better supplies whose only apparent merits from application notes are superior heat dissipation to T0220s. But if the circuits draw around 50mA and the preamps run class A, each reg is only dissipating a steady 0.7W (assuming a 14V drop from 38V to 24V) which ain't alot. Only a small heatsink would do (say 13W/o)to keep the regs within a very comfortable operating temperature, and if the dissipation is steady there shouldn't be a thermal noise issue due to fluctuations in temperature with varying current draw. This would imply Naim think the regs work best at a low temperature. Any thoughts on this thinking from a theoretical point of view?
Ced
Hey Andy,
Great explanation and I'm with you all the way. The earthing arrangement is exactly how I've done it and the rationale behind it is logical so I don't reckon that's the problem. My bet is now on a something wrong with my transformer (of which I have my suspicions) or I've got a dud batch of LM317ATs. Think I'll try the LT317ATs like you suggest.
Out of interest what are your thoughts on heatsinking though? I find it interesting that Naim use hefty heatsinking and T03 packages in their better supplies whose only apparent merits from application notes are superior heat dissipation to T0220s. But if the circuits draw around 50mA and the preamps run class A, each reg is only dissipating a steady 0.7W (assuming a 14V drop from 38V to 24V) which ain't alot. Only a small heatsink would do (say 13W/o)to keep the regs within a very comfortable operating temperature, and if the dissipation is steady there shouldn't be a thermal noise issue due to fluctuations in temperature with varying current draw. This would imply Naim think the regs work best at a low temperature. Any thoughts on this thinking from a theoretical point of view?
Ced
Ced:
I've been following you and Andy through this thread.
From screwing around with my own and my Brother-in-law's HICAP power supplies I totally agree with your comments.
No matter what I seem to do in the way of tweaks and a few major open hearts, the sound of the system just doesn't change.
With the exception of using underrated Power Transformers or no smoothing caps, no matter what component I change there just isn't the major dramatic effects that others seem to experience. If any effects I have noticed I put down to component variation and environmental temperature. Generally just using another component of the same type and batch and re-checking generally gives the answers to why.
For the hell of it this is what I've been up to:
1. Initial attempt to change the transformer design to using Bifilar secondary windings to reduce inductance, since the rectifier currents are harmonic rich, but hold-on the load current is only about 340mA or a about whopping 2AMP peak! . Anyway the Toroid was custom wound using dual primary windings for 220V. VA rating was increased to 650VA due to fill factor (from 500VA), DC resistance of original NAIM transformer was around 0.036 ohm per secondary winding, I managed to get this down to 0.012 max per winding! Heavy!
Neither myself or my NAIM convert brother-in-law could here the difference. Initially he said he could here greater depth, then I did a blind swap test or two between my PSU and his factory standard HICAP, after that he called me a "******".
2. What about the bridge rectifier and that evil turn-off current and the nice HF harmonics it produces. Well cool I'll get myself some soft recovery diodes as used in most SMPSs. Much to our disappointment the sound remained the same, even with the new 650VA Toroid. Nothing jumped out at us as having DRAMATICALLY improved, like what they say in those reviews.
3. Then it dawned on us we're using linear regulators, and by Ohm's law they pose a dynamic resistance/impedance in series with our super low impedance Toroid and our pre-amps. So 35V into the LM317K regulators, 24V about on the output, so 11 volt difference and an output current into our pre-amp (NAC 82) of around 0.1AMP, so the regulator is giving us a dynamic resistance of small 110 ohms. No wonder the Toroid wasn't having an effect.
Also the bridge rectifier goes along way to increasing the impedance due to the dynamic switching action. Cool, what next?
4. The Regulator....
These liner regulators don't like back EMF, their impedance (not resistance) is also pretty poor, dramatically rising very quickly, like an inductor, but the regulator is never an inductor.
I guess that's why I always have to keep using those 0.1uf decoupling caps in my designs at work.
So why not lump some capacitance on the output regulator to reduce the impedance, cool! We added some 4700uf and 0.22uf Polypropylene caps on the output.
Then it dawned on us again that the series inductance of the cable from the HICAP to the PREAMP really defeated the purpose of the caps. Since it effectively isolates the capacitors by it's own series inductance, the XLs might be smaller in magnitude than the XCp, but it still there between your PSU and the pre-amp - bugger. What the hell, just power it up and listen.
Results were about the same as our previous tests, no noticeable effects, even swapping NAIM cables from SNAIC's to the older GREY ones, and then to a NORDOST Black something. ($$$). Just wasn't happening. Tried it for a weak. No noticeable change.
