onsemi & the old motorola only give an NF value and that is 1.5dB @ 10k Rg
That does not look good, but as earlier may be no good to us for audio.
Fairchild do post a variety of graphs for noise but they are in a different format and I don't know how to read them. eg. they all seem to show that very low Ie gives lower noise than higher Ie. That's opposite to what is usually seen, so clearly I am not understanding their noise data. They also state an identical NF of 1.5dB @ 10k
For the bc550 I can only find NF values and these vary from a low of 0.6dB to a max of 10dB from Mot, ON & Fair.
That does not look good, but as earlier may be no good to us for audio.
Fairchild do post a variety of graphs for noise but they are in a different format and I don't know how to read them. eg. they all seem to show that very low Ie gives lower noise than higher Ie. That's opposite to what is usually seen, so clearly I am not understanding their noise data. They also state an identical NF of 1.5dB @ 10k
For the bc550 I can only find NF values and these vary from a low of 0.6dB to a max of 10dB from Mot, ON & Fair.
Last edited:
Your maths hold up I feel. One should always aim to have near 0R for the preamp out to get really ideal measurements. If using an op amp use a class A buffer like a fancy version of the class B stages often in text books. You will be able to drive headphones also so worth a slightly complex solution. I bet anything better than a real Naim pre amp of the NAC32(.5 ) type. An NE5534 will be almost that in itself. A NE5532 in paralell could be excellent.
On reading it is said the base current is critical. Too much or too little won't work. The graphs at least say that. The noise of the input resistor and transistor are not quite the same. The resistor for want of a better word is nicer. The 2SC1845 is as far as I know an audio device pure and simple. It was a vast market back in the day. Alas I doubt we will ever see advanced devices for audio in the future unless a spin off device ( BF 720/721 ) . The market died when CD came along. Whilst I can not prove the link I suspect computers alone did not kill audio. I am 60 and still a young man in audio!!! I am seeing some who are 30 and they buy vinyl !
Combining transistors might do more than reducing noise. It might turn a blue spectrum to white ( noise cancellation due to random phase ). That will be percieved. 2 x BC550C might be enough, 4 per LTP. It should sound more like a resistor.
Blue is also the sound of bad DAC's. No one ever says that. I suspect that's what it is. That is plenty of unmusical extra spectrum in the hiss levels.
I must admit to being in the same boat. I have never really looked into noise performance of transistors for power amps, I always assumed they were all quiet enough for the relatively high signal levels. However, Nigel's comments have got me thinking.
Take a look at the 2 graphs on P4 here:
https://www.rcscomponents.kiev.ua/datasheets/2sc2240.pdf
and read the 2nd answer here:
amplifier - Understanding Noise Figures for BJT transistor - Electrical Engineering Stack Exchange
We need to be in or near the "blank" area in the middle of the graphs. Less current means more noise voltage and more current means more flicker (current) noise.
It is interesting that the increase in LTP current from NAP140 to NAP200 would appear to make the flicker noise worse (although nobody seems to have the noise curves for BC239C to confirm this).
The 2SC2240 looks much better than BC239C wrt noise: 2dB versus 4 dB max.
http://www.onsemi.com/pub_link/Collateral/MJE13007-D.PDF
This is a transistor that might be used in a MC preamp. It's noise figue is 0.37nV per root Hz. It is said it is at the limit of the possible. As the capacitance is 80pF it might not be ideal here. In the 1970 and early 1980's preamps like Lentek used similar devices. Gain is about like MPSA42, so not out of the question to use.
Only added for interest. BC337-40 is only 0.65 nV!
Gain isn't even that now I read the graphs. Were it not for someone discovering this I doubt it would ever have been thought of. BF459 was another. It's gain > 25 and capacitance nicely low.
This is a transistor that might be used in a MC preamp. It's noise figue is 0.37nV per root Hz. It is said it is at the limit of the possible. As the capacitance is 80pF it might not be ideal here. In the 1970 and early 1980's preamps like Lentek used similar devices. Gain is about like MPSA42, so not out of the question to use.
Only added for interest. BC337-40 is only 0.65 nV!
Gain isn't even that now I read the graphs. Were it not for someone discovering this I doubt it would ever have been thought of. BF459 was another. It's gain > 25 and capacitance nicely low.
Last edited:
Hi,
I've just build a pair of Nap140 clones, the blue LJM boards. And I run into a few problems on the way. This information could already be here, but the search function didn't help much, so I'm sorry if everyone already know this.
On the boards that I got about 6 month ago, (yes, I'm slow) there's a layout problem. The boards are made to fit the fake transistors, that come with the kit. If you like me want to use genuine semis, you run into trouble. The 5551 transistors that come with the kit have reversed C-B pins. A real one should be EBC, but these have ECB. And since the boards are made for the wrong layout, using a genuine 5551 or MPSA06 will blow up your input transistors. So on both locations where 5551's are indicated, you have to twist the legs to make the right connection.
