sound quality of different transistors?

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

The early effect is in fact stronger for higher h_FE transistors. In cascode configuration, however, this should not matter. Whether low h_FE is a problem depends mostly on the impedance of the previous stage .

Excuse my ignorance, but who is NTE?

------------------------------------------------------------------------------------

Capslock,

Thanks for your insight on that. I guess I need to go back to the books to make sure I understood that correctly.

As another posted, http://www.nteinc.com. They also make great resistors and capacitors up to 450 volts for lytics and much higher for poly, as well as pots and other devices.

Thanks!
Gabe
 
A few years ago, I worked for an NTE distributor. This is what was true then.

NTE is does not manufacture semiconductors. They buy in bulk and then put the items in the little bag, for your convience. They cost more, but you get of a choice of over 100 different semiconductors at the local parts store.

As far as substitutions, NTE guarantees that the replacement is "as good or better" than your original. Some parameters that might be "better" are: higher bandwidth, hfe, power disapation, maximum voltages, etc.

It is possible that the first little bag will have a Brand X part and the next little bag on the peg would have Brand Y part. They could have different spec's, such as the ones noted above. Both will have part number NTE 123.

I doubt that the above story is very likely. I think these guys buy in very big bulk, if the date codes match, then you probably have the same parts. It might be a mistake to assume that the NTE part you bought last year will be the identical NTE part available today.

Aud_Mot
 
Aud_Mot,

You may be right. But what I mentioned about them is what I read when they first came on the scene in the 1980's. New factories with laser cut semicondutcors.

I looked on the 'net under "semiconductor manufacturers" and they showed up on a couple of pages.

Of course, those sources could be incorrect also. However... as you may know they bought ECG, so if they weren't before... they are now manufacturers.

Gabe
 
All,

I just got off the phone with NTE in New Jersey.

NTE does NOT manufacture any semiconductors.

They buy in bulk, have them labled and packaged.

So if you thought the NTE cross of a transistor sounded better, it might be because it was a Toshiba part, or it might be a Motorola part, or some other manufacturer.

When I have seen "laser cut" (or laser trimmed, or laser some verb'ed) it is describing resistors.

Regards,

Aud_Mot
 
Hi Greg,

Along with Tomcat, I suggest you use the larger boards from RCS.
I was involved in building quite a number of the DIGI 125s about
10 years ago for a country TV station. We used the Mk. 5 board
at the time. I have been using a pair of modules for the past 10
years for my own amplifier. I was probably the first one to replace
Q3 ( a BC546) with a BF469 as it runs quite hot an needs a heatsink. If you can fabricate a heatsink for the BC546, I suggest you retain it, but it's not easy to heatsink a TO92 device. Other devices you can use are MJE340 and 2N3019. In the AKSA circuit which superficially at least, appears similar to the DIGI, a 2SC1819 is used in that position. I think Tomcat stated that the 2N3019 doesn't sound as good as the BF469 but it's Ft and HFe characteristics are probably the closest to the original BC546!

As others have stated on this forum, if you change to transistors with differing characteristics you may need to make further changes to th circuit. With the benefit of hindsight I probably should have changed the value of C2 when replacing Q3 but at the time I was rather rushed for time! I would suggest a smaller
value Polystyrene or Silver Mica cap to replace the 150pF ceramic if you intend using the BF469.

Whilst building the boards for the TV station, a colleague discovered that R6 (2K2) runs quite hot and should be replaced with at least a 1 watt type or preferably a 2 watt device as I use.
I passed these mods on to many others over the years and Tomcat has probably already mentioned them.

Another more recent mod is to increase the quiescant current by adding a third diode to D1 and D2. The original quiescant current
level was only 40mA and a third diode makes a noticeable improvement in sound quality. But remember that as these diodes are not mechanically coupled to the output devices heatsink, they can not compensate thermally for rising temperatures in the output stage.
 
Member
Joined 2002
Paid Member
900gripen thanks a lot for your recommendations. I think the best way to describe the Digi125 is its the generic 3 stage EF amplifier as shown in the text books with a few bits missing. I think the beauty of this amp is its simplicity.

I don't know anything about amp design and obviously don't want to criticise an experienced designer like Graham Dicker, but from what I've read the bias for a EF output stage should be about 2.6 volts not 1.4 volts which is about right of CF output stage, so I guess that's why it works better with three diodes. Have you tried 4 diodes or could a LED work?

Apparently the value C2 is critical to correct operation of the amp and from looking at other similar designs is usually in the 50pf to 100pf range.

Also, I was wondering why R1 and R5 are different values, I thought it was a requirement to make these values the same.

Greg
 
Hi Greg,

I haven't tried 4 diodes but I did try green LEDs, but couldn't find 2 that matched properly, so one channel had a higher bias current than the other! You really need precision LEDs like HP types if you want to try this. Alternately you can use 3 diodes an about a 200 ohm muti-turn trimpot and adjust the quiescant current to ideally
about 100mA. Remember to set the pot to minimum resistance before adjusting! As I think I mentioned, the DIGI has no mechanical coupling of these diodes so they do not "sense" the output devices getting warm. You can try coupling them in a similar
fashion to th AKSA. I am using 3 diodes and I have to agree that more bais current would make an improvement.:cool:
 
Member
Joined 2002
Paid Member
Hi 900gripen,

Thanks for the information, I didn't realise that LEDs would differ that much. I will have to look closely before using them.

I actually have an AKSA 55, and the circuit has changed a fair bit since the one you are referring too when it used diodes for biasing. It uses a FET circuit now with the transistor mounted on the heatsink.
 
I agree with you Nelson, neurosis is the right term for a lot of these things, like running things in, some of which is pie in the sky stuff. For example, cables, resistors, etc,... Speakers yes, but passive devices?, mostly I think is some wishfull thinking, allied with instead the ear getting used to the different sound due to whatever changes one had made with extended listening, which incidentally can also (to contradict myself) apply to speakers as well. That is, unless one has an unused reference pair to hand to compare them with. I also agree with your view on topoligy, only I don't think there has been as much work or investigation done into the audible difference between similar devices that basically do the same thing in the same circuit like others have done with tubes... Thats all I was trying to determine when I started this thread...
And now that I have blasphemed and set the cat amongst the pigeons, I better sign off, lest I be smitten by on high...

Let sanity reign supreme and peace be with you...
Cheers,
tomcat
 
To grege.

Hi Greg,

I note your comments on the AKSA. I haven't seen the current circuit, only an old concept drawing. A suggestion for the DIGI biassing is to replace 2 diodes with transistors connected as diodes
and mount them on the main heatsink. Mount a third diode (1N914/1N4148) on the board with a 200 Ohm multi-turn trimpot
and ajust the quiescent current for 100mA. The two 'diodes' mounted on the heatsink will regulate the biasing when the output devices get hot as Vbe for the two diodes we have made will drop and throttle back the quiescent current. The third diode mounted on the board with the trimpot will hold some biassing so that even when the Vbe drops to minimum on the devices mounted on the heatsink, there will still be some biasing applied to the circuit as not to increase distortion too much! I suggest BD139s for these transistors.

I have tried the Toshiba devices as Tomcat suggested. They are definitely worth the effort of replacing. They put the DIGI-125 ina different league in my opinion.
 
Yes there are huge differences in components - I've even seen top end amps use copper pipe rather than wire for the connections between output devices and speaker terminals. Speaker cables, interconnect, resistors, capacitors, running temperature, proximity to rf, etc. etc. etc. is all a part of voicing. I hope to design another amp soon and I'm very interested in difference in components.
If anyone finds a list that someone may have already made in the past of good vs bad I'd sure like to know.

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