If one aims to NOT make a Naim clone then the 22k should be the same value as the 1k across Tr4.
AND
The effect of Tr4 base current must be corrected by equalising the EMITTER currents of Tr1 and Tr2.
This is done by measuring the Tr1 Tr2 difference in emitter terminal voltage, above the two emitter resistors. If the emitter resistors are set to 0r0, then one cannot check the balance of the input stage.
Small adjustment of the 24k at the input is where one makes the small Vbe adjustment of the two input transistors. It may have to be very slightly higher, or very slightly lower and that depends on how well matched for both Vbe and hFE are the two input transistors. If Tr1 was identical to Tr2, then the 24k+ 2k7 would have to be adjusted up very slightly to equal the 27k (NFB upper).
One can restore the input stage transconductance by increasing the stage tail current, i.e. make the Tr3 emitter resistor smaller. Try 270r, or 240r, or even 220r.
AND
The effect of Tr4 base current must be corrected by equalising the EMITTER currents of Tr1 and Tr2.
This is done by measuring the Tr1 Tr2 difference in emitter terminal voltage, above the two emitter resistors. If the emitter resistors are set to 0r0, then one cannot check the balance of the input stage.
Small adjustment of the 24k at the input is where one makes the small Vbe adjustment of the two input transistors. It may have to be very slightly higher, or very slightly lower and that depends on how well matched for both Vbe and hFE are the two input transistors. If Tr1 was identical to Tr2, then the 24k+ 2k7 would have to be adjusted up very slightly to equal the 27k (NFB upper).
One can restore the input stage transconductance by increasing the stage tail current, i.e. make the Tr3 emitter resistor smaller. Try 270r, or 240r, or even 220r.
Last edited:
You must get the TR1 1K to >600 mV as your prime target. If not the whole point of the 22K is redundant. I put in the small TR4 resistance as the only tool in the box if the Vbe is < 600 mV. All the 22K does is ensure a nice harmonic structure. It doesn't add vast amounts of distortion. Kodachrome film made red a priority colour. Naim isn't Kodachrome. CD will sound better with Naim's intended balance. Listening to my friends NAP 250 it needs that colour. To my ears it is still a bit blue. Maybe 700 mV would be best (22 R ??). I might have some valve measurements that shows this and how easy it is to truely kill a good thing. Personally I have no valve or transistor preference. They need very different engineering solutions. Valves seem slightly better devices and transistors more practical.
This is a two valve design with 16.5 dB feedback ( and a gain stage in one ECC81 ) . The feedback is as direct as possible and based on the PYE Mozart deisgn without the need for a special transformer. The one I call best is using 82% triode connection via a tap on the transformer. All the other options are not so good including the favourite of many the pure triode. This option is most like the 22K and 600mV balance drifting to 500 mV when pure triode.
The interesting thing is many valve experts think the bad spectra are caused by the feedback and transformer combination. In a way yes, mostly no. It took me months of testing to find that feedback and valves can work. The bonus is mostly the damping factor is improved. I doubt many feedback valve amps look this good. - 54 dB is true hi fi when the nastier 1/3 harmonic is - 60 dB as in the " best " graph. 8 watts from one KT88 is not bad. It is class A that not even a Krell can offer.
I have developed a habit of downloading datasheets of devices of interest and keep them in a folder. This makes it easy to compare the merits of transistors against one another for whatever parameter is in consideration.
Last question - for a horses mouth answer see http://waltjung.org/PDFs/Sources_101_P1.pdf and
http://www.waltjung.org/PDFs/AX_Letters_0907.pdf
By all means get hold of your mate with the spectrum analyser and get your test subject transistors lined up for testing for IMD - presumably he has a signal generator.
One way to do this would be to use one channel as a test subject and the other as the reference starting point. If one result is an improvement, upgrade the reference and repeat the process if or as required.
The 22k collector resistor for TR2 develops significant Miller capacitance and shorting this out may reduce the effectiveness of the C.dom capacitor whose value might need adjusting - early Naim amplifiers were made without the 22k resistor. These things may be co-dependent.
