I did like you do 🙂 My first project is making AKSA 55 clone. I tried to understand how Mr. Hugh Dean though when he made AKSA 55. But I think I achieved it not in AKSA 55 clone, but in "Perkutut" amplifier (not the free version). The mid's quality is sub verb. Here, people said: "ada emosinya" may be can translate as "engage" in English. But I am not satisfied with the quality of the low and high's frequency. The distortion at high frequency is not low enough, it can not produce the bell ring's sound believable. But when I used "blameless" topology with TMC compensation, the low and high frequency's quality is much better. Although the mid is not warm as "Perkutut" amplifier. When reproducing a complex music, "blameless" with TMC is better and sound believable to me.
NAP200 has a more spacious layout than 140. And look at all the vacant board space in a NAC152.
Naim obviously need to come to you to learn some "tricks".😉
It is incredible after 80 years plus of understanding how an amplifier works to really seem so lost when discussing sound quality. Mostly the problem is to hold the speaker cone in a way that is like the input signal. Whereas the input is mostly thought of as a voltage varying in time the output is a facimile which also has an interesting current component ( one can see it in the emitter resistors ).
My car is a VW Golf diesel TDi 90. When it was new it went up hills like they didn't exist. Now it's 15 years old it doesn't, The reason is the turbo which would take time and maybe money to fix. I would liken how it is now to how many amplifers are from day one, subtly broken. Fine on most music, but not as good as others on the hills. One could be forgiven if thinking my TDi 90 is as good as the day when it was new. By changing gear it almost is. I remember when it was a champion.
It is very likely a LM1875 built to be a little like the LAB 47 design would impress those who like Naim. The thing to understand is a Naim sound is 80% power supply and 20% the amplifier if Julian is allowed to remind us. If an LM1875 was put with a Nait power supply I suspect the difference would fool most Naim lovers. What you are hearing is the the pendulum of PSU and speaker.
My ex father in law had an Audi A6 with the same engine as my Golf giving > 140 BHP. It was a much larger car and much faster still with moderate fuel use. They were mostly different cars with the hand of VW in there somewhere. I would suggest the circuit we look at is as important that engine.
My car is a VW Golf diesel TDi 90. When it was new it went up hills like they didn't exist. Now it's 15 years old it doesn't, The reason is the turbo which would take time and maybe money to fix. I would liken how it is now to how many amplifers are from day one, subtly broken. Fine on most music, but not as good as others on the hills. One could be forgiven if thinking my TDi 90 is as good as the day when it was new. By changing gear it almost is. I remember when it was a champion.
It is very likely a LM1875 built to be a little like the LAB 47 design would impress those who like Naim. The thing to understand is a Naim sound is 80% power supply and 20% the amplifier if Julian is allowed to remind us. If an LM1875 was put with a Nait power supply I suspect the difference would fool most Naim lovers. What you are hearing is the the pendulum of PSU and speaker.
My ex father in law had an Audi A6 with the same engine as my Golf giving > 140 BHP. It was a much larger car and much faster still with moderate fuel use. They were mostly different cars with the hand of VW in there somewhere. I would suggest the circuit we look at is as important that engine.
In 1947 DTN Williamson ( who was 19 years old I think) invented the Worlds first amplifer that tried to answer all the questions, the Leak TL12 and a few others also were around. He realised that a feedback amplifier could cause problems and gave insights still as true today. I looked but did not find his quote about the sound of keys making the sound keys do as a test via microphone. It is said he designed with Peter Walker the Quad ESL 57 and Quad 2/22. I know for a fact the Marantz Model 9 was inspired by this amplifier. Hope you get the time to read some of it in the PDF with oscillosope diagrams. The sound of the amplifier built correctly sounds to be the lowest distortion ever I heard and the bass is very good. A little bit too much like a transistor amplifer if liking a typical valve sound, it's a bit dry. When I say distortion I mean music and not the statistics we often talk about. One nasty trait is getting 1.5 watts extra power by having high value grid resistors to the KT66. I would play with that. You can see it when the feedback loop is removed as a sudden leap in distortion at the 13 watt level. Nait One was a similar output level. Julian said it was the compromise he could live with.
