That would be great. Here is how mismatch works. Sorry it has to be valves. Some buy matched valves. You don't have to. Note the PSU has moved to a better place. It is a FET capacitance multiplier. Valve people would like that. No, it's just the right device to use as BU208 is a low gain device. The 2 x 47 R is useful in many ways. It allows measurement as a bonus.
That would be great. Here is how mismatch works. Sorry it has to be valves. Some buy matched valves. You don't have to. Note the PSU has moved to a better place. It is a FET capacitance multiplier. Valve people would like that. No, it's just the right device to use as BU208 is a low gain device. The 2 x 47 R is useful in many ways. It allows measurement as a bonus.
Thanks Nigel, I note the Avondale Audio circuit recommends selection of a higher current gain transistor for TR1.
The look of the graph is exactly like a NAP 160 from memory. Some how I lost that one.
The Nait series power amplifier sections began with NAP90 when NA005 custom output transistors, similar to BD743, were introduced. ZTX652/752 were used as both VAS and drivers there. Small signal types were either BC550 or BC239 in at least Nait1&2 models.
BC546 is Avondale's substitution suiting their products, as is 2N5551 and the lethargic MJ15003 output transistors.
The main differences to other early NAP designs for anyone reading now, is that the phase correcting networks in the driver bases were omitted and TR2's collector resistor was increased to 27k for these lower power models. Tail currents were adjusted across all models for the particular rail voltages by changing the emitter resistor accordingly. VI limiters needed different parts too. NAP140 rails are nom. 34V, NAP90 30V and Naits are around 24V, based on standard transformer windings of 25, 22 and 18VAC respectively.
Here's a link to a hand drawn Nait circuit, apparently "reverse-engineered" from a real Nait1. There's an error - Q3 56R emitter resistor should read 560R.
http://www.diyaudio.com/forums/atta...d1133975938-naim-nait-2-power-amp-circuit.gif
Current balance in the pair is critically dependent on the TR1 collector resistor among other things, such as setting sufficient tail current, having enough Hfe in the pair, base loading, output loading - even more when AC response is considered too. However, TR2 collector resistor is almost useless. I suspect it's high to ensure that remains the case or perhaps to tweak audio response
(Note - TR numbering is from the schematics posted here, to avoid confusion).
It seems to have been missed, that we are considering different amplifiers. Even though the same basic circuit may be used, it will have differing operating conditions and adjustments have to be made. If TR1 collector resistor is increased to 1.6k in my earlier NAP140 sim. for example, DC current balance is completely restored to 1:1 with about 0.45 mA in each leg. The rail voltages in NAP140 are nothing like 43V so other types won't be the same balance without some tinkering.
The point here, is that each different model has to be considered on its specifics and what we understand by the term "clone". I have relied previously on a generic NAP circuit, commonly said to apply to the NAP250, which has 2 * 2 amplifiers in bridged mode with regulated 40V supplies. Now we have DaveS's amplifier with 43V rails, the NAP140 34V and Naits 24V. It does make a difference when builders fit any available supply they choose and should not be surprised to wind up with different results to what they expected.
Since this is a clone kit thread, I'll venture that nowhere in any Ebay kit documentation seen here so far, are directions for using different supply voltages, transistor types etc. So what have unwitting builders got? Typically, a widely varying DC imbalance and not exactly the sound they expected, I think. I have a couple of prebuilt and tested LJM clone boards with me now. With 32V supplies (my bench supply limit) there is a significant measured current imbalance:
- TR1 collector current - 0.61mA - Vce 25.8V
- TR2 collector current - 0.31mA - Vce 32.4V
To be fair in any comparison, the line up is:
BC556 input stage - 2SD647/B647 VAS - 2N5551/5401 for tail current source, Vbe mult. and VI limiters. 2SC5171/A1930 drivers - Sanken 2SC3858 outputs and yes, so far these prove to be genuine parts in several built versions now, from trusted sellers.
If you like the Naim sound that you get to an extent with these clones, experiment with the LTP balance a little and see what happens. The VAS types are just wrong if not the proper ZTX types IMHO and that of others such as Ruwe, who has offered a lot of good advice and experience here. I listen to a clone myself on occasion, when I want to hear some extra breathy vocals, more dynamic transients, more details than were intended and a diffused stereo image to boot. As I age, these little features start to sound good 😀.
Last edited:
Small error above and - timed out again! In the line up of clone parts, input transistors are BC546B (naturally) 😱
Not sure where I said my rails were 43V. They are actually +/-40V.
At +/-32V the LTP balance is poor:
0.45mA versus 0.55mA.
At +/-40V the balance is pretty good - as per my figures earlier.
My amps are the official NCC200 built with all the Avondale recommended bits.
My other amp balances up differently - it uses different transistors and a bit of emitter degen on the VAS. It's hardly surprising it needs a different TR1 collector resistor to restore DC balance in the LTP.
If the clones don't balance then it just shows the pitfalls of buying unqualified stuff off Ebay - buyer beware. OTOH I've bought quite a few Ebay amps and they usually require only a modicum of sorting (DC operating conditions or compensation tweaks).
At +/-32V the LTP balance is poor:
0.45mA versus 0.55mA.
At +/-40V the balance is pretty good - as per my figures earlier.
My amps are the official NCC200 built with all the Avondale recommended bits.
My other amp balances up differently - it uses different transistors and a bit of emitter degen on the VAS. It's hardly surprising it needs a different TR1 collector resistor to restore DC balance in the LTP.
If the clones don't balance then it just shows the pitfalls of buying unqualified stuff off Ebay - buyer beware. OTOH I've bought quite a few Ebay amps and they usually require only a modicum of sorting (DC operating conditions or compensation tweaks).
The LTP current dropping that far is not really a problem . Tail current usually is 0.7/R or 0.7/56 = 12.5 mA from the so called J V dia. More realistically 560 R,opps. 620 R in dia used here, which might be 1 mA as the 22K shared CCS resistor is a bit undersized .
0.7 V / 1 K = 700 uA ( approx ) 1000 - 700 uA @ 22K = 300 uA. It would be intersting to know a voltage at 22 K. I doubt it will be exactly ohms law as the transistor base draws some current ( VAS ).
If 32V is used try this. 22K x 32/40 = 17K6. 18 K will do. This is setting the original current through the CCS transistor bases.
0.7 V / 1 K = 700 uA ( approx ) 1000 - 700 uA @ 22K = 300 uA. It would be intersting to know a voltage at 22 K. I doubt it will be exactly ohms law as the transistor base draws some current ( VAS ).
If 32V is used try this. 22K x 32/40 = 17K6. 18 K will do. This is setting the original current through the CCS transistor bases.
One thing if bored you can try is adding an extra 1N4148 diode to the CCS bases ( 3 in total ). Double the emittrer resitors to 2 x 620R LTP ( 1K24 ) and 2 x 68 R (136 R ). The amp should behave as before except have slightly less second harmonic. It will have a tiny fraction less power ( -0.22 dB at a guess ). Humans are sensetive to 1 dB at best I would say.
Yes, on rechecking, I must have read that elsewhere. Apologies for misquoting you. Perhaps I assumed it from standard transformer rectified voltages many here opt for.
In my understanding, a 43V supply gives balance with any typical clone based on the NAP140 (or rather 250) circuit, being the basis of Ebay clones. If given the appropriate +/-34V supply, it doesn't balance and the sound is usually slightly richer, more characteristic of the original NAP140. Avondale's modules have some departures that set them aside I think, but the only NCC200 modules I have seen (they don't export), couldn't be checked out. Yet with a 40V supply, they were still cleaner sounding than early NAPs (Chrome bumper or older) I'm familiar with.
Perhaps we are more in agreement here, given your last figures.
While I've focused on the input stage, Miller capacitance mayhem in the VAS is also a source of more harmonic distortion, meaning the NCC200 has its share too. I don't know how this squares with your 002CCN variant, but I can't see it having particularly low distortion if ZTX parts are used there too.
I said earlier if bored you could try this. The idea being to take the transistors away from the Early effect. In my valve amps I found this to be about 1.2 V in the emitter resitors. I have always infered it was Early effect or to be even simpler the bias lifts the transistor above it's on point. As it is a constant current source the transistor will be at the same temperature as long as the conditions are kept in the same relationships. Thus if we go from 0.7 to 1.4 V across the emitter resistor we need double the resistor value to keep the same current.
We could possibly use a 1.5 V alkaline battery instead. Perhaps 1K and 110 R if so. The 22K resistor would be removed. AA batteries I would guess to be 2 AH ( soldered to the poles , not spring clips, some batteries have tags ). As the bases draw very little current this might last a long time and sound wonderful. If a comparator was set up on the battery a warning LED could light at 1.3 V. If a LM324 another at 1.6, 1.5 and 1.4 if you like. LM324 makes a fine comparator or LM339 a real one. Both can drive LED's. 1.5 V should already be fighting the switch on knee.
I used the Julian Vereker example as it shows it well enough. The Baxendale diode is a very significant improvement to all NPN output stages. It almost settles debate about the weakness of all NPN and offers the speed advantages which were very true in the late 1970's.
I tried to find something worth reading on Early effect. Alas I failed. It was like reading about Malaria without any interest in practical measures to live with it.
Julian Vereker Naim Audio 1989 : Top Audio Gear
We could possibly use a 1.5 V alkaline battery instead. Perhaps 1K and 110 R if so. The 22K resistor would be removed. AA batteries I would guess to be 2 AH ( soldered to the poles , not spring clips, some batteries have tags ). As the bases draw very little current this might last a long time and sound wonderful. If a comparator was set up on the battery a warning LED could light at 1.3 V. If a LM324 another at 1.6, 1.5 and 1.4 if you like. LM324 makes a fine comparator or LM339 a real one. Both can drive LED's. 1.5 V should already be fighting the switch on knee.
I used the Julian Vereker example as it shows it well enough. The Baxendale diode is a very significant improvement to all NPN output stages. It almost settles debate about the weakness of all NPN and offers the speed advantages which were very true in the late 1970's.
I tried to find something worth reading on Early effect. Alas I failed. It was like reading about Malaria without any interest in practical measures to live with it.
Julian Vereker Naim Audio 1989 : Top Audio Gear
There is a good explanation of Early Effect in "The Transistor" by E.F. Wolfendale. I see someone has one on offer here Transistor - AbeBooks. If you remember the 30 watt Bailey Amplifier from 1968 (there is a pdf of this and the Baxandall diode article on Keith Snook's website) you will see this quoted as the reference in regard to Bailey's observations about Early effect.
Last edited:
This is a copy of the original hand drawn sketch from Julian.
Output transistors that he suggested to use was BDY58.
Last edited:
Thanks for posting that msdin.
Interesting to note the low gain, heavy compensation and no TR2 collector resistor. I guess they were still developing the "Naim sound" at that point in time.
Interesting to note the low gain, heavy compensation and no TR2 collector resistor. I guess they were still developing the "Naim sound" at that point in time.
On Early effect. I think I might have guess right from this interesting answer. I have never seen a Cascode CCS. For my valve amps it could be very interesting. Second harmonic is sometimes something you want so perhaps it is wrong to cure it.
early effect voltage in current mirror
early effect voltage in current mirror
Without speaking ill of someone who is no longer here I wonder if Julian did pen the NAP design. Alan Mornington West is said to be it's author. No doubt Julian develloped the design once he had learnt it. I design things every day. I can't stop myself and my boss wouldn't want me to. 90 % of what I design is in my way of doing things be they valve, MOS FET or bipolar. The way J V kept with this design does not suggest a designer. I knew Julian and can tell you his facination for engineering was something he took trouble to explain. So odd he was so narrow when this amplifer. Even Peter Walker produced 3 designs of note in his lifetime. Quad told the BBC they had to sell 25 000 units to start making money. For them 3 designs since the 1950 to 1980 ( circa ) is defendable. The one they had most praise for was the least good. The Quad 405 opened the door for Julian Vereker. Between Quad and Naim was Sugden. Like Garrard 401 and Lenco no one gave a dam about Sudgen then. J V had Sugden test gear when I visited Salt Lane ( as was ). J V said " Sugden make good amplifers ". The A48 is a bargain.
I can scan the relevant pages of "The Transistor".
In a common emitter amplifier with low voltage swings Early effect can be regarded as non extant.
Last edited:
I can only confirm the LTP current imbalance that Ian and mjona mentioned. I didn't know, though, that this imbalance disappears with increasing power supply voltage. I just didn't have adjustable power source at the time.
I ran my clone from +/-31V. What I did (very easy) was to insert trimmer in the LTP pair current source. This way I experimented with resistances above and below 560ohms. You can reach perfect balance or really bad imbalance in the LTP currents by playing with that. Of course the price is change in LTP total current, say from 0.8 to 1.2mA. In my opinion the total current should be around 1mA. I settled with 560 ohms. Some people use 620 ohms, but that depends again on the Vbe in the particular CCS transistor.
Apart from the academic exercise, I don't think there is anything in tweaking the LTP. You need good wide BW, low noise, small signal transistors of suitable voltage and highest possible gain.
I ran my clone from +/-31V. What I did (very easy) was to insert trimmer in the LTP pair current source. This way I experimented with resistances above and below 560ohms. You can reach perfect balance or really bad imbalance in the LTP currents by playing with that. Of course the price is change in LTP total current, say from 0.8 to 1.2mA. In my opinion the total current should be around 1mA. I settled with 560 ohms. Some people use 620 ohms, but that depends again on the Vbe in the particular CCS transistor.
Apart from the academic exercise, I don't think there is anything in tweaking the LTP. You need good wide BW, low noise, small signal transistors of suitable voltage and highest possible gain.