From my experience (listening to ready-made amps), most discrete LTP-based quasi amps sound awful, compared to the normal Lin topology (It's probably the distortion that tends to be higher).
Circuits like Naim shows that quasi can actually be more symmetrical than most complementary output, which is not automatically symmetrical.
Circuits like Naim shows that quasi can actually be more symmetrical than most complementary output, which is not automatically symmetrical.
Member
Joined 2009
Paid Member
Dacs
You might be interested in the following:
Budget DAC Review: Schiit Modi 2 ($99) | Audio Science Review (ASR) Forum
Budget DAC Review: behringer UMC204HD | Audio Science Review (ASR) Forum
You might be interested in the following:
Budget DAC Review: Schiit Modi 2 ($99) | Audio Science Review (ASR) Forum
Budget DAC Review: behringer UMC204HD | Audio Science Review (ASR) Forum
Member
Joined 2009
Paid Member
Bare board on ebay.
A search of ebay for "naim 140 pcb" returns several relevant pcb boards. It looks like there is a choice among at least 4 boards:
- Light green pcb
- Dark green pcb
- Blue pcb
- Red pcb
Is there any quality differences between these? I understand that some have incorrect silkscreen printing, but that does not bother me too much. At this stage I am also flexible between the Sanken MT-200 and TO-3 output transistors.
A search of ebay for "naim 140 pcb" returns several relevant pcb boards. It looks like there is a choice among at least 4 boards:
- Light green pcb
- Dark green pcb
- Blue pcb
- Red pcb
Is there any quality differences between these? I understand that some have incorrect silkscreen printing, but that does not bother me too much. At this stage I am also flexible between the Sanken MT-200 and TO-3 output transistors.
You'll have to provide links to the various PCBs/kits you describe by colour, maxf. There are large differences between them because some are genuine attempts to clone the original as nearly as possible with available components. Others can be something that only has part of the design in common. In fact, some kits even have the expensive parts as recycles from old products. Others will be cut-down versions of the original design, using cheapest possible generic components that just don't work correctly or sound good as we could expect for a clone.
Your choice might depend on whether you just want an old style quasi-complementary amplifier kit that is cheap and easy to assemble or your aim is higher - say, to build a clone that with care in parts selection and full implementation, will go close to the original product's sound quality - i.e. early audiophile class.
If faced with using expensive, obsolete and often fake MT200 Sanken transistors, you might consider their MT100 (TO3P) versions which will be virtually the same as the original type, within the limits of their dissipation rating.
Your choice might depend on whether you just want an old style quasi-complementary amplifier kit that is cheap and easy to assemble or your aim is higher - say, to build a clone that with care in parts selection and full implementation, will go close to the original product's sound quality - i.e. early audiophile class.
If faced with using expensive, obsolete and often fake MT200 Sanken transistors, you might consider their MT100 (TO3P) versions which will be virtually the same as the original type, within the limits of their dissipation rating.
Last edited:
DIY DIY Ebay Kit
I'm looking for a bare pcb that corresponds the original Nap 250 schematic. I'll buy my own parts for reliable quality.
Here are links for the 4 I'm considering on ebay:
Light green pcb
Dark green pcb
Blue pcb
Red pcb
The red looks good but is sold as a single board and is 3x the price of the others. I'm willing to consider others, but those are the ones I've found so far.
I'm looking for a bare pcb that corresponds the original Nap 250 schematic. I'll buy my own parts for reliable quality.
Here are links for the 4 I'm considering on ebay:
Light green pcb
Dark green pcb
Blue pcb
Red pcb
The red looks good but is sold as a single board and is 3x the price of the others. I'm willing to consider others, but those are the ones I've found so far.
All of the boards you refer to appear to have the VI limiter and phase correction networks included. Those are the usual omissions on cheaper boards and builders often omit the associated components anyway, apparently because they read derogatory comments on the internet about their sound effects. That's a quandary though, because Naim rose to its status and retains it with those same circuits in place.
The physical board differences will be in overall dimensions, layout and the amount of copper remaining on the PCB, whether its double sided etc. This can make audible changes to electrical noise levels and even to the audio itself if not done correctly or at least as a facsimile of original boards. However, we are now touching on expert level differences that few of us would be able to nit-pick without reference to measurements.
As a comment, you might notice that part numbers aren't just different between all boards, some are repeated, wrong values for the model you may think you are cloning or just confused, so don't assume anything regarding the accuracy of printed overlays - check the copper traces if in doubt.
The physical board differences will be in overall dimensions, layout and the amount of copper remaining on the PCB, whether its double sided etc. This can make audible changes to electrical noise levels and even to the audio itself if not done correctly or at least as a facsimile of original boards. However, we are now touching on expert level differences that few of us would be able to nit-pick without reference to measurements.
As a comment, you might notice that part numbers aren't just different between all boards, some are repeated, wrong values for the model you may think you are cloning or just confused, so don't assume anything regarding the accuracy of printed overlays - check the copper traces if in doubt.
Last edited:
Member
Joined 2009
Paid Member
My 2c....As a commercial product the output protection has some merit but I bet th3 modern versions are far better and hence less intrusive. For the old NAP I see no need for DIY to include it. Good idea to include the phase networks.
With the different kinds of hf hash on mains supply these days I wonder if subtle differences in clone design are going to be worthwhile worrying over ?
With the different kinds of hf hash on mains supply these days I wonder if subtle differences in clone design are going to be worthwhile worrying over ?
Bare ebay PCB reviews
After looking at the options further, here is my uniformed review of the choices:
Light green. Poor quality pcb material. Layout is similar but the the same as the original (see the extra ground connection on the input). Traces are drawn with modern cad tools. Footprint for MT-200 output devices.
Dark green. Good quality pcb materials with, perhaps, ENIG flash. Layout is similar but the the same as the original (see the extra ground connection on the input). Modern cad-drawn traces. Some of the footprints have been modernized (e.g., trimmer pot). Footprint to TO-3 output devices.
Blue. Actually now black in the description. Small boards with a layout emphasis on size not sound.
Red. Good quality pcb material. Layout is the closest to the original but still not the same. Modern cad-drawn traces. Footprints for either MT-200 or TO-3 outputs. Sold for 3x the price of the other options.
I've gone ahead and ordered the dark green boards.
I intended to include the SOAR circuitry. I can't think of a good reason to omit it.
I understand that for a myriad of reasons my efforts will result in a poor copy and in no way a "clone" of the original but I'd like to try to stay true to the original. Just because you can't step in the same river twice doesn't mean you don't like to swim.
After looking at the options further, here is my uniformed review of the choices:
Light green. Poor quality pcb material. Layout is similar but the the same as the original (see the extra ground connection on the input). Traces are drawn with modern cad tools. Footprint for MT-200 output devices.
Dark green. Good quality pcb materials with, perhaps, ENIG flash. Layout is similar but the the same as the original (see the extra ground connection on the input). Modern cad-drawn traces. Some of the footprints have been modernized (e.g., trimmer pot). Footprint to TO-3 output devices.
Blue. Actually now black in the description. Small boards with a layout emphasis on size not sound.
Red. Good quality pcb material. Layout is the closest to the original but still not the same. Modern cad-drawn traces. Footprints for either MT-200 or TO-3 outputs. Sold for 3x the price of the other options.
I've gone ahead and ordered the dark green boards.
I intended to include the SOAR circuitry. I can't think of a good reason to omit it.
I understand that for a myriad of reasons my efforts will result in a poor copy and in no way a "clone" of the original but I'd like to try to stay true to the original. Just because you can't step in the same river twice doesn't mean you don't like to swim.
You'll probably find the differences between those particular NAP boards isn't a big deal but the semis you use probably will be. The original parts were often close matched pairs/complements (and some were intentionally mis-matched too). Several times in the thread, It's been found that you really should use the original semis and grades where possible. Exceptions are the output transistors which have gone through several big changes without issues over the decades in production and I guess the general purpose MPSA types used for the current source and limiters could be substituted with BC546B/56B without drama too.
The input LTP are BC239C or possibly 2N5089 because of their higher than usual hFE. I have found it to be ≥ 450 in some original Naim amps, which may not be possible to do from small quantities of recommended types like BC550C or whatever else you can buy in low-noise transistors. Types I can buy now seem to fit a spec of only about 300-450. Sure, they should work OK but Naim were fussy about the numbers and known to reject batches that didn't meet their spec.
The input LTP are BC239C or possibly 2N5089 because of their higher than usual hFE. I have found it to be ≥ 450 in some original Naim amps, which may not be possible to do from small quantities of recommended types like BC550C or whatever else you can buy in low-noise transistors. Types I can buy now seem to fit a spec of only about 300-450. Sure, they should work OK but Naim were fussy about the numbers and known to reject batches that didn't meet their spec.
The BC550 datasheet says that the C types have hfe 420 to 800.
https://www.mouser.com/ds/2/149/BC550-888526.pdf
https://www.mouser.com/ds/2/149/BC550-888526.pdf
Perhaps it's only the ex-Fairchild (or ...gulp...fake) product that I measured. I have a couple of lots of 20 ea. Motorola BC239C that I intend to use now for reference parts and they appear untouched but still, they arrived loose packed and could have been carefully picked clean of high hFE examples before I bought them. I measured these mostly at 450 - 560 with a few scattered examples above that, up to 680 I think.
I have a hundred each, cut taped Fairchild BC550C from Mouser and a Taiwanese supplier. Samples measured about the same and I rarely see anything measure much over 600 with a succession of meters I've used over the years. Most of it seems to fall below 560 and is about the same from both suppliers, different batches.
My concern is just that if we buy only the quantity needed for a project, it's reasonable that you won't get a good match and probably none at all if you required high range hFE parts. Actually, its a slight mismatch you need to trim by selection of the semis in order to null any DC offset at the output and that may require a few more parts than you reckon on. 'seems amateurish but that's essentially how Naim do or at least did it, back in the day.
For clones I was helping assemble and test about 10 years ago, I gave away a load of Philips BC549C from a remnant box I was given. Those semis proved to have a high range spread that even exceeded 1000 in a few examples but most worked nicely, with very low noise. Methinks I should have kept them instead.
I have a hundred each, cut taped Fairchild BC550C from Mouser and a Taiwanese supplier. Samples measured about the same and I rarely see anything measure much over 600 with a succession of meters I've used over the years. Most of it seems to fall below 560 and is about the same from both suppliers, different batches.
My concern is just that if we buy only the quantity needed for a project, it's reasonable that you won't get a good match and probably none at all if you required high range hFE parts. Actually, its a slight mismatch you need to trim by selection of the semis in order to null any DC offset at the output and that may require a few more parts than you reckon on. 'seems amateurish but that's essentially how Naim do or at least did it, back in the day.
For clones I was helping assemble and test about 10 years ago, I gave away a load of Philips BC549C from a remnant box I was given. Those semis proved to have a high range spread that even exceeded 1000 in a few examples but most worked nicely, with very low noise. Methinks I should have kept them instead.
The following list of transistors is taken (mostly) from the original Nap 250 Schematic:
Transistors
Original:
Q1,Q2: BC239C Obsolete.
Q3,Q7: MPSA06 Active, $0.30 Digikey
Q4: ZTX753* Active, $0.95 Digikey
Q5: ZTX108 Obsolete.
Q6: ZTX653* Active, $0.85 Mouser (not at Digikey)
Q7: MPSA56 Active, $0.30 Digikey
Q8: MJE243 Active, $0.56 Digikey
Q9: MJE253 Active, $0.59 Digikey
Q11,Q12: 2SC2922/NA001 Active, $7.17 Digikey
(* not original?)
It is fascinating that all but 2 are still active devices and all but 3 are available from my main supplier.
I’ll need to find substitutes for BC239C, ZTX108, ZTX653/753, and a TO-3 output transistor.
Transistors
Original:
Q1,Q2: BC239C Obsolete.
Q3,Q7: MPSA06 Active, $0.30 Digikey
Q4: ZTX753* Active, $0.95 Digikey
Q5: ZTX108 Obsolete.
Q6: ZTX653* Active, $0.85 Mouser (not at Digikey)
Q7: MPSA56 Active, $0.30 Digikey
Q8: MJE243 Active, $0.56 Digikey
Q9: MJE253 Active, $0.59 Digikey
Q11,Q12: 2SC2922/NA001 Active, $7.17 Digikey
(* not original?)
It is fascinating that all but 2 are still active devices and all but 3 are available from my main supplier.
I’ll need to find substitutes for BC239C, ZTX108, ZTX653/753, and a TO-3 output transistor.
Is your theory that the unbalanced LTP deliberately increases the distortion of the amplifier?
Member
Joined 2009
Paid Member
I believe its well established that this is the case - D. Self has published much on the topic. But even with matched devices, without any emitter degeneration of the LTP or current mirrors you are not going to get great balance because the circuit that includes the VAS is not symmetrical. Secondly, the LTP interacts with the output signal via the low PSRR of the input stage to produce further harmonics. There's no simple approach to viewing the design in terms of it's sound, there are a lot of interactions going on and they are frequency dependent too.