Look at the board. Where PG+ and PG- go is actually one small plane on the board. The voltages at + and - negate each other and we end up with 0V which is what we need for GND. On the MiniA PSU there are 5 holes near the letters GND. They all are 0V. You can take one of those to either PG+ OR PG- as it is already 0V and we dont need to add a - voltage and a + voltage to get zero anymore, it already happened on the PSU board.
Uriah
Uriah
I don't know how efficient my speakers are (no info seems to be out there on them), but I'm using a similar PSU to the BrianGT design (but with 15,000uF main caps) and my own LM3886 design. With either a 2x18VAC toroid or a 74VAC centre tap EI transformer, it's impossible to hear anything by putting your ear to the tweeter. I've measured about 1% ripple at idle at the chip. I agree it's more about the power supply and grounding than anything.
Thanks! I was thinking "amplifier output snubber", not "passive crossover inductance flattener", but now I see the light.
This actually reduces the impedance a bit, by putting things in parallel with the voice coil. Given the chipamps' dislike for low impedance and high current, is that a good thing?
The detailed Zobel as in the paper, indeed flatten the impedance. IF this nominal impedance is to low, it can be a bad thing, it depends on the power supply voltages of your chipamp. An other downside of this Zobel is that you get some looses, which you actually don't want.
You can keep things simpler and stick to the simple Zobel with a resistor and capacitor to ground, and calculate an approximately right value that fits the specific driver.
The Zobel resistor (Rz) should be equal the Series resistance (in series with th voice coil induction) Rs
The Zobel capacitor (Cz) you calculate by: Cz = Ls/(Rs*Rs)
Those specifications you can find in the T/S parameters sheet from the driver. Keep in mind that in real world of course those T/S parameters will change by heat and ambient temp. enz. This is a near approach. In our case it stabilize the amplifier more, then just add the standard Zobel.
Active is way to go, and gives the benefit that you know the driver the amplifier will see, and you can truly design the amplifier for the driver.
With kind regards,
Bas
Then this comes a bit close...This is what I am listening to right now
With kind regards,
Bas
Nice... I am tempted to give it a try... any additional details you are allowed to share?
why are u using the op-amps in non-inverting configuration? they work better when used in inverting mode...
BTW, let me say that the "MyRef" design is much better than this. There's not only THD, noise floor and amazing (as well as meaningless) DF figures.
In the output circuit there are many other resistances: speaker itself, cross-over, cables, connections, etc. Summed up, these are in the order of several Ohm!
Thus, as far as the amplifier output impedance is negligible with respect to that (and even a few 100s of mOhm is!), its actual value does not really make any practical difference anymore.
That is, to all practical effects, "a dampingfactor of over a million" cannot do any better than a DF of "just" 80 (or even less).
'cause the real, effective damping factor is limited by the other resistances in the circuit and, even in most favorable case, it can never be more than ~ 10 (referred to 8 ohm).
Besides, as shown by Mauro (and others), it's much more important how the output impedance of the amplifier varies with frequency in amplitude & phase rather than it's actual value.
Carefully controlling that "reverse driven" amplifier behavior is the main (and most important) "secret" of the MyRef. And the main reason why it sounds so much different than a "basic", simple chip amp.
Last edited:
I can see that my thread still attracts some attention
I have managed to get hold of some Black Gate capacitors, and been playing with different configurations, let me share my findings:
So I will definitely be going with BGN 100uF (with or without 4.7uf BGN) for Mid and Hi channels.
I'm still considering my choices as to what to use for Low. Possibly BG STD on their own, or with 4.7uF BGN - haven't tested these combinations yet. If I don't like them, I'll go for 1500uF FC + 4.7uF BGN.
I have managed to get hold of some Black Gate capacitors, and been playing with different configurations, let me share my findings:
Panasonic FC 1500uF - Very good - to me, the point of reference for all the others. Pleasant, exciting and spacious sound. Very neutral. Bass control is somewhat lacking.
Panasonic FC 1500uF + Panasonic FC 10uF in PSU - Bass control improves, and the sound generally becomes more focused, although some of the "air" and ambience is lost. Overall, I'd consider it an upgrade over 1500uF on their own.
Panasonic FC 1500uF + Black Gate N 4.7uF in PSU - an electrifying combination, very exciting sound. Bass precision is hard to believe; it's almost sculpted in the air (although because of that, it may give an impression of there being less bass weight overall). Rendering of transients is very sharp, bordering on edginess. Magnificently reproduced timbres. As with FC 10uF, it does take away some air, but not so much so, and the gains are well worth it overall.
Black Gate N 100uF - Initial impression was that they are hardly an improvement over the above combination - the sound was somewhat less "exciting", although very smooth. After some more listening it became clear that BG is the better capacitor though. The amplifier sounds more "HiFi". Bags of detail, sound is very holographic, airy. There is something correct, natural about the way it reproduces music. Bass response is also clean and natural; sometimes you might wish for a bit more bass "meat", but it still sounds very pleasant.
Black Gate N 100uF + Black Gate STD 1000uF at the PSU - Still very good detail, although compared to 100BGN the sound is a bit muffled. Bass has clearly more power, but the way it is reproduced is "rounded". It makes me think of a big, furry bear - It packs a punch and it's definitely not clumsy - but still, it's not like the vicious tiger I'd like it to be. Otherwise very similar to 100BG - very "HiFi" kind of sound. All in all, I preffered 100BGN on their own though.
Black Gate N 100uF + Black Gate N 4.7uF at PSU - adding 4.7BGN has effect somewhat similar to what it does for FCs (described above), but the gains are much less pronounced, and I found it hard to decide whether it's an improvement or not. I switched back and forth between this and the plain 100BGN configuration many times. 100BGN on their own sound more coherent and have a little bit more detail - adding 4.7uFs seems to take some of the background detail, but in exchange gives you slightly better attack and timbres, and a bit tighter bass (it is a bit like lens - shifting the focus from wide to narrow).
So I will definitely be going with BGN 100uF (with or without 4.7uf BGN) for Mid and Hi channels.
I'm still considering my choices as to what to use for Low. Possibly BG STD on their own, or with 4.7uF BGN - haven't tested these combinations yet. If I don't like them, I'll go for 1500uF FC + 4.7uF BGN.
Last edited:
I have successfully built a P2P GC by the way:
(For the curious - pictured above are: LM3875TF V1, Caddock and Riken resistors, BG N 100uF and silver-plated copper heatsink; the feedback resistor is soldered right at the very base of chip legs, the loop can physically be only about 0.5mm shorter - I'm quite proud of that one)
I'm far from being an expert, but from my practical experience so far, I can say that for anyone wanting to make a great sounding Gainclone these things should be kept in mind:
1. Quality of components definitely has fundamental influence on the sound. When experimenting, the resulting changes can range from very slight (changing a single resistor or using a better wire) to quite substantial (changing capacitors) - but ALWAYS changing a component will influence the sound. The trick here is - you may not always like the result, even if it is supposed to be an "upgrade".
2. Cable arrangement is CRITICALLY important for achieving good sound. The difference between a "spaghetti" and a perfect layout is easily as significant as a major upgrade in components. In fact, I believe it is impossible to achieve great sound even with the best components ever made by humanity unless the cable layout is good.
My advice here:
- in a cable pair (plus and minus), keep both cables parallel and equidistant for their entire length. I know a lot of people like twisting, but perfectly straight and parallel is the way to go in my opinion - it just sounds correct, "right" somehow.
- keep every pair of cables as far as possible from other cables
- especially keep signal cables away from output cables
- avoid cable pairs running parallel to other cables; if it is unavoidable that cables cross, make them form a 90deg angle.
- use as short cable runs as you can get away with (obviously)
- in fact - use as little metal as you can get away with; this way you will minimise eddy currents
And before someone shouts at me for not following my own advice - it is just a prototype in the above picture, not a finished amp I will probably use DNM cables throughout in the final version.
An externally hosted image should be here but it was not working when we last tested it.
(For the curious - pictured above are: LM3875TF V1, Caddock and Riken resistors, BG N 100uF and silver-plated copper heatsink; the feedback resistor is soldered right at the very base of chip legs, the loop can physically be only about 0.5mm shorter - I'm quite proud of that one)
I'm far from being an expert, but from my practical experience so far, I can say that for anyone wanting to make a great sounding Gainclone these things should be kept in mind:
1. Quality of components definitely has fundamental influence on the sound. When experimenting, the resulting changes can range from very slight (changing a single resistor or using a better wire) to quite substantial (changing capacitors) - but ALWAYS changing a component will influence the sound. The trick here is - you may not always like the result, even if it is supposed to be an "upgrade".