A similar situation might arise by substituting a different Vas transistor, so the two modifications might need some tuning to get the best result.
If you have supply rail fuses you might consider removing them and inserting a pair of 100R 5 watt safety resistors in their place. A convenient way to do that is to use a couple of spare fuses, find a way to blow them, and solder the resistor to the fuse end caps. I use this approach to monitor the voltage drop across resistors to deduce the current level whenever making any change to a power circuit.
Apart from changing transistors and testing with the 22k resistor shorted, you might consider investigating the IMD with different nfb blocking capacitors - electrolytic by various types, voltage, esr ratings and manufacturers.
Swapping capacitor types might be made easier if a pair of inline snap off socket were temporarily soldered into the test board in lieu of the standard part you are using.
Last edited:
I respectfully disagree with the above. The problem is the gate capacitance.
This is a current output. To see voltage on the load you need current to flow. For the current to flow the Vgs voltage must rise, i.e. the gate "capacitor" to charge. No matter what's at the output.
I'm sure you can charge the gates with 2mA, but it will be one slow amplifier (through gate resistors) or one with ugly uneven rising profile without them. It's even worse if real speaker load is considered.
Audio FETs have pretty low gain and need about 1Vgs change for 1A increase at the source.
There is a very nice Application Note from Semelab on that.
I haven't run calculations, but I would agree 5.6mA is most likely OK for a single pair of FETs up to may be 50W. Which is enough for this amp anyway. For two pairs, all the VAS current will be robbed by the gates. Not good!
So Ruwe you are saying Hitachi got it wrong? Below is the real life proof and not that coming out of the makers of Bipolar devices. There is plenty of false data out there. Look carefully. No Bipolar amp can do this and if it can it will be very complex.
What you may have overlooked is running tweeters at 100 watts seldom is more than a microsecond event before destruction. Most modern railway locomotives will do 500 kPH, reality is they cannot with safety. The French proved this recently ( safely even ).
I very very very strongly suggest the low current VAS for MOSFET's should not be rejected. I equally stronghly suggest the higher current route is not to all teastes the best. My friend also called Nigel who introduced me to the Hitachi amp ran 5 sets of devices on the circuit below. To his ears it was the best. He was getting 1000 watts. The HH-1200 as used by the BBC also ( 600 constant ). It used MPSA 92/42 which are nowhere near as good as the Hitachi devices.
I run ribbon tweeters on some speakers I designed that will do 45 kHz flat. I have to limit them to abnout 32 kHz as most DAC's do not reconstruct the wave well enough. I uses the old Crystal 20 bit DAC. That seems the exception. Most DAC's use something much like a VAS at the output. Ring any bells regarding both? Thus for people to choose to insert a pole in the response in the amplifer is not so daft ( FET with low current VAS is a filter ). The problem is the FET's themselves are far too able and willing to go higher. To deliberately make a FET harsh sounding is a shame. The thing to understand is, it is a choice and nothing more. Some people are too timid to try. Thus convention will stop them having a sound they might prefer. That is awful. That is like saying all sweet wines are awful. One or two are not and cost plenty. The lower current VAS will be more stable as a bonus. 2 mA is enough for real music. I suspect it will be - 3dB 25 kHz @ 100 watts ? Don't use IRF for simulations. BU900 etc is about the nearest I think that has Spice data. Remember I was saying the Naim clone at 9 mS " may" be too much. That's why I put double devices on the drawing. See also on the graphs below just how fast the FET is. This is for this circuit in the early days. We must assume the devics are quicker now.
All these graphs below relate to the Hitachi circuit which is much like this one. The Hitachi used slightly higher values than 3K9. Please note the full power output at 100 kHz and the miniscule distortion. Note also how good the double VAS balance is. Did Hitachi get it wrong ?
