PMA, don't the make 411's anymore? That should work, OK, sort of. I used to use it for servos before the 711.
I do realize this, but we are not speaking of twenty cent IC's either. I don't know how many $10.00 IC's need to be made a day to make it viable.syn08 said:
I used to work for Harris (telephony not semiconductor) for years the Harris semi people made a living selling $75 A to D chips. Eventually they bought RCA because they wanted to have a bread and butter line.
Unfortunately they didn't fully understand the difference between marketing commodities and products and soon went under.
I certainly claim no expertize in this area. I tend to think of Dinosaurs who couldn't adapt fast enough and died out.
hermanv said:
It's 20 cents because they do 5 billion, not the other way around.
Wasn't Harris at that time mostly selling to the US military? If so, the $75 they were selling for was really a bargain.
Yes, but in todays dollars it's probably closer to $250. At the time they and Burr Brown were selling many devices as modules we would call hybrids today.syn08 said:
If you wanted performance in a small space, you paid their price.
Still I'm basically with John on the discrete solution, it allows you to optimize that which is most important for your application.
The twenty cents and 5 billion is inter-related you can't have either one without the other. No matter how cheap, a 5 billion piece market must also exist.
Highest quality usually costs more. 'Cherry picking' IC's to send to the military, pays many initial costs, I should think. We get the rejects, usually. Back in 1971, we got 1000 rejected military ceramic pack IC's for $0.25 ea when they would have cost perhaps $10 with their fancy package, if they had passed their temp range. Please remember that $5/hr was an engineer's wage at the time. The parts worked for what we wanted to do with them, which included studio boards, guitar electronics, etc. I still have lots of them in my office.
I seem to remember a studio mixer board. 741 Op-Amps for both gain buffers and 1/5th octave equalizers. 100's of them in total - sounded god awful.
I worked aerospace at the time, paid better than audio. Ended up in telephony because I felt that devoting my talents (even tangentially) to the War (Vietnam) was wrong. I remember meeting my first audiophile, one of the engineers built speakers into the walls of his house out of bricks. I thought he was nuts, maybe that's where I caught the disease.
Modems were new, exciting 110 baud, wow! Then quadrature coding took us all the way to 2800 baud, thrilling stuff.
98% discrete design then, remember trim pots? Some designs used them by the handful all along the edges of the PCB's, Technicians still existed, we needed those skills to set all the trim pots - whee.
Sorry for the diversion and revealing my age.
OK, everyone back to Blowtorch.
I worked aerospace at the time, paid better than audio. Ended up in telephony because I felt that devoting my talents (even tangentially) to the War (Vietnam) was wrong. I remember meeting my first audiophile, one of the engineers built speakers into the walls of his house out of bricks. I thought he was nuts, maybe that's where I caught the disease.
Modems were new, exciting 110 baud, wow! Then quadrature coding took us all the way to 2800 baud, thrilling stuff.
98% discrete design then, remember trim pots? Some designs used them by the handful all along the edges of the PCB's, Technicians still existed, we needed those skills to set all the trim pots - whee.
Sorry for the diversion and revealing my age.
OK, everyone back to Blowtorch.
We used Radiation Inc RA-911 op amps. They were dielectric isolation, ran on +/- 24V supplies into 600 ohms and had a +5/-2 V/us slew rate. They measured even better when I opened their open loop bandwidth with a resistor on pin 8, as well. Dick Burwen also used this op amp big time. 741's didn't compare. Still, discrete beat it every time.
john curl said:
😀
I still have a small stock of 411.
Many opamps work good as servos. 627 (bit pricey at this position), also 134.
Interesting discussions so far. Now we're on the subject of op-amps, it would be good to have John's take on which types would be best to use for the circuit which this thread is all about.
The servo return-point has not been explored as far as I know, correct me if I'm wrong, but it seems to me that when using separate servo's (with separate functions) the return-point in the already proposed topology can only be useful in the second or output stage (Mosfets) avoiding the signal path?
John could you shed some light on this? Thanks in advance.
The servo return-point has not been explored as far as I know, correct me if I'm wrong, but it seems to me that when using separate servo's (with separate functions) the return-point in the already proposed topology can only be useful in the second or output stage (Mosfets) avoiding the signal path?
John could you shed some light on this? Thanks in advance.
PMA said:
PMA,
I was thinking about an opamp where one could adjust the (class-A) standing current in the output stage, not turning it in to a single-ended output stage by assymmetrical loading. I don't think you can do that with an AD844 either.