Okay what are we doing wrong?
The noise of the regulator was way small, about -82dB down.
Since it was broadband, as noise is, being made up of many frequencies - no resonance hear, the lumped capacitors turned it more into 1/f pink noise. It's quite interesting what you see with a spectrum analyser that goes from DC to 500MHz.
We could only see subtle noise changes as we swapped the Toroids around may be +/- 1dB change.
5. What about the filter caps.
Those SNCE French caps CO38 15000uf are pretty bog standard, average performers really, the impedance of the caps rises as with BC, RIFA, and the 13Amp-ripple current isn't bad. But the CO31SP 47uF 40V caps used in the pre-amp, with impedance as low as 0.07 are pretty special 10 years ago.
Well anyway we thought to get ourselves some 15000uF & 22000uF EPCOS Sikorels. BABY these are just the ticket, big black cans, with that smooth gold writing, and the weight, these just have to sound better, they f**k'n cost enough at 75DM each!
The output noise of the regulators should be way down at audio now! Since noise is broadband signals, ISN'T IT!
These things have a ripple current of 32AMP at 125 degrees C! ESR is way down, and the just "feel" better.
The Toroid was definitely going to hate us at switch-on, the inrush current to the caps, was theoretically, about a whopping 2000Amp! Thank good for high-impedance primaries and small power cords, let alone the cable in the wall!
Well, we got a real good crack sound out of the power switch, reckon it will be stuffed after few a on/off attempts. Dam, must of switched on at the positive peak of the mains.
Anyway, we couldn't wait to hear the improvement, it was my Brother-in-laws investment in those caps and I had to show something for it!
I was getting kind of worried, we'd left the power amp on (idle) and the system sounded like it used to? What's going on here?
We left things as they were for about a week; my brother-in-law was starting to get "pissed" with me. I promptly reminded him it was his idea in the first place.
Well four months on, I have one hell of an expensive HICAP and little to show for it.
Ced, mate, next we're going to start changing them pre-amp-circuits, why? Cause my brother-in-law wants to?
I've been following you and Andy through this thread.
From screwing around with my own and my Brother-in-law's HICAP power supplies I totally agree with your comments.
No matter what I seem to do in the way of tweaks and a few major open hearts, the sound of the system just doesn't change.
With the exception of using underrated Power Transformers or no smoothing caps, no matter what component I change there just isn't the major dramatic effects that others seem to experience. If any effects I have noticed I put down to component variation and environmental temperature. Generally just using another component of the same type and batch and re-checking generally gives the answers to why.
For the hell of it this is what I've been up to:
1. Initial attempt to change the transformer design to using Bifilar secondary windings to reduce inductance, since the rectifier currents are harmonic rich, but hold-on the load current is only about 340mA or a about whopping 2AMP peak! . Anyway the Toroid was custom wound using dual primary windings for 220V. VA rating was increased to 650VA due to fill factor (from 500VA), DC resistance of original NAIM transformer was around 0.036 ohm per secondary winding, I managed to get this down to 0.012 max per winding! Heavy!
Neither myself or my NAIM convert brother-in-law could here the difference. Initially he said he could here greater depth, then I did a blind swap test or two between my PSU and his factory standard HICAP, after that he called me a "******".
2. What about the bridge rectifier and that evil turn-off current and the nice HF harmonics it produces. Well cool I'll get myself some soft recovery diodes as used in most SMPSs. Much to our disappointment the sound remained the same, even with the new 650VA Toroid. Nothing jumped out at us as having DRAMATICALLY improved, like what they say in those reviews.
3. Then it dawned on us we're using linear regulators, and by Ohm's law they pose a dynamic resistance/impedance in series with our super low impedance Toroid and our pre-amps. So 35V into the LM317K regulators, 24V about on the output, so 11 volt difference and an output current into our pre-amp (NAC 82) of around 0.1AMP, so the regulator is giving us a dynamic resistance of small 110 ohms. No wonder the Toroid wasn't having an effect.
Also the bridge rectifier goes along way to increasing the impedance due to the dynamic switching action. Cool, what next?
4. The Regulator....
These liner regulators don't like back EMF, their impedance (not resistance) is also pretty poor, dramatically rising very quickly, like an inductor, but the regulator is never an inductor.
I guess that's why I always have to keep using those 0.1uf decoupling caps in my designs at work.
So why not lump some capacitance on the output regulator to reduce the impedance, cool! We added some 4700uf and 0.22uf Polypropylene caps on the output.
