Have a look at the Sony TA-F555ES (mark 1, the first ever ES lineup) output stage. It uses two MT200 cased LAPTs fro, Sanken and no emitter resistors at all. The design is very thermally stable, no doubt because it uses double thermal feedback, from the main heatsink with the output BJTs (bias servo Vbe multiplier BJT) and from the drivers which are on separate small heatsinks (a diode in the C-B part of the Vbe multiplier).
This approach however cannot be used for multiple output pairs in parallel as some means for current balancing and thermal stabilization has to be provided, also it cannot be used with just any output BJT, however most modern ones will work like that (tripple diffused, LAPT), because various structures are used to combat secondarybreakdown which all result in some built-in emitter resistance. If there is an Ube/Ib plot, you can actually see the effect of this resistance linearising the Ube/Ib curve, and this can in fact be used to estimate the resistance value. Now, here comes the not so obvious part. because the resistance are part of the diffusion process, it is not the same in the PNP and NPN part. A good look at the beta vs Ic plot for the complementary parts and the estimated value of the resistors might give rise to the idea that emitter resistors should not be the same value on the NPN and PNP part if one wants maximum symmetry of the output stage (which, granted, is not always the goal).
This approach however cannot be used for multiple output pairs in parallel as some means for current balancing and thermal stabilization has to be provided, also it cannot be used with just any output BJT, however most modern ones will work like that (tripple diffused, LAPT), because various structures are used to combat secondarybreakdown which all result in some built-in emitter resistance. If there is an Ube/Ib plot, you can actually see the effect of this resistance linearising the Ube/Ib curve, and this can in fact be used to estimate the resistance value. Now, here comes the not so obvious part. because the resistance are part of the diffusion process, it is not the same in the PNP and NPN part. A good look at the beta vs Ic plot for the complementary parts and the estimated value of the resistors might give rise to the idea that emitter resistors should not be the same value on the NPN and PNP part if one wants maximum symmetry of the output stage (which, granted, is not always the goal).
yes, those re's at output stage are for dc bias stability, prevents thermal runaways...
there is a good explanation for emitter resistors in the 1972 RCA transistor, thyristor & diode manual,
i wonder if there is a scan of that book anywhere in the internet...
i have seen AKSA vas stage without those....
there is a good explanation for emitter resistors in the 1972 RCA transistor, thyristor & diode manual,
i wonder if there is a scan of that book anywhere in the internet...
i have seen AKSA vas stage without those....
For reference please: Top one is (a), Bottom left is (b), and (c) is bottom right.
Fig. 7. Shaw's improved quasi-complementary design from 1969, which used a diode to improve symmetry, is seen at (a). At (b), Baxandall's variation further improves symmetry, and (c) shows author's use of small capacitor to simulate effect of base/emitter capacitance.
Attachments
Last edited:
So I remembered wrongly.
I was attributing the resistor to Shaw. He suggested the diode.
At least I got his name right so that you could find the example to keep me right.
I think this more points to builders getting the voltages and heatsink wrong.
quasi-complementary mosfet output stage
John,
I certainly value your opinion. But could you give some detailed reasoning on why quasi-complementary was a 'stopgap'? I remember that in many of your posts you have stated postively for the Naim topology, which is a quasi-complementary design, and you praised the stability.
Even today, the vast majority power transistors are NPN/N-channel Mosfets. If somebody wants build a mosfet output stage, what circuit topology and components will you suggest to him to use?
Jazz
John,
I certainly value your opinion. But could you give some detailed reasoning on why quasi-complementary was a 'stopgap'? I remember that in many of your posts you have stated postively for the Naim topology, which is a quasi-complementary design, and you praised the stability.
Even today, the vast majority power transistors are NPN/N-channel Mosfets. If somebody wants build a mosfet output stage, what circuit topology and components will you suggest to him to use?
Jazz
designers build with anything that was available at the time, i think that it is as simple as that...those designers did not know then that it was a stop gap measure....they were making products based on available semis...
it was a stop gap now because today we have better and superior devices that were not available to early designers...
who will not use superior devices when it was available?
it was a stop gap now because today we have better and superior devices that were not available to early designers...
who will not use superior devices when it was available?
It would be appreciated that you lay out a list of the superior output devices in your opinion. I am curious whether the list will include any of the mosfets, especially those vetical ones.
For me, I like 2SC5200/A1943, 2SC2922/A1216, IRF740, IRFP240, 2SK313, etc.
By the way, I think class-D is one of the direction for future hi-fi amps. But let us put aside the devices for class-D for this discussion.
Last edited:
i have very limited experience with mosfets so i am unable to comment, low beta droop in the 2sc5200 snf high Ft's make them superior as with other Sanken devices favored by JC....
may i ask what is your take on the other triples, like the second trannie in the triple connected as Sziklai instead of a regular 3 stage darlington?
Naim amplifiers were (prior to 1990s) hardly stable with a typical 20-30 min. warm up before it was near the factory setting. You could say they were praiseworthy in that if you waited that long, they would then remain remarkably stable at typical domestic volume level. In hot weather, it was a different story because of the weak thermal coupling between bias controller and output stage.
All quasi complementary designs suffer from a compromise of bias current where one transistor requires about 13 mA and the other around 100 mA for optimum performance. Both however, have to share a poor compromise (around 30 mA) where neither is at its best.
I haven't considered designs where the asymmetry of quasi and mixed device output stages is intentional - simply to get a higher proportion of even harmonic distortion from it for sound enrichment. I think that becomes a subjective matter of audiophile taste and a design skill (or rather art) beyond this discussion.
Last edited:
I am totally agree that Maim has only a weak thermal coupling between the vbe multiplexier and the output transistors. But it seems that weak coupling rarely leads to any damage, strange? Not only that, the Naim circuit, if my spice simulation is correct, has the merit that the THD remains virtually the same when the quiescent current vary from a few tens of milliamp to a few amps. My simulation tells that this circuit has a THD 1KHz <0.02%.
You are correct that it is a subjective matter to have some special THD from quasi-complementary, in fact a characteristic of quasi-complementary output. The question is how many people can hear distortion below 0.1% and how many speaker systems have distortions below 0.1%
I will appreciate that you let me know how you measure THD of your amp, and how you judge an amp soundwise.
Your opinion will help to understand why there are still a lot who enjoy the sound of single end and tube amps, which, in my opinion, for sure have higher THD and lower slew rate.
Can somebody enlighten me why an amp with two 2SC5200 cannot beat an amp with a complementary pair of 2SC5200/2SA1943?
Certainly an amp has to amplify the positive and negative signal portion symetrically, but from the datasheet of all the NPN/PNP pair, people rarely see there is a perferct beta/ft match
Certainly an amp has to amplify the positive and negative signal portion symetrically, but from the datasheet of all the NPN/PNP pair, people rarely see there is a perferct beta/ft match
This is true for 'classic' quasi designs where one side is a darlington, the other a CFP (incl. variations). There are ways to avoid this using different totem-pole output approaches, however they are better suited to MOSFET output designs. Even with that said, it's practically impossible to use these topologies with BJTs in a follower configuration, and extremely difficult verging on impossible doing the same with high gm MOSFETs.
Still, even with complementary BJTs, there is the issue of the degree of complementarity. This is, I assume, why John mentioned to keep the emitter resistors, and in the general case i would definitely agree. I've just mentioned one possible exception, where they are actually still present, though not necessarily as an external component.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.