2. Cable arrangement is CRITICALLY important for achieving good sound. The difference between a "spaghetti" and a perfect layout is easily as significant as a major upgrade in components. In fact, I believe it is impossible to achieve great sound even with the best components ever made by humanity unless the cable layout is good.
My advice here:
- in a cable pair (plus and minus), keep both cables parallel and equidistant for their entire length. I know a lot of people like twisting, but perfectly straight and parallel is the way to go in my opinion - it just sounds correct, "right" somehow.
- keep every pair of cables as far as possible from other cables
- especially keep signal cables away from output cables
- avoid cable pairs running parallel to other cables; if it is unavoidable that cables cross, make them form a 90deg angle.
- use as short cable runs as you can get away with (obviously)
- in fact - use as little metal as you can get away with; this way you will minimise eddy currents
And before someone shouts at me for not following my own advice - it is just a prototype in the above picture, not a finished amp
I will probably use DNM cables throughout in the final version.
Last edited:
Congratulations!
But I have just one nit to pick, with what you said:
I think that for all wire or conductor pairs, it is very important to keep them as close together as possible. Otherwise, they can easily have currents induced in them by time-varying magnetic and electro-magnetic fields, such as those from your AC power components and conductors (and your outputs), which could cause hum (and does, very often) or other noise. Twisting them tightly together is usually better, too, since it helps to prevent things from coupling into them in other ways, and also keeps them as close together as possible. For inputs, especially, it would be even better to use a shielded cable i.e. with two conductors inside plus shield outside, with the shield grounded only at the input end, to the chassis.
I do not see how keeping an input signal/ground wire pair "equidistant" from each other could have a good (or much of any) effect. But not keeping them AGAINST each other could definitely have a bad effect.
Also, do you recommend not tightly twisting together the AC wire pairs, nor the DC power supply pairs? I believe that all pairs should be tightly twisted together. Many people have had problems when they were not.
But everything else you said sounds correct.
But I have just one nit to pick, with what you said:
I think that for all wire or conductor pairs, it is very important to keep them as close together as possible. Otherwise, they can easily have currents induced in them by time-varying magnetic and electro-magnetic fields, such as those from your AC power components and conductors (and your outputs), which could cause hum (and does, very often) or other noise. Twisting them tightly together is usually better, too, since it helps to prevent things from coupling into them in other ways, and also keeps them as close together as possible. For inputs, especially, it would be even better to use a shielded cable i.e. with two conductors inside plus shield outside, with the shield grounded only at the input end, to the chassis.
I do not see how keeping an input signal/ground wire pair "equidistant" from each other could have a good (or much of any) effect. But not keeping them AGAINST each other could definitely have a bad effect.
Also, do you recommend not tightly twisting together the AC wire pairs, nor the DC power supply pairs? I believe that all pairs should be tightly twisted together. Many people have had problems when they were not.
But everything else you said sounds correct.
I don't going to argue the meaning of damping factor again. If you think it is meaningless, fine
I think different. If you choose inverting configurations, you also need higher resistor values with increase the noise-floor. Then one would argue that noise-floor isn't meaningful as well. Again, I think different
I am glad you share your enthusiasm for Mauro's design, but before building my design idea in real life, you can't simple say one is better then another. I never build Mauro's so I can't say. But I can say, that personally I never liked the many compensation poles needed in his design.
THD, Noisefloor and other measurements might not say everything, I build and experiment with this kind of circuits by ears
With kind regards,
Bas
Last edited:
I can see that my thread still attracts some attention
I have managed to get hold of some Black Gate capacitors, and been playing with different configurations, let me share my findings:
Did you "burn in" any of these capacitors before doing your listening tests? Reports indicate the BG caps take a long time, and I know the Panasonics (which I use exclusively for the 1000uF-2200uF range in my amps) take a good two days to sound better.
Of course I'm sure to get some people saying they don't believe in burning in capacitors, but I've certainly heard the difference (harshness in particular lessens) and so have many others.
@gootee - re. parallel vs twisted
I read a bit about Eddy currents, and did some experimenting on this... And it seems that negative effect of increased exposure to external electromagnetic "pollution" in parallel configuration is far outweighed by decreasing the amount of Eddy currents, compared to twisted. DNM cables are built to minimise them, their website briefly explains the principle.
I actually bought DNM interconnect to see if it would be any good, despite being exposed, unshielded etc. - It turned out to be by far the best cable I've used (although I've never used used any overpriced "HiEnd" cables). Music is so flowing, easy to listen to. It just sounds right. It has an "old NAD" quality to it (in the sense that it's very musical and pleasant). The interconnect is made of OFC copper and insulated in LDPE, so it's definitely not due to some fancy materials either.
