@marigno 
I do indeed appreciate this: finally we have realistic bill of materials. I have one pair of bare PCBs left and I will place an order for components these days in accordance to your BOM. Also there was a hear-say theory that servo is deterrent to the sound, and obviously we now know that that was yet another fake news. I will ad servo too.
I am just a modest amateur and will not even attempt to replicate your total makeover as presented in your previous post.
By the way, I have already suspected, based on the Ohm Laws, that R17 and R19 were overrated. Indeed, you suggest 0,6 watts. Thanks for that.
Thanks again
I do indeed appreciate this: finally we have realistic bill of materials. I have one pair of bare PCBs left and I will place an order for components these days in accordance to your BOM. Also there was a hear-say theory that servo is deterrent to the sound, and obviously we now know that that was yet another fake news. I will ad servo too.
I am just a modest amateur and will not even attempt to replicate your total makeover as presented in your previous post.
By the way, I have already suspected, based on the Ohm Laws, that R17 and R19 were overrated. Indeed, you suggest 0,6 watts. Thanks for that.
Thanks again
I cannot locate the resistors you say. R17/19 are not homolog, perhaps you are referring to another schematic.
Referring to my schematics WHA-217 CuE, the correct rating of R14/17 and R22/23 is 0.25W, according to Ohm. If you use the original 39R value for R22/R23, it is 0.5W. But you have to keep in mind that everything will work in a cabinet, where the temperature raises, becoming far higher than the ambient one. So I'm used rating resistor power around 4 times the calculations. Also, you must consider the ppm derating in an important point of the circuitry. So "abundare melius quam deficere" ("cum grano salis"), in order to have a stable working point. So I placed them 2W, not less, not more.
At the time of purchasing, I could not find any 2W Vishay metal oxide at 1% tolerance so you have to buy a lot of 10 at least and select matching values among them.
Have in mind that my BOM is intended for 70+70V, read the BOM carefully, there is explained where I changed values in function of the increased voltage.
IMPORTANT, important indeed: you MUST replace all polyester capacitors in the servo with polypropylene ones, not over 5% tolerance. I have somewhere my BOMs at Digikey and Mouser, (I'm an Italian living in US), if you are interested I can find a way to share them or extract and publish them here. I had a hard time searching and finding components fitting the PCBs. However, in the pictures I posted, you can see what I'm talking about.
Referring to my schematics WHA-217 CuE, the correct rating of R14/17 and R22/23 is 0.25W, according to Ohm. If you use the original 39R value for R22/R23, it is 0.5W. But you have to keep in mind that everything will work in a cabinet, where the temperature raises, becoming far higher than the ambient one. So I'm used rating resistor power around 4 times the calculations. Also, you must consider the ppm derating in an important point of the circuitry. So "abundare melius quam deficere" ("cum grano salis"), in order to have a stable working point. So I placed them 2W, not less, not more.
At the time of purchasing, I could not find any 2W Vishay metal oxide at 1% tolerance so you have to buy a lot of 10 at least and select matching values among them.
Have in mind that my BOM is intended for 70+70V, read the BOM carefully, there is explained where I changed values in function of the increased voltage.
IMPORTANT, important indeed: you MUST replace all polyester capacitors in the servo with polypropylene ones, not over 5% tolerance. I have somewhere my BOMs at Digikey and Mouser, (I'm an Italian living in US), if you are interested I can find a way to share them or extract and publish them here. I had a hard time searching and finding components fitting the PCBs. However, in the pictures I posted, you can see what I'm talking about.
https://www.diyaudio.com/forums/solid-state/134362-dartzeel-amp-schematic-build-15.html#post5418630
This reasoning has encouraged me to select MF resistors with lower wattage, 1W perhaps. Yet, I’m not sure if some nasty transients can occur there even though that is a part of the schematics with rather constant currents due to zener diodes which serve as Constant Current Generators.
With lower rated components there is more choices among the desired metal film resistors.
During the ongoing weekend I will sit down and prepare a purchase list for my new pair of Dartzeel clones. I might also try to make a replica of your advanced version.
I like these photos very much. Especially the one with the amplifier on the cart tied with a rope. Cool.
I hope once I will be able to do similar designs.
Goodfella!
In your schematics these two resistors are denoted with R18 (R21) and R22 (R23) all four with 2W ratings. In the Chinese schematics in post #141 these resistors are denoted with R17 and R18 (2W) and R18 and R20 (5W). I was forced to use wire wound resistors, selection that I don’t like. After some calculations and measurements I have realized that R19 and R20 conduct about 110 to 120 mA, which indicates about 500 mW. Similarly, R22 and R23 conduct about 10 mA, which indicates also about 500 mW.
