MOD was a sequence of recorded sounds.
Primitive, yet impressive things were done.
You seem pretty stuck on this front end.
I forget now, what was the original point
of the unique topology? I recall something
about offset cancellations? But I still see
manual offset adjust and transistors that
don't add up to a cancellation of emitter
drops, that couldn't have been it....
You also claim to use best transistors from
your junk box. I am curious why you have
not yet tried 0.21V emitter drop Germanium
for Q12 with scaled down output resistors?
Is it just the lack of Spice model?
Primitive, yet impressive things were done.
You seem pretty stuck on this front end.
I forget now, what was the original point
of the unique topology? I recall something
about offset cancellations? But I still see
manual offset adjust and transistors that
don't add up to a cancellation of emitter
drops, that couldn't have been it....
You also claim to use best transistors from
your junk box. I am curious why you have
not yet tried 0.21V emitter drop Germanium
for Q12 with scaled down output resistors?
Is it just the lack of Spice model?
Last edited:
I still can't find those germaniums, they dropped off the face of the earth.
Offset cancelation does work on this amp. If you keep the rails steady, offset usually stays within 1mV.
The frontend is unique in three ways, the NTP usage, the lack of B-E resistor for the output transistor, and the use of shunt compensation.
The 2SC5171 would probably be best here. I was using the 2SC4370 before though, and I got it sounding very warm and smooth. However with faster outputs like the 2SD669, there is more definition. So I plan to test with the 2SC5171, because it's Beta rage extends to 300mA and it is fast, which makes it an ideal candidate.
- keantoken
Offset cancelation does work on this amp. If you keep the rails steady, offset usually stays within 1mV.
The frontend is unique in three ways, the NTP usage, the lack of B-E resistor for the output transistor, and the use of shunt compensation.
The 2SC5171 would probably be best here. I was using the 2SC4370 before though, and I got it sounding very warm and smooth. However with faster outputs like the 2SD669, there is more definition. So I plan to test with the 2SC5171, because it's Beta rage extends to 300mA and it is fast, which makes it an ideal candidate.
- keantoken
Member
Joined 2009
Paid Member
Kean,
It's looking very nice.
I went back to post 1 go re-read a few of the basics again regarding the benefits of the NTP. I'm thinking of a desktop amp, low power and started off with a question about the front end. I'm bored with LTPs. A question - you compared the NTP with an LTP, how would you compare the NTP with a Singleton input in terms of pro's and con's ?
It's looking very nice.
I went back to post 1 go re-read a few of the basics again regarding the benefits of the NTP. I'm thinking of a desktop amp, low power and started off with a question about the front end. I'm bored with LTPs. A question - you compared the NTP with an LTP, how would you compare the NTP with a Singleton input in terms of pro's and con's ?
I've never compared it sonically to an LTP. However I will say that the amp has an enjoyable sound that I haven't heard from anything else. The Bass is rather nice. I've been going through some orchestrals listening to my soundcard and I was reminded of the way the amp sounded back when the prototype was up and running. Hardly clinical at all, especially with 2SC4370 outputs.
Here are a few advantages I know of from a technical standpoint:
1: higher dynamic range
2: less compressive harmonics (good for bass I think)
3: No CCS to worry about
As far as the singleton goes, the singleton wins - as long as you keep a low-impedance at its emitter (the feedback network acts like degeneration). This is difficult. And if you succeed in this, it will be blazingly fast, and may be hard to control. This may necessitate a large Cdom, which will cause inordinate HF distortion. It will also exhibit a very nonlinear transconductance, which can cause instability with swings in the BJT current. In 2-BJT input stages this is offset by a halved transconductance. I think most of this can be remedied by using highish bias current and lowering gain with degeneration, but this will make output offset more difficult to control without a servo.
I think this is why you don't see singletons often and if you do, they use an output capacitor.