Then it dawned on us again that the series inductance of the cable from the HICAP to the PREAMP really defeated the purpose of the caps. Since it effectively isolates the capacitors by it's own series inductance, the XLs might be smaller in magnitude than the XCp, but it still there between your PSU and the pre-amp - bugger. What the hell, just power it up and listen.
Results were about the same as our previous tests, no noticeable effects, even swapping NAIM cables from SNAIC's to the older GREY ones, and then to a NORDOST Black something. ($$$). Just wasn't happening. Tried it for a weak. No noticeable change.
Okay what are we doing wrong?
The noise of the regulator was way small, about -82dB down.
Since it was broadband, as noise is, being made up of many frequencies - no resonance hear, the lumped capacitors turned it more into 1/f pink noise. It's quite interesting what you see with a spectrum analyser that goes from DC to 500MHz.
We could only see subtle noise changes as we swapped the Toroids around may be +/- 1dB change.
5. What about the filter caps.
Those SNCE French caps CO38 15000uf are pretty bog standard, average performers really, the impedance of the caps rises as with BC, RIFA, and the 13Amp-ripple current isn't bad. But the CO31SP 47uF 40V caps used in the pre-amp, with impedance as low as 0.07 are pretty special 10 years ago.
Well anyway we thought to get ourselves some 15000uF & 22000uF EPCOS Sikorels. BABY these are just the ticket, big black cans, with that smooth gold writing, and the weight, these just have to sound better, they f**k'n cost enough at 75DM each!
The output noise of the regulators should be way down at audio now! Since noise is broadband signals, ISN'T IT!
These things have a ripple current of 32AMP at 125 degrees C! ESR is way down, and the just "feel" better.
The Toroid was definitely going to hate us at switch-on, the inrush current to the caps, was theoretically, about a whopping 2000Amp! Thank good for high-impedance primaries and small power cords, let alone the cable in the wall!
Well, we got a real good crack sound out of the power switch, reckon it will be stuffed after few a on/off attempts. Dam, must of switched on at the positive peak of the mains.
Anyway, we couldn't wait to hear the improvement, it was my Brother-in-laws investment in those caps and I had to show something for it!
I was getting kind of worried, we'd left the power amp on (idle) and the system sounded like it used to? What's going on here?
We left things as they were for about a week; my brother-in-law was starting to get "pissed" with me. I promptly reminded him it was his idea in the first place.
Well four months on, I have one hell of an expensive HICAP and little to show for it.
Ced, mate, next we're going to start changing them pre-amp-circuits, why? Cause my brother-in-law wants to?
I was reading the whole tread from beginning and there are a couple of ideas I'd like to share.
First of all I would like to suggest an idea about PS differences discussed earlier. Can it perhaps be that the half wave rectifier PS version is better simply because of the frequency content of ripple voltage. With half wave you get 50Hz and with full wave 100Hz. Well if one takes the harmonics of these into consideration they would be 2nd 100Hz, 3rd 150Hz, 4th 200Hz etc for half wave rect. as opposed to 200Hz, 300Hz, 400Hz etc. for full wave. It is easy to see that the full wave rectifiers harmonics come much closer to critical mid band and the most sensitive part of our hearing. On the other hand if there is any effect relating to this, it should show up on spectral analysis....... If any of you have the Naim's and is willing to compare and analyse the idea it would be interesting to know. I can only look at Naims on pictures and read about them as I'm pretty certain there is no Naim equipment here in Estonia : (
As for the spectral analysis idea I would like to suggest the version I'm using. It is a Cheap Leadtek Winfast soundcard (Cmedia 8738 IC) with optical in/out and Sony MDS-JE520 (bought as second hand) for AD conversion. The whole thing cost me about 170$, is better in THD and noise figures than any soundcard in that price range and gives a great means for DIY measurements. Include the measurement mic and mic amp and also the acoustival measurements... I'm using that too.
