I used 220uF right @ the OutPut transistors, i can't tell if it made any difference at all... so i removed them...
Now i want to de-couple again, this time against possible RF @ radio range frequencies... means ~68nF+5uF decoupling on outputs!
Reason is simple, all the rest, mostly 90's amplifiers i have dealt with had some kind of decoupling network @ output transistors (starting from ceramics and ending with 10uF electrolytes).
From my experience using TIP41 and Faster CSS bjt will result in punchy bass
reproduction.
Using MPSA and MJE series, will gain you a huge benefit to Vocals, guitar, piano and dynamics.
Drums are more loose (to me more real)
Here, bass gets more relaxed...
Unfortunaly, you can't have both benefits with this setup...
Now i want to de-couple again, this time against possible RF @ radio range frequencies... means ~68nF+5uF decoupling on outputs!
Reason is simple, all the rest, mostly 90's amplifiers i have dealt with had some kind of decoupling network @ output transistors (starting from ceramics and ending with 10uF electrolytes).
From my experience using TIP41 and Faster CSS bjt will result in punchy bass
reproduction.Using MPSA and MJE series, will gain you a huge benefit to Vocals, guitar, piano and dynamics.
Drums are more loose (to me more real)
Here, bass gets more relaxed...
Unfortunaly, you can't have both benefits with this setup...
I would read Self (e.g. section 5.1.2 at http://www.douglas-self.com/ampins/dipa/dipa.htm) on the behaviour of LTPs. Current balance in the pair remains roughly proportional to the collector resistor ratio R1:R2 but it is forced into balance by the action of global negative feedback and this doesn't alter the basic circuit operation nor make it a balanced LTP. Model the input stage in isolation or even break the feedback loop and check this out.
Last edited:
To my understanding (and Selfie's explanation seems to concur) the collector current the input transistor is "regulated" by the Vbe of the VAS maintaining about 0.6V across the collector resistor. The common base side on the LTP than takes the tail current minus the input transistor collector current. By your logic fig 8b cannot work.
Anyway the point is that the NCC/NAP LTP is properly balanced. Andrew claimed it is not. I just wanted to set the record straight.
Dave S,
you set the record wrongly.
The LTP pair need to be balanced for least distortion. I, you and D.Self agree.
That requires the balance to be maintained when the devices are operating.
To do this one must balance all the currents and voltages and thus temperatures.
The Ic needs to match, the Ib needs to match, then the Ie will also match.
The Vce needs to match, the Vbe needs to match , then the Vcb will also match.With matching currents AND matching voltages the powers will also match. Thus Tj matches.
Those are the pre-requisites for a balanced LTP.
Here are your values again
Then look at your example of decent balance. It is even worse !
you set the record wrongly.
The LTP pair need to be balanced for least distortion. I, you and D.Self agree.
That requires the balance to be maintained when the devices are operating.
To do this one must balance all the currents and voltages and thus temperatures.
The Ic needs to match, the Ib needs to match, then the Ie will also match.
The Vce needs to match, the Vbe needs to match , then the Vcb will also match.With matching currents AND matching voltages the powers will also match. Thus Tj matches.
Those are the pre-requisites for a balanced LTP.
Here are your values again
Note how different the voltages and powers are.
Then look at your example of decent balance. It is even worse !
Last edited:
It's easy to get tied up with semantics and worrying about what "excellent" and "decent" mean to different individuals.
My point is that your initial maths involved an LTP current of 2mA. Maybe you are looking at a different schematic but all the schematics I have seen use 1 1N41418 forward diode drop (or more pedantically 2 X VD - 1 X VBE) across 620R to set the LTP current - that's nominally 1mA.
The NCC has degen resistors so mild imbalance is more tolerable. In addition to this some designers like to add a bit more 2HD e.g. Pass has cited that falling levels of harmonics with frequency is a desirable trend.
In your original post you stated "Hardly qualifies as a long tail PAIR!"
IMO the LTP is balanced well enough for practical purposes.
My point is that your initial maths involved an LTP current of 2mA. Maybe you are looking at a different schematic but all the schematics I have seen use 1 1N41418 forward diode drop (or more pedantically 2 X VD - 1 X VBE) across 620R to set the LTP current - that's nominally 1mA.
