TI r0xx0rs
Jun 15, 2002 at 3:17 AM Post #16 of 32
Darn it, didn't you see my reply?! My point was that by using the EL2001s in the META42, you and tangent have purchased a lot of buffers!!!
 
Jun 15, 2002 at 3:25 AM Post #17 of 32
Nezer, the reason I got a bit annoyed is because of the tone you had in your first post. It was kind of a " I got free stuff because I can" kind of thing.

It would be a bit less suspicious if you had only gotten one of each, because bascily all you are doing is an a/b comparision. If you used a buffer to drive the opamps from the same input you could have one opamp per channel and be able to hear the differences between them. That would be what I would do to compare chips.
 
Jun 15, 2002 at 3:37 AM Post #18 of 32
Quote:

Originally posted by tangent
Nezer,

I hope you don't think I'm accusing you of being an immoral SOB. I personally wouldn't have placed an order for samples like that, and my later post said why. However, TI made the rules, and you're playing within them -- I wouldn't call your order "abuse". It's just that, given the cost of the parts, I'd have to stop and think a bit to find justification. If you've justified it to yourself, then that's what really matters. [/B]


While I don't think I'm 'abusing' the system I'll be the first to admit I'm taking advantage of it. After all, that's why it's there. :wink:

Anyhow, I really won't be using these after I'm done testing (whatever I don't fry) and even if I did use them I would only need a 2 of one of the types at most. If in the end I decided to use a dual-opamp-stage setup (as oppsed to a buffer follower) I'm anal enough to want to have all four single-channel opamps to come from the same batch so I would just order four anyway even if I already had 3 lying around.

It is a fair amount of money in parts but it gives me the opprotunity to do something I otherwise couldn't afford. Things like parallel an input stage with drastically different opamp types just to see what the results would be). It also gives me the chance to compare the 627A to the B to the 637 A and B without plopping down $120 for a pair of each. Let's face it, the 627s and 637s are a HUGE portion of this order in terms of money. I personally think that says a lot that TI thinks these things are so good they are willing to give a few away to get you hooked. Theres a drug dealer in my area like that and he has 85% of the local market covered. :wink:

I want to also get som LM617x and a few others too to make a fair comparison. I could grab the 6171s from my Corda I suppose.

What other opamps should I look into playing with while I 'll have the test board lying around?

As for the test board I'm going to keep the design simple--a basic CMOY with a buiffer follower (EL2001s look intresting, do you offer free samples Tangent? :wink:) and I will probably setup a circuit to bias the opamp into class-a operation. I have a funny felling this will change the charastic on some of the opamps drastically (particuarlly making an awful-sounding opamp listenable).
 
Jun 15, 2002 at 4:06 AM Post #20 of 32
Quote:

Originally posted by eric343
Of course Tangent does free samples! Shipping and handling is just $3.50/buffer extra...


LOL, when I get the cash I'm going to pick up a bunch of these and probably about 10 of the Meta boards.

First, I need to make a power supply to fry all these free opamps. Cerafine 