Scott makes an important point about current mirrors. Yes, they are useful, yes, they are important, but the first current mirror that I ever saw was in an IC chip, the UA709, in 1966. It might have also been designed into the ua702, but I can't find a schematic to verify it.
At the same time I was charged by Friden to evaluate the ua709's in our possession, we also made a lot of stuff with germanium transistors, and we didn't use any stinkin' current mirrors in those designs, too expensive. However, we still had to evaluate the circuits with the computer for worst case analysis, and sometimes even transient analysis.
It took a roomful of IBM mainframe computer and 1/2 hr running time for a given set of results. Still, it beat a mechanical calculator or a slide rule. As is often said: 'In the world of slide rules, an IBM computer is KING!'
Bob Widler, a true design genius, introduced the current mirror concept to the general engineering public, but it is possible that it was known and used earlier in hybrid circuit design by Dick Burwen and others.
At the same time I was charged by Friden to evaluate the ua709's in our possession, we also made a lot of stuff with germanium transistors, and we didn't use any stinkin' current mirrors in those designs, too expensive. However, we still had to evaluate the circuits with the computer for worst case analysis, and sometimes even transient analysis.
It took a roomful of IBM mainframe computer and 1/2 hr running time for a given set of results. Still, it beat a mechanical calculator or a slide rule. As is often said: 'In the world of slide rules, an IBM computer is KING!'
Bob Widler, a true design genius, introduced the current mirror concept to the general engineering public, but it is possible that it was known and used earlier in hybrid circuit design by Dick Burwen and others.
I regards to ALPHA, you might read up on point contact devices, which were still used in 1953 and beyond. They were faster than any diffused device at the time. I have data sheets to prove it. ALPHA in point contact devices could be significantly greater than one. Even diffused devices could be troublesome, if care was not exercised in manufacture. Just read 'Shea' for more details.
The point of this conversation is to bring up different ways of looking at the same problem and offering different solutions. One problem with what Scott says, is that the voltage approach leads to a certain kind of design approach, but precludes others. It is interesting that in 'Electronics Design' 'EDN' and other publications, many qualified 'ideas for design' come from people in different countries, some relatively exotic. While I realize that the 'prize' for winning a design competition might be more important to a person from a 3'rd world country, still the approaches are often unique, and show a certain 'open-mindness' lost by people who all went to MIT or some such close knit clique of engineers.
It reminds me of my first contact with Bob Pease, an MIT graduate, and his 'dressing me down' for 'improper' definition of what I was trying to describe. He was actually rude! I traced this back to his learning 'electronic definitions' in his early years at MIT, and intolerant of any other words of description of the same thing.
Joshua sometimes gets the same treatment here, on this website.
It reminds me of my first contact with Bob Pease, an MIT graduate, and his 'dressing me down' for 'improper' definition of what I was trying to describe. He was actually rude! I traced this back to his learning 'electronic definitions' in his early years at MIT, and intolerant of any other words of description of the same thing.
Joshua sometimes gets the same treatment here, on this website.
Thank you John. I did not know that Russia was a 3'rd world country. 😎
However, due to central planning of everything through the entire huge country some dumb ideas sometimes dominated: for example, they decided that following IBM and Digital hardware paths we can avoid software design costs because everything will be compatible. As the result, some very unique and promising computer programs lost financing, and digital design was pre-programmed to copy already obsoleted hardware.
Edit: there were some exceptions; some enthusiasts kept working on supercomputers; they were taken by Sun Microsystems after Soviet Union collapsed.
However, due to central planning of everything through the entire huge country some dumb ideas sometimes dominated: for example, they decided that following IBM and Digital hardware paths we can avoid software design costs because everything will be compatible. As the result, some very unique and promising computer programs lost financing, and digital design was pre-programmed to copy already obsoleted hardware.
Edit: there were some exceptions; some enthusiasts kept working on supercomputers; they were taken by Sun Microsystems after Soviet Union collapsed.
Lumba Ogir said:
In 1969 I went to the library to find something special for a lab project. I found Barrie Gilbert’s most famous 1968 paper (one of the top ten referenced IEEE papers of all time) and yes Wavebourn I tried unsuccessfully to apply IC technology to discretes checking out an unusual quantity of 2N3904 and 2N3906’s in the process, but for me the rest is history. So may I quote from the master, this BTW is from the “PIM” article that John keeps wanting everyone to read…
“The Level-0 model for the BJT is simply a voltage-controlled current-source (VCCS) having an exact exponential relationship between its collector current, IC, and its base-emitter voltage, VBE; this is the heart of the BJT:
IC = IS exp (VBE/VT) (1)
...
The Level-0 model also conveniently omits such pesky set-backs as the finite base current and Early voltage of a BJT, its ohmic resistances and parasitic capacitances, base transit time and other effects. Accordingly, we set BF = BR = VAF = VAR = 1E6, and most other parameters to zero. Crazy? Not really: This is just what first-order textbook analyses do, without drawing attention to the fact. It is nonetheless surprising just how much of the reality of an IC's behavior emerges from the application of this simple translinear model to circuit analysis.”
There's a bit of a difference. Bob actually has an understanding of electronics and physics.
Just out of curiosity, what definitions/terms set him off?
Well, maybe, IC's, Scott. However, I first used Barrie's multiplier for the first time time in 1970 and understood its operation, well enough, with my archaic and obsolete knowledge. 
scott wurcer said:
Hi Scott,
I always wondered about using power devices for this type of application. What part numbers have you found that work well for this sort of thing?
Thanks,
current driven?
It's not only a matter of 'very wrong thinking', but also it may easily lead to a flawed design.
Cheers,
Edmond.
scott wurcer said:
It's not only a matter of 'very wrong thinking', but also it may easily lead to a flawed design.
Cheers,
Edmond.
Re: current driven?
Can you give some examples of that flaws?
Edmond Stuart said:
It's not only a matter of 'very wrong thinking', but also it may easily lead to a flawed design.
Can you give some examples of that flaws?
It seems to me that 'success' tells a lot more than anything else. At the end of my life, I must admit that I would not have been a great IC designer. I need 'good' parts, complementary fets, etc, to do my 'magic'. It should be realized that there are literally 1000's of different IC op amps, made by 100's to 1000's of engineers. Where are most of them today? In a land fill?
I can design something and it has value, even after 35 years, just like a classic car. So much for 'behind the times' engineering, and the sneers of some designers, here.
I can design something and it has value, even after 35 years, just like a classic car. So much for 'behind the times' engineering, and the sneers of some designers, here.
John;
let's wait for Edmond Stuart's answer.
After that I'll ask him where to get voltages to control that voltage controlled devices if all of them are current sources. 😎
let's wait for Edmond Stuart's answer.
After that I'll ask him where to get voltages to control that voltage controlled devices if all of them are current sources. 😎
AD797 model
So my home brew model using MAT02 as input devices (see: http://www.diyaudio.com/forums/showthread.php?postid=1814785#post1814785 ) was totally wrong. This (erroneous) choice was only based on the noise figures.
Looking at the sizes of the MAT04 transistors seems to be a better choice. How about that?
Below the (estimated) sizes of the AD797 input trannies and in the next post the sizes of two MAT04 trannies.
Cheers,
Edmond.
scott wurcer said:
So my home brew model using MAT02 as input devices (see: http://www.diyaudio.com/forums/showthread.php?postid=1814785#post1814785 ) was totally wrong. This (erroneous) choice was only based on the noise figures.
Looking at the sizes of the MAT04 transistors seems to be a better choice. How about that?
Below the (estimated) sizes of the AD797 input trannies and in the next post the sizes of two MAT04 trannies.
Cheers,
Edmond.
Attachments
Re: Re: current driven?
No, as it might be rather embarrassing. So I suffice to say that relying on the assumption that beta is well defined, constant, flat or whatsoever, is not a good idea.
Cheers,
Edmond.
Wavebourn said:
No, as it might be rather embarrassing. So I suffice to say that relying on the assumption that beta is well defined, constant, flat or whatsoever, is not a good idea.
Cheers,
Edmond.
- Status
- Not open for further replies.