http://www.radiomuseum.org/dsp_multipage_pdf.cfm?pdf=the_williamson_amplifier.pdf
This was designed by me in about 2000 ( In Houston Texas in summer by a small river watching the wildlife ) and was made in China called Tallon. I haven't looked to see if it's correct ( NPN when PNP etc ) . It's just my notes. It was on my shed wall for years before I made this copy. I am told it is still sold on the Chinese domestic market. My friend who wanted it said make it good, but not something you mind giving away. It had a very open and detailed sound ( Due to the 47R to VAS, play with it to give various sounds ). The AKSA 55 was new to me untill today. Notice I already had raw and shunt regulated DC to the sections. The input is slightly wrong. I corrected that in the real one. I seem to remember the current sources were tweaked. The Chinese company had no idea how it worked until we said " like an op amp "! They asked how the mid point was set !
Last edited:
There is nothing wrong with vacant board space. Even if the board is big, you will notice that certain parts are grouped close to each other. If the board is small, those who dont understand will less likely make a mistake because everything is close to each other.
Member
Joined 2009
Paid Member
hmmmm. well nfb implies three things, a) there must be enough gain to throw some away on the feedback, b) there must be loop compensation of some sort to ensure stability, c) there must be an error amplifier with a voltage divider from the output
My own experience is that b), the compensation, is the most important.
When Paul Stephenson took over Naim Audio I am told the designs changed. Do we have a circuit?
That circuit of 2000 of mine was done in 10 minutes to send with the friend. It was concept that wasn't much changed. I forget why the rush. The design was given up as the 50 sample amps were not made to a high standard. The ones that worked were fine. There was a dispute over the brand name also.
80% of that amp was from the Chinese company including a relay and resistor volume control. We didn't like their zero loop feedback power amp which seemed to have too little current to drive the output transistors. We took their excellent idea and made it more to our taste. One thing theirs had was real verve. It also sounded like singing through a comb with paper over it, Kazoo sound.
That circuit of 2000 of mine was done in 10 minutes to send with the friend. It was concept that wasn't much changed. I forget why the rush. The design was given up as the 50 sample amps were not made to a high standard. The ones that worked were fine. There was a dispute over the brand name also.
80% of that amp was from the Chinese company including a relay and resistor volume control. We didn't like their zero loop feedback power amp which seemed to have too little current to drive the output transistors. We took their excellent idea and made it more to our taste. One thing theirs had was real verve. It also sounded like singing through a comb with paper over it, Kazoo sound.
Last edited:
[wiki=]%[/wiki]
I think it was Ian who mentioned in another thread (broken guitar amp) that TDA2030 has high distortion but having sweeter sound. I think this is more in line with Naim sound spirit. LM1875 has much lower thd but the sound is not "organic".
Unfortunately, both types have inconsistent quality from product to product (not same batch of course).
I think it was Ian who mentioned in another thread (broken guitar amp) that TDA2030 has high distortion but having sweeter sound. I think this is more in line with Naim sound spirit. LM1875 has much lower thd but the sound is not "organic".
Unfortunately, both types have inconsistent quality from product to product (not same batch of course).
The Nait xs2 i heard thru youtube, is it paul or JV's? I like that one.
I think many of the ancient naim sound is related with the sound of tantalum. A better Naim should use better parts than that.
I think that also. The TDA 2030 is no more as far as I know nor the very cheap PCB from RS 434-576.
I often wondered what L165 would sound like. There is a hint that crossover distortion was thought about. It looks like a LM324N or LM339 input stage. It is unity gain stable which is a good quality.
I often wondered what L165 would sound like. There is a hint that crossover distortion was thought about. It looks like a LM324N or LM339 input stage. It is unity gain stable which is a good quality.
Yes because it is very critical in affecting the sound quality. But the detail is in how it can affect the sound.
Your circuit uses fast transistors but why do you use the same compensation with nap250? My prediction is you dont need that base stopper + RC values at all with fast transistor. Different transistors usually call for different compensation.