This reasoning has encouraged me to select MF resistors with lower wattage, 1W perhaps. Yet, I’m not sure if some nasty transients can occur there even though that is a part of the schematics with rather constant currents due to zener diodes which serve as Constant Current Generators.
With lower rated components there is more choices among the desired metal film resistors.
During the ongoing weekend I will sit down and prepare a purchase list for my new pair of Dartzeel clones. I might also try to make a replica of your advanced version.
I like these photos very much. Especially the one with the amplifier on the cart tied with a rope. Cool.
I hope once I will be able to do similar designs.
Why so important? Not saying PP is not better than KP, it usually is, but in the original amp a wide use of polyester caps can be seen, as both coupling and decoupling. Why would the servo be so important in comparison? And why low tolerance?
https://www.diyaudio.com/forums/solid-state/134362-dartzeel-amp-schematic-build-15.html#post5418630
Goodfella!
In your schematics these two resistors are denoted with R18 (R21) and R22 (R23) all four with 2W ratings. In the Chinese schematics in post #141 these resistors are denoted with R17 and R18 (2W) and R18 and R20 (5W). I was forced to use wire wound resistors, selection that I don’t like. After some calculations and measurements I have realized that R19 and R20 conduct about 110 to 120 mA, which indicates about 500 mW. Similarly, R22 and R23 conduct about 10 mA, which indicates also about 500 mW.
This reasoning has encouraged me to select MF resistors with lower wattage, 1W perhaps. Yet, I’m not sure if some nasty transients can occur there even though that is a part of the schematics with rather constant currents due to zener diodes which serve as Constant Current Generators.
With lower rated components there is more choices among the desired metal film resistors.
During the ongoing weekend I will sit down and prepare a purchase list for my new pair of Dartzeel clones. I might also try to make a replica of your advanced version.
I like these photos very much. Especially the one with the amplifier on the cart tied with a rope. Cool.
I hope once I will be able to do similar designs.
The resistors feeding the zener, according to Ohm, must be 0.5W (10KR @ 65V) or 0.3W (6.8KR @ 45V).
For the same reason, the resistor on the CCSs must be 0.25W (100R @ 46 mA) or 0.6W (39R @ 118mA).
So, in both cases, it is a good idea to use 2W rating resistors and no bigger ones. I wouldn't use a 1W rating, in order to have a not so high temperature on the resistors. Remember everything must work in a closed cabinet and you do want good stability of all working points. I know 1W power rating allows a wider catalog but I recommend 2W.
You can follow my schematic in order to have a higher power, or, at least, a harder final stage with a lower impedance. But, if you want to improve the original Chinese clone without adding another stage, use the mod about Baxandall super couple. And do not duplicate the output pairs!
You will be able, no doubt! I am an accountant with a passion for music and Latin and could accomplish what I had in mind.
The "bravetta" cart was needed to transport the almost 60Kg fellow up/down stairs!
I always prefer MKP, I always try to avoid MKT in my projects. The only exception is the input capacitor that has very good reviews. And I do not particularly care about originality, if I could improve I would do. Then we must say that MKT capacitors on the servo are low-quality ones, like resistors and OpAmps. And, according to most of you, the servo should not be used. So, perhaps, it is due to the poor quality of components. I changed everything.
The tolerance of capacitors and resistors, in such a low pass integrator, has to be as low as possible in order to have it work well, I read it in different places over the net when I started working on this Cu Edition.
Thanks dear Marigno for sharing your knowledge.
After we have approached a sound solution regarding the sizing of these larger resistors I have realized that all my fears from inductance are unfounded. Completely. Let me explain. Obviously, these resistors are part of Constant Current Generators, hence they conduct almost purely DC current, well, except some noise. In this case impedance equals almost purely resistance of the device whilst inductance approaches zero at the steady state conditions.
For me this conclusion is valuable because my ultimate goal isn't to make a cheap amplifier - the way I am doing this will cost me a lot but what I want is to understand it and to improve my listening experience.
After we have approached a sound solution regarding the sizing of these larger resistors I have realized that all my fears from inductance are unfounded. Completely. Let me explain. Obviously, these resistors are part of Constant Current Generators, hence they conduct almost purely DC current, well, except some noise. In this case impedance equals almost purely resistance of the device whilst inductance approaches zero at the steady state conditions.