First of all I would like to suggest an idea about PS differences discussed earlier. Can it perhaps be that the half wave rectifier PS version is better simply because of the frequency content of ripple voltage. With half wave you get 50Hz and with full wave 100Hz. Well if one takes the harmonics of these into consideration they would be 2nd 100Hz, 3rd 150Hz, 4th 200Hz etc for half wave rect. as opposed to 200Hz, 300Hz, 400Hz etc. for full wave. It is easy to see that the full wave rectifiers harmonics come much closer to critical mid band and the most sensitive part of our hearing. On the other hand if there is any effect relating to this, it should show up on spectral analysis....... If any of you have the Naim's and is willing to compare and analyse the idea it would be interesting to know. I can only look at Naims on pictures and read about them as I'm pretty certain there is no Naim equipment here in Estonia : (
As for the spectral analysis idea I would like to suggest the version I'm using. It is a Cheap Leadtek Winfast soundcard (Cmedia 8738 IC) with optical in/out and Sony MDS-JE520 (bought as second hand) for AD conversion. The whole thing cost me about 170$, is better in THD and noise figures than any soundcard in that price range and gives a great means for DIY measurements. Include the measurement mic and mic amp and also the acoustival measurements... I'm using that too.
Naim mods and clones
Hi,
I have been with Naim clones since 1984. So I might share some results.
First, idle current in power amps is set way to low. naim amps really sing with 20mAmps idle current, but will sound harsh when set at zero mAmps, like they are from stock.
Second, the anti-ringing compensation in the power amps is 70ies art, and together with omitting the output series inductor will lead to ringing, when using cables with high parallel capacitor. Furthermore the output stage compensation couples power supply noise into the power amp circuit. Just reroute all the compensation stuff in 90ies manor, and you get far better and smooth treble operation with immuness against cables. My favourite reading book on operational amplifier internal design is "Dostal: Operationsverstaerker", of East German origin (!).
I tried that double regulators like in SuperCap 15 years ago, and while it gave super regulation, my preamp did not sing any more.
I find it quite funny, that I do not like Naim amps from the 90ies. They lost the tube-like singing quality, but gained some kind of electronic neutralness.
regards,
Hartmut from Munich
Hi,
I have been with Naim clones since 1984. So I might share some results.
First, idle current in power amps is set way to low. naim amps really sing with 20mAmps idle current, but will sound harsh when set at zero mAmps, like they are from stock.
Second, the anti-ringing compensation in the power amps is 70ies art, and together with omitting the output series inductor will lead to ringing, when using cables with high parallel capacitor. Furthermore the output stage compensation couples power supply noise into the power amp circuit. Just reroute all the compensation stuff in 90ies manor, and you get far better and smooth treble operation with immuness against cables. My favourite reading book on operational amplifier internal design is "Dostal: Operationsverstaerker", of East German origin (!).
I tried that double regulators like in SuperCap 15 years ago, and while it gave super regulation, my preamp did not sing any more.
I find it quite funny, that I do not like Naim amps from the 90ies. They lost the tube-like singing quality, but gained some kind of electronic neutralness.
regards,
Hartmut from Munich
hicap clones
HI T-hill (can't find your proper name or personal email so excuse the formality),
Just thought I'd drop you a line to say I've finally cracked the psu problem (well pleased) and was wondering if you'd had any more sucess yourself?
For me what did it was using really thick 30A hookup wire (thanks to Jason Hector for that one), thick copper tracks on my pcb and follwing a hicap's wiring schematic really closely particularly where leads go on the earthing bus bar between the reservoir caps (thanks to Andy Weeks for all his research and emphasising that importance and providing general inspiration).
I don't know which one or if all of these changes made a difference but who cares. It works nicely.
If you're interested I can send you the details. I don't know how it rates comparing A-B wise to a hicap but its a very nice improvement and judging from listening to a top flight system recently which I found not an earth shaker for the money involved, -I now understand what people mean about diminishing returns the higher up the ladder you go- quite significant.
cheers and good luck
Cedric
HI T-hill (can't find your proper name or personal email so excuse the formality),
Just thought I'd drop you a line to say I've finally cracked the psu problem (well pleased) and was wondering if you'd had any more sucess yourself?
For me what did it was using really thick 30A hookup wire (thanks to Jason Hector for that one), thick copper tracks on my pcb and follwing a hicap's wiring schematic really closely particularly where leads go on the earthing bus bar between the reservoir caps (thanks to Andy Weeks for all his research and emphasising that importance and providing general inspiration).
I don't know which one or if all of these changes made a difference but who cares. It works nicely.
If you're interested I can send you the details. I don't know how it rates comparing A-B wise to a hicap but its a very nice improvement and judging from listening to a top flight system recently which I found not an earth shaker for the money involved, -I now understand what people mean about diminishing returns the higher up the ladder you go- quite significant.
cheers and good luck
Cedric
Let's continue with this thread
Cedric,
as the person who started this thread "ages" ago, I thank you for bringing it back to life.
Please share your details with all of us, there are more persons interested in your results.
