Lumba Ogir said:
How's this? I though correct information prevents myths and misconceptions.
Could you mention a few?
Re: AD797 input trannies.
Busy weekend, Yes Edmond that will be closer. Unfortunately there will still be slight noise disadvantage due to the fact that the 797 broke the design rules to lower rbb below the minium allowed by the design rules. I can't find it now but there is a very nice explicit calculation of rbb' from basic process parameters somewhere. You consider the base pinch resistance under the emitter with uniform injection. It shows the advantage of stripe geometries (as narrow as possible) and why thin base transistors like super beta ones have high rbb'. Disk drive head amps used a trick of having emitters shaped like rows of small ellipses, the idea being the ends get thin (below the min allowed by the drawn geometery of a stripe). I used an extreme version of this.
Edmond Stuart said:
Busy weekend, Yes Edmond that will be closer. Unfortunately there will still be slight noise disadvantage due to the fact that the 797 broke the design rules to lower rbb below the minium allowed by the design rules. I can't find it now but there is a very nice explicit calculation of rbb' from basic process parameters somewhere. You consider the base pinch resistance under the emitter with uniform injection. It shows the advantage of stripe geometries (as narrow as possible) and why thin base transistors like super beta ones have high rbb'. Disk drive head amps used a trick of having emitters shaped like rows of small ellipses, the idea being the ends get thin (below the min allowed by the drawn geometery of a stripe). I used an extreme version of this.
Re: Re: AD797 input trannies.
Patent #4586072
Been there, done that, nice trick 🙂
scott wurcer said:
Patent #4586072
Been there, done that, nice trick 🙂
Re: Re: Re: minimum stage count
Let me guess... Are this four thingies related to some very familiar topology with some complementary thingy that suppose to generate symmetrical currents from differences of 2 voltages, and is this thingy designed to drive some another thingy with a unity gain, as high as possible input resistance, and as low as possible output resistance? 😀
Edmond Stuart said:
Let me guess... Are this four thingies related to some very familiar topology with some complementary thingy that suppose to generate symmetrical currents from differences of 2 voltages, and is this thingy designed to drive some another thingy with a unity gain, as high as possible input resistance, and as low as possible output resistance? 😀
Re: Re: AD797 input trannies.
Thanks a lot Scott.
I's quite interesting to learn a bout the special features of the input trannies. So you combined the advantages of two different geometries: Low rbb' and high hFE.
As for noise, I've just copied RB, RC, and RE from the MAT04 model. Now the noise figure should be close to 0.9nV/rtHz. (BTW, I also lowered CJC and CJE).
One more question about the output stage (I you don't mind). There are four large trannies at the left of the picture. Do they represent the two output trannies Q22 and Q30? If so, are they spread across the chip for better thermal balance?
Cheers,
Edmond.
scott wurcer said:
Thanks a lot Scott.
I's quite interesting to learn a bout the special features of the input trannies. So you combined the advantages of two different geometries: Low rbb' and high hFE.
As for noise, I've just copied RB, RC, and RE from the MAT04 model. Now the noise figure should be close to 0.9nV/rtHz. (BTW, I also lowered CJC and CJE).
One more question about the output stage (I you don't mind). There are four large trannies at the left of the picture. Do they represent the two output trannies Q22 and Q30? If so, are they spread across the chip for better thermal balance?
Cheers,
Edmond.
G.Kleinschmidt said:I once made a hfe meter whose design was base on beta and it worked well so there!
Long time ago in one institute of semiconductor devices I had access to a thingy that drew nice hooks on the screen! When graduated in volts it drew one hook; when graduated in amps it drew many hooks!
I miss that thingy, because I care of Beta. Where to find one? I suppose some plugin can be bought for PC, or for MAC?
I have one here, Wavebourn, a TEK 577. If you are nice, you might get a chance to use it, sometime.
Re: Re: Re: AD797 input trannies.
Yes, that is one way to assure thermal symmetry.
Edmond Stuart said:
Yes, that is one way to assure thermal symmetry.
Re: Re: Re: Re: minimum stage count
You better not guess, instead, let me guess. You're talking about an amplifier, right? 😀
Well, I must admit, it has something to do with it, but I've no objection if you call it a strait wire with gain. Are you satisfied now?
Wavebourn said:
You better not guess, instead, let me guess. You're talking about an amplifier, right? 😀
Well, I must admit, it has something to do with it, but I've no objection if you call it a strait wire with gain. Are you satisfied now?
Re: Re: Re: Re: Re: minimum stage count
Hmmm... Now it looks like a snake pretending to be a rope. 😉
I've seen a very similar thingy in To-3, long time ago...
Edit: http://www.datasheetcatalog.org/datasheet/elantec/ELH0021K.pdf
Edmond Stuart said:
Hmmm... Now it looks like a snake pretending to be a rope. 😉
I've seen a very similar thingy in To-3, long time ago...
Edit: http://www.datasheetcatalog.org/datasheet/elantec/ELH0021K.pdf
At the risk of boring everyone, once again, I want to give a quick history of Rbb' and its evolution in discrete bipolar devices.
By 1968, 40 years ago, small signal bipolar devices were generally general purpose with low betas (50-150) and medium Rbb' (40-150), or 'low noise' devices with high betas (200-500), and high Rbb' (150-400 ohms).
It was in fact possible to parallel a number of 40-50 ohm devices and achieve an effective noise of 10 ohms or 0.4nV/rt Hz at the time.
Motorola was a leader in making audio quality devices, but they seemed to stop doing any new research in discrete bipolars from about 1970, and even made a pie graph showing the tradeoffs.