Today I looked at possible VAS transistors for an improved version of this amplifier. There are none! It might be 2N5401 and MPSA92 as only choices soon as ZTX753 might be on death row. They would be on their limits heat wise.
A Cascode might make sense ( read link on linearity bonus ). I selected BC560 and BF721 with a small resistance between collector BC560 and emitter BF721 ( Magic resistor, ). I am very pleased to see someone has written it up. I suspect the old BF470 is the same device. If anyone is worried about the biasing voltage loss there is enough voltage in the BF's to play with if we want to get more voltage swing, 1.5 watts dissipation is possible. BC546 for inputs perhaps to make life easy.
This set up might need the 22K 1K LTP pair changed to perhaps 33K 1K to restore the H2 H3 balance of distortion. The higher gain of BC560C might allow better slewing. The Cobo of the Cascode also helps.
One thing worth thinking about is making the amplifier input senstivity slightly higher. If 500 mV CD should drive it directly ( via a pot ). Stability should improve into the bargain. That might be enought stabilise the reduced VAS capacitance.
Self has additional info on Cascode VAS. He feels a Cascode current source has no value to replace ZTX653. I aggree.
BF720 might repalce the ZTX653. Even these might vanish soon as I suspect they were used in CRT monitors in the past. Hopefully their specs are rather good so might find new uses.
Although I wasn't intending to talk about Memory Effect it is worth saying a little. A speculation which seems almost to be fact is that a VAS whilst charging has a very different THD than when the same signal has passed many times, it is different when music. Perhaps that's right. The Cascode VAS is said to reduce memory effect. Sorry if I over simplifiy the idea.
Memory Distortion Philosophies - Part 5 : Circuits, continued
A Cascode might make sense ( read link on linearity bonus ). I selected BC560 and BF721 with a small resistance between collector BC560 and emitter BF721 ( Magic resistor, ). I am very pleased to see someone has written it up. I suspect the old BF470 is the same device. If anyone is worried about the biasing voltage loss there is enough voltage in the BF's to play with if we want to get more voltage swing, 1.5 watts dissipation is possible. BC546 for inputs perhaps to make life easy.
This set up might need the 22K 1K LTP pair changed to perhaps 33K 1K to restore the H2 H3 balance of distortion. The higher gain of BC560C might allow better slewing. The Cobo of the Cascode also helps.
One thing worth thinking about is making the amplifier input senstivity slightly higher. If 500 mV CD should drive it directly ( via a pot ). Stability should improve into the bargain. That might be enought stabilise the reduced VAS capacitance.
Self has additional info on Cascode VAS. He feels a Cascode current source has no value to replace ZTX653. I aggree.
BF720 might repalce the ZTX653. Even these might vanish soon as I suspect they were used in CRT monitors in the past. Hopefully their specs are rather good so might find new uses.
Although I wasn't intending to talk about Memory Effect it is worth saying a little. A speculation which seems almost to be fact is that a VAS whilst charging has a very different THD than when the same signal has passed many times, it is different when music. Perhaps that's right. The Cascode VAS is said to reduce memory effect. Sorry if I over simplifiy the idea.
Memory Distortion Philosophies - Part 5 : Circuits, continued
Last edited:
Member
Joined 2009
Paid Member
I looked at memory distortion articles. The main problem I have with this is that the solutions are more complex circuits that are clearly much more linear than the originals. When people try the improved circuits they quickly find the resulting amplifier outperforms, technically speaking, the original. But there's no way to separate out the impact of removing he memory distortion (by operating devices at constant power) from the overwhelming impact of improved circuit linearity. Subjectively at least, we can't be sure we had a problem with memory distortion in the first place.
Ouch! What if the supply rail reduced to about 24v, still none?
Im very interested to create the better clone but unfortunately i dont have a reference or comparator. Or may be i should use my other different amp topology for subjective preference comparison?? 🙁
Maybe a new project for the future, a clone of Lavardin? Here's one:
http://bas.elitesecurity.org/epsilon.html
Member
Joined 2009
Paid Member
I will disclose that my current employer made the items shown in the attached photos (I used to own a set of those AR38's until the foam rotted out)
Attachments
Last edited:
- Status
- Not open for further replies.