Hartmut,
please tell us more about your experiences with your "clones".
A bit frustrated by my search for the "ideal" transistor I just started to make a new low power version of this amp with BD241Cs, mainly because I have a bunch of them lying around. At least there should be a big difference to the BD911, but the SOA is probably to small. Anyway, I will try it.
What are your experiences with "the transistor", which ones did you use ?
regards
Klaus
Cedric,
as the person who started this thread "ages" ago, I thank you for bringing it back to life.
Please share your details with all of us, there are more persons interested in your results.
Hartmut,
please tell us more about your experiences with your "clones".
A bit frustrated by my search for the "ideal" transistor I just started to make a new low power version of this amp with BD241Cs, mainly because I have a bunch of them lying around. At least there should be a big difference to the BD911, but the SOA is probably to small. Anyway, I will try it.
What are your experiences with "the transistor", which ones did you use ?
regards
Klaus
details then
Hi Lokh,
sure.
Basically I was having the same problems as T-hill in that I built a DIYcap based pretty closely on the Neil McBride design (see www.neilmcbride.co.uk) and got nothing- it worked but provided no improvement in sound over the internal supply of My Nap140 poweramp. This was based on a 625Va toroid from Antrim giving 32v at the caps (should have been 38V but they messed up the spec and I didn't know what I was doing at the time) and the windings paralleled. Hookup wire was maplins 6A stuff (about 1mm2).
I Tried his discreet hexfred diode rectifier circuit instead of a bridge in full wave configuration and blew several 6amp ones until I went for 35A ones (I still don't know how his shottkeys survived as he said he'd even used 1A ones successfully) and then discovered about inrush currents. This first attempt actually changed the sound slightly in that it was minutely more detailed but thinner and less musical and I perfered the sound without my psu. I changed the hexfred bridge for a cheap 35A bridge and the sound returned to the same as the NAP140 internal supply so left that in. Then tried 2 fullwave bridges (one for each transformer winding) feeding 30 000uf each and one dual regulator circuit each. No change. I then read that the psu sounded better in centertapped half wave rectification format so I went back to one bridge and centertapping. No change. Then tried making a star earth between all 6 x 10000uf (63V aerovox als30) caps and connecting all earths to this. Still no change. Got fed up with messing around with 6 reservoir caps so went down to 2 of them. Still no change. Gave up. At this point I have to say that my wiring was a bit of a mess too.
But kept looking around and emailing a few people. Jason Hector said he'd also had loads of trouble but found the greatest improvement by using really thick hookup wire and Andy week's postings have emphasised the importance of earthing and hints on the naim forum as to the importance of wiring looms for low noise.
So I thought I'd give it one last try. I'd also been looking at specific components like trying to track down the blue tantalums that naim use, their specific capacitors, talema transformers that they use, etc etc under the assumption that it was the properties of particular components that were vital. But I thought I'm not spending more money before I get some sort of change. IMHO now, specific components are of relatively little importance (although there may be gains to be had)- it is how it's all hooked up.
So I managed to get hold of a good hirez pic of the innards of a hicap which gave a clear view of their wiring system and decided to follow that exactly. Basically cheat. I used 30A hookup wire (2.5mm2 or 50/0.25) which is not far short of speaker wire thickness. I also etched new regulator boards (ditching the LM329 voltage reference configuration) with really wide tracks to simple datasheet specifications, but reused all the components I'd already bought. And bingo.
For me the key things have been a) thick hookup wire and pcb tracks, b) 2 reservoir caps only c) specific wiring looms d) correct earthing across the reservoir caps
What has worked for me can still be improved on easily but currently its as follows. 625Va toroid providing 32V dc (36-38Vdc would be better I'm sure) at the caps feeding a 35v silicon bridge in half wave formation. 2 x 10 000uf caps with a star earth 'bus bar' between the negative terminals made from the thick solid core of a mains cable. The 2 ends of the windings which form the centertap from the transformer are not joined together as they emerge from the transformer but connect to the earth bus bar at either end. The rest of the earth returns from the regulator boards and the din sockets all connect to this between these 2. All the cables are neatly bound together with cable ties and are cut exactly to size to make the neat wiring looms you see in naim equipment. Its a bugger to solder and get the lengths right but it looks great and works so I think its worth the effort.