BTW. If I give a test result of my own it is always real. The valves yesterday were real. One graph was supurb and a circuit you might think OK was not very good. The not very good one was the favourite of a forum and it's endless speculations. The designer was banned in the end as he was so frustrated by people not begining to understand his design. I managed to transform his design for pennies into true hi fi. He wouldn't have liked that I am sure. I still totally like this man for his inventive ability. Alex Kitic and RH88 design. I added feedback to the ECC81 cathode by placing the output transformer secondary in series antiphase ( it looks crazy, from PYE Mozart ). I then made a gain of 26 triode amplifier from the spare ECC81 section. This gave 16.5 dB feedback 0.22% THD 8 watts. 1.26V in for full power. 2 valve bottles per channel. The other guys took months basically to say Alex is wrong and stole the design from 1938 ( he didn't ). Some said the ECC81 is a RF valve and hopeless for Audio. My spectrum analyser didn't agree with that. The ECC81 with a cathode bootstrap gives a gain of about 100. Without the bootstrap about 26. That is the little trick to get it to work well. There is no bootstrap on the Kitic input, it would stop it working which most refused to understand. That make feedback easier that he didn't. Kitic uses the ECC81 and KT 88 much like the I to V conveter of a DAC or VAS. It looks so much like convention that most people never understood that it was weirdly different. To use a valve this way is crazy and rather good. If you like the KT88 and ECC81 became a super triode. That's why adding feedbabck for once actually worked. The output transformer usually kills that.
What you may have overlooked is running tweeters at 100 watts seldom is more than a microsecond event before destruction. Most modern railway locomotives will do 500 kPH, reality is they cannot with safety. The French proved this recently ( safely even ).
I very very very strongly suggest the low current VAS for MOSFET's should not be rejected. I equally stronghly suggest the higher current route is not to all teastes the best. My friend also called Nigel who introduced me to the Hitachi amp ran 5 sets of devices on the circuit below. To his ears it was the best. He was getting 1000 watts. The HH-1200 as used by the BBC also ( 600 constant ). It used MPSA 92/42 which are nowhere near as good as the Hitachi devices.
I run ribbon tweeters on some speakers I designed that will do 45 kHz flat. I have to limit them to abnout 32 kHz as most DAC's do not reconstruct the wave well enough. I uses the old Crystal 20 bit DAC. That seems the exception. Most DAC's use something much like a VAS at the output. Ring any bells regarding both? Thus for people to choose to insert a pole in the response in the amplifer is not so daft ( FET with low current VAS is a filter ). The problem is the FET's themselves are far too able and willing to go higher. To deliberately make a FET harsh sounding is a shame. The thing to understand is, it is a choice and nothing more. Some people are too timid to try. Thus convention will stop them having a sound they might prefer. That is awful. That is like saying all sweet wines are awful. One or two are not and cost plenty. The lower current VAS will be more stable as a bonus. 2 mA is enough for real music. I suspect it will be - 3dB 25 kHz @ 100 watts ? Don't use IRF for simulations. BU900 etc is about the nearest I think that has Spice data. Remember I was saying the Naim clone at 9 mS " may" be too much. That's why I put double devices on the drawing. See also on the graphs below just how fast the FET is. This is for this circuit in the early days. We must assume the devics are quicker now.
All these graphs below relate to the Hitachi circuit which is much like this one. The Hitachi used slightly higher values than 3K9. Please note the full power output at 100 kHz and the miniscule distortion. Note also how good the double VAS balance is. Did Hitachi get it wrong ?