The NCC has degen resistors so mild imbalance is more tolerable. In addition to this some designers like to add a bit more 2HD e.g. Pass has cited that falling levels of harmonics with frequency is a desirable trend.
In your original post you stated "Hardly qualifies as a long tail PAIR!"
IMO the LTP is balanced well enough for practical purposes.
The schematic, based on the generic NAP 250 design, as is the NCC200, shows a 2 diode bias string and that's what I find to be normal. It often results in 1.5 -2 mA tail current but that's not really the issue here. See the schematic for rensli's H140 clone at #1407. It's essentially the same, barring the hand drawn modifications.
Naim (and some other old designs of the 70's) have been criticised by engineers and experts for decades over their intentionally unbalanced LTP design. As you may be suggesting, this is all about recovering the 2nd harmonic distortion that is selectively cancelled by a properly balanced LTP. Make of this technique what you will but balancing the LTP here, turns the amplifier into a generic one with only ho-hum sound quality. In fact, it's worse than ho-hum and yes, I've tried it - also with current mirror and more than once.
So the last thing the designer wants here is a balanced LTP and that's why its worth rabbiting on about what makes a truly balanced LTP. AndrewT and Self, I assume, are correctly arguing the case for true balance and what is required to make it so but it needs to said (again) that this is irrelevant to a Naim clone based on the early models as are the clone kits discussed here and the NCC200.
Naim (and some other old designs of the 70's) have been criticised by engineers and experts for decades over their intentionally unbalanced LTP design. As you may be suggesting, this is all about recovering the 2nd harmonic distortion that is selectively cancelled by a properly balanced LTP. Make of this technique what you will but balancing the LTP here, turns the amplifier into a generic one with only ho-hum sound quality. In fact, it's worse than ho-hum and yes, I've tried it - also with current mirror and more than once.
So the last thing the designer wants here is a balanced LTP and that's why its worth rabbiting on about what makes a truly balanced LTP. AndrewT and Self, I assume, are correctly arguing the case for true balance and what is required to make it so but it needs to said (again) that this is irrelevant to a Naim clone based on the early models as are the clone kits discussed here and the NCC200.
Must admit I've never seen >1.5mA out of that design of current source with a 620R setting resistor. More like 1mA.
Maybe we need to clarify if we are talking DC balance of AC? (I was talking about DC)
If I understand correctly the reason for the 22K TR2 collector resistor is that it is part of the compensation scheme. My own (practical) observations are that it slows the LTP down a bit and actually works against the dominant pole compensation scheme. It has very little effect on DC balance e.g. shorting it makes very little difference to the DC conditions.
Any ideas? Are you saying it's raison d'etre is to add 2HD?
Maybe we need to clarify if we are talking DC balance of AC? (I was talking about DC)
If I understand correctly the reason for the 22K TR2 collector resistor is that it is part of the compensation scheme. My own (practical) observations are that it slows the LTP down a bit and actually works against the dominant pole compensation scheme. It has very little effect on DC balance e.g. shorting it makes very little difference to the DC conditions.
Any ideas? Are you saying it's raison d'etre is to add 2HD?
Yes.
Here are the measurements from both channels at quiescent with +/-40V supplies.
Voltages measured across resistors:
1K 0.500V 0.501V
22K 10.75V 10.66V
620R 0.632V 0.628V
68R 0.716 0.705V
220R in + rail 2.52V 2.49V
220R in - rail 2.86V 2.82V
This implies
LTP current ~1mA
VAS current ~10.3mA
That 1k voltage drop must be the Vbe of the VAS transistor.
What current is passing the ZTX that allows a 0.5Vbe?
Compare to the ZTX653, it is passing ~10.4mA
It should have a Vbe ~ 2*Diode drop - ~0.71 this will be around 620 to 650mVbe
There is no way the VAS is passing 10.3mA @ 500mVbe
It appears from your numbers that the CCS sink is being fed from the bases of the drivers, probably due to very unequal base currents.
You have resistors there that you can check.
You also have the LTP degen resistors where you can check Ie.
What current is passing the ZTX that allows a 0.5Vbe?