For me this conclusion is valuable because my ultimate goal isn't to make a cheap amplifier - the way I am doing this will cost me a lot but what I want is to understand it and to improve my listening experience.
Actually, I was a passionate musician actively too for about my initial 18 years, I have played multiple instruments and lived for music until I had to decide on my future - professional career. A friend of mine, a professional musician, told me briefly: "Listen, music is a great hobby but a lousy profession". I have decided for chemistry and I don't regret. Quite often I "pray God to turn me into an ear (totum ut te faciant aurem)" whilst listening to great musical performances.
You are right, if you already have them, you can use WW resistors in these particular applications.
My WHA-217 is driven by an Aikido pre, made out of tubes. I don't recognize myself anymore, I'm appreciating a lot Gregorian Chants too, a new discovery for me, not boring at all. This HiFi chain is very good for its realism: voices, single and choirs, are outstanding. Goosebumps! You can touch the singer. I was an ICT senior analyst, now I'm retired so I can practice my hobbies, music, and electronics. I like sharing my knowledge, it is the best way to increase it!
My WHA-217 is driven by an Aikido pre, made out of tubes. I don't recognize myself anymore, I'm appreciating a lot Gregorian Chants too, a new discovery for me, not boring at all. This HiFi chain is very good for its realism: voices, single and choirs, are outstanding. Goosebumps! You can touch the singer. I was an ICT senior analyst, now I'm retired so I can practice my hobbies, music, and electronics. I like sharing my knowledge, it is the best way to increase it!
Thank you Marigno for all the information you provided!
I will use your circuit as a base. I will not use a servo, no feedback cap (but I will leave a space in the board for the capacitor in case I need it), and probably remove the input cap as well, as in my simulation the dc at the input is around 50 mV and most preamps/DACs have output capacitors, and will use 1 pair of output transistors as per the original, operating at a lower voltage. I will also put a stronger RF filter.
We'll see how that goes.
It's worth mentioning that I plan to use a dc protection circuit. Don't want to damage my (expensive) speakers.
I will use your circuit as a base. I will not use a servo, no feedback cap (but I will leave a space in the board for the capacitor in case I need it), and probably remove the input cap as well, as in my simulation the dc at the input is around 50 mV and most preamps/DACs have output capacitors, and will use 1 pair of output transistors as per the original, operating at a lower voltage. I will also put a stronger RF filter.
We'll see how that goes.
It's worth mentioning that I plan to use a dc protection circuit. Don't want to damage my (expensive) speakers.
Last edited:
With the 4 2N5xx1 thermally coupled and no servo, my complete assembly, in the open air (not in the cabinet), shows +/- 1mV offset, stable for 20 mins, after this time I turned it off. So, perhaps, it is worth a try.
About the input cap: I was in doubt to remove it. Then I decided to leave it. Without it, the assembly, having no total DC feedback loop, is sensitive to the DC. Some pre, with PP output stage, can have their output directly coupled and you must rely on their offset to be zero. If the pre has an output cap, you must rely on it having absolutely no DC loss. I always keep in my mind that I'm not a HiFi factory and wouldn't tempt fate too much. It is so frustrating having to demount everything in order to modify something.
I bought my speaker protection on eBay, you need two of them in order to build-up a dual-mono assembly with separate GNDs. I also bought 2 relays 4 ways (used paralleled), 10A a way, whose coils are driven by the protection board, after removing their own relays.
What I have understood, about your project, is that, actually, you will use my circuit as a base only for the Baxandall super pair tweak, the 24KOhm resistor in place of the 2 x 12 KOhm, and the remotion of the trimmer, leaving the final stage original NHB-108's. Then I suggest you increase the CCSs current. But you should not decrease their resistor value nor increase the zener voltage (that limits the signal range). Rather study on a solution with a depletion mosfet, with the goal to obtain an impedance close to the original. The impedance of the CCSs (multiplied by drivers hfe) affects the VAS: a low impedance improves its open-loop bandwidth but could be an undesired load for the VAS itself, and increases 2nd harmonic distortion (virtually nulled by the PP). This as per my vintage knowledge, someone could have better suggestions.
About the input cap: I was in doubt to remove it. Then I decided to leave it. Without it, the assembly, having no total DC feedback loop, is sensitive to the DC. Some pre, with PP output stage, can have their output directly coupled and you must rely on their offset to be zero. If the pre has an output cap, you must rely on it having absolutely no DC loss. I always keep in my mind that I'm not a HiFi factory and wouldn't tempt fate too much. It is so frustrating having to demount everything in order to modify something.