BTW. If I give a test result of my own it is always real. The valves yesterday were real. One graph was supurb and a circuit you might think OK was not very good. The not very good one was the favourite of a forum and it's endless speculations. The designer was banned in the end as he was so frustrated by people not begining to understand his design. I managed to transform his design for pennies into true hi fi. He wouldn't have liked that I am sure. I still totally like this man for his inventive ability. Alex Kitic and RH88 design. I added feedback to the ECC81 cathode by placing the output transformer secondary in series antiphase ( it looks crazy, from PYE Mozart ). I then made a gain of 26 triode amplifier from the spare ECC81 section. This gave 16.5 dB feedback 0.22% THD 8 watts. 1.26V in for full power. 2 valve bottles per channel. The other guys took months basically to say Alex is wrong and stole the design from 1938 ( he didn't ). Some said the ECC81 is a RF valve and hopeless for Audio. My spectrum analyser didn't agree with that. The ECC81 with a cathode bootstrap gives a gain of about 100. Without the bootstrap about 26. That is the little trick to get it to work well. There is no bootstrap on the Kitic input, it would stop it working which most refused to understand. That make feedback easier that he didn't. Kitic uses the ECC81 and KT 88 much like the I to V conveter of a DAC or VAS. It looks so much like convention that most people never understood that it was weirdly different. To use a valve this way is crazy and rather good. If you like the KT88 and ECC81 became a super triode. That's why adding feedbabck for once actually worked. The output transformer usually kills that.
I don't think Hitachi or Semelab got it wrong. It's a matter of approach to the design - how much will be heard or not heard, at what power, what application, load etc.
I agree with you that 100kHz full power is not really necessary. And also that things have to be tried in real life. I personally try everything. I'm paid to do that. I think theory should give practical results, if it just give more theory I lose interest. We have quantum physics for that 😉
Many manufacturers make amplifiers linear to 400kHz. On the other hand I'm perfectly happy with my tube amp doing 35kHz@-3db.
I just wanted to point out that FETs are not completely voltage regulated devices as people often thing, and gate currents should be expected and considered in design phase. I work on power inverters with IGBT output. I wish I could show you how complex and powerful is the gate drive circuit at only few kHz switching frequency.
Sorry for the off-topic post. No Naim info here.
I agree with you that 100kHz full power is not really necessary. And also that things have to be tried in real life. I personally try everything. I'm paid to do that. I think theory should give practical results, if it just give more theory I lose interest. We have quantum physics for that 😉
Many manufacturers make amplifiers linear to 400kHz. On the other hand I'm perfectly happy with my tube amp doing 35kHz@-3db.
I just wanted to point out that FETs are not completely voltage regulated devices as people often thing, and gate currents should be expected and considered in design phase. I work on power inverters with IGBT output. I wish I could show you how complex and powerful is the gate drive circuit at only few kHz switching frequency.
Sorry for the off-topic post. No Naim info here.
Hard switching of high voltage compound devices like trench IGBTs is very different to operating a pair of lateral fets in linear mode where the gate capacitance is mostly bootstrapped.
No doubt about that. Especially at 2MW as is in our inverter 🙂
To finish this off-topic: What is an acceptable drive, in your opinion, for lateral FETs? How much current, or no current? I just want to know the consensus, my experience with lat FETs audio otputs is limited to my own version of the clone and the Simple Symetrical Amplifier elsewhere in this forum.
I found the Semelab application note to be in agreement with several other sources in the way they calculate gate drive in FETs. Based on power and also bandwidth requirements. That's the way it should be IMO.
Last edited:
In my experience the Hitachi circuit @ ~6mA works very well along with JLH 80W @ 20mA. I found emitter follower buffers gave no perfomance advantage with 1 or 2 pairs of lateral FETs.
How about a PrimePACK 3+ quasi output stage? 1800A @ 1700V should be adequate for most speakers 😱
How about a PrimePACK 3+ quasi output stage? 1800A @ 1700V should be adequate for most speakers 😱
Thanks, Dave. That's in line with the expected. I think ~10mA is a value that will suit most of the FET outputs for reasonable audio power and BW.
... We use similar to PrimePack devices to generate 60Hz sinewave only. It would be interesting to hear it through a speaker 🙂 At 2000Arms they can supply power to heavy loads 😉 3 of them, actually, being 3-phase inverter...
... We use similar to PrimePack devices to generate 60Hz sinewave only. It would be interesting to hear it through a speaker 🙂 At 2000Arms they can supply power to heavy loads 😉 3 of them, actually, being 3-phase inverter...
The Fairchild KSA datasheet in Figures 1 and 2 both contain an identical error.
The Collector Current values on the vertical axis are incorrectly scaled in [A].
