The LH series was made in the 'cold war' to design equipment to resist 'Soviet intimidation'. It is ironic that Wavebourn has a few surplus units. Kind of like me having Russian made Teflon caps, which I do.
john curl said:
Some people don't see the difference between stages designed for class A and biased in class A. I prefer to run stages designed for class AB in class AB, because class A means totally different approach.
john curl said:
The best devices to resist "Soviet intimidation" were those that played nice music. The best way to win was to show us your humanism, because our propaganda painted monsters who squeeze last drop of blood from poor American workers to get little bit more of profit out of them.
'Workers of the world, unite! All you will lose is your chains!' Yes, we are still oppressed by the Capitalist pigs.
(This is meant in jest, or at least in the same spirit as Wavebourn offers)
(This is meant in jest, or at least in the same spirit as Wavebourn offers)
syn08 said:
Thanks, I have had samples of both almost 2 years.
To me, the AD844 is the most natural sounding opamp. I mentioned AD811 because of its high output stage bias current and excellent HF linearity.
scott wurcer said:
National and the Fairchild sibling are about the worse when it comes to samples. Give it a shot, best case you'll get them for free, worst case they'll ask for more than DigiKey is charging. Usually they settle for the shipping fee from the Singapore warehouse. Worth for these buffers (they are very expensive), not so worth for other opamps, etc...
I also prefer class A in my amps. But when designing with OPamps
I sometimes have to skip that preference to get the job done.
Sigurd
I sometimes have to skip that preference to get the job done.
Sigurd
john curl said:
Sigurd Ruschkow said:
AD8597/8599 are much slower compared to AD797. Otherwise, they are very good.
Sigurd Ruschkow said:
That business is all handled in CA now I'm doing 10G comms stuff these days. You might notice some of the 1nV parts have total supply currents just over 4mA, the processes are so fast that the rest of the circuit can be starved to death, JC must hate it.🙂
I will have to scold Joanne (sp?) about it. Actually, you guys keep me in business by compromising your designs, so that I have to either: fix them, or design around them.
john curl said:
That's with an e on the end. I don't think she'll ever forget our dinner with Jim Bongiorno. I think this part is someone else actually.
For the record, Scott, and others: I have already made the basic phono design, and it will be in production once I finalize a few parameters. While, I appreciate your op amp selections, I am stuck with the realities of the design layout as it stands. I am using a mini-dip package, it has to have 1nV/rt Hz noise, reasonably low bias current, and low 'higher order odd' distortion. I, already have a slot for a fet load to offset the output stage, IF I find it useful. I must work at +/- 15V or slightly more, for headroom and my effective load is 800 ohms above 2KHz.
I find the AD844 intriguing. Barrie's design, isn't it?
I find the AD844 intriguing. Barrie's design, isn't it?
I would like to talk about why I am so concerned about Iq in an IC output stage. and the difference between class A and class AB.
So far as I usually design, myself, almost any class AB design can be biased into class A, IF the heatsinks are big enough, in an ideal sense. This is especially true with preamp output stages.
This is because my output stages are invariably complementary push pull, either bipolar, jfet, mos fet, or V fet. I have made all 4 versions and have them working in equipment today. Power dissipation is the major factor in setting the Iq, but sometimes Idss, second breakdown, or some other mechanism must be included.
Now, WHY class A, when the distortion is essentially unmeasurable, due to lots of feedback?
Well, first, there is a problem with the measurements. Normal HD testing does NOT signal average (I do, by the way) and the distortion residual AT LOW LEVELS of output voltage is LOST in the noise. It doesn't go away, and it is not necessarily inaudible, it just can't be measured.
The most nasty kind of distortion in these conditions is x-over distortion, due to one device turning off before the complementary device turns on. This happens a lot, especially with cheap IC's. However, a good design, even with .5ma of Iq can minimize x-over distortion, and I expect the IC designers here, take care in making the x-over zone and smooth as practical.
