Dear @cumbb,
please let us know exactly what your load looks like.
Which driver, which housing, which electrical, passive corrective measures / eq ... etc.
In the simplest case, carry out an impedance recording with DATS-3 and post the data or the file.
#
Regarding the impedance number, the load, I would like to point out that the current Musical Fidelity A1 is completely unsuitable for 4 Ohm types --> so here too:
better no impedance minimum below 8 Ohm ..!
please let us know exactly what your load looks like.
Which driver, which housing, which electrical, passive corrective measures / eq ... etc.
In the simplest case, carry out an impedance recording with DATS-3 and post the data or the file.
#
Regarding the impedance number, the load, I would like to point out that the current Musical Fidelity A1 is completely unsuitable for 4 Ohm types --> so here too:
better no impedance minimum below 8 Ohm ..!
Choosing between the original and the BD441 442 as output transistors turns out much more challenging for me than I thought it would be. At night and especially higher quality music I want to listen with BDs. Louder, lower quality (what I listen to most), during the day time, the original transistors seems to be more fun listening experience, because more impactful bass, more body and meat on the bones in general. Louder capability of playing by maybe 6 dB without anomalies (which is too loud anyway but still I note). The highs are much less nuanced, but still good.
Tomorrow I may have time to build my headphone cables and listen with headphones. Although I suspect BD442 441 will be clear winners here. 1 stage follower probably better than either in 2 stage configuration. Last few days I've been playing with driver transistors. Both output transistors to me in the current configuration make this a very fun listening device. In many ways much better than my Tandberg 3026A. Although Tandberg had a slight edge in the way it would Flood you with quite nuanced sound. This is much cleaner though and even the bass is much tighter.
I have already achieved what I set out to do originally - replace the Tandberg with something equal or better. This is much better (not even close to Tandberg before modifications). I learned a lot and had a lot of fun along the way. I wish I could spend a bit more time on audio, but I don't know if I can justify it since I do have a very nice way of listening now already (which was the reason I started) and the time we live in is so incredibly nuanced and exciting.. I also need to give more time for the healing and training of the body as this will be of use for the rest of my life... I will work more on the amplifiers, since I clearly enjoy it so much! But please pardon me if I'm not as quick as you guys. This has been an amazing journey so far and the music really plays!
Tomorrow I may have time to build my headphone cables and listen with headphones. Although I suspect BD442 441 will be clear winners here. 1 stage follower probably better than either in 2 stage configuration. Last few days I've been playing with driver transistors. Both output transistors to me in the current configuration make this a very fun listening device. In many ways much better than my Tandberg 3026A. Although Tandberg had a slight edge in the way it would Flood you with quite nuanced sound. This is much cleaner though and even the bass is much tighter.
I have already achieved what I set out to do originally - replace the Tandberg with something equal or better. This is much better (not even close to Tandberg before modifications). I learned a lot and had a lot of fun along the way. I wish I could spend a bit more time on audio, but I don't know if I can justify it since I do have a very nice way of listening now already (which was the reason I started) and the time we live in is so incredibly nuanced and exciting.. I also need to give more time for the healing and training of the body as this will be of use for the rest of my life... I will work more on the amplifiers, since I clearly enjoy it so much! But please pardon me if I'm not as quick as you guys. This has been an amazing journey so far and the music really plays!
The NAD is a fine example of the fact that audio electronics should be current-related: an optimal sound with low or high currents, or in between. It is a small living room amplifier that should be able to handle a party, but cannot withstand a disco, gym or open air;-)
But I would suggest the vast majority to start by getting to grips with their loudspeakers. As many as 99 out of 100 loudspeakers on the market do not take the simplest mechanical principles into account in order to reproduce a clean sound)-;
I'll tie in with my listening experience pnp = black, colorless: I now compare this hearing to culture, motif, milieu Phototgraphy. It is extremely demanding to listen with these: as with a black and white photo, you have to know the objects depicted in order to recognize them. And even a different distortion behavior of different pnps among each other, which was recognizable as metal clanking in the npns, cannot be assigned at first, is not perceived as such noise but as some kind of distortion. Only when I became aware of the metal distortion of TO-220 compared to TO-92 or TO-126, for example, was I able to assign it and "subtract" it from the image. Comparable: You can't just photograph anything without knowing the addressee and their "cultural" background.