I bought my speaker protection on eBay, you need two of them in order to build-up a dual-mono assembly with separate GNDs. I also bought 2 relays 4 ways (used paralleled), 10A a way, whose coils are driven by the protection board, after removing their own relays.
What I have understood, about your project, is that, actually, you will use my circuit as a base only for the Baxandall super pair tweak, the 24KOhm resistor in place of the 2 x 12 KOhm, and the remotion of the trimmer, leaving the final stage original NHB-108's. Then I suggest you increase the CCSs current. But you should not decrease their resistor value nor increase the zener voltage (that limits the signal range). Rather study on a solution with a depletion mosfet, with the goal to obtain an impedance close to the original. The impedance of the CCSs (multiplied by drivers hfe) affects the VAS: a low impedance improves its open-loop bandwidth but could be an undesired load for the VAS itself, and increases 2nd harmonic distortion (virtually nulled by the PP). This as per my vintage knowledge, someone could have better suggestions.
Last edited:
Nice work, it's not a clone, so you could experiment with various amounts of feedback from the output stage. Do you have any measurements to show? It's 150mA total, for 11 output pairs?
I am very glad to you for your appreciation.
At the beginning of this adventure, I was searching, in order to modify it, for a base-board with no feedback loop at all. That would have been very complicated with SS technology. I'm thankful I could find this NHB-108 that has an internal loop and no loop from the speaker output. I will never close any feedback loop from the output, a decision taken by personal experience, my last one was comparing a cross-coupled tube main amp with no internal/external feedback loop to a Williamson. The cross-coupled literally ate the Williamson. The only lack is about the deep bass (affected by a low DF), which is nothing compared to the authenticity and realism of the sound.
My whole measuring equipment is still in Italy, here I have only some multimeters and my ears. So I couldn't conduct any measurement over current and voltage with multimeters. The 150mA on the final stage is a total amount, estimated multiplying by 11 the 13.5mA average, measured emitter/emitter of two opposite transistors.
I'm so sorry, no measure to show!
At the beginning of this adventure, I was searching, in order to modify it, for a base-board with no feedback loop at all. That would have been very complicated with SS technology. I'm thankful I could find this NHB-108 that has an internal loop and no loop from the speaker output. I will never close any feedback loop from the output, a decision taken by personal experience, my last one was comparing a cross-coupled tube main amp with no internal/external feedback loop to a Williamson. The cross-coupled literally ate the Williamson. The only lack is about the deep bass (affected by a low DF), which is nothing compared to the authenticity and realism of the sound.
My whole measuring equipment is still in Italy, here I have only some multimeters and my ears. So I couldn't conduct any measurement over current and voltage with multimeters. The 150mA on the final stage is a total amount, estimated multiplying by 11 the 13.5mA average, measured emitter/emitter of two opposite transistors.
I'm so sorry, no measure to show!
Last edited:
Pretty much what you describe for the changes, although I plan to keep the DC voltage adjustment parts (with different values).
All good points about the CCS.
in the current form, the collector current of the output transistors is about 220 mA. So that needs to be increased. As far as I know, the original circuit operates at around 350 mA
I never measured the current of the output pair of the original. I wouldn't increase it, on the contrary, having no emitter resistors, I would take advantage of this and decrease the idle current as low as possible. Unfortunately, this is a bad idea because, to achieve a lower idle current, you should increase the value of 27Ohm resistors, getting a worse driving capability of the speaker. After all, I wouldn't touch any value, 27 Ohm and the diode are a magic couple. Search for this topic some posts ago, perhaps a couple of months in the past, by Analog_SA. What should be increased, in order to have a better driving capability of speakers, is the CCSs current. And you should get also a higher idle current (that IMHO is not a goal at all).
мои эксперименты с увеличением тока покоя положительно повлияли на звук. В этом выходном каскаде ток покоя задается не только базовыми резисторам 27 ом, но и соотношением размеров кристаллов выходных и пред выходных транзисторов, а также зависит от их коэффициента усиления. Так как параметры транзисторов могут отличатся в зависимости от партии, резисторы 27 ом могут быть пересчитаны, для обеспечения более высокого тока покоя, но оставляя транзисторы в зоне безопасной работы. Таким образом искажения падают еще больше и это заметно при прослушивании в моем случае.