It's also possible to use the diodes in the bias spread and then add a little feed forward around the output transistors.. this will also keep the transistors from ever shutting off...the value of the feed-forward reisitors to the emitter resistors will determine the minimum current through the output stage
No. An emitter could sense what the lightbulb was doing. A collector cannot.
What you have there are bulbs fed back to the Early effect of the collectors.
Thats not much of a feedback, certainly not comparable to the emitter.
If Marantz had cadmium sulfide cell feedback to the bias spreader, those
bulbs might do something. Otherwise, I see nothing but a big fuse with a
thermally variable voltage drop. If it protects the output at all, it does so
by starving the collector of voltage, and causing it to clip at less than full
output swing...
What you have there are bulbs fed back to the Early effect of the collectors.
Thats not much of a feedback, certainly not comparable to the emitter.
If Marantz had cadmium sulfide cell feedback to the bias spreader, those
bulbs might do something. Otherwise, I see nothing but a big fuse with a
thermally variable voltage drop. If it protects the output at all, it does so
by starving the collector of voltage, and causing it to clip at less than full
output swing...
Last edited:
Not immediately. Q108 and Q111 would still be there. And the global loop
would shut the opposing channel down before offset became dangerous.
But that protection wouldn't last long...
I'd be more worried the driver Q108 or Q111 would eventually melt down
without assistance of it's compound current dumper and blown bulb.
Then you are gonna have a DC offset for sure.
Last edited:
You simply need a 15R bleeder resistor across the output diode drops, nothing fancy.
Remember, Marantz' spreader and drivers are holding this voltage near 1.4V constant.
The outputs will then always conduct 100mA minimum current. Problem solved.
Refer to R10 in my simplified circuit post#19. 150mA minimum current at all times...
This current completely bypasses the load, so nothing wacky the load might do
can ever turn off one of the transistors. Worst, it might turn off a diode, and
even the most primitive diode recovers much quicker than a transistor.
With Schottkys as output devices, it becomes a complete non-issue.
Last edited:
Oops.. yes you are right! The bulbs are before the junction of the output collectors and the feedback to the compound transistor so yes they act only on the collectors.
Interestingly the Son of Ampzilla has no diodes across emitter resistors. The 1st thing I did was removing these Schottky diodes in the Ampzilla. They were blown and unobtainable. It sounded better without the diodes, more tight bass.
According to an old white paper entitled The Ultimate Amplifier by Bongiorno those diodes were implemented to lower power losses across the emitter resistors. The ultimate amplifier example was then the SAE IIICM.
According to an old white paper entitled The Ultimate Amplifier by Bongiorno those diodes were implemented to lower power losses across the emitter resistors. The ultimate amplifier example was then the SAE IIICM.
You would have to take a guess about the transistor part numbers - Marantz did not publish it, only their own part numbers.
My guess would be 2N3055 for the output. It was once state of the art (hard to believe). The Model 15 come out circa 1965 and the 2N3055 in 1962, AFAIK.
Last edited:
The voltage drop across severely heated emitters of a workhorse device
probably should not be in the equation that ultimately determines current.
Presents a challenge to the voltage spreader to hold the same thermal
curve, but drop out well before the emitter when things get really hot.
Monitoring the drop only across external diodes and/or resistors at ambient
temperature makes far better sense...
probably should not be in the equation that ultimately determines current.
Presents a challenge to the voltage spreader to hold the same thermal
curve, but drop out well before the emitter when things get really hot.
Monitoring the drop only across external diodes and/or resistors at ambient
temperature makes far better sense...
Done some simulations....
Very interesting!
For the same Iq the THD is lower by about 15 % (roughly, visual from graph)
But more importantly the harmonic structure is totally different with 2nd now dominant above 3rd and any higher odd order harmonics virtually gone!
This is worthy of further investigation
For comparison I also simmed 0R in place of the diodes (used 0.22R in the resistors versus diodes sim). This gave slightly lower THD than either of the above but still had the odd order predominance of the emitter resistors.
Very interesting!
For the same Iq the THD is lower by about 15 % (roughly, visual from graph)
But more importantly the harmonic structure is totally different with 2nd now dominant above 3rd and any higher odd order harmonics virtually gone!
This is worthy of further investigation
For comparison I also simmed 0R in place of the diodes (used 0.22R in the resistors versus diodes sim). This gave slightly lower THD than either of the above but still had the odd order predominance of the emitter resistors.
I'd be suspicious (or at least cautious) about those results: a 15% change in THD is almost insignificant in reality, and more so in sim.Done some simulations....
Very interesting!
For the same Iq the THD is lower by about 15 % (roughly, visual from graph)
But more importantly the harmonic structure is totally different with 2nd now dominant above 3rd and any higher odd order harmonics virtually gone!
This is worthy of further investigation
For comparison I also simmed 0R in place of the diodes (used 0.22R in the resistors versus diodes sim). This gave slightly lower THD than either of the above but still had the odd order predominance of the emitter resistors.
And if the harmonic structure is changed, there must be another hidden effect somewhere, because if you start from a symmetrical circuit and make changes, also symmetrical, the even harmonics shoudn't change in absolute level.
And a 15% global change is insufficient to alter the colour of the global mix.
The even harmonics arn't all that different! Big drop in odd ones though....
FYI all the above are open loop tests on just a darlington emitter follower output stage using some Bob Cordell spice models.
Since you have an almost insignificant drop in the global THD, this means the even ones were already dominant.The even harmonics arn't all that different! Big drop in odd ones though....
In a PP more or less symmetrical amplifier, this is probably a symptom of something, like a device starved in bias current at a positive or negative excursion.
This doesn't mean the improvement you mention is worthless, but it is mostly masked by a dominant factor.
Could you share your sim?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.