My next stage will be to swap the 2 x 10000uf caps for 2x33 000uf caps (no more messing around with loads of reservoir caps and I suspect this could be detrimental to earthing if you don't get it exactly right) which should improve the sound further and then convert to Andy Weekes' Din socket earth mod for further gains I hope. I'll post results when I get them. And the sound improvement? Well, all the classic things I've heard about; more warmth, detail, PRT, sense of instrumental seperation, basically more groovy. I like it.
I can't vouch for it's performance vis a vis a highcap or supercap but from what I remember it trounces a flatcap and listening to a couple of high end naim systems recently it's pretty damn good. Saying that other DIYers are probably further down the road and have created better performance psus.
But if anyone is struggling to make the damn things work at all, I can mail them a pdf of my reg boards and detailed jpg of the psu layout and wiring to at least get off the ground and you can work from there. Drop me a line at nospamcedric_taylor@buildergroup.co.uk (nospam is obviously not part of my email).
Oh and listening was via a cheap Aiwa tape deck! (until the CDX appears), 32.5/140/credos.
Hope this helps anyone and sorry for the tedious essay.
cheers
Cedric
Hi Lokh,
sure.
Basically I was having the same problems as T-hill in that I built a DIYcap based pretty closely on the Neil McBride design (see www.neilmcbride.co.uk) and got nothing- it worked but provided no improvement in sound over the internal supply of My Nap140 poweramp. This was based on a 625Va toroid from Antrim giving 32v at the caps (should have been 38V but they messed up the spec and I didn't know what I was doing at the time) and the windings paralleled. Hookup wire was maplins 6A stuff (about 1mm2).
I Tried his discreet hexfred diode rectifier circuit instead of a bridge in full wave configuration and blew several 6amp ones until I went for 35A ones (I still don't know how his shottkeys survived as he said he'd even used 1A ones successfully) and then discovered about inrush currents. This first attempt actually changed the sound slightly in that it was minutely more detailed but thinner and less musical and I perfered the sound without my psu. I changed the hexfred bridge for a cheap 35A bridge and the sound returned to the same as the NAP140 internal supply so left that in. Then tried 2 fullwave bridges (one for each transformer winding) feeding 30 000uf each and one dual regulator circuit each. No change. I then read that the psu sounded better in centertapped half wave rectification format so I went back to one bridge and centertapping. No change. Then tried making a star earth between all 6 x 10000uf (63V aerovox als30) caps and connecting all earths to this. Still no change. Got fed up with messing around with 6 reservoir caps so went down to 2 of them. Still no change. Gave up. At this point I have to say that my wiring was a bit of a mess too.
But kept looking around and emailing a few people. Jason Hector said he'd also had loads of trouble but found the greatest improvement by using really thick hookup wire and Andy week's postings have emphasised the importance of earthing and hints on the naim forum as to the importance of wiring looms for low noise.
So I thought I'd give it one last try. I'd also been looking at specific components like trying to track down the blue tantalums that naim use, their specific capacitors, talema transformers that they use, etc etc under the assumption that it was the properties of particular components that were vital. But I thought I'm not spending more money before I get some sort of change. IMHO now, specific components are of relatively little importance (although there may be gains to be had)- it is how it's all hooked up.
So I managed to get hold of a good hirez pic of the innards of a hicap which gave a clear view of their wiring system and decided to follow that exactly. Basically cheat. I used 30A hookup wire (2.5mm2 or 50/0.25) which is not far short of speaker wire thickness. I also etched new regulator boards (ditching the LM329 voltage reference configuration) with really wide tracks to simple datasheet specifications, but reused all the components I'd already bought. And bingo.
For me the key things have been a) thick hookup wire and pcb tracks, b) 2 reservoir caps only c) specific wiring looms d) correct earthing across the reservoir caps
What has worked for me can still be improved on easily but currently its as follows. 625Va toroid providing 32V dc (36-38Vdc would be better I'm sure) at the caps feeding a 35v silicon bridge in half wave formation. 2 x 10 000uf caps with a star earth 'bus bar' between the negative terminals made from the thick solid core of a mains cable. The 2 ends of the windings which form the centertap from the transformer are not joined together as they emerge from the transformer but connect to the earth bus bar at either end. The rest of the earth returns from the regulator boards and the din sockets all connect to this between these 2. All the cables are neatly bound together with cable ties and are cut exactly to size to make the neat wiring looms you see in naim equipment. Its a bugger to solder and get the lengths right but it looks great and works so I think its worth the effort.
