Wow, that's an big piece of copper but overall, heat dissipation would be much better with an all-aluminium cabinet, as per the original. The aluminium base alone, is now enclosed under the amplifier so it won't be capable of dissipating much heat through the steel cabinet. Be cautious with how hot the base becomes when you turn up the volume. I suspect you won't be able to get full power with safety for more than short periods before thermal runaway occurs.
I thought that closing the cabinet underneath so as not to let the heat out immediately was a good idea because it was made of steel. Keep in mind it's a 3U because I wanted to build a 4-input selectable pre-integrated on the Naim Nac 152XS with dual AK4493 DAC board so there is room internally. At this point I was taking a completely closed aluminum cabinet like the original Naim. I don't know whether to tack on a class D with TPS 3255 at this point or Hypex ...
Also on the original Naim Nap 200 there is a dissipation bar on the 4 power transistors. I understand the completely closed cabinet that Naim have but basically it's about getting the amplifier to reach a certain constant thermal equilibrium. The heat is not lacking with three toroidal transformers for a total of 600VA and the heat generated by the amps. It's not like it can be totally closed like the original Naim. The construction philosophy is different and I am willing to make calibrated thermal adaptations.
Class AB amplifiers can only have a real thermal balance when there is no output. Ideally, the bias control would be very fast, so the thermal circuit would need to have very low mass. However, NAP series output transistors have a low bias current setting (only ~ 30mA flows through the output transistors) which means that they run quite cool until they are producing significant power. Then the heat dissipation begins to bounce about with the audio output current but there is really only a delayed average bias control range which maintains bias within a safe range, rather than one that is anywhere near optimal for linear audio.
That can true for other designs too but a quasi-complementary design has a significant problem in that the thermal model for either the EF or CFP transistor is different, so bias for the pair can never be set and held to any optimum level - not simply anyway. So, as a compromise, we set a current that is appropriate for a Sziklai (CFP) pair and happily, this coincides with safe operation and lowest THD for the Quasi pair.
However, with the high mass of your amplifier and consequent very slow thermal loop, I think you are going to have problems with holding a stable bias for long periods of time. Some of the heat generated will, as you say, be related to transformer losses too but this may have little to do with the required bias. Air vents at the top may be necessary but they won't help much with without a corresponding inlet at the bottom and again, that will be counter-productive to control.
I would have chosen a standard EF amplifier design with finned heatsinks for that cabinet but we all have different aims and likes so I understand what you are trying to achieve. Still, I do think you will need to be careful to avoid bias problems as the listening sessions get longer and you try higher sound levels.
That can true for other designs too but a quasi-complementary design has a significant problem in that the thermal model for either the EF or CFP transistor is different, so bias for the pair can never be set and held to any optimum level - not simply anyway. So, as a compromise, we set a current that is appropriate for a Sziklai (CFP) pair and happily, this coincides with safe operation and lowest THD for the Quasi pair.
However, with the high mass of your amplifier and consequent very slow thermal loop, I think you are going to have problems with holding a stable bias for long periods of time. Some of the heat generated will, as you say, be related to transformer losses too but this may have little to do with the required bias. Air vents at the top may be necessary but they won't help much with without a corresponding inlet at the bottom and again, that will be counter-productive to control.
I would have chosen a standard EF amplifier design with finned heatsinks for that cabinet but we all have different aims and likes so I understand what you are trying to achieve. Still, I do think you will need to be careful to avoid bias problems as the listening sessions get longer and you try higher sound levels.
Thank you for the valuable technical overview of the problem. I am thinking of possible remedies in order to reconcile the need to combine everything in a single cabinet and the correct regulation of the bias even at high power levels. For example, I could enclose the part relating to the Nap 200 in a closed container or at least delimited by an aluminum sheet by a couple of millimeters in order to have a controlled environment separated from the rest. In your opinion, can this be a solution?
I'm sure that is a possible solution but there are no numbers for a model that could be used to design something that would work reliably from the start. What we do know, is that the thermal control loop for Quasi-complementary amplifiers is usually loose and very slow by comparison with standard emitter-follower output stages. Even CFP output stages were problematic until recent times when Douglas Self published an easy way to monitor the driver transistor case temperature rather than the output transistor case temperature as the bias feedback element. Otherwise, trying to design a CFP bias controller with guesswork or copying other designs, usually meant an expensive failure.
You may like to experiment and have some fun trying your ideas out but I would make sure first, that you understand how the bias might best be controlled and how to maintain safe operation. You might read an old edition of Self's book "Audio Power Amplifier Design handbook" online, such as the 4th edition, chapter 13 - particularly the CFP type bias controller design: https://www.desmith.net/NMdS/Data/B...- Audio Power Amp Design Handbook 4th Edn.pdf
There may even be Italian translations out there but you should be able to use a translation engine with the downloaded PDF text to make it easier.
You may like to experiment and have some fun trying your ideas out but I would make sure first, that you understand how the bias might best be controlled and how to maintain safe operation. You might read an old edition of Self's book "Audio Power Amplifier Design handbook" online, such as the 4th edition, chapter 13 - particularly the CFP type bias controller design: https://www.desmith.net/NMdS/Data/B...- Audio Power Amp Design Handbook 4th Edn.pdf
There may even be Italian translations out there but you should be able to use a translation engine with the downloaded PDF text to make it easier.
Here is a link to an entertaining and informative article on Naim amplifier biasing from Pinkfishmedia Forum in the UK. It refers to testing the performance of a Nait2 model but the electronic design and requirements are quite similar. I've posted this link before but I hope you enjoy and appreciate the findings as much as I still do: Just one thing about music - when it hits you feel no pain
At the left margin, scroll down to "more Naim stuff" and click on "bias-settings in Naim amps".
At the left margin, scroll down to "more Naim stuff" and click on "bias-settings in Naim amps".
The gist of the speech at the end is this:
So - based on a sample of one(!) - the recommendation is simple: make sure your Naim's output stage displays at least 4.5mV across a 0R22 emitter resistor that is, Iq around/up to 20mA here; more accurately - aim for 8.0 to 8.9mV across BOTH emitter resistors. Provisos are 1) just don't worry if it's within the 4-6mV range and 2) don't waste a lot more bias current in the hope of lower distortion, it only risks thermal runaway and does not improve matters. That's all!
Seems easy...I hope...
Yes, it does look easy. However the Nait2 amplifier is quite small, closed and was also pre-conditioned by applying a sine wave with a fixed load to stabilise its temperature prior to testing. That makes it look easy to get repeatable test results but it isn't what happens when you casually listen to music. I simply turn an amplifier on, select the program and play but it takes time for Naim amps to settle at a stable bias level, particularly at modest volume levels which are necessary in my home. The sound isn't quite "right" until it does settle, with some fanatics claiming at least half an hour is necessary or that you should leave the amplifier on 24/7 which is high-risk For DIY and other simple designs.
I think the recommended bias setting for NAP models is actually 6.5mV or 13mV across both resistors but as the author shows, it's not really critical if the requirement is just lowest measured THD.
I think the recommended bias setting for NAP models is actually 6.5mV or 13mV across both resistors but as the author shows, it's not really critical if the requirement is just lowest measured THD.
Last edited:
You can find here bias adjustment measurements; Naim, NAPs, Naits, distortion and bias-current settings
The bias current that passes through both output transistors and their emitter resistors, is all the same. Therefore the voltage drop across each emitter resistor will be virtually the same. This means that you can just as easily (and more precisely with basic digital voltmeters ) measure the total voltage drop of 13mV across both resistors. There are options for how you measure the bias current, but applying Ohm's law to the total 0.44 ohm resistance will be slightly better.
@mdardeniz - yes, see post #29. That is the same article discussed here.
@mdardeniz - yes, see post #29. That is the same article discussed here.
Perfect. Thank you for the clarification. I just have to get to work and cross my fingers.
BC546A/556A do actually a better job despite the smaller Hfe that do not matter much anyway, their higher VCE (smaller parasistic collector/emitter and collector/base resistances) reduce the distorsion of the input stage.
Otherwise there s japanese items that cumulate all advantages, high VCE and very high Hfe, notably the 2SA872(A)/2SC1775(A) that are only 300mW though, and the 2SA992/2SC1845.
In principle I would agree with your suggestion though perhaps with Naim product, we misunderstand the design and assume it is intended to be just another low distortion design from the 1970s. However there is clearly an imbalance of the LTP currents to encourage even harmonic distortion and there are a few more sound effect "tricks" in there too. All early Naim product follows this line of thinking and design topology, which was the basis of their early popularity and success.
Good day but sound is not good
)Bc239c better
I think Chinese gents have a very mistake with mpsa18 because naim use 2n59xx
so if you change In yours pcb mpsa to bc you need check output mVolts and if this over 30mv you correcting output bias power
You can have Chinese BC239C labelled semis but they won't be BC239 anything, will they? They will be generic, small signal substitutes for BC, 2SC1815 or 2N3904 type semis, labelled according to the specified order types.
Actually, a couple of old Naim videos make a point of the selection of mis-matched rather than matched pairs of LTP transistors, if you want to be specific about what really made original Naim sound. Of course, a by-the-book engineer would eliminate that and fit an op-amp or allow the tape-and-reel production sequence to select closest matches but with the sad trade-off of lower THD numbers for sound quality - a sad but necessary shift from the unique NAP and NAIT sound back to mainstream sound quality. I imagine it suits the current market for high end consumer audio product and Focal's interests better, though.
Actually, a couple of old Naim videos make a point of the selection of mis-matched rather than matched pairs of LTP transistors, if you want to be specific about what really made original Naim sound. Of course, a by-the-book engineer would eliminate that and fit an op-amp or allow the tape-and-reel production sequence to select closest matches but with the sad trade-off of lower THD numbers for sound quality - a sad but necessary shift from the unique NAP and NAIT sound back to mainstream sound quality. I imagine it suits the current market for high end consumer audio product and Focal's interests better, though.
Last edited:
Good day
Dear Lan
Yes Chinese transistors is relabelled but BC239C and BC238C (on semi - Motorola) is still sales in Russian stocks
Last edited: