Then it would be easy for you to answer to this question, i suppose:
WHO was the first to use a mos-fet transistor to drive bipolar transistors to circumvent the BE capacitance that lowers the slew rate in the final stage?
Attachments
You should be able to eliminate the output capacitors and one supply too with balanced operation...
Is your project public?I'd like to see it.
It's 4k7 +2k2 trim pot...not 6k trim. With matched arrays you shouldn't need more than 500 ohms trimpot.
i just realized that i never heard this type of music 🙂...
Some of it is really funny although i don't understand all the words those guys tell there:
YouTube
Yes, if you can afford a NiCr layer. Not so, if you have nothing but quench resistors
or poly. And thick film is probably not chip design. Thick film lasering won't determine
the tail current of your 797.
But at least, no people are involved to tune it. I have seen people with a Dremel-like
tool adjusting pots under control of a test program.
At Verigy, I had my own mixed signal wafer tester. I mean, no one but me, give or take
now and then a test if old programs still run as they used to. Customers like AD would
calculate quite precisely if it's OK to spend one extra second of it's time on an op amp
offset.
Probably not for a thing like
< TS461CDT STMicroelectronics | Integrierte Schaltungen (ICs) | DigiKey > ,
and these are legion.
BTW thanks for the literature list. That's like drinking from a hydrant.
Last edited:
In a way...yes, but anyway i'm not coming from an English or German like society, we're deep into the Balkan's culture...I just realized that there must be something with those clips as Google wanted to check my phone number and identity shortly after 🙂)
Last edited:
I hope you realize that your op-amp is 0.3...0.4 dollars and ad797 costs 10 dollars...I couldn't afford to buy more than two of them, along with another 100 op-amps in the last two years ... ad844, opa1632 and ada4898 simply killed my budget . Better that ad797 be up to its datasheet !!!
Last edited:
I don't know who used complementary power fets for drivers first, I just adapted what Parasound was already using in 1989. However, in 1976, I modified a circuit for Mark Levinson which allowed bipolar transistors for the output, and I drove them with complementary jfets. This became useful for the LNC-2 crossover, I believe that it was used for that.
Just to see some measurements for this amp topology made by a guy who didn't know anything about electronics...but scaled for higher power
Jean Hiraga's Super Class-A Amplifier
Jean Hiraga's Super Class-A Amplifier
Hm. that opamp was just on s.e.d. in usenet as an example of el cheapo. (BTW usenet may have it's share of ar**holes, but judged by the messages I feel worth to archive, it is 30 dB better than this website here.)
Those high priced AD op amps touch an open wound here currently. The AD4898-2 used to be a good deal, but now at €7.50 per pop it starts to hurt. And I noticed in my 10 * 4898-2 parallel amplifier that the noise went up from 200 pV/rtHz to 600 pV at 1 MHz. Unacceptable. I found out that this was nothing but the skin effect on the board. I soldered in some parallel wires, et voila, effect gone.
But then I measured an object that could deliver current without end. I mean, really without end, with Lithium batteries involved. The aluminium case was fastened with 6 screws, but the stench, OMG. I took it out to the balcony to open. Near one of the ADA4898-2, the board was black as coal. Wires were molten. I had soldered one of the wires to the input side of the that wet slug tantalum cap. Now, I'll send another €80 to digikey to bring it back alive 🙁
Grief, Gerhard
That's not paradise. It is called work. You must be able to defend your stuff. Design reviews and so. There is no pardon. The others won't forget that you found bugs in their stuff. They will find your's. And that's good. The goal is that the Falcon rocket does not fly for nothing. And you don't want to be the culprit if that should happen. regards, Gerhard
Last edited:
I'm not sure if i understood your text , but i did have problems with ada4898-1 melting which determined me to look into paralleling cheaper op-amps in i/v sections.In the end it seems to be a better idea .
Ada4898 has too much idle current, works too hot for my taste, but i found a way to use it in i/v stages which is a bit smarter as i don't care about that 130db thd+noise that the new dacs offer.
For me, the old 90's standard of 100db SNR is ok... I simply lower the gain in the transconductance stage by 2x and lower the supply rails to +-5v. so that they work colder , the current noise which is kinda high is generating half the voltage noise now (indeed transconductance stages need the Rfb to be higher for lower noise, but in the end all adds up and the result is fine.I can thus use furter op-amps buffers with a little bit of gain which gave me more choices , plus their psrr and cmrr get better with a little bit of gain.
Ada4898 has too much idle current, works too hot for my taste, but i found a way to use it in i/v stages which is a bit smarter as i don't care about that 130db thd+noise that the new dacs offer.
For me, the old 90's standard of 100db SNR is ok... I simply lower the gain in the transconductance stage by 2x and lower the supply rails to +-5v. so that they work colder , the current noise which is kinda high is generating half the voltage noise now (indeed transconductance stages need the Rfb to be higher for lower noise, but in the end all adds up and the result is fine.I can thus use furter op-amps buffers with a little bit of gain which gave me more choices , plus their psrr and cmrr get better with a little bit of gain.
Last edited:
No, it was not the 4898's fault that it died (very probably I cannot trust the other 9 that did not produce a crater in the board). I measured the noise of 4 18650 Lithium cells, about +16V with 30A short circuit current. I was not happy with the noise of an IF3602 amplifier that was supplied from the Lithia. 350pV/rtHz for one FET. Worse than the data sheet, even with more drain current. Not bad for a FET, and respectable 1/f corner, but definitely worse than specc'ed.
16V/30A or so into an op amp running at +-6.5V is a bad thing. mea culpa about that shorted input cap. You can see the bad wire in the photo. It is kinda unfair that it survived.
The ADA4898-2 can get burning hot when one of it's inputs is open. That would belong into the data sheet. At least it won't die on the spot.
mea maxima culpa. Gerhard
16V/30A or so into an op amp running at +-6.5V is a bad thing. mea culpa about that shorted input cap. You can see the bad wire in the photo. It is kinda unfair that it survived.
The ADA4898-2 can get burning hot when one of it's inputs is open. That would belong into the data sheet. At least it won't die on the spot.
mea maxima culpa. Gerhard
Attachments
Last edited:
Attached is an AES paper collecting many of the latest info on high resolution audio. It talks about dynamic range and what is needed based on some good research and some of the proposed "why's" of extended bandwidth. Its open so no copyright issues and an interesting read.
Undoubtedly different world I live in. Here, despite all the work done, ADI need to trim every AD797 to meet spec due to limitation in semiconductor process control.
Don't be tense Gerhard, it's a joke. 😀
I used software calibration in 1989. You feed into the device under cal a voltage (+-10% is fine long as it is quiet), measure it with a very accurate meter like a Fluke or Agilent 7 or 8 digit instrument, compare the two and store the correction facto for each input type and gain range. That cals out the voltage reference error and the gain errors.
The calibration s/ware sat on the instrument itself (6805 MCU IIRC - software written by another guy - I only did the hardware and all the flow charts etc), which had a universal front end for mV, TC, SG etc.
Today you can buy fully integrated analog S-D front end IC’s from ADI, TI etc now for about $10 - I did it the hard way with a dual opamp, 4051 muxes and Teledyne 16 bit dual slope A-D. The opamp was configured as a 3 bit PGA so the total range of the instrument was c 19 bits. The second opamp was configured as a Howland current pump (which initially oscillated so one of the guys called it a ‘Howling current pump’).
Last edited:
- Status
- Not open for further replies.