Pass's Patents.

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

Following the instructions under Advanced Search I entered the argument IN/pass-nelson-s and got seven hits. The text of one of them is not available on-line though, because it is prior to patent number 4,000,000 where their database starts. Is seven the correct number of your patents? And what does the S. stand for? (I am just curious...)

;)
 
7 patents. Cool ones.

1. The latest issued is Aleph CCS Patent (1998). The Aleph series has discontinued now, why the patent appear the latest?
2. The "Optocoupler Bias" Patent seems tobe development of "Smart Bias" patent. Does the "Optocoupler Bias" also have properties of maintaining "non-turnoff" in final stages or not?
3. Mr.Pass, are you the one behind the "Current Dumper" power amps? The "Cascode" patent.
 
The one and only
Joined 2001
Paid Member
1. That's just the way it came out.

2. The optocoupler bias is designed to maintain a more
constant bias over temperature. It is independent of
Class A/B/ or AB

3. The Current Dumping (an unfortunate name, I think) is the
product of Peter Walker of Quad, and a brilliant concept I might
add.
 
Mr. Pass,

I like the "Smart Bias" patent so much, and have been trying to get deeper into it (for almost a month now), but not understand it wholly.
In a glance, the "Smart Bias" is so similiar to Hawksford EC, but after looking at those deeper and deeper, they are quite different.

You said your patent is for maintaining all output transistor on for whole cycle, and not intended for Error Correction (but it really does EC)

Hawksford said, his circuit is intended for EC, not for maintaining output stages on all cycles, and non-turnoff is just a "good side effect" (if that non-turnoff properties really exist in Hawksford EC)

Your "Smart Bias" patent surely can maintain all transistor not turning off all the cycle, but I cannot see (until now) that Hawksford EC can maintain non-turnoff.

In Hawksford EC, the Bias drop is rigidly fixed by VBE drop. In your patent, the bias drop can be made sliding, because the VBE drop has other path to flow current.

I also have been trying to reconstruct what happened when you are digging this "non-turnoff" issue. From dates of patents (the smart bias and the optocoupler bias), the dates of the Threshold and Statis amps schematics available to see, but I cannot see in what form the quest is ended.
(Threshold and Statis series amps are really wonderfull design to my eyes. I wonder why they don't have too much discussion about?)

The latest Threshold (you said when you are about to left) just uses one VBE multiplier drop/the ordinary one. Is this mean you are forgetting the quest of "non-turnoff" in class AB power amp?

Mr. Pass,

I conducted a simple experiment, that is in class AB amp, the feedback point is taken from output stage and from VAS.
The sonics really different.

I can think about 2 possible issue, but I dont know which is the right one. Could you tell me?

1. The difference is caused by the CrossOver distortion (that always occur in class AB power amp when using ordinary VBE multiplier) is fed back to the differential. If we take the feedback from VAS (excluding the class AB output stage), the feedback signal does not have CrossOver distortion in it.

2. The difference is mainly caused by LoudSpeaker's back EMF, not by CrossOver distortion. Taking feedback point from VAS does not include this speaker's back EMF into the feedback signal.

3. You are still making class AB amp now, that is the X series amp (X1000, X650, X350, X250, etc). Are you using oridinary bias (one VBE drop), or "Smart Bias" (like your patent) or "Optocoupler Bias" (like your patent) in the X series amps from PassLabs?
 
I believe Nelson is using "regular bias" in the X amplifiers.
There are problems with sliding bias schemes. One is that the devices never really reach thermal equilibrium. Another is that changing the bias on a device also changes the operating characteristics. Just yesterday, I was torturing a circuit and increased the bias on the JFET differential in the front end. As a result, the transconductance increased. Now, had this happened in real time, the differential would have increased gain slightly as the bias went up. This counts as distortion. Yes, the same thing happens in output stages.

Grey
 
We know that NP prefers classA instead of classAB when he can. But from the traces, he really tries to make good class AB. Minimal from 197.. to 199.. (thats 20 years) with Threshold/Statis line amp. (Now with classAB X amp line from PassLabs)

I couldn't imagine all that "Brain Energy Sucking" issue (the non-turnoff / classAB Output stage CrossOver distortion issue) is just left unused. It sucks my brain tries to understand it, I imagine how difficult it is to draw it for the first time.

I believe Nelson is using "regular bias" in the X amplifiers.
I also tought that is a possibility of the final conclusion of the "ClassAB output stage CrossOver distortion" issue. Using ordinary 1transistor VBE multiplier, and just exclude the output stage for classAB power amp in the feedback loop.

Altough I did not put so much brain energy in it (NP surely does), but what a waste not to use it (smart bias scheme) anymore.

There are problems with sliding bias schemes
I notice there is L1=300uH in Threshold 800a schematic. This is not usual, putting L in the bias generator cct.

I also notice how hard it is to optimise sliding bias. In Statis S150 and Adcom 555, NP uses another not usual bias scheme. Double VBE bias, but not forming VBE multiplier, plus ordinary VR drop for bias.

In the latest Threshold, NP just use ordinary 1 transistor VBE multiplier, put 4k7 from bias generator to output, and take feedback from VBE multiplier. Is this how the "ClassAB CrossOver distortion" elimination ended?
 
The one and only
Joined 2001
Paid Member
lumanauw said:
I couldn't imagine all that "Brain Energy Sucking" issue (the non-turnoff / classAB Output stage CrossOver distortion issue) is just left unused.

I dropped dynamic bias primarily because the imitators in the
marketplace gave the technique a bad name. It works well if
you are trying to improve the efficiency by a reasonable margin,
for example reducing a 100 watt amp from 200 to 100 watts
idle dissipation. Much of what appeared on the market idled at
less than 10 watts, and it is my opinion that an amp that runs
cold is not a Class A amp, even if you maintain some tiny forward
conduction.

In any case, I always keep a hat full of new designs, and I
decided to do one of those instead.

A few years later dynamic biasing resurfaced as an error
correction feedback mechanism in the bias circuit in Hawksford's
circuit. As to whether it improves the subjective performance,
it is currently my opinion that raw idle current best improves
the subjective performance.
 
Ah, I see now. Thanks for the explenation, Mr. Pass.

I know you make a "Diplomatic" answer when you are asked which Threshold / Statis amp you like the best. You said you wouldn't differ your son over your daughter, they all your sons.

Statis S300 series technology is bought by Nakamichi, so some people do likes S300 series.

But from the sonic point of view (Threshold/Statis/Adcom classAB amp lines uses many kind of biasing technique), which one you like the most ?(assuming the front stage and final stage are the same).
 
Hi, Mr. Pass,

Threshold SA3.9e is the one which uses thr_os89 output stage and thr_fe89 front stage? This uses optobias (very complicated, I will try to understand it)

Mr. Pass,

In this cct, I have difficulties in understanding it. Between base of Q5-Q10 the difference is 2xVBE. But between Q6-Q11, the difference is 4XVBE, because there is D7 and D8 there.
The base of Q5 is connected to base of Q6 and base of Q10 is connected to base of Q11.
There is VBE difference here between left transistors and right transistors. In the patent, you said it is necessary to put these diodes to make sure that the current will dominantly flow in the left transistors.

I can't understand that until now. What makes you decide to have left and right transistors to not-have exact the same VBE drop (Putting D7-D8 there)?

Does a transistor or diode will conduct if the VBE is less than 0.6V?
 

Attachments

  • dynamicbias.jpg
    dynamicbias.jpg
    32.6 KB · Views: 428
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.