Bipolars for low noise input - Alternatives to 2SC2240

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I wonder about bipolar transistors which are best suitable for moving coil input stages.

I have here 2SC2240/2SA970, 2SC2546/2SA1084 (and many other bipolars) lying around. The 2SC2240 sounds better in my amps so far, but the Rbb is about 40R which is too high for low signal inputs. The 2SC2546 has much lower Rbb, but I do not like it soundwise.

Somewhere in diyaudio.com I read about transistors which sound better than 2SC2240 in input stages, but I cannot find this place anymore.

I want to drive the input stage at 5 to 10mAmps idle current each transistor, to keep voltage noise low down for my 0.1mV output moving coil cartridges (Ortofon 7500, Accuphase AC2, Fidelity Research). I have in mind a simple (non-paralleled, non mirrored) balanced LPT, or alternatively a balanced zero-impedance input, all without feedback.

After many years of playing around, I myself dislike the sound of JFETs and opamps in general, so these devices should be excluded rather.

Thanks in advance for discussion.

Hartmut
 
The 2SA1084 and 2SC2546 series that you dislike are pretty much what is commonly available.

The much lamented Rohm 2SB737 is still available at a price.......£2 each from Gareth Connor of the Signal Transfer Company near Cleckhuddersfax, can be googled or via the Self site.

Transistors such as the 2SC2546 etc can be safely paralleled WITHOUT the use of emitter balancing resistors to reduce the effective rbb' and hence an improvement in noise figure. Especially if they have been purchased as a single lot. Any difference in rbb' will cause an imbalance in current, but that is not a problem :) And they are a lot less than £2 ;)
 
Toshiba recommended the 2SA1316 and 2SC3329 for MC head amps. Toshiba no longer manufactures transistors so they may not be available. You could also try some of the DIODES (ZETEX) medium power transistors as shown here. http://www.diodes.com/_files/design_note_pdfs/zetex/dn11.pdf

Ray

I am afraid you are correct, a quick google will list them with suppliers, but they are always out of stock or made from unobtanium. I certainly have not been able to find any in the UK.

It all comes down to rbb', and I can only go back to 1084 and 2547 and for which the data sheets show comprehensive plots of the noise figure curves. These imply a low rbb' but still a dB or two or three above 737, 1316 and 3329.

The mention of Zetex is good, they used to do a range of switching transistors for things like print heads, small package TO92 for low duty cycle, but high peak current rating.
 
Member
Joined 2009
Paid Member
You should probably be careful with the Zetex Design Note. While looking for transistors with low rbb I came across the following, where there is a quiz around the Design Note. See post #9 and 10.

Ancient ultra low noise narrow band preamplifier - Geotech Forums

Not exactly a positive evaluation!

I am leaning towards the 2N4401 (or rather the SMD variants of this like the MMBT4401), but this is for a 200-250ohm source impedance and since it is for mass production, the cost is also an important parameter. I know the 2N4403 should be better, but I have some other constraints in my design making me prefer the NPN device.
By the way, does anyone here know the rbb of the 2N4401? Reference?
 
Interesting comment on the Zetex design example. Quad used the Zetex transistors in the MC preamp of the Quad 34 and Quad 44. I forget what transistor Quad was using prior to redesigning the MC preamp to use the Zetex transistors. I don't think Quad would have made the switch if the Zetex transistors weren't better. Of course, that was in the 1980s and transistors have improved since then.
 
Balanced zero impedance MC stage

This stage is 12 components, just 5 components more than the well known Hiraga LePrepre.

Of course, because of both floating inputs being on the same voltage level, one could dispense the trimmer. But I feel safer, when both inputs are really on 0.000V level, in case of the cartridge leads getting contact to ground.

This is my first sketch, and I will try to build it later in the evening. I consider replacing the output cap with a simple servo later.

Hartmut IMG_7903_MCstage1.jpg
 
Last edited:
I just searched on these little fellas and can find nothing other than the Fairchild data sheet with very brief info and certainly nothing relating to rbb or noise.. Do you have a link to info?.

The noise specification is quoted on the first page of the data sheet. Only it's expressed as a voltage so would have to be evaluated with regards to a specific design, as the Toshiba figures are.
 
The noise specification is quoted on the first page of the data sheet. Only it's expressed as a voltage so would have to be evaluated with regards to a specific design, as the Toshiba figures are.

Thank you found it! I wasn't looking for mV's!


That will take a little while to calculate back, A bandwidth of 1KHz incorporating all the low frequency flicker. A source impedance of 100K and referenced to an 80dB gain. That won't be much of a guide to a MC phono amp with an Rg of 3 ohm.

On one sheet the operating parameters were missing!

NV Noise Voltage V​
CE = -5.0V, IC = -1.0mA,
R
G =100KW, GV = 80dB,
f = 10Hz to 1.0KHz

25 mV typ. 40 mV max

 
Could somebody explain the meaning of this noise specification? I don't understand it. Probably Rg is a source resistance of 100k, not 100kw. Is gv the voltage gain so you have to divide the specified noise level by the gain to reference it to the transistor input?
http://www.fairchildsemi.com/ds/KS/KSA992.pdf


My own interpretation is the same as your own, the datasheets are actually incompetently written. On the other one, the 1845, the test conditions are not mentioned.
It is very bad practice to use the data of a specific test in a data sheet. Data sheet info should be referred back to inputs and generalised into a useable format. But... Hey transistor manufacturers never ever bothered to do that.:smash:
 
Hello Hartmut,

could you possibly elaborate on what exactly you dislike about the sound of these transistors? I have a bunch of 2SC2545E/2SA1085E that I have not tried yet but that I wanted to use in the future, so I am really curious.

From what I have seen in circuits so far they were usually biased at ca. 3-4mA, which is supposedly optimal from a noise perspective, so maybe if you parallel to fit your current needs? Also, did you play with load and voltage across the transistors? NP has an interesting article "The Sweet Spot" where he shows how much optimising the operating point effects distortion. Re. BJTs: "The reduction in distortion is less dramatic, but it's still about 7 or 8 dB better ...". So maybe a different operating point would help?

Btw. Sanyo 2SC3504/2SA1433 and 2SC3601/2SA1407 should have a similarly low Rbb. Syn08 posted some measurements some time ago (should still be in his WEB site).
 
Last edited:
Hello Hartmut,

could you possibly elaborate on what exactly you dislike about the sound of these transistors? I have a bunch of 2SC2545E/2SA1085E that I have not tried yet but that I wanted to use in the future, so I am really curious.

...

Martin,

I had tried them in discrete opamps, Mark Levinson style and the like, biased at 50uA to 125uA, but this was about 20 years ago. So my experiences don't apply to this case, where they are to be used at 3..6mA in a non-feedback configurtion.

BTW, the Swiss producer DNM uses exactly these transistors in their preamps.

Hartmut
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.