It makes sense when you think about it - sound is AC, and will induce currents in every conductor near it. Parallel arrangement keeps the resulting distortion to minimum, and in my experiments parallel actually sounded better. Hence my recommendation.
Keeping conductors as close as possible is by no means the best way to go either - choosing the right distance between conductors is all a play of trade-offs. You may reduce inductance, but you will increase capacitance as a result. And there's Eddy currents, skin effects, and all the rest of it to keep in mind...
@Redshift187
In my notes I do mention that on initial listening tests Black Gates were somewhat disappointing - it could have been due to lack of burn-in. After the first test, I left the amp playing for a day and it indeed sounded better when I came back. If I were to describe my subjective impressions, after a few days the sound got smoother, more full-bodied, but also more delicate somehow, easier to listen to.
In other words, yes, I do think there was some kind of burn-in process going on.
By the way, I should have probably mentioned what were the conditions for my listening tests:
- amp: classic, non-inverting stereo Gainclone, with Caddock and Riken resistors and Noble pot
- CD: modified Marantz CD52 MKII Special Edition
- speakers: Mordaunt Short System 442 - hi end speakers from early 90' (they improved no end after replacing old capacitors in crossover - see my earlier posts)
- cables: DNM Interconnect, Ecosse 3.2 (tri-wired)
Music:
vocal: Norah Jones, Michael Jackson*, Phil Collins, Elton John
acoustic/instrumental: Dave Brubeck, Chuck Mangione*, The Corrs (unplugged), Dan Ar Braz
symphonic: Mike Oldfield (Music of the Spheres), Mozart
bass/dynamics: Justin Timberlake, Nelly Furtado, Black Eyed Peas, Mortal Kombat Soundtrack
*my main "testers"
I read a bit about Eddy currents, and did some experimenting on this... And it seems that negative effect of increased exposure to external electromagnetic "pollution" in parallel configuration is far outweighed by decreasing the amount of Eddy currents, compared to twisted. DNM cables are built to minimise them, their website briefly explains the principle.
I actually bought DNM interconnect to see if it would be any good, despite being exposed, unshielded etc. - It turned out to be by far the best cable I've used (although I've never used used any overpriced "HiEnd" cables). Music is so flowing, easy to listen to. It just sounds right. It has an "old NAD" quality to it (in the sense that it's very musical and pleasant). The interconnect is made of OFC copper and insulated in LDPE, so it's definitely not due to some fancy materials either.
It makes sense when you think about it - sound is AC, and will induce currents in every conductor near it. Parallel arrangement keeps the resulting distortion to minimum, and in my experiments parallel actually sounded better. Hence my recommendation.
Keeping conductors as close as possible is by no means the best way to go either - choosing the right distance between conductors is all a play of trade-offs. You may reduce inductance, but you will increase capacitance as a result. And there's Eddy currents, skin effects, and all the rest of it to keep in mind...
@Redshift187
In my notes I do mention that on initial listening tests Black Gates were somewhat disappointing - it could have been due to lack of burn-in. After the first test, I left the amp playing for a day and it indeed sounded better when I came back. If I were to describe my subjective impressions, after a few days the sound got smoother, more full-bodied, but also more delicate somehow, easier to listen to.
In other words, yes, I do think there was some kind of burn-in process going on.
By the way, I should have probably mentioned what were the conditions for my listening tests:
- amp: classic, non-inverting stereo Gainclone, with Caddock and Riken resistors and Noble pot
- CD: modified Marantz CD52 MKII Special Edition
- speakers: Mordaunt Short System 442 - hi end speakers from early 90' (they improved no end after replacing old capacitors in crossover - see my earlier posts)
- cables: DNM Interconnect, Ecosse 3.2 (tri-wired)
Music:
vocal: Norah Jones, Michael Jackson*, Phil Collins, Elton John
acoustic/instrumental: Dave Brubeck, Chuck Mangione*, The Corrs (unplugged), Dan Ar Braz
symphonic: Mike Oldfield (Music of the Spheres), Mozart
bass/dynamics: Justin Timberlake, Nelly Furtado, Black Eyed Peas, Mortal Kombat Soundtrack
*my main "testers"
Last edited:
Hi,
Looks nice.