Funnily enough the 5401 that comes with the kit, does have the right layout. So in that location you don't have to do anything.
I also had a problem with one of the boards going full rail voltage on the output, when turned on. It seems like this construction is very sensitive to which output transistor has the highest gain. Because the problem was solved by simply switching the two transistors around.
The only thing left is to get the right idle current set up. The information in this thread and the information on the board, does not agree with each other. So can someone tell me what the correct idle current is ?
On the board it says, "450mV" between Base and Emitter on the output device. But that will only give a little shy of 7mA of idle current. Which can't be right. But maybe they are trying to prevent their bad transistors from blowing up.
Hope this can help someone else. I did get a few extra grey hairs from it.
I've just build a pair of Nap140 clones, the blue LJM boards. And I run into a few problems on the way. This information could already be here, but the search function didn't help much, so I'm sorry if everyone already know this.
On the boards that I got about 6 month ago, (yes, I'm slow) there's a layout problem. The boards are made to fit the fake transistors, that come with the kit. If you like me want to use genuine semis, you run into trouble. The 5551 transistors that come with the kit have reversed C-B pins. A real one should be EBC, but these have ECB. And since the boards are made for the wrong layout, using a genuine 5551 or MPSA06 will blow up your input transistors. So on both locations where 5551's are indicated, you have to twist the legs to make the right connection.
Funnily enough the 5401 that comes with the kit, does have the right layout. So in that location you don't have to do anything.
I also had a problem with one of the boards going full rail voltage on the output, when turned on. It seems like this construction is very sensitive to which output transistor has the highest gain. Because the problem was solved by simply switching the two transistors around.
The only thing left is to get the right idle current set up. The information in this thread and the information on the board, does not agree with each other. So can someone tell me what the correct idle current is ?
On the board it says, "450mV" between Base and Emitter on the output device. But that will only give a little shy of 7mA of idle current. Which can't be right. But maybe they are trying to prevent their bad transistors from blowing up.
Hope this can help someone else. I did get a few extra grey hairs from it.
The BC550Cs are now in the amp and all seems to be stable. I've not had chance to have a proper listen yet, again it's getting into night time here so that will have to wait. The trimmers have also been replaced with (genuine) Bourns parts but they seem so incredibly sensitive to even the slightest adjustment.
As an aside, both before and after substitution of the LTP transistors I noticed that on both channels the ZTX753 and ZTX653 devices (TR4 and TR6 respectively) get almost unbearably hot to touch. I'm not sure if this is normal or not or whether there is something not quite right. I can't see them having a particularly good life span when running so hot. Would it be worth adding a small heatsink to each of these devices?
As an aside, both before and after substitution of the LTP transistors I noticed that on both channels the ZTX753 and ZTX653 devices (TR4 and TR6 respectively) get almost unbearably hot to touch. I'm not sure if this is normal or not or whether there is something not quite right. I can't see them having a particularly good life span when running so hot. Would it be worth adding a small heatsink to each of these devices?
Just had chance for a listen now that the amp has warmed up and I would say that the NAP200-clone sound is quite significantly improved as a result of the substitution of the LTPs. I've used just two BC550Cs for each LTP in each channel, not multiples as suggested by Nigel. Maybe another time.
Obviously it's subjective, but on a few reference tracks that I frequently use the sound seems more dynamic and it's easier to pick out individual instruments in the soundstage than I feel it was before the substitution. The amp sounds more 'Naim-like', a quality that I thought was previously lacking in comparison with my Nait 3.
More listening is required, but a positive first impression remains.
2N5401 have the same EBC pinout as 2N5551 and many BC type transistors, depending on the manufacturer. The Philips input pair supplied in several LJM kits I have bought, are reversed to US style ECB pinout as the component overlay shows, having a CBE pinout and hence the confusion for folk who do not realise that there can be different pinouts for the same type devices.
The different type parts for TR4,6 used in the voltage amplifier, i.e. 2SB647/2SC667, are obviously Japanese ECB pinout rather than original types ZTX653/ZTX753 which have EBC pinout.
The Vbe multiplier TR5, could be any type but LJM NAP140C PCB has another BC546B, so the pinout must be CBE if you follow the overlay but you could just as easily fit 2N5551 by rotating it 180°
This problem has been with all kits that have been available over the last 10 years, even those with the original PCB layout. The fact is that the original parts are too expensive to source for cheap kits and different suppliers use different parts sources which may require pinout adjustments. It's not difficult to fit these substitutes or the original types if you wish but do check the datasheets first.
The 450 mV instruction is strange, it doesn't appear on any of my PCBs. I believe it refers to the voltage between TR11 emitter and TR12 collector. This is the same as reading across both R29+R30 0R22 2W resistors. That voltage should be initially set to no more than 7+7 mV or 14mV and this means approx. 32 mA bias current. For budget DMMs, reading across both resistors will be more accurate.
The ZTXs do indeed run hot. 10mA X 40V = 0.4W
Rth(j-a) is 175 therefore the junction is ~70C above ambient.
Anything over ~50C feels too hot to touch, but the plastic insulates the chip from your fingers a bit.
However the power dissipation if set by the rail voltages and the current source resistor (68R IIRC). It should not change with changes to the LTP.
The ZTXs running at 100C will outlive you . It's thermal cycling that kills them and they will only see that on power up/down or if you drive rail to rail at <~20Hz for long periods of time.
cbe pin out is the standard for BC type and for ElineHi,
I've just build a pair of Nap140 clones, the blue LJM boards. And I run into a few problems on the way. This information could already be here, but the search function didn't help much, so I'm sorry if everyone already know this.
On the boards that I got about 6 month ago, (yes, I'm slow) there's a layout problem. The boards are made to fit the fake transistors, that come with the kit. If you like me want to use genuine semis, you run into trouble. The 5551 transistors that come with the kit have reversed C-B pins. A real one should be EBC, but these have ECB. And since the boards are made for the wrong layout, using a genuine 5551 or MPSA06 will blow up your input transistors. So on both locations where 5551's are indicated, you have to twist the legs to make the right connection.
Funnily enough the 5401 that comes with the kit, does have the right layout. So in that location you don't have to do anything.
I also had a problem with one of the boards going full rail voltage on the output, when turned on. It seems like this construction is very sensitive to which output transistor has the highest gain. Because the problem was solved by simply switching the two transistors around.
The only thing left is to get the right idle current set up. The information in this thread and the information on the board, does not agree with each other. So can someone tell me what the correct idle current is ?
On the board it says, "450mV" between Base and Emitter on the output device. But that will only give a little shy of 7mA of idle current. Which can't be right. But maybe they are trying to prevent their bad transistors from blowing up.
Hope this can help someone else. I did get a few extra grey hairs from it.
ebc pin out is standard for mpsa and for 2N
Collector in the middle is standard for all 2sa/c including the power devices.
But manufacturers do make specials. Check each transistor before assembly.
It appears the PCB layout was designed to allow 2sa/c types to be used, It appears that the Fake supplier knows so little about the product he is supplying that he supplied a different pin out without checking that the fake matched the marking!. (he probbaly does not know how to check). What does that say about the supplier's "Customer Support"?
Why do we buy from these obvious fakers?
Why do we provide that missing support to those that keep these fakers in business?
Why do we allow fakers to be Members here?
Last edited:
I took a picture to show the rather unclear instruction. The wire is hiding some of the text, but all it says is: "V BE< 450mV NOTE". And then there's the indication of two "test points" pointing at the output transistors Base and Emitter legs.
I won't, not anymore at least. It just confused me for a while, because of the other trouble I had. But now it's just another useless information.
I only mention it, to prevent other people getting their project derailed because of it.
But thanks for you help.
If you do, just use 4 BC550C. Listen to the hiss with ear to the speaker I would guess. It should sound softer more than being less. The hiss from sources will be higher. It is quality of hiss alone. When valves this would be too obvious to even state. Transistor people don't allow themselves to think like this. Silly really as it could be important to them. Naim apart from it's tantallum capacitors is a more valve like sound with PA style punch.
I bought a very cheap meter to have in the car from CPC. It was less than £5. It has usable transistor gain and diode tests. If the transistor is allowed to settle in a constant temperture for 1 minute both Vbe and hFE tests good enough for the Naim 140 is possible. All we need is the same numbers. Some transistors take longer to settle. No idea why. Holding a transistor is not helpful. I have used a plastic bag and water at 40 C before now. A cheap poundshop thermometer is fine as it is just to maintain the temperature. Some boiling water to top up is what it requires and watching the alcohol is all it needs (+/- 1C is possible). A shoe box is OK and the heat of the room. Most important that you stop air currents.
Below is how two tranistors are converted into diodes for the NE5534 op amp. This should protect the inputs to maintain low noise ( read PDF ). Even testing transistors for Vbe can cause problems. The fact Philips fitted them says it matters to be careful. The NE5534 is more or less a small power amp. It's complexity makes it easier to build out of the silicon wafer. Notice it is a double LTP inc double VAS. That allows for a very simple LTP 1 balance via 2 x 12K. It has a very simple LTP current source and is not unlike the Niam except it has balance and second harmonic rejection as a result.
Analog Devices : Analog Dialogue : Input Protection
Andrew. Some help. Would LED's be OK to protect our inputs? I have a hunch you might have tried this? I take them to be 1.5V minimum if red. That should just about allow 1Vrms . 1V rms should be 100 watts 8R which means the power amp clips first. I suspect after some listening tests it could be OK? I have never seen power amps with input protection as far as I can remember. A silly idea?
It is implied that even bipolar transistor hate static charges. A friend says it shows up later and is cumulative.
It is implied that even bipolar transistor hate static charges. A friend says it shows up later and is cumulative.