When the Levinson JC-1 was introduced in 1973, it utilized what was known and available at the time to make a product with 0.4nV/rt Hz and it caused a lot of notice, because it was so quiet, perhaps 20 dB below most designs, without using a transformer. Today, it is obvious how it was done.
This problem was in Rbb', which was then taken up by the Japanese who became very competitive with each other as to what could be done. While Rbb' in most American devices remained about 40 ohms, some devices were 'accidently' developed as output stages for transistor radios that were very low Rbb', 5 ohms or so. These devices, made on the same assembly line as low noise devices remained very quiet and relatively free of 1/f noise. The biggest tradeoff was the medium beta, and high input capacitance of these devices.
About this time, 1978, the Japanese made some remarkable devices, culminating in a complementary pair of Hitachi devices (not available today) that had just about everything: Hi beta, low capacitance, very low Rbb', high voltage, almost no 1/f noise. To my knowledge, these devices have never been beaten, but they are also not generally available, either.
In the next few years, ROHM came up with a good, complementary bipolar pair that were almost as good as the Hitachi devices, with Rbb' as low as 2 ohms for the PNP, and 4 ohms for the NPN. Slightly higher capacitance and slightly lower beta however. Still, a very good part pair.
Back in America, National introduced the LM394 at some point, the MAT 02 and MAT 03 became available as well. The rest is well known to most here.
By 1968, 40 years ago, small signal bipolar devices were generally general purpose with low betas (50-150) and medium Rbb' (40-150), or 'low noise' devices with high betas (200-500), and high Rbb' (150-400 ohms).
It was in fact possible to parallel a number of 40-50 ohm devices and achieve an effective noise of 10 ohms or 0.4nV/rt Hz at the time.
Motorola was a leader in making audio quality devices, but they seemed to stop doing any new research in discrete bipolars from about 1970, and even made a pie graph showing the tradeoffs.
When the Levinson JC-1 was introduced in 1973, it utilized what was known and available at the time to make a product with 0.4nV/rt Hz and it caused a lot of notice, because it was so quiet, perhaps 20 dB below most designs, without using a transformer. Today, it is obvious how it was done.
This problem was in Rbb', which was then taken up by the Japanese who became very competitive with each other as to what could be done. While Rbb' in most American devices remained about 40 ohms, some devices were 'accidently' developed as output stages for transistor radios that were very low Rbb', 5 ohms or so. These devices, made on the same assembly line as low noise devices remained very quiet and relatively free of 1/f noise. The biggest tradeoff was the medium beta, and high input capacitance of these devices.
About this time, 1978, the Japanese made some remarkable devices, culminating in a complementary pair of Hitachi devices (not available today) that had just about everything: Hi beta, low capacitance, very low Rbb', high voltage, almost no 1/f noise. To my knowledge, these devices have never been beaten, but they are also not generally available, either.
In the next few years, ROHM came up with a good, complementary bipolar pair that were almost as good as the Hitachi devices, with Rbb' as low as 2 ohms for the PNP, and 4 ohms for the NPN. Slightly higher capacitance and slightly lower beta however. Still, a very good part pair.
Back in America, National introduced the LM394 at some point, the MAT 02 and MAT 03 became available as well. The rest is well known to most here.
In late 70'th struggling with noises in dynamic microphone preamplifier I opened a plastic bag with new unmarked orange devices and found enormously low Icb:
http://www.chipinfo.ru/dsheets/transistors/1626.html
...and gave them a try. The result was like the amp did not work... Grave silence...
They could be marked with any last letter, depending on the plan: often to get right numbers of production letters designating low Uce and H21e were stamped on devices that were made from so clean crystals, so superseded all specs. Everything had to be made according to Plan.
People who designed them said that they decided to make an emitter that looked nice on the picture. May be they were kidding, but may be not. 😉
http://www.chipinfo.ru/dsheets/transistors/1626.html
...and gave them a try. The result was like the amp did not work... Grave silence...
They could be marked with any last letter, depending on the plan: often to get right numbers of production letters designating low Uce and H21e were stamped on devices that were made from so clean crystals, so superseded all specs. Everything had to be made according to Plan.
People who designed them said that they decided to make an emitter that looked nice on the picture. May be they were kidding, but may be not. 😉
Re: Re: Re: Re: Re: Re: minimum stage count
I doubt that's a monolithic device, but anyway, here's the modern incarnation:
LT1210 in TO220, 1.1A output, current feedback, 35MHz, 900V/uS, 3nV/rtHz noise, stable in 10nF output load.
http://www.linear.com/pc/downloadDocument.do?name=1210fa.pdf
Even John could be happy with this piece of dirty sand, it takes 35-65mA from the power supplies, in standby.
Wavebourn said:
I doubt that's a monolithic device, but anyway, here's the modern incarnation:
LT1210 in TO220, 1.1A output, current feedback, 35MHz, 900V/uS, 3nV/rtHz noise, stable in 10nF output load.
http://www.linear.com/pc/downloadDocument.do?name=1210fa.pdf
Even John could be happy with this piece of dirty sand, it takes 35-65mA from the power supplies, in standby.
Re: Re: Re: Re: Re: Re: Re: minimum stage count
No, it is completely different thingy (indeed, a hybrid one). I mean the schematic, it reminds me Edmund's snake-rope... 😎
...and your servo.
syn08 said:
No, it is completely different thingy (indeed, a hybrid one). I mean the schematic, it reminds me Edmund's snake-rope... 😎
...and your servo.
syn08,
The BJT is a (non-linear) voltage controlled device.
Anywhere.
The BJT is a (non-linear) voltage controlled device.
Unfortunately, it does not. Some information seems to be coming from layman pocket-sized paperbacks and marketing pamphlets, spreading heresies. Be on your guard.
And your point is...?- Status
- Not open for further replies.