This should read (mA) as can be seen in the Sanyo 2SA1381 equivalent where identical graphs are correctly scaled.
That's the point guys. Tell people what is safe to try it and why others might not like it ( remember, in matters of taste you are always King as long as it is safe). I was doing some thinking about the NAP VAS. If we call the current 9 mA and due to heating the VAS Re = 25/9(mA ) since the transistor will be at 50C 30/9 seems about right allowing for Boltzman's constant. Thus we get 3R3. We might guess a gain of 100 typical for TR4. This says the VAS sits at 330R with luck. This is fed by 1K. This is not ideal. In radio design we usually use 50 or 75 R as our prefered loads and match the input to the output. You will often see a resistor added to do exactly that. This is called transconductance and offers the maximum transfer of power. If you put an oscilloscope to the VAS base you will see a highly distorted wave yet the wave from the collector is not! How come? This is the bit I don't like in text books. The input to the VAS is pure current drive therefore it doesn't matter they say. It looks like sharks teeth.
This can be verified if you have a modern scope. A 1R resitor can be fixed between TR4 base and TR1 collector. The " floating input " scope clipped to the 1R and not to ground or rails. If the statement of the text book is correct the wave will be perfect. I bet you it won't be. 1R is small enough not to change reality too much.
If we were to reduce the NAP VAS current we might just get what we want. Lets take the transistors TR7 to 10 to be gain of 50 at sensible output current. We might get get a compound gain of 2500. If we only had 3 mA VAS current we might just get the 3 amps we need with a small reserve. Typical all in one devices state beta of >750. Into 4R that's if lucky 2 watts. 3 mA would satisfy our transconductance, but is unlikely to sound forceful. Don't reject it. TR6 68R would be about 200R to offer transconductance into TR4.
Lets do some simplistic gain approximations. If we pretend the VAS collector load is a simple resistor we find if 80V 80/9mA = 8K9. If we divide that by Re (about 3R ) we get a voltage gain of about 2700. Now lets do 3 mA 80/3 = 26K7 and Re of 10R. As you can see the gain is unchanged. The real gain is massively more as the TR6 might be >100K due to the TR6 current source. Even if 2 x 4K3 is used with 100 uF as an old style bootstrap CCS the impedance of the TR4 collector load is almost infinite. The only thing that is bad is the non linear TR7 to TR10 load. It is fun to measure the amplifer without TR7+8. It will be nearly perfect. Class A amps help a little as the load is dynamically more constant. Did anyone calculate the NAP loop gain? I guess it to be about 300 000 at 1 Hz ( with very large caps )? That looks to be just about enough to correct at 10 kHz. If you like 30% as good as a LM741 op amp. I don't mind being wrong about that, so just a ball park figure.
Here is a practical solution to one problem. If the we add a resistor to the TR4 emmiter we loose the VAS gain. Below perhaps 10 kHz this really isn't a big problem. TR4 Cdom of 47 pF is the cause of that problem. We have no choice because reducing it's value invites instability and smoke! With a scope we might go to 27 pF. Lets say we need 16R to nicely make the Naim balance happen ( 600 mV 1K TR1 ). 16R at lets say 1 kHz is 10uF. That is a nice value in 100 V polyester ( on wires ). Realistically a Panasonic 10uF 50V FC grade will be fine. 22uF perfect. That would bering the loop gain to almost no change at 10 kHz. If you prefer 220uF that's fine. If you have some dance music have a 0R option. Mine was a plug in jumper.
Looking to the gain of the transistor. If we say a 2SA device gain tested at 10 mA might give a gain of 200 Zin might be 600R. If we reduced the VAS current to 6.5 mA using 100R to TR6 emitter we would reach transconductance. Without changing the transistor TR4 we could try that. As we found the loop gain is unchanged we know we are only really testing the gain compromise. The load does alter it some.
We can times 3 the input pair TR1+2 current. That is ideal. We would then need a scope. If we went for 3mA 1K becomes 330R. We should add 2 x 33R to TR1+2 emitters to restore linearity and gain stability. The scope should prove it to be fine. The DC offset will get closer to 50 mV, that to me is OK.