But I would suggest the vast majority to start by getting to grips with their loudspeakers. As many as 99 out of 100 loudspeakers on the market do not take the simplest mechanical principles into account in order to reproduce a clean sound)-;
I'll tie in with my listening experience pnp = black, colorless: I now compare this hearing to culture, motif, milieu Phototgraphy. It is extremely demanding to listen with these: as with a black and white photo, you have to know the objects depicted in order to recognize them. And even a different distortion behavior of different pnps among each other, which was recognizable as metal clanking in the npns, cannot be assigned at first, is not perceived as such noise but as some kind of distortion. Only when I became aware of the metal distortion of TO-220 compared to TO-92 or TO-126, for example, was I able to assign it and "subtract" it from the image. Comparable: You can't just photograph anything without knowing the addressee and their "cultural" background.
Attachments
My Tandberg used to run really hot. It was too hot to run in a room in the summer. I think this may be the slight edge that it had that I was hearing - the Flooding or Swelling as I described it. Maybe I was listening to it mostly in class A, even if it's a class AB amplifier.
There was some kind of presense that was emotional and unrelenting and it captured a lot of detail too. I can't describe it any better at this time.
There was some kind of presense that was emotional and unrelenting and it captured a lot of detail too. I can't describe it any better at this time.
I quickly made a 2-stage "crossed" follower. Initially without anything, I would only react if problems arose. The sound is more overcast, more tired, more doughy than a single-stage follower, less "flowery". But very balanced and fine. The BD 441 sounds very focused, the BD442 even tighter: this follower also shows these characters: it is tighter, more focused and has more spatial depth with uncorrected connection compared to almost all other power amplifiers. If it is too clear, one should move the speakers slightly apart.
I think it will be one day. A bit here and there, but it will be round enough;-)
I think it will be one day. A bit here and there, but it will be round enough;-)
Attachments
![DSCN0606[1].JPG](/community/data/attachments/1266/1266687-0efc5c47c8b080a4d41032b3c6eb3f27.jpg?hash=DvxcR8iwgK)
![DSCN0607[1].JPG](/community/data/attachments/1266/1266688-5bad280b700bc050075364443782577b.jpg?hash=W60oC3ALwF)
So you turned it into a diamond buffer? That’s not actually a bad idea, since it does have thermal stability advantages over a regular EF. But all 4 trannies really need to be heat sinked together to gain that advantage.
IIRC, some NADs actually use diamond buffers. They also use “diamond buffer like” front ends, and run them as conventional (push pull, not SE input like the 3020 and it’s cousins) CFAs.
IIRC, some NADs actually use diamond buffers. They also use “diamond buffer like” front ends, and run them as conventional (push pull, not SE input like the 3020 and it’s cousins) CFAs.
Dear @cumbb !
For better or worse, one have to realize that your tinkering has absolutely nothing in common or to do with a NAD amplifier.
#
🙁
For better or worse, one have to realize that your tinkering has absolutely nothing in common or to do with a NAD amplifier.
#
With a request for a comprehensive technical explanation of the mechanical relationships.
🙁
I think it's just the S300, not really a NAD design, that had a diamond output.
Got me thinking you may be onto something with that diamond buffer. The BD441/2 are perfectly adequate for an easy to drive 8 ohm load (one that doesn’t dip to 6) as far a power handling. Three in parallel would beat a 2N3055. Maybe a DB, with one driving three, and you have something that’s technically sound for a +-30V rail and real 4/ohm loads. Switch to MJE243/253, and then you even have their fT advantage. All with cute little TO-126’s. And you even get the fun of matching 4 transistors - well, 16 really because you get to do it twice per channel. Matching would be recommended. And the closer NPN is to PNP the better too, as that reduces input bias current.
Wouldnt be the first time somebody just replaced a NAD output stage. @poundy used LatFets and was pleased with the result.
Wouldnt be the first time somebody just replaced a NAD output stage. @poundy used LatFets and was pleased with the result.
Oh my God, this is such an unbelievably good headphone amplifier! I've heard many things, but I have never heard my headphones play this nicely, nor have I heard commercial amps play this cleanly and powerfully either. Although I have not heard many high end amps designed for headphones, but I heard some that were reviewed as giant killers and they were really bad.
However I already immediately hear some high frequency fuckery with the channel with BD442 BD441 as output transistors. Only when music is playing, but there is a kind of a high frequency static that sounds kinda like electromagnetic interference on cables, but it's not that. This I only noticed with headphones and noticed quite immediately. Original output transes play clean and the before described characters remain. BD442 441 more nuanced especially on higher frequencies, voices, while original ones have more body, bigger bass, with headphones it sounds more dimensional, less flat in presentation than the BD. Not much detail is really lost, more a character change, but due to the noise I hear when music is playing, unless I am able to fix it - the original output transes are CLEAR winners for my use cases. I may try to change which channel plays which transes to see if the issue travels to the other channel successfully.
Also today is a great day. After years of mostly failed attempts, today I became a man capable of soldering litz wires. 999 pure Tin works perfectly. This is a great wire here. Advertised as 6 or 7 N OCC silver plated copper from Furakawa Japan, but in any case it sounds as good as some of the best cables I've ever heard. Very alive and preserves the bass nicely. Similar character to Nordost, but way more conductor for the price.
However I already immediately hear some high frequency fuckery with the channel with BD442 BD441 as output transistors. Only when music is playing, but there is a kind of a high frequency static that sounds kinda like electromagnetic interference on cables, but it's not that. This I only noticed with headphones and noticed quite immediately. Original output transes play clean and the before described characters remain. BD442 441 more nuanced especially on higher frequencies, voices, while original ones have more body, bigger bass, with headphones it sounds more dimensional, less flat in presentation than the BD. Not much detail is really lost, more a character change, but due to the noise I hear when music is playing, unless I am able to fix it - the original output transes are CLEAR winners for my use cases. I may try to change which channel plays which transes to see if the issue travels to the other channel successfully.
Also today is a great day. After years of mostly failed attempts, today I became a man capable of soldering litz wires. 999 pure Tin works perfectly. This is a great wire here. Advertised as 6 or 7 N OCC silver plated copper from Furakawa Japan, but in any case it sounds as good as some of the best cables I've ever heard. Very alive and preserves the bass nicely. Similar character to Nordost, but way more conductor for the price.
Attachments
Last edited:
Just so it's clear, are you referring here to the two stage follower you just made or are you describing here the one stage follower as opposed to the one you just made and listened?
Last edited:
I'll have to run this through AI a few times to understand, but are you saying the 302 has a PP input? If I have a balanced DAC, could I use advantage of this instead of using the RCA inputs?
Just yesterday reworked my RCA cables - sound like a dream. Six silver cores per pole in teflon tubes. Short as that greatly reduces noise floor in unbalanced cables. Much more effective than any screen! The silver is from Latvian folk named Lavricables. Not the best (some OCC silver is on another level) but it's great for the price. And it's as good or better than any copper I've heard regardless of price!
Attachments
Last edited:
Both as an SE amp and follower and as a single-stage pp follower and also as a two-stage pp follower and as diodes. And all others;-) Also this character can be heard always. And it is also easy to prove using the example of speaker placement, by correcting the placement in order to balance the sound image again.
Sir, when matching the transistors, what should I focus on most, the Ube, Ie or hFE? My first batches of BD442 441 had completely different hFE and Ie, only the Ube was close enough to find matches. Now my new batch of BD442 have almost two times smaller hFE and Ie so a possibility of matching on that opens up. Well more the current, less so amplification, but what should I focus on most when matching?
I now connected less sensitive headphones - the Hifiman HE6SE v2 (~83 dB/mW at an impedance of 50 Ohm, this translates to a sensitivity of 96 dB at 1 V) and the issue is mostly gone although still audible. So this is closely related to sensitivity. Previously I played with planars (Hifiman Edition XX) that even a laptop or a phone can power to produce sound to appreciable volume. Just reporting this, because I made it sound like a horrible problem. It is a problem, but not horrible. This also explains why I wasn't able to hear it on my speakers even though they are very resolving.
This amp puts a smile on my face.
It's been a while since last I listened to headphones and really enjoyed it.
Last edited:
If you are doing a diamond buffer, it is necessary to match the NPNs to each other and the PNPs to each other, for vbe and hFE. Putting in degeneration resistors in the output pair removes this requirement for thermal stability, but they would be nice to at least minimize. Scaling of the bias currents between the two stages is done by the ratio of device areas in ICs, for discretes, that’s paralleling. Mismatch between NPN and PNP vbe results in an offset voltage. Mismatch between hFE results in a net input bias current. It is nice to eliminate both, but neither is a show stopper. Paralleled transistors need to be matched for sure, if parallel outputs are used. All the transistors need to be at the same temperature to maintain any of the advantages of a diamond buffer and to be thermally stable at all. All on one heat sink, co-located to the extent possible.