Still, that is only 1/2 the answer. It just so happens that when an output stage transitions from class A to class B, due to a rising output current, the inherent even order distortion cancellation of class A is not possible. Well, where does the 2'd harmonic distortion go? It converts to higher order odd harmonic distortion, and sums with the existing 3'rd harmonic distortion. Is this good, tube fans? I would prefer to ignore it, by increasing Iq.
Now, it CAN be avoided by either increasing the Iq of the output stage, or NOT loading down the output with a smallish resistance. Unfortunately, then you cannot utilize the inherent low noise of a quality IC, because the feedback resistance will dominate the noise. Is this what we want? Do you see my dilemma?
So far as I usually design, myself, almost any class AB design can be biased into class A, IF the heatsinks are big enough, in an ideal sense. This is especially true with preamp output stages.
This is because my output stages are invariably complementary push pull, either bipolar, jfet, mos fet, or V fet. I have made all 4 versions and have them working in equipment today. Power dissipation is the major factor in setting the Iq, but sometimes Idss, second breakdown, or some other mechanism must be included.
Now, WHY class A, when the distortion is essentially unmeasurable, due to lots of feedback?
Well, first, there is a problem with the measurements. Normal HD testing does NOT signal average (I do, by the way) and the distortion residual AT LOW LEVELS of output voltage is LOST in the noise. It doesn't go away, and it is not necessarily inaudible, it just can't be measured.
The most nasty kind of distortion in these conditions is x-over distortion, due to one device turning off before the complementary device turns on. This happens a lot, especially with cheap IC's. However, a good design, even with .5ma of Iq can minimize x-over distortion, and I expect the IC designers here, take care in making the x-over zone and smooth as practical.
Still, that is only 1/2 the answer. It just so happens that when an output stage transitions from class A to class B, due to a rising output current, the inherent even order distortion cancellation of class A is not possible. Well, where does the 2'd harmonic distortion go? It converts to higher order odd harmonic distortion, and sums with the existing 3'rd harmonic distortion. Is this good, tube fans? I would prefer to ignore it, by increasing Iq.
Now, it CAN be avoided by either increasing the Iq of the output stage, or NOT loading down the output with a smallish resistance. Unfortunately, then you cannot utilize the inherent low noise of a quality IC, because the feedback resistance will dominate the noise. Is this what we want? Do you see my dilemma?
john curl said:
Hello John
Would this be a solution to add in the signal path an additional unity gain buffer biased in class A (even for low impedence loads) , this would also remove the loading issues on the output stage of the opamp ( low Iq bias). Is this what you do to get around the problem .
Regards
Arthur
john curl said:
All valid concerns, but if you put some numbers and look at the modern opamp specs you'll find it's not that bad as you expect.
E.g. integrated DSL drivers have an amazing output current drive with extremely low distortions and noise. This is essential, 'cause any distortions means wasted bandwidth for the same date rate. Disadvantages: low supply voltages (+/-5V) and rather large input bias current. Look at THS4275 fig. 5 and up.
Other ICs have still very high drive current capability, also very high supply voltages. Take a look at OPA551 JFET input, with +/-30V supply and 200mA output current capability. Disadvantage: not very fast, distortions and noise performance are average.
If you really want the best of all worlds, get a current feedback, high output current opamp, like LT1210 Supports +/-15V supplies, has 1.1Amps output current, 55MHz bandwidth, 900V/uS slew rate, voltage noise 3nV/rtHz, current noise (on the non-inverting input, typical for a current feedback configuration) 3pA/rtHz. Comes in TO220 to allow heatsinking. The only disadvantage that I can think of is price: not the cheapest around, but definitely not outrageous at $17 in singles. The sibling LT1206 with only 250mA output current sells in singles for about half of that, which is already affordable.
If these won't do it for you, then I don't know... BTW, do you really care for distortions in a MC/MM preamp? C'mon, the whole vinyl process/media/cartridge chain has 3 orders of magnitude more distortions from what even an average opamp can provide.
syn08 said:
There are plenty of 24V DSL drivers, and remember the AD815 was the first DSL driver (we did the first proof of concept with a Cambridge think tank using 4-AD811's in parallel in 1995 or so). Good rate and reach has been achieved over barbed wire.
- Status
- Not open for further replies.