This can be verified if you have a modern scope. A 1R resitor can be fixed between TR4 base and TR1 collector. The " floating input " scope clipped to the 1R and not to ground or rails. If the statement of the text book is correct the wave will be perfect. I bet you it won't be. 1R is small enough not to change reality too much.
If we were to reduce the NAP VAS current we might just get what we want. Lets take the transistors TR7 to 10 to be gain of 50 at sensible output current. We might get get a compound gain of 2500. If we only had 3 mA VAS current we might just get the 3 amps we need with a small reserve. Typical all in one devices state beta of >750. Into 4R that's if lucky 2 watts. 3 mA would satisfy our transconductance, but is unlikely to sound forceful. Don't reject it. TR6 68R would be about 200R to offer transconductance into TR4.
Lets do some simplistic gain approximations. If we pretend the VAS collector load is a simple resistor we find if 80V 80/9mA = 8K9. If we divide that by Re (about 3R ) we get a voltage gain of about 2700. Now lets do 3 mA 80/3 = 26K7 and Re of 10R. As you can see the gain is unchanged. The real gain is massively more as the TR6 might be >100K due to the TR6 current source. Even if 2 x 4K3 is used with 100 uF as an old style bootstrap CCS the impedance of the TR4 collector load is almost infinite. The only thing that is bad is the non linear TR7 to TR10 load. It is fun to measure the amplifer without TR7+8. It will be nearly perfect. Class A amps help a little as the load is dynamically more constant. Did anyone calculate the NAP loop gain? I guess it to be about 300 000 at 1 Hz ( with very large caps )? That looks to be just about enough to correct at 10 kHz. If you like 30% as good as a LM741 op amp. I don't mind being wrong about that, so just a ball park figure.
Here is a practical solution to one problem. If the we add a resistor to the TR4 emmiter we loose the VAS gain. Below perhaps 10 kHz this really isn't a big problem. TR4 Cdom of 47 pF is the cause of that problem. We have no choice because reducing it's value invites instability and smoke! With a scope we might go to 27 pF. Lets say we need 16R to nicely make the Naim balance happen ( 600 mV 1K TR1 ). 16R at lets say 1 kHz is 10uF. That is a nice value in 100 V polyester ( on wires ). Realistically a Panasonic 10uF 50V FC grade will be fine. 22uF perfect. That would bering the loop gain to almost no change at 10 kHz. If you prefer 220uF that's fine. If you have some dance music have a 0R option. Mine was a plug in jumper.
Looking to the gain of the transistor. If we say a 2SA device gain tested at 10 mA might give a gain of 200 Zin might be 600R. If we reduced the VAS current to 6.5 mA using 100R to TR6 emitter we would reach transconductance. Without changing the transistor TR4 we could try that. As we found the loop gain is unchanged we know we are only really testing the gain compromise. The load does alter it some.
We can times 3 the input pair TR1+2 current. That is ideal. We would then need a scope. If we went for 3mA 1K becomes 330R. We should add 2 x 33R to TR1+2 emitters to restore linearity and gain stability. The scope should prove it to be fine. The DC offset will get closer to 50 mV, that to me is OK.
Thanks for that. I've always assumed the same regarding buffers. I really do need to build an amp for me. Against my better judgement I might build the single VAS FET amp, at least it's my own design ( intended for a 15 inch bass unit only ). I will test it to see if I can get 70kHz - 3dB at 3 mA VAS as I guess. That seems to be my typical preference.
Looking at typical specs ( at 1 MHz Exicon , Alas BUZ900 gives no useful data ).
Ciss 500 to 700 pF
Cdss 300 pF
Crss 10 to 25 pF.
Sadly no Cgd. We might infer it to be about 200 pF?
The Crss is impressive. That comes into play in class B designs. The gate looses it's charge very quickly when switching.
I'm not quite sure why, possibly an addiction, but I am currently building a pair of P101s:
Project 101 - High Power, High Fidelity Lateral MOSFET power amplifier
VAS current seems to ~5mA for 2 pairs of lateral fets in this design.
