NAP-140 Clone Amp Kit on eBay

very interesting - so the gain stage from the NAC is therefore an essential part of the 'whole' chain


Date: September 22, 1999 05:45 PM
Author: julian vereker
Subject: speed

I have read some rubbish in my time, but I think that some of the
contents of this thread is right up there.

Ross, Our amps are designed to amplify without audibly altering the
music signal at all. The idea that we could somehow "emphasise the
leading edge of transients" is ludicrous - this would imply a rising
HF response which would be clearly measurable and contrary to our
published specs or curves published in reviews.

Although our amps were very 'fast' when I designed them in 1970, to
day it is possible to design much faster amplifiers, for example the
NAP500 is some 5 times faster.

But the real art (which seems to elude you, Ross) is to ensure that
the amplifier is not called upon to drive a load with a signal outside
its operating envelope, and we go to great lengths to band limit (or
reduce the HF content of) the signal fed to the power amp to make sure
this does not happen.

Our amps are not "forward or edgy", they merely amplify the signal fed
to them unless one 'clips' them or uses them with speakers that have
such a low impedance that the protection circuit becomes operational.

julian

(404 Not Found
D=38131)
 
Most have a "guestimate" approach at best or simply copy another design. It's complicated.

Yes, it is complicated because each capacitor is different but there is a rule of thumb that can be followed. I'm aware of the theory on how to calculate the capacitance but if we use ears we know right away that it doesn't 'work'.

We cannot parallel too many capacitors. The more capacitance the cap has, the less number of parallels it can have. It is a proper approach that the F3 has 4 caps for each channel instead of 8 caps for both channels.

I have never found a 22,000uF cap, good enough for paralleling with another 22,000. The F3, parallel of 2x8,200uF makes more sense than 3x10,000uF Ultima Legion is using for his Naim clone (which is a class AB amp unlike the F3). I didn't say this is impossible as Ultima is using high quality caps i have never tried.
 
Among the Russian lovers of sound Hi-End is the most famous and informal in their views people. This is Yuri Makarov. His main creations are SE tube amplifiers with a power of no more than 6W, but working with such "heavy" acoustics as Montana WAS!

His personal formula, is constantly used by him: "The capacity in the power supply of the output cascade in class A, must ensure a drop in the peak of a pulse with a duration of T=1 s by no more than 1% - Makarov Yury Anatolyevich (wizard) - and is calculated by the formula C = 50 * I / U (F / A / V). "
YouTube
 
I asked because I wanted to touch on how people choose how much capacitance to use and how much transformer VA. Most have a "guestimate" approach at best or simply copy another design. It's complicated.

I'm still very much in the guesstimate/copy camp.

I do model the supplies in PSUD2 and/or LTSpice, but I'm not sure what sorts of ripple and impedance I should be targeting. It was less critical with the F3 because it's got a big C-multiplier on each channel, which leaves a residual ripple of 1.2mV.

On my F5 (a two-rail design) I went with a pair of 33,000uF caps per rail, followed by channel-decoupling at the resistors, followed by a single 33,000uF cap per channel per rail. (So 8 caps in total.) Residual ripple is 100mV. (That's a lot higher than the F3, but the F5 is push-pull and so has a better PSRR than the single-ended F3.)
 
Another thing to consider is the charging current. With a full-wave rectifier the reservoirs get recharged every 10ms (assuming 50Hz mains). If, for example, your amp draws 1A continuously, then it drains 10mC between recharges. Now, the recharging time will depend inversely on the reservoir capacitance. It could have, for example, 2ms in which to recharge the 10mC and this would require 5A (I=Q/t). So your transformer is having to provide 5A + 1A amp draw for 2ms every 10ms. The larger the reservoir capacitance the shorter the recharge time and the higher the transformer current.
 
Last edited:
Yes, it is complicated because each capacitor is different but there is a rule of thumb that can be followed. I'm aware of the theory on how to calculate the capacitance but if we use ears we know right away that it doesn't 'work'.

We cannot parallel too many capacitors. The more capacitance the cap has, the less number of parallels it can have. It is a proper approach that the F3 has 4 caps for each channel instead of 8 caps for both channels.

I have never found a 22,000uF cap, good enough for paralleling with another 22,000. The F3, parallel of 2x8,200uF makes more sense than 3x10,000uF Ultima Legion is using for his Naim clone (which is a class AB amp unlike the F3). I didn't say this is impossible as Ultima is using high quality caps i have never tried.


I absolutely agree.

When it comes to electrolytic capacitors, together with a careful and correct sizing of the total capacitance (to avoid a "swollen" and "not very fast" sound) it is always better to avoid parallelism of more than 2/3 units maximum.


It is no coincidence that the exact scheme I used for the linear power supply stage of each single channel of the stage driver and power BJT of the Naim clone is the following (C108 and C109 are two capacitors, possibly in Polypropylene, 0.33/0.47 uF):


T2-VBBHXf8-XXXXXXXXX-67890995.jpg


The one with six capacitors (three for each rail) posted before, is the one I used for another project, the DartZeel NHB-108 clone with two pairs of BJTs of power, which needs a higher power reserve in dynamic mode.
 
Last edited:

I think it was said already but just to clarify that this two diode rectifier is actually full wave NOT half wave. Secondly, it's implemented in this link with a choke input filter with a small cap to bring the voltage up - choke input would normally be 0.9 x Vac but a small cap before the choke can be used to tune it upwards, at the expense of higher charging spikes than a pure choke input filter. This approach is wasteful on power transformer copper - you need twice the winding when not using a bridge rectifier so you get twice the copper losses. For low current tube amps it's less of an issue. I used a two diode cap filtered psu approach on a Sansui Tuner when I upgraded the power supply but it's relatively low current - Grasshopper (you'll see a Naim circuit in that thread too).
 
Last edited:
The 250 DR spec says 80W per channel into 8 ohms. I take that to mean average sine wave power, implying the rails are around +-40V or so.

Some use a rule of thumb that the transformer VA should be twice the speaker output power. In this case 320VA. The Naim transformer is quite a bit higher than this by the look of it.
 
Last edited:
Member
Joined 2010
Paid Member
Note that the 250 models are linear regulated which requires higher voltage AC windings and more current than a simple rectifier and smoothing caps used on most other NAP models. To be scientific, you'd need to do a lot of static load testing if you expected any parity with the real products. Or you could search out the spec for the transformer - perhaps it's even written or labelled on later models.

The general description, including the transformer of the early versions of the NAP 250 was, according to Ced Taylor, as follows: "Nap 250: 70 watts into 8. 450VA transformer running at 56Vdc, 2 windings, 2x22,000uf caps feeding 2 regulator boards and pushing out 40V rails. A stereo amp- both channels use the same supply! Dodgy star ground at the speaker 0V terminals."

I would guess that the DC stated as 56V feeding the regulators would be an unloaded figure and you probably will need to calculate by the transformer's regulation, what that falls to at full rated load with the actual transformer, caps, rectifier and circuit resistances you wind up with.
 
Last edited:
Member
Joined 2010
Paid Member
To follow on, NAP250/2 was a dual mono arrangement and I'd be surprised if the 250DR version wasn't also. The transformer is still a single core but both sets of secondary windings are wound separately on it so loading effects are shared to some extent and the capacitance is now 2x10,000 uF for each channel's raw DC supply. This is the same arrangement as most later (black fascia series onward) models.
 
I think a more scientific way of deciding VA is in order.


:D:D

Of course!

In fact when designing a power supply stage of any amplification stage, the determination of the transformer power would dare to say that it is the main aspect, which must be addressed well beyond any empiricism.

In the case of a power amplifier, we're off:

1) fixing the maximum polarization point of the power transistor in static mode,

2) is measured according to the DC voltage with which we have decided to supply our amplification stage, which is the corresponding value of maximum current allowed by S.O.A. (Safe Operating Area) of the power BJT.


SOA.jpg

For example, in the case of a 2SC5200 that I have chosen for my NAIM clone, in correspondence to the DC voltage of about 40 Volts, with which I have decided to power my amplifier, a current of about 4 Amps is detected.

3) once determined according to our power BJT and to the chosen DC supply voltage, we calculate the maximum static power required: P = V x I = 40 Vdc x 4 A = 160 VA which obviously is multiplied by x 2 (in the case of stages with a pair of power amps as in the case of the NAIM clone) obtaining 320 VA of power for each channel.

4) Since the Power is now known, we proceed by dividing it by the AC voltage of each individual supply rail (before the rectification stage) and we obtain the current that each secondary of the supply transformer must be able to deliver for each single amplifier channel: I = P : V = 320 VA : 30 Vac = 10.67 Ampere.

5) Now we are able to establish that our power transformer must have two secondary transformers with output each: 30 Vac Voltage, 10 Amperes Current for a total of 600 VA.

So if you want to get out all the power available from a pair of high-performance 2SC5200 (which I remember have a maximum collector current of 15 Amps), any transformer between 500 and 600 VA, will do more than fine.

Besides it is absolutely not worth going as Power because especially in the case of the clone NAIM type H-140, without the protection / current limiter, we seriously risk, in the case of demanding speakers and low impedance module listened to at extreme volumes, to send in crisis the final power transistors.
 
Member
Joined 2010
Paid Member
The primary power limiting factor for the transformer in the 250 models is the safety current limit of the series regulators. They can't just be allowed to pump out whatever current the load may demand like a transformer, as that would soon result in failure of the reg, if not the output stage of the amplifier as well. The regulator current limit and transformer specs would be matched for each amplifier model/revision, I would think.

VI limiting circuits are the front line of overload protection but they're principally to prevent SOA excursions of the output stage. Unfortunately, lots of unthinking tinkers remove the limiters in order to prevent any possible degradation of the sound and that removes the front line, leaving the regulators and fuses as backline disaster prevention - just hope you never need it! If you have an unregulated model, you'll need to use the amp at low power only, fit smaller fuses and really beefy semis and PCBs if you omit the limiters.