One idea you could try is the dual singleton, which is better known as a current-feedback amplifier (a malunderstanding I think). This is an NPN and PNP singleton, emitters joined at the feedback node. A symmetrical VAS is driven by the two collectors. Just two singletons in complimentary, really. Yes it will be dominant odds like the LTP, but still different...
- keantoken
Here are a few advantages I know of from a technical standpoint:
1: higher dynamic range
2: less compressive harmonics (good for bass I think)
3: No CCS to worry about
As far as the singleton goes, the singleton wins - as long as you keep a low-impedance at its emitter (the feedback network acts like degeneration). This is difficult. And if you succeed in this, it will be blazingly fast, and may be hard to control. This may necessitate a large Cdom, which will cause inordinate HF distortion. It will also exhibit a very nonlinear transconductance, which can cause instability with swings in the BJT current. In 2-BJT input stages this is offset by a halved transconductance. I think most of this can be remedied by using highish bias current and lowering gain with degeneration, but this will make output offset more difficult to control without a servo.
I think this is why you don't see singletons often and if you do, they use an output capacitor.
One idea you could try is the dual singleton, which is better known as a current-feedback amplifier (a malunderstanding I think). This is an NPN and PNP singleton, emitters joined at the feedback node. A symmetrical VAS is driven by the two collectors. Just two singletons in complimentary, really. Yes it will be dominant odds like the LTP, but still different...
- keantoken
Always keep it at 100% and use foobar to lower it. Most other players are crap.
It also may help to disable system sounds.
That's why I said to set the system sounds to "no sound".
You can check the programs that you run normally if they have some sounds, but for instance outlook gets silenced that way.
Hey all.
I just got my last prototype working again. It turns out I accidentally used 12R instead of 1k for R21 and R23! D'oh...
Wow, this amp is still interesting. I've learned a lot since then and figure I can make it stable enough if I want...
It sounds like crap - but it sounds so goooood...
I can more realistically listen to this amp now that I've heard a simple MOSFET 800mA class A using IRF250's. Voices "speak" to me in an involving way, and speech is clearer - and it doesn't seem superficial when the singer sings "emotionally"... Same with some musical passages. With the MOSFET amp most things were just "hmm, and interesting combination of notes there, but I've heard that dozens of times in other songs...". With this amp, the compositions "speak" to me emotionally. I feel like I'm hearing it as it was intended to be heard.
However there is a lack of background silence and the music has that familiar smudged background, whereas the MOSFET amp was totally silent and encouraged me to separate the instruments. This amp seems to combine the instruments in a way that, although it decreases resolution, makes them more involving. I cannot be sure I'm not confusing causation here though... I think I can increase the resolution without losing this effect.
Treble is sharp and dull, sometimes painful. I could remedy this if I could decrease C2, but it gets too unstable. I now know how to fix this problem, though I'm not sure I'm ready to rework the prototype yet.
My first impression was that this amp sounded more "boxy" than the MOSFET amp. I think this is mostly because of the treble issues.
- keantoken
I just got my last prototype working again. It turns out I accidentally used 12R instead of 1k for R21 and R23! D'oh...
Wow, this amp is still interesting. I've learned a lot since then and figure I can make it stable enough if I want...
It sounds like crap - but it sounds so goooood...
I can more realistically listen to this amp now that I've heard a simple MOSFET 800mA class A using IRF250's. Voices "speak" to me in an involving way, and speech is clearer - and it doesn't seem superficial when the singer sings "emotionally"... Same with some musical passages. With the MOSFET amp most things were just "hmm, and interesting combination of notes there, but I've heard that dozens of times in other songs...". With this amp, the compositions "speak" to me emotionally. I feel like I'm hearing it as it was intended to be heard.
However there is a lack of background silence and the music has that familiar smudged background, whereas the MOSFET amp was totally silent and encouraged me to separate the instruments. This amp seems to combine the instruments in a way that, although it decreases resolution, makes them more involving. I cannot be sure I'm not confusing causation here though... I think I can increase the resolution without losing this effect.