I've also got a stereo one of these on the go:
MJR7-Mk5 Mosfet Power Amplifier
Looks like ~20mA in the driver stage for 1 pair of FETs.
Sorry for another OT post!
Project 101 - High Power, High Fidelity Lateral MOSFET power amplifier
VAS current seems to ~5mA for 2 pairs of lateral fets in this design.
I've also got a stereo one of these on the go:
MJR7-Mk5 Mosfet Power Amplifier
Looks like ~20mA in the driver stage for 1 pair of FETs.
Sorry for another OT post!
I wouldn't be too concerned. This thread has been trashed to the point now, where you have to scan through many pages of O/T to find anything useful about completing the clone kits, let alone from anyone like Atupi, who has recently built one.
Every design deserves a separate discussion thread and both P101 and MJR7 are interesting but an MJR7 build would be especially interesting. I would have begun one myself some time back, but for lack of a PCB. Why not start a thread about it and post your progress as you go? 🙂
An externally hosted image should be here but it was not working when we last tested it.
hi all.how to add power output for Avondale ncc200?and if I add one pair of power output what supply voltage range I can use?tq all
Hi sasek. The present version of NCC200 design is specified for 30-50V max. rails and one pair of MJ15003 output transitors. It could be modified with 2 pairs of higher gain output transistors such as MJ21194, MJ21196, MJL21194, MJL21196 if you want more power for 4R loads but I would not alter the Avondale module design or maximum voltage if you want to keep the sound quality, which is really the only reason to choose this old design.
Note that you should refer to the latest design details as many schematics posted in this thread are altered, to illustrate different ideas: Amplifier Modules | AVONDALE AUDIO
The last published original schematic is posted at #1492 or you can find it on the web.
P.S. Note that this a Quasi-complementary design. It could well be unstable and require further mods to prevent oscillation with 2 pairs. Its not a simple procedure.
Note that you should refer to the latest design details as many schematics posted in this thread are altered, to illustrate different ideas: Amplifier Modules | AVONDALE AUDIO
The last published original schematic is posted at #1492 or you can find it on the web.
P.S. Note that this a Quasi-complementary design. It could well be unstable and require further mods to prevent oscillation with 2 pairs. Its not a simple procedure.
Last edited:
Hi Ian.Thanks for reply.
So as your conclusion, add more power output can affect the sound quality.Ok I accept it.
What I can say, I very proud sound of Avondale NCC200, very dynamic sound.
Can you suggest another amp+pcb layout that have 2 pair power output or higher?
My project have done is..
1.hiraga
2.sr200
3.lm3886
4.dxblame
So as your conclusion, add more power output can affect the sound quality.Ok I accept it.
What I can say, I very proud sound of Avondale NCC200, very dynamic sound.
Can you suggest another amp+pcb layout that have 2 pair power output or higher?
My project have done is..
1.hiraga
2.sr200
3.lm3886
4.dxblame
Looking at the variety of designs you have built so far, I think you would already have a better idea of what you might prefer. However, I am convinced that simply scaling up any small amplifier is unlikely to be successful unless you limit the type to very low distortion designs because they simply produce lower distortion for a given output power, as you increase the number of output devices. e.g. blameless model amplifiers or any of the larger designs by Dadod, Vzaichenko, Ostripper et al.
I believe that Ostripper's Honey Badger (see the sticky thread) is among the best ways to get started with higher power and plenty of support docs and PCB products from the DIYaudio store. Many builders like its hint of "sound quality" and noticeably higher power capability. Increasing an amplifier from 80W to say, 145W won't do much (<3dB), since perceived volume is in a logarithmic relationship to power so you need to really increase the power amplifier size to make a worthwhile difference. Unless you just want more capability to drive low impedance loads, 3 pairs of output devices would be a more appropriate size to adopt.
http://www.diyaudio.com/forums/solid-state/211905-diyab-amp-honey-badger-build-thread.html