Bob Cordell Interview: Error Correction

KSTR said:
Now I found that patent application again, http://www.pat2pdf.org/patents/pat20070030033.pdf

Compare Fig.1 with Fig.29 in the AD843 spec sheet 😉

"Fast Peak Detector" (not fullwave rectifier, sorry), that also was what I had googled for when I found that striking similarity by accident, as both pop up in the search results. And I suspect any researcher/inventor would google quite a bit and then that old and known circuit would have popped up, as established prior art. Especially when he's talking suitable opamps, where ADI sure has some candidates.

Probably the application won't make it anyway... and it's no big deal at all. At first I was thinking about making an objection, but then again... why should I care?

- Klaus
This patent, what was the invention really? This peak circuit can be found since the 80's at least in every datasheet of opamps, especially NS. This was a good example of bothering to whole patent process of something with zero commercial value. Isn't it the idea of a patent, of making money of it?

I saw an another patent about a headphone amp. Equally stupid.
 
Re: Re: Re: Re: ODNF in plain old English...

snoopy said:
For this to be error correction Rb needs to be in series with the base emitter voltage error rather than across it as implemented in the patent. As it is the gm stage is just bootstrapping Rb which effectively makes the impedance since by A1 much higher and hence the gain is higher and feedback factor is greater thus reducing distortion. But still no error correction and also the possibility of instability.

Hi Snoopy,

I don't see a serious discrepancy between the block diagram and the real implementation, except that one of the inputs of the error amplifier (gm stage) is tied to the emitters of the pre-drivers (via R109 and R109), instead of the bases of the latter.

Do I miss something?

Cheers,
Edmond.
 
Re: Re: Re: Re: Re: ODNF in plain old English...

Edmond Stuart said:


Hi Snoopy,

I don't see a serious discrepancy between the block diagram and the real implementation, except that one of the inputs of the error amplifier (gm stage) is tied to the emitters of the pre-drivers (via R109 and R109), instead of the bases of the latter.

Do I miss something?

Cheers,
Edmond.

Yes this is to provide some buffering.

Looking at the following implementation on Sheet 6 using a single transistor instead of the differential pair this looks very similar to the Zero Distortion Rule circuit as used in some of the Yamaha amplifiers. However Yamaha have done it differently again in order not to breach the patent.

All circuits appear to use some sort of positive feedback in order to artificially increase the forward path gain.
 
Re: patent 4785257

Edmond Stuart said:


Hi David,

In the mean time I had a closer look at this patent and discovered that it only works if the two 2kOhm resistors between input and output of the OPS are the only parts that load the VAS ouput. This means that you need a VAS with a very high output impedance. OK, that can be done with a Hawksford cascode. BUT..... you also need some kind of Miller compensation (absent in the patent!), which lowers the VAS output impedance considerable. Under these conditions, the EC circuit is totally ineffective and without Miller compensation, this amp is highly unstable.

I leave the conclusions to you.

Cheers,
Edmond.


Hi Edmond,

These are good points, but I think that the concern about loading the VAS with a Miller compensation capacitor would be eliminated if the capacitor was tapped off from the pre-driver emitter follower that comes after the VAS.

Cheers,
Bob
 
Edmond Stuart said:


Hi Jan,

I guess it's unlikely that an European patent grant authority will accept such invention, but in the US, even a patent on a rocket, fueled by human farts, has been granted.

Cheers,
Edmond.


Hi Edmond,

That's so funny, but, unfortunately sometimes true.

I am not a patent attorney, but have been involved closely in many patents. The truth is, you can patent almost anything, BUT the key issue is the breadth of the claims. If one makes the claims narrow enough (as often ends up being the case), one will often be able to get a patent in the US. There are, however, sinfully broad patents sometimes issued that sneak by the examiners.

Many of us, with the benefit of hindsight, would consider the Darlington connection and the Widlar current mirror to be obvious to one with ordinary skill in the art.

Cheers,
Bob
 
Re: Re: patent 4785257

Bob Cordell said:
Hi Edmond,

These are good points, but I think that the concern about loading the VAS with a Miller compensation capacitor would be eliminated if the capacitor was tapped off from the pre-driver emitter follower that comes after the VAS.

Cheers,
Bob

Hi Bob,

Hmm... not a bad idea at all. The day after tomorrow, I will try that (i.e. simulate it). If it really works, you might apply for a patent. 😉

Cheers, Edmond.
 
Edmond Stuart said:


I'm afraid that doesn't work either. At the Philips company, for example, you can only sell your ideas if you hold a patent.


Hey, I've got a 70's Electronics Australia magazine with review and schematics of a Philips bookshelf speaker with motional feedback. Your handy work?

Cheers,
Glen
 
Re: Re: patent 4785257

Bob Cordell said:
Hi Edmond,

These are good points, but I think that the concern about loading the VAS with a Miller compensation capacitor would be eliminated if the capacitor was tapped off from the pre-driver emitter follower that comes after the VAS.

Cheers,
Bob
Edmond Stuart said:
.........
Hmm... not a bad idea at all. The day after tomorrow, I will try that (i.e. simulate it). If it really works, you might apply for a patent.
.........

Hi Bob,

I'm back home again and I've tried the alternative compensation and guess...... no improvement at all. :sad:
Sorry.

Cheers,
Edmond.

@Jan, I'm still waiting for more details.
 
Re: Re: Re: Re: patent 4785257

Bob Cordell said:
Then I guess your concern about loading of the VAS with the Miller capacitor was unfounded and not the source of distortion.

Cheers,
Bob

Bob, I'm sorry to say, but I had never expected such response like this, in particular from you, as it's totally obvious that loading the VAS will spoil the the EC circuit (because it disrupts the so called balance equation).
Now it's your turn to think a bit harder and explain (to the rest of us) why a Miller cap, even when one leg is tied to the the pre-driver, doesn't work either.

Cheers,
Edmond.
 
Re: Re: Re: Re: Re: patent 4785257

Edmond Stuart said:


Bob, I'm sorry to say, but I had never expected such response like this, in particular from you, as it's totally obvious that loading the VAS will spoil the the EC circuit (because it disrupts the so called balance equation).
Now it's your turn to think a bit harder and explain (to the rest of us) why a Miller cap, even when one leg is tied to the the pre-driver, doesn't work either.

Cheers,
Edmond.


Huuugghhh????
 

Attachments

  • tesla.jpg
    tesla.jpg
    97.2 KB · Views: 592