Compare to the ZTX653, it is passing ~10.4mA
It should have a Vbe ~ 2*Diode drop - ~0.71 this will be around 620 to 650mVbe
There is no way the VAS is passing 10.3mA @ 500mVbe
It appears from your numbers that the CCS sink is being fed from the bases of the drivers, probably due to very unequal base currents.
You have resistors there that you can check.
You also have the LTP degen resistors where you can check Ie.
Last edited:
I can't say how the NAPs were developed, i.e. the NAP200, back in the early 1970's. Nigel Pearson has a theory about a pro. designer being involved yet this may simply be the outcome of Mr Vereker's self-taught design skills. It was then, only a one-man business. You can see the relatively high level of H2 on any soundcard spectrum analyser (c.f. Self's generic design amplifier). It's not so impressive now, but in a '70s world of hard sounding transistor amplifiers, this is surely what gave the NAPs so much appeal and strong demand. Then there are those novel filters in the driver bases but that's another story.
Last edited:
Measurements:
TR1 Emitter resistor V 53mV 53mV
TR2 Emittor resistor V 46mV 48mV
TR6 VBE 0.516V 0.522V
Don't forget that the ZTXs are probably running Tj of ~100degC.
Agreed!!!
WRT the novel filters, the original RCA note says they are to improve transient response. The lower one actually does, the upper one seems to do nowt. However, transient response is improved more by just using a sensible (e.g. 100R) in place of the entire network.
The original reason for the high resistor values seems to be to limit fault current to protect the VAS. I'm not convinced this extra precaution is necessary and neither do most amp designers, judging from the lack of them in most/all modern amps.
OTOH the RC on the base of Tr6 is rather cunning.
That's interesting for me at least, as I can't locate a reliable copy or reference to the "RCA note". Care to share it here or post a link? I'd agree on the overdone driver base limiting - I think it's retained mainly for originality.
VAS protection often is useful for anything with ≥50V rails and mandatory when there is an emitter follower driving it. These days, a small transistor samples VAS current and shunts base current to prevent it being fried.
I think the "RC" is the traditional 1k dividing the reference voltage for the 2 current sources and so reduce signal feedthrough from the IPS to the VAS. I would have thought that the 220pF then couldn't do much more than reduce noise at the voltage amplifier's current source.
I understand the Naim power amp design originated from an RCA handbook in 1971:
http://img46.imageshack.us/img46/479/rcaamppo8.jpg
http://img46.imageshack.us/img46/479/rcaamppo8.jpg
Sorry I don't have a copy of the RCA note, I just saw a copy of some pages on the web.
My theory on the RC is that the R is there to exaggerate the Early effect in the VAS. The C limits this effect to audio frequencies. I assume this provides a bit of linearisation at high +ive output swings.
My theory on the RC is that the R is there to exaggerate the Early effect in the VAS. The C limits this effect to audio frequencies. I assume this provides a bit of linearisation at high +ive output swings.
Thanks for the link df_genius, It reminds me of the many precursors to the RCA design I recall but I'm almost certain it's not the design that used to grace Avondale Audios's previous web pages. Les Wolstonholme used another RCA circuit on his old webpages to point out the similarities to his NCC200 but I'd fail to recognize this one as similar.
Here we have no current source for the LTP, a bootstrapped VAS and the VI limiters are simple current limiters. C3, a bit of OI feedback and compensation, is an interesting old feature. Perhaps there is an archive for the site in the Wayback Machine?
DaveS, I have my doubts but it wouldn't hurt to breadboard the front end and check various values of C out with scope etc. on the relative safety of a workbench. I'd suggest simulation as more useful but it isn't always cool to suggest that in reference to Naim products
Here we have no current source for the LTP, a bootstrapped VAS and the VI limiters are simple current limiters. C3, a bit of OI feedback and compensation, is an interesting old feature. Perhaps there is an archive for the site in the Wayback Machine?
DaveS, I have my doubts but it wouldn't hurt to breadboard the front end and check various values of C out with scope etc. on the relative safety of a workbench. I'd suggest simulation as more useful but it isn't always cool to suggest that in reference to Naim products
You were right df_genius, the image in your link is what Les posted on his old site and thanks to Wayback Machine, I'm presently eating my hat. It makes me wonder what he saw as similarities though. By the 1970s, there were more people than RCA distributing app. notes for their semis etc. and to recognizably more modern designs too.