Could you explain the physical diferences between a "switching die" and an "audio die" in bipolar transistors?
I don't get it.
Also, in my opinion bipolar transisros work better in switching applications when:
- They keep good current gain up to high current levels, so that not a too high base current is required to achieve a low saturation Vce
- Capacitances are low, capacitances are evil in switching applications (losses, parasitic turn on, etc...)
- Transition frequency is high, which tends to imply quick saturation and de-saturation, the opposite of storing huge charge in the junctions
In fact, I think the Japanese audio bipolar transistors could be great for switching too...
I don't get it.
Also, in my opinion bipolar transisros work better in switching applications when:
- They keep good current gain up to high current levels, so that not a too high base current is required to achieve a low saturation Vce
- Capacitances are low, capacitances are evil in switching applications (losses, parasitic turn on, etc...)
- Transition frequency is high, which tends to imply quick saturation and de-saturation, the opposite of storing huge charge in the junctions
In fact, I think the Japanese audio bipolar transistors could be great for switching too...
Last edited:
Video transistors often have very low gain and very non-linear Ic/Vce curves. Choosing with care also means not to generalize.
jd
Last edited:
Hugh,
was that a reply to sajti's post where he provided an attachment of the test setup for the ON-Semi transistors, then I just wonder how did you come to the conclusion the Iq is 240 mA?
I downloaded the datasheet but couldn't see anything speced.
I would have assumed, according to the theoretical proper biasing for "lowest" crossover distortion, it would be ~52 mA.
Wahab,
could you perhaps simulate with Iq = 52 mA, would be interesting to know the THD figures just out of curiosity, thanks.
Cheers Michael
Last edited:
the figures are higher with low quiescent current, furthermore at these
levels, close to clipping , the transfer function is heavily (literaly) distorded,
and there s no feedback to balance it...so the only mean to reduce
distorsion is to increase quiescent current, as the degeneration of
the 0.5R emitter resistance is anecdotic...
here the spectrum for 52 ma quiescent current..
levels, close to clipping , the transfer function is heavily (literaly) distorded,
and there s no feedback to balance it...so the only mean to reduce
distorsion is to increase quiescent current, as the degeneration of
the 0.5R emitter resistance is anecdotic...
here the spectrum for 52 ma quiescent current..
yes, steve, but iddle currents are higher in these sims,
about 250 mA...somhow hard class AB...
Steve,
Yes, but my post concerned post #71, or perhaps do you know what is the Iq?
Wahab said:
Wahab,
thanks for making the simulation, very kind of you!
Sorry for not understanding, by saying "clipping" did you mean with lower Iq it's closer to "turned off" output device or as in general understood, clipping into the rails voltage?
Cheers Michael
since the rail voltage is +-50V, at 100W RMS,the output is close to clipping..
the emitter 0.5R degeneration bring a 2.5 V /rail loss..
if we add the Vbe at these current, we end with no more than a
margin of 6.5 V before saturation of the conductive device is effective....
the other device has reached switch off state long before the conductive
device hit the 40V peak..
Hi Eva,
that´s a weird designation, I know. When optimizing transistors for some application (switching, video, current amplification, HF, and so on), comparatively, certain qualities will inevitably be degraded. You can`t have everything at the same time. A change in one parameter will effect a large number of other parameters. A highly-accurate manufacturing process is equally important and no one can beat the Japanese technology. You see, it`s pointless studying the data sheet if you actually have something else on hand.
To be continued...
that´s a weird designation, I know. When optimizing transistors for some application (switching, video, current amplification, HF, and so on), comparatively, certain qualities will inevitably be degraded. You can`t have everything at the same time. A change in one parameter will effect a large number of other parameters. A highly-accurate manufacturing process is equally important and no one can beat the Japanese technology. You see, it`s pointless studying the data sheet if you actually have something else on hand.
To be continued...
I have heard a wide range of comercial audio hifi power amplifiers. Such with low speed low Ft devices (MJE340/350 for VAS/predriver and 2N3055/MJ3055 and MJ15003/15004 in the power output stage) and such with high speed and very high ft (go to post #1 fore some examples)
But until this day I cannot confirm, that amplifier stages equipped with last kind of semiconductors definitely sounds better.
I recall follow:
To investigate sound effects I have built in some years ago different semiconductor types by two faulty exemplares of an "A1" from Musical Fidelity (only power amp unit), a special circuit developed from Mr. Tim de PARAVICINI - go to
Musical Fidelity A1 - Technical
The output stage runs with an idle current of 350mA
First exemplare I fitted with old Motorola versions of MJ2955/2N3055 and second with 2SA1095/2SC2565 from Toshiba (a little trouble through other outline). Because the supply voltage is low, most power devices would be match. As long as the speaker current was below arround 700 mA, there are no audible different. Otherwise (i. e. above this current value) by the version with Toshiba's 2SA1095/2SC2565 I note very slightly advantages (a little bit less harshness THD-Value identical, but less low-order and little more high order distortion).
With an idle current reduction to 100 mA, the same effect was observed, even at small volume. By compare of both exemplares in that kind, that I use 350mA idle current by the MJ/2N and 100mA by the 2SA/2SC, this one with 350mA sounds clearly better (in this case I have compare classA to class AB).
Unfortunately I have not make documentation about this, because I haven't cad supported distortion measurement equipment.
But it is to assume, that high ft by power devices only exhibit slightly audible advantages by typical A/B output stages and not by class-A.
But in generall, all in all, should not be overlooked that the quality of circuit development and the PCB layout developement significantly greater influence on the final sonic and measure result than the fact, whether fast Japanese (special made for audio) or slow American or European power devices are in use !!!
The proof is simple. A very well-developed power amplifier circuit (especially, if there is Class-A) with the vintage-2N3055 or older MJ types - still in production, is still hard to beat, even with the best today's available power devices for audio.
If they have very reliable P-Spice parameters, it must be possible through simulatiom, even with transistors that are not designed for typical audio applications to achieve superior sonic transmission results and low IM/THD
But until this day I cannot confirm, that amplifier stages equipped with last kind of semiconductors definitely sounds better.
I recall follow:
To investigate sound effects I have built in some years ago different semiconductor types by two faulty exemplares of an "A1" from Musical Fidelity (only power amp unit), a special circuit developed from Mr. Tim de PARAVICINI - go to
Musical Fidelity A1 - Technical
The output stage runs with an idle current of 350mA
First exemplare I fitted with old Motorola versions of MJ2955/2N3055 and second with 2SA1095/2SC2565 from Toshiba (a little trouble through other outline). Because the supply voltage is low, most power devices would be match. As long as the speaker current was below arround 700 mA, there are no audible different. Otherwise (i. e. above this current value) by the version with Toshiba's 2SA1095/2SC2565 I note very slightly advantages (a little bit less harshness THD-Value identical, but less low-order and little more high order distortion).
With an idle current reduction to 100 mA, the same effect was observed, even at small volume. By compare of both exemplares in that kind, that I use 350mA idle current by the MJ/2N and 100mA by the 2SA/2SC, this one with 350mA sounds clearly better (in this case I have compare classA to class AB).
Unfortunately I have not make documentation about this, because I haven't cad supported distortion measurement equipment.
But it is to assume, that high ft by power devices only exhibit slightly audible advantages by typical A/B output stages and not by class-A.
But in generall, all in all, should not be overlooked that the quality of circuit development and the PCB layout developement significantly greater influence on the final sonic and measure result than the fact, whether fast Japanese (special made for audio) or slow American or European power devices are in use !!!
The proof is simple. A very well-developed power amplifier circuit (especially, if there is Class-A) with the vintage-2N3055 or older MJ types - still in production, is still hard to beat, even with the best today's available power devices for audio.
If they have very reliable P-Spice parameters, it must be possible through simulatiom, even with transistors that are not designed for typical audio applications to achieve superior sonic transmission results and low IM/THD
Last edited:
Let me see, I had a second look to try understand how the information should be interpreted given in the data sheet and test circuit in attachment of post #71
It says 28,3 V RMS and 100 Watt @ output into 8 Ohm..
Let's see if we do the Ohms law calculation the 28,3^2 = 800...
P = U^2/R >>> 800/8 = 100 Watt
Upk (for sine) = U RMS * sqr2 >>> 28,3 * 1,414 = 40 Volt peak
alternatively we can calculate the power also as follow:
28,3 U RMS / 8 Ohm = ~3,54 A RMS
28,3 * 3,54 = 100 Watt
As I can see we have 10 Volts of headroom to the 50 Volt rails supply before clipping occures.
Wahab, correct me if I'm wrong here!?
If we do the simulation again with 80 Volt pk-2-pk and for both 240 mA and 52 mA what how would the distortion look then?
(Re for output transistor as in pots #71, eg. 0,5 Ohm)
Cheers Michael
It says 28,3 V RMS and 100 Watt @ output into 8 Ohm..
Let's see if we do the Ohms law calculation the 28,3^2 = 800...
P = U^2/R >>> 800/8 = 100 Watt
Upk (for sine) = U RMS * sqr2 >>> 28,3 * 1,414 = 40 Volt peak
alternatively we can calculate the power also as follow:
28,3 U RMS / 8 Ohm = ~3,54 A RMS
28,3 * 3,54 = 100 Watt
As I can see we have 10 Volts of headroom to the 50 Volt rails supply before clipping occures.
Wahab, correct me if I'm wrong here!?
If we do the simulation again with 80 Volt pk-2-pk and for both 240 mA and 52 mA what how would the distortion look then?
(Re for output transistor as in pots #71, eg. 0,5 Ohm)
Cheers Michael
Last edited:
This is a specious argument. They DO in fact make transistors for audio. Just ask Charles Hansen. You are apparently unaware of the thermal bias stability issues in BJT output stages that these transistors greatly mitigate.
Cheers,
Bob
You know not of what you speak. Have you worked at Moto? BTW, not all transistors designed primarily for the switching market are unsuitable for audio. This same ignorance applies to those who think you cannot make a good audio amplifier out of HEXFET vertical MOSFETs. Ask Nelson Pass. You are dealing in too many bland generalities.
Cheers,
Bob
I would have thought that the MF A1 runs normally at considerably greater quiescent current than 350mA....
I used to work for MF and I well recall carrying out tests one day to see if it really was class A.
From memory, it gave about 14Watts RMS per channel in class A.
The surprise was the A370 which gave only 18W RMS in class A! A bit less than the often touted 185W!!
I used to work for MF and I well recall carrying out tests one day to see if it really was class A.
From memory, it gave about 14Watts RMS per channel in class A.
The surprise was the A370 which gave only 18W RMS in class A! A bit less than the often touted 185W!!
Jan,
This matter would require a comprehensive answer, you should have followed the discussions about it in the other threads.
As usual, I strongly disagree. The essential device properties for video and voltage amplification coincide. Please take a look at the data sheet of the earlier mentioned 2SC3600/2SA1406 as a good example, some features will not be immediately apparent.
This matter would require a comprehensive answer, you should have followed the discussions about it in the other threads.