I just (for fun) posted my own years ago build (also hardwired) here:
http://www.diyaudio.com/forums/chip-amps/79303-chip-amp-photo-gallery-29.html#post2227576
Another suggestion, instead of DNM, try some solid siver, my style...
Ciao T
Looks nice.
I just (for fun) posted my own years ago build (also hardwired) here:
http://www.diyaudio.com/forums/chip-amps/79303-chip-amp-photo-gallery-29.html#post2227576
Another suggestion, instead of DNM, try some solid siver, my style...
Ciao T
Thanks Thorsten!
All of DNM cables are solid core, but unfortunately made of copper. I would love to use silver version instead, but they don't make it (and even if they did, it would've been prohibitively expensive - their copper cables are quite expensive already).
I'm actually seriously considering making silver "knock-offs" of their cables. Parameters like wire diameter and spacing are easy to copy, I'm only somewhat unsure as to microphonics/vibration control (which they claim was among their design considerations).
Another thing I don't like about their cables is that they are insulated with plastic - I very much prefer natural materials. In my version I would probably use paper as insulator (which could also be used to make the spacing ribbon). I find it hard to imagine that good old paper would have worse vibrational properties than some nasty polyethylene...
All of DNM cables are solid core, but unfortunately made of copper. I would love to use silver version instead, but they don't make it (and even if they did, it would've been prohibitively expensive - their copper cables are quite expensive already).
I'm actually seriously considering making silver "knock-offs" of their cables. Parameters like wire diameter and spacing are easy to copy, I'm only somewhat unsure as to microphonics/vibration control (which they claim was among their design considerations).
Another thing I don't like about their cables is that they are insulated with plastic - I very much prefer natural materials. In my version I would probably use paper as insulator (which could also be used to make the spacing ribbon). I find it hard to imagine that good old paper would have worse vibrational properties than some nasty polyethylene...
Then this comes a bit close...
With kind regards,
Bas
Sebastian, care to share your schematic including the DC-servo's? Im interested in this design and in need of a summer project
Hi,
Yes, I know the Cables, the company and Dennis & Son.
Go for it. How about using PTFE Tubing and then PTFE Strips?
Well, they also only cook with water. The commercial cables I used to make for Realhifi where several steps up sonically from DNM.
Interesting to try. In this case use goldplated silver and use some kind of cardboard.
I like PE insulation, it is quite soft and does not seem to sound very plasticky.
Ciao T
Yes, I know the Cables, the company and Dennis & Son.
Go for it. How about using PTFE Tubing and then PTFE Strips?
Well, they also only cook with water. The commercial cables I used to make for Realhifi where several steps up sonically from DNM.
Interesting to try. In this case use goldplated silver and use some kind of cardboard.
I like PE insulation, it is quite soft and does not seem to sound very plasticky.
Ciao T
After much reading and experimenting, I too came to the conclusion that a flat ribbon with equal spacing of conductors is best. Minimal dielectric is also key. I was going to use solid 5-9s silver sewn into cotton for the best dielectric solution. As a proof of concept I built some using clear packing tape. The result was so good, I never moved on to cotton.
I buy pure silver wire at Myron Toback mail order from NYC. It's a jewelry supply house that sells by the ounce. an once of pure silver is $20, and at 28 gauge that is ALOT of wire... Page 51 - Myron Toback 2009 Catalog
Based on research by Jade on audio asylem, I went nuts and added one strand of pure gold on the + conductor + 2 strands of silver all at 28 gauge. The neg has 1 strand of 24 gauge silver. JadeD claims pure gold is the best. At $10 a foot, gold gets expensive, and I'm not sure its that much better.
At any rate, the interconnect is amazing. I also built speaker cables the same way with a lot more conductor and I think they are pretty decent too.
Attachments
I like the way you put it!
Of course, I do realise DNM cables are not the end of the road, but on the other hand - they definitely got something right. So I'm going to do some experimenting based around their concept.
@wlowes - great to hear that someone actually went and tried this!
I agree with your point about minimising dielectric - that's why I wanted to use thin paper insulation (although I was considering cotton too).
One problem I have though, is that silver is not very stable when not sealed... And natural insulation materials are almost always loose in some way, so I'm not really sure which way to go from here. I thought of oiling paper - but then it becomes extremely difficult to work with, and rather messy, to be honest. And 100% seal is still not guaranteed.
Thanks for the link to the silver suppliers - unfortunately I live in UK, so I will probably be better off sourcing something more locally.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.