Properly implemenTed, it is probably more thermally stable than the original emitter-resistor-less EF2 that was there in the first place. In practice, I’d probably put .047 ohms in the emitters which would allow safe paralleling and allow for some temperature gradient without runaway. If you used three outputs for one driver, that’s a 120W output device which you can actually beat on.
I added a couple more of 2.8 mF PEH227 to the PSU. The noise floor is perhaps a little lower and a lot smoother, but it persists. Other than that the sound seems to have become even more enjoyable and powerful and serious and detailed. Well, detail didn't change, but everything's more effortless, more musical so it may be perceived as higher resolution. Or I still can't get over how amazing my headphones sound with this amp truly awesome. For sure the noise floor has improved (smoother, less audible) and overall enjoyableness factor has raised or at the very least hasn't become reduced.
Now I tried making left channel into 1 stage follower with BD output transistors, I tried it as 2 stage with BD outputs, I changed it into newer batch BDs, I changed the output transistors into the original ones, but the high frequency fuckery that is static like is still very present with the sensitive headphones. That's a bit unfortunate, because all other factors being equal between the channels, I have no idea where the issue may be. On the positive side, I know it's not the BD output transistors that are causing this so I feel free to use them if I want at the end of the day. But this issue will make it much harder to do critical listening and comparisons with headphones rather than just one speaker. Oh well. I might have damaged some part of this channel. Most won't ever notice it. It's subtle. Unless you listen through sensitive headphones. On most speakers this will be completely inaudible.
Still even with speakers this complicates doing comparisons between different configurations because I lost ultimate trust in the left channel. So deciding on 1 vs 2 stage follower or which output transistors to use will be post poned for a later time, possibly with another amp entirely unless I find the culprit. The headphones will be very valuable in making this decision.
Although I did most of the modifications recommended by cumbb there's more to be refined. However, I don't feel like doing further refining before I even figured out which kind of output agreement I want to use. Besides diagnosing this issue or starting from scratch, I don't know what else to do with this project. Probably most important and valuable steps have been taken, I doubt it can sound much greater than it already does unless every corner and fascet has been polished and refined, for which this doesn't feel like the right timing. I guess I'll just try to study the diagrams and the basics of what is happening here at all and put down the physical tools for the time being.
Now I tried making left channel into 1 stage follower with BD output transistors, I tried it as 2 stage with BD outputs, I changed it into newer batch BDs, I changed the output transistors into the original ones, but the high frequency fuckery that is static like is still very present with the sensitive headphones. That's a bit unfortunate, because all other factors being equal between the channels, I have no idea where the issue may be. On the positive side, I know it's not the BD output transistors that are causing this so I feel free to use them if I want at the end of the day. But this issue will make it much harder to do critical listening and comparisons with headphones rather than just one speaker. Oh well. I might have damaged some part of this channel. Most won't ever notice it. It's subtle. Unless you listen through sensitive headphones. On most speakers this will be completely inaudible.
Still even with speakers this complicates doing comparisons between different configurations because I lost ultimate trust in the left channel. So deciding on 1 vs 2 stage follower or which output transistors to use will be post poned for a later time, possibly with another amp entirely unless I find the culprit. The headphones will be very valuable in making this decision.
Although I did most of the modifications recommended by cumbb there's more to be refined. However, I don't feel like doing further refining before I even figured out which kind of output agreement I want to use. Besides diagnosing this issue or starting from scratch, I don't know what else to do with this project. Probably most important and valuable steps have been taken, I doubt it can sound much greater than it already does unless every corner and fascet has been polished and refined, for which this doesn't feel like the right timing. I guess I'll just try to study the diagrams and the basics of what is happening here at all and put down the physical tools for the time being.
Attachments
#92
I've been meaning to do this. Removed J425, J412 and put in bridges at C440, C439. Put the driver transistors back in one more time.
I've heard too many different configurations today to be able to tell if there's any difference. Plays great. Perhaps even better, but also likely I'm still not used to how amazing my headphones play with this amplifier. Truly awesome.
Last edited: