TGM10 - based on NAIM by Julian Vereker

Status
Not open for further replies.
Bigun,
Have you got all your parts yet...will this weekend be the big build? 🙂
Regarding the LTP. You'll want the transistors to be identical and you'll want their Ic to be the same. Ha! You can use your trimmer to adjust the Ic and if you get too much dc offset you could try hand-matching parts from a small pool of transistors.

🙁 nope, not this weekend. My schedule is too tight at the moment. I will be going on a business trip this coming week and need to be ready. I'll actually be in your neck of the woods at the following two weekends (Ross on Wye) visiting my parents, with the week in-between spent visiting business connections for my company. There's a lot going on, I don't think it'll lighten up until well into May so the usual frenzy of activity for the hobby will be a bit here-and-there.

I have been remiss in not ordering parts, I thought I'd have the boards some time ago and wanted to see them before ordering - Digikey is very fast.

So, back to the LTP... the LTP is inherently unbalanced because it has to feed the base-bias current of the VAS and at a.c. it has to drive Cdom. The prevailing 'wisdom' suggests that I plant one of the devices into the circuit and then select-on-test the other device to minimize dc-offset at the output ?
 
Bigun,
Have you got all your parts yet...will this weekend be the big build? 🙂
Regarding the LTP. You'll want the transistors to be identical and you'll want their Ic to be the same. Ha! You can use your trimmer to adjust the Ic and if you get too much dc offset you could try hand-matching parts from a small pool of transistors.

The Naims seem to need a very unbalanced LTP to produce the THD that JV thought necessary for "his sound".

If one is building a Naim clone then one NEEDS to copy all of it.
If one wants true "reproduction" then choose a better reproducer, not an effects machine.

What are you talking about?
Replying to your post316.
 
My experience with NAP clones:

The main contributor to the "Naim sound" is the unequal LTP collector resistors. This warms up the sound noticeably compared to shorting the 22k (or 5K6 with correspondingly higher LTP current in later amps). Shorting this resistor makes a tiny difference to the dc offset and stability of the amp so it's an easy and safe (albeit at your own risk) experiment.

Front end rail decoupling....
This is an interesting one. I prefer the sound the way Naim does it i.e. with no extra decoupling.. I assume they do this to keep the ground traces free of charging currents, of course there is a possible tradeoff with stabiliity, but in practice this does not seem to cause any issues.

The phase correction networks...
The idea of relatively larger base stoppers on the drivers transistors was to protect the drivers in the event of an output device failure. The problem with the large base stoppers was that is slowed down the output stage and therefore they added the "phase advance" networks to improve the transient response. I found it was better still with "modern" values of base stoppers (which don't benefit from or need the phase advance networks).

I don't think there is any particularly special about the 00X transistors. They were Sanken die selected and packaged by Semelab (not sure of the exact period). My guess is that the newer ones use Semelab's own chips.

As others have said, the amps' sound is quite dependant on the power supply used. I am not a great fan of toroids and I suspect Naim know how to specify them to minimie the issues. I also suspect this is the most difficult part for DIYers to get right, when trying to emulate the Naim sound.
 
So, back to the LTP... the LTP is inherently unbalanced because it has to feed the base-bias current of the VAS and at a.c. it has to drive Cdom. The prevailing 'wisdom' suggests that I plant one of the devices into the circuit and then select-on-test the other device to minimize dc-offset at the output ?
Viva Ross Vegas. :up:
Wisdom that prevails; what a concept. Yes, I recommend that you get the collector currents the same and select on test for dc offset. That subtractor should be as accurate as you can be bothered to make it.
 
The Naims seem to need a very unbalanced LTP to produce the THD that JV thought necessary for "his sound".

If one is building a Naim clone then one NEEDS to copy all of it.
If one wants true "reproduction" then choose a better reproducer, not an effects machine.

I cant believe that this is all about imbalance LTP current. If we change arbritrary resistor, it will change the sound, but it doesnt mean that this "sound" is about that resistor.

I believe that to clone the Naim sound one must listen to the sound carefully then decide which part or parts of the sound that people really want. Of course it is not the lousy HF!

Then one must predict what causing that unique sound. It doesnt have to be with LTP input!

Most unique sounded amps with cult followers have unique bass quality. The typical amp with Naim topology does have unique bass quality. And I dont believe this bass signature can be cloned with 2SC5200, the transistor Bigun is using. Yes, the HF will be smoother but the bass could be lighter too.
 
2pair of 100W devices will give a reliable 100W amplifier driving severe reactance speaker if the heatsink maintains a sensible Tc.

I noted one criteria: amp reliability to drive severe reactance! Isnt this part of the design quality?! A bad two pairs are not better than a good one pair.

The heatsink is not a limit. It can be as big as we want it. So, how many transistors in total, 14?
 
Last edited:
I cant believe that this is all about imbalance LTP current..

The current balance in the LTP is quite good. It's the collector voltages that are unbalanced.
The reason for this is to achieve the DC 'balance' of an LTP along with the sonic characterisitcs of a single ended class A amp at signal frequencies.
It is such an easy experiment to try the amp with the "distorting" collector resistor fitted and shorted, that anybody who builds this really ought to try it. With it shorted, the amp sounds much more like any other ss amp. The most interesting question is why does an "effects box" can sound more right that a lower distortion amp?
 
The current balance in the LTP is quite good. It's the collector voltages that are unbalanced.
Yes.
The reason for this is to achieve the DC 'balance' of an LTP along with the sonic characterisitcs of a single ended class A amp at signal frequencies.
It is such an easy experiment to try the amp with the "distorting" collector resistor fitted and shorted, that anybody who builds this really ought to try it. With it shorted, the amp sounds much more like any other ss amp. The most interesting question is why does an "effects box" can sound more right that a lower distortion amp?
Check your assumptions. You are assuming that the resistor adds significant distortion and that is why it sounds better. Of course this begs your "most interesting question". But are you asking the wrong question because of a false premise?
 
I gave you an answer: 2pair of 100W devices. (=4 devices)
Why are you talking about 14 devices?

Read again. There was a reason why I asked the question.

It is easy to achieve 0.001% THD at 100W even with latfet amp. The question is how we can achieve that. By sacrificing other performance parameter? By throwing in unnecessary parts and complexity? Number of transistors used in the circuit is just a clue about the method used.
 
The most interesting question is why does an "effects box" can sound more right that a lower distortion amp?

Traderbam is right, you made a wrong assumption. Amplifier sound is not just about THD. The answer to your question is: when you improve thd, other parameters may suffer, so it is normal ta sacrifice thd a bit for the sake of other parameters.

You may ask: what parameters? There are many. But I will give you an example that may sound controversial, so people will understand that these parameters are not always simple variable like DF, noise, PM, etc.

Linearity can have a lot of meanings in amplifier. All or most of them are important. Suppose we design an amplifier and we are faced with 2 choices. Amp A has lower distortion than amp B, but amp B sounds better. Is it possible? Why not? In amp B, an increase in voltage (at listening spl) will be followed by a reduction in thd. In amp A, an increase in voltage will be followed by an increase in thd.
 
Check your assumptions. You are assuming that the resistor adds significant distortion and that is why it sounds better. Of course this begs your "most interesting question". But are you asking the wrong question because of a false premise?

It is entirely possible I have made an incorrect assumption. I will have to have a think about this, unless of course you would be happy to enlighten me.
 
Ask how the LTP in this amplifier works in comparison to one with a constant current load with a buffer transistor feeding the Vas transistor and one where the collector of the feedback transistor connects to a supply voltage rail.

How good is each of these in reducing harmonics generated in transmission through internal stages - particularly the output which is capable of generating odd order harmonics.

With no input signal opposition in the last case the feedback signal for this will be larger in relation to others present in both LTP inputs.

In this amplifier the input transistor runs more emitter current than the nfb return transistor emitter and the nfb feedback path is between these two points.

How well does this compare with a situation where there are equal emitter currents and one where the nfb transistor has the greatest share of emitter current.

While each transistor in a LTP is a common emitter amplifier the base voltage changes according to emitter current.

There is Miller capacitance in both LTP transistors of the Naim circuit this is an inverse function of the voltage at the collector in each case - there will be less capacitance when the voltage is high. This will have more effect on the nfb transistor but in either case it will be variable with signal inputs.
 
Last edited:
a matched pair of LTP input transistor can only remain matched if they operate at the same voltages (Vbe, Vce, Vcb) and operate at the same currents (Ib, Ic Ie) and operate at the same junction temperatures and see the same impedances from each of their leads.

If any of those fall out of balance, then the input pair are no longer matched.
This unbalancing is a deliberate act on the part of the Naim Designer.
A Naim clone must copy that out of balancing to stand any chance of matching the performance of the original Naim one is trying to clone.
 
Status
Not open for further replies.