My next stage will be to swap the 2 x 10000uf caps for 2x33 000uf caps (no more messing around with loads of reservoir caps and I suspect this could be detrimental to earthing if you don't get it exactly right) which should improve the sound further and then convert to Andy Weekes' Din socket earth mod for further gains I hope. I'll post results when I get them. And the sound improvement? Well, all the classic things I've heard about; more warmth, detail, PRT, sense of instrumental seperation, basically more groovy. I like it.
I can't vouch for it's performance vis a vis a highcap or supercap but from what I remember it trounces a flatcap and listening to a couple of high end naim systems recently it's pretty damn good. Saying that other DIYers are probably further down the road and have created better performance psus.
But if anyone is struggling to make the damn things work at all, I can mail them a pdf of my reg boards and detailed jpg of the psu layout and wiring to at least get off the ground and you can work from there. Drop me a line at nospamcedric_taylor@buildergroup.co.uk (nospam is obviously not part of my email).
Oh and listening was via a cheap Aiwa tape deck! (until the CDX appears), 32.5/140/credos.
Hope this helps anyone and sorry for the tedious essay.
cheers
Cedric
loopy thoughts
I agree with what ced said.
A couple of rules of thumb that I have learned that echo ced's experiences:
1) Electrical currents flow in loops. It is very important to identify these loops so that you can ensure your wiring optimises the characterisitcs of these loops. For instance Naim's very neat wiring looms are not just there for visual appeal - by keeping the area of the current loops as small as possible the loop inductance is minimized.
2) Beware of two loops sharing a common wire. Sometimes this doesn't matter but usually it does. The common currents combined with the wire impedance cause an additive voltage drops along the wire and effectively mix the two signals. This can be mitigated by the use of very low resisitance connections or star earthing.
3) All voltages are relative between one point and another. Your circuit must maintain the integrity of the music signal between one point and another. What the voltages of these two points is relative to other points is not so important.
I also agree with ced that most of the components in Naim circuits are nothing special. I wouldn't worry about combing the planet for blue coloured tantalums. If you compare Naim stuff to other high-end mfrs you will see it is cheap, cheap, cheap. The mfg costs are kept to an absolute minimum. They only spend exceptional money on certain key components like output transistors. IMO Naim have done an extraordinary job of getting priorities right to maximize sound quality against mfg cost - even down to using non-standard DIN connectors. This is a laudable achievement although it opens their designs to many improvement opportunities. The lesson for the DIY'er is that if you get your priorities right you needn't spend a lot of money or use complex circuits to get great sound.
I agree with what ced said.
A couple of rules of thumb that I have learned that echo ced's experiences:
1) Electrical currents flow in loops. It is very important to identify these loops so that you can ensure your wiring optimises the characterisitcs of these loops. For instance Naim's very neat wiring looms are not just there for visual appeal - by keeping the area of the current loops as small as possible the loop inductance is minimized.
2) Beware of two loops sharing a common wire. Sometimes this doesn't matter but usually it does. The common currents combined with the wire impedance cause an additive voltage drops along the wire and effectively mix the two signals. This can be mitigated by the use of very low resisitance connections or star earthing.
3) All voltages are relative between one point and another. Your circuit must maintain the integrity of the music signal between one point and another. What the voltages of these two points is relative to other points is not so important.
I also agree with ced that most of the components in Naim circuits are nothing special. I wouldn't worry about combing the planet for blue coloured tantalums. If you compare Naim stuff to other high-end mfrs you will see it is cheap, cheap, cheap. The mfg costs are kept to an absolute minimum. They only spend exceptional money on certain key components like output transistors. IMO Naim have done an extraordinary job of getting priorities right to maximize sound quality against mfg cost - even down to using non-standard DIN connectors. This is a laudable achievement although it opens their designs to many improvement opportunities. The lesson for the DIY'er is that if you get your priorities right you needn't spend a lot of money or use complex circuits to get great sound.
Ohhhh ...... I agree!
In fact Nelson also hints at this in the opening description of the Son of Zen. People in HiFi get "sucked-in" to thinking so much about the minor-league items when these are swamped into insignificance by larger factors.
Retailers trying to promote their own product, selling BS to the unwary.
However, it is also true that the 1-unit DIY nut (an affectionate term in these circles) can "go the extra yard". The gain is often minimal in truth .... but why the hell not 🙂
mark
In fact Nelson also hints at this in the opening description of the Son of Zen. People in HiFi get "sucked-in" to thinking so much about the minor-league items when these are swamped into insignificance by larger factors.
Retailers trying to promote their own product, selling BS to the unwary.
However, it is also true that the 1-unit DIY nut (an affectionate term in these circles) can "go the extra yard". The gain is often minimal in truth .... but why the hell not 🙂
mark
lohk said:
lohk,
I got here from doing a search after your mention of this thread in a more recent thread. I was wondering if you had made any progress on this. I'd be happy to help with graphics & HTML if that would help you along.
dave
Dave,
no, I did not, I am sorry.
Mainly because I was - as far as the restricted time permitted - engaged in other projects.
And also because the discussion started already in this and several other threads. I also "took my mouth too full" - I am not the naim expert that it may seem. Although I used, modified and "cloned" them since several years now, my "main experience" is just to listen to music play via them everyday without thinking too much about mods and tweaks.
I use to sell and repair those amps in the eighties - quite a long time ago, although the main schematics do not seem to have changed that much.
But thank you anyway. Html and graphics are not the problem, the problem is time and "clever organized content" - I did not even succed having all my other projects and activities on the net yet...
Ced,
I hope you read this, my apologations about not having answered your thread. In spring the DIYaudio forum grew bigger so quick, I lost my overview - I could not read all the good posts, hidden somewhat behind all that babble (and English is not my mother tongue anyway).
So I do not remember having read your post before already, I am very sorry, also because it is a really good description of "that special experience".
Dave,
your suggestions about "fusing" all those "Naimed" activities together somehow? With ALWs experiments and gains concerning PSUs (which have al lot to do with studiying this stuff) it is quite a lot yet.
I cannot do the proposed page at the moment.
Hartmut,
let me apply here also for your support and your mods please.
Klaus
no, I did not, I am sorry.
Mainly because I was - as far as the restricted time permitted - engaged in other projects.
And also because the discussion started already in this and several other threads. I also "took my mouth too full" - I am not the naim expert that it may seem. Although I used, modified and "cloned" them since several years now, my "main experience" is just to listen to music play via them everyday without thinking too much about mods and tweaks.
I use to sell and repair those amps in the eighties - quite a long time ago, although the main schematics do not seem to have changed that much.
But thank you anyway. Html and graphics are not the problem, the problem is time and "clever organized content" - I did not even succed having all my other projects and activities on the net yet...
Ced,
I hope you read this, my apologations about not having answered your thread. In spring the DIYaudio forum grew bigger so quick, I lost my overview - I could not read all the good posts, hidden somewhat behind all that babble (and English is not my mother tongue anyway).
So I do not remember having read your post before already, I am very sorry, also because it is a really good description of "that special experience".
Dave,
your suggestions about "fusing" all those "Naimed" activities together somehow? With ALWs experiments and gains concerning PSUs (which have al lot to do with studiying this stuff) it is quite a lot yet.
I cannot do the proposed page at the moment.
Hartmut,
let me apply here also for your support and your mods please.
Klaus
Let me drop in some infos here, as goodies:
NAIM output transistors (npn):
NAIT and NAP90: BD743
NAP140 (prob. the new NAP150 too): 2SC2922 (Sanken)
historic: the old NAP250 used BDY58R and the old NAP110 used BUV20
diodes antiparallel to the output devices: FR105 or BYF406
enjoy
Klaus
NAIM output transistors (npn):
NAIT and NAP90: BD743
NAP140 (prob. the new NAP150 too): 2SC2922 (Sanken)
historic: the old NAP250 used BDY58R and the old NAP110 used BUV20
diodes antiparallel to the output devices: FR105 or BYF406
enjoy
Klaus
Naim o/p devices
Older NAP140's did use 2SC2922, but later units were updated to the Naim custom devices (NA001).
The current NAP150 does use the 2SC2922 (I use one daily!), but has different topology and circuit, based on the newer NAP500.
It will be interesting to see what the new products contain (there's a new lineup due soon, hence your local dealer is clearing his old stock - could be some bargains for those interested). The original o/p devices (NA001 / NA002 etc) are now being reserved for repair support only, the NAP500 uses new custom devices from Semelab, but I suspect these may be too expensive for the lower end products.
A.
Older NAP140's did use 2SC2922, but later units were updated to the Naim custom devices (NA001).
The current NAP150 does use the 2SC2922 (I use one daily!), but has different topology and circuit, based on the newer NAP500.
It will be interesting to see what the new products contain (there's a new lineup due soon, hence your local dealer is clearing his old stock - could be some bargains for those interested). The original o/p devices (NA001 / NA002 etc) are now being reserved for repair support only, the NAP500 uses new custom devices from Semelab, but I suspect these may be too expensive for the lower end products.
A.
- Status
- Not open for further replies.