An OPA860 is an example where you can set the quiescent current (and many other parameters) with an external resistor.
Jan Didden
Jan,
I meant to omit opamp output stage completely and to use the external one. Not SE, but p-p, class A biased.
I meant to omit opamp output stage completely and to use the external one. Not SE, but p-p, class A biased.
PMA said:
OK, that'll work indeed.
@Scott: Now that we are on the AD844, I noticed that it takes about 7mA from the supply. Where does this go to? Are the output transistors in the current conveyor biased that high, or is it the ol output buffer?
Jan Didden
janneman said:
Barrie might remember, I tried to find a schematic for Charles Hansen once and ended up poping the lid on my last ceramic one to answer some question.
john curl said:
Dick still claims he co-founded ADI just to make op-amps for his stereo. I think it was the AD211 circa 1966 that turned into the JE-990. The inductors across the input degeneration are a give away. This trick gives even more OLG at audio, does nasty things to settling time.
Sigurd Ruschkow said:
Hi Sigurd, I bought a number of AD745 op amps in the DIP8 package a while back from Rochester, but the adapter is an acceptable method.
Gary Galo designed a nice low-noise MM phono preamp using the OPA637.
Best, Chuck Hansen
Dick may have co-founded ADI, but he used the Radiation Inc (Harris) 911's in everything that he designed for Mark Levinson, and in many of his products. However, he did select the best of the 911's for the audio path, and used the rejects (xover distortion) for signal processing such as compressors, expanders, peak detectors, etc. I love Dick Burwen's early discrete designs. He was way ahead of his time.
The inductor trick was cute, but I tried to get Deane Jensen to change to fets, in order to return a favor that he had given me (AC analysis program for the APPLE 2 computer), but Jensen just scoffed at me. Later, just before he died, he was reconsidering my input. It shows how 'some' people have difficulty in accepting ideas different from theirs, even when it is handed to them for free.
PS Chuck, I think that Walt Jung makes the best designs in IC's. Especially his best efforts.
The inductor trick was cute, but I tried to get Deane Jensen to change to fets, in order to return a favor that he had given me (AC analysis program for the APPLE 2 computer), but Jensen just scoffed at me. Later, just before he died, he was reconsidering my input. It shows how 'some' people have difficulty in accepting ideas different from theirs, even when it is handed to them for free.
PS Chuck, I think that Walt Jung makes the best designs in IC's. Especially his best efforts.
scott wurcer said:
I still have that old correspondence from 2001. I was asking how the null pins were connected, as they are not shown in the simplified schematic on the data sheet.
Scott:
The plot thickens though, a schematic does not
seem to exist anywhere in the company. Hard to believe, but true. I found a
lone ceramic AD844 at the bottom of our group junk drawer and popped the lid.
When I can find a microscope I will be able to see how the pins are connected.
The 844 came at a time when we were transitioning to fully automated design,
and along with a few other parts has no electronic documentation.
Charlie:
Wow!! This is really exciting, kind of like uncovering buried treasure! I didn't
know you could interpret an IC just by looking at the die itself, but I suppose that
the AD844 is a *relatively* simple part... Talk about going the extra mile — I
really appreciate it.
Scott:
The null pins appear to have ~10k series resistors and connect to the junction
of ~1k resistors (tied to the plus rail) and the emitters of pnp current
mirror connected transistors.
Charlie:
I didn't realize that you had the capability to also measure the resistor
values. Cool!
The simplified schematic on the data sheet (figure 28 on page 8) doesn't show
any resistors. Just to make sure I'm following you, there is a modified Wilson
current mirror above the square terminals labeled "2" and "5" in figure 28. The upper
pair of these transistors is shown connected directly to the positive supply rail. I
assume that in reality there are some ~1k resistors between these emitters and the
rail, which is where the ~10k resistors then go to the null pins.
If I have understood you correctly, this scheme makes perfect sense. Thanks once
again for going the extra mile to dig up this information for me.
~~~~~~~~~~
So with 1 kohm resistors on the current mirrors, we can safely assume that the front end is running well under 1 mA -- this is confirmed by the fact that the input impedance of the inverting input is 50 ohms, implying a bias current of ~250 uA. That means that the bulk of the 7 mA supply current is in the output stage.
Doing the same analysis with the LT1028, I found that there was almost no extra current available for the output stage. The trend, it seems, is to starve the output stage. We are going backwards. We might as well update the UA709, with its class C output stage. 😕
- Status
- Not open for further replies.