Treble is sharp and dull, sometimes painful. I could remedy this if I could decrease C2, but it gets too unstable. I now know how to fix this problem, though I'm not sure I'm ready to rework the prototype yet.
My first impression was that this amp sounded more "boxy" than the MOSFET amp. I think this is mostly because of the treble issues.
- keantoken
I am regulating the amp with these:
http://www.diyaudio.com/forums/powe...ntokens-cfp-cap-multiplier-5.html#post2425305
Except with the LEDs replaced with normal diodes. The circuit is more stable now, I think because of less resonance on the rails because of long leads to main reservoir caps.
Wavebourn's music totally rocks with this amp! So does Buena Vista Social Club...
- keantoken
http://www.diyaudio.com/forums/powe...ntokens-cfp-cap-multiplier-5.html#post2425305
Except with the LEDs replaced with normal diodes. The circuit is more stable now, I think because of less resonance on the rails because of long leads to main reservoir caps.
Wavebourn's music totally rocks with this amp! So does Buena Vista Social Club...
- keantoken
The regulation seemed to take away the emotional engagement quite a bit, but this amp still has a very unique sound... Why is there such a strong correlation between good sound and instability!?
You don't recognize it unless you're relaxed and used to the sound already, but it has it's own specific character, which I find agreeable. The sound did in fact change a lot depending on which output transistor I used. This only makes sense because the feedback loop is heavily founded on Hfe. The 2SC4370 had the best sound and was in the first prototype I was raving at. However the datasheet doesn't show any curves so I have no clue why that may be.
I'm considering changing transistors. Cordell's SPICE models have shown me what great transistors the 2N5551/5401 are, and how the BC550/560 don't have very great Hfe linearity (but which ones did he test, phillips or fairchild?).
The amp is stable, but has issues with capacitive loads below 100nF. So ultimately my objective hasn't been fulfilled. I wanted the amp to sound good and be rock-stable into capacitive loads. I know how to fix this now I'm pretty sure, but I need my signal generator and it's broken. I need about 8 50V/1500uF caps , since the originals are dying (with one destroyed rectifier diode.
- keantoken
You don't recognize it unless you're relaxed and used to the sound already, but it has it's own specific character, which I find agreeable. The sound did in fact change a lot depending on which output transistor I used. This only makes sense because the feedback loop is heavily founded on Hfe. The 2SC4370 had the best sound and was in the first prototype I was raving at. However the datasheet doesn't show any curves so I have no clue why that may be.
I'm considering changing transistors. Cordell's SPICE models have shown me what great transistors the 2N5551/5401 are, and how the BC550/560 don't have very great Hfe linearity (but which ones did he test, phillips or fairchild?).
The amp is stable, but has issues with capacitive loads below 100nF. So ultimately my objective hasn't been fulfilled. I wanted the amp to sound good and be rock-stable into capacitive loads. I know how to fix this now I'm pretty sure, but I need my signal generator and it's broken. I need about 8 50V/1500uF caps , since the originals are dying (with one destroyed rectifier diode.
- keantoken
Here is the schematic I'm currently looking to build. The reduced reliance on Q9's Vbe should increase reliability.
R15 will allow me to use a smaller C2. I will change this resistor to see which capacitor is best for the position of C2.
The prototype I'm using right now has a 33R degen for R15, and a ~45R+15nF network for C2 and R2. Those are the only changes since the last modification. Oddly enough the sound of everything has changed, especially the strings. Again I'm bumping into the issue of not ever having been to an orchestra or any kind of real concert. It still sounds better than it did before, so I will continue.
- keantoken
R15 will allow me to use a smaller C2. I will change this resistor to see which capacitor is best for the position of C2.
The prototype I'm using right now has a 33R degen for R15, and a ~45R+15nF network for C2 and R2. Those are the only changes since the last modification. Oddly enough the sound of everything has changed, especially the strings. Again I'm bumping into the issue of not ever having been to an orchestra or any kind of real concert. It still sounds better than it did before, so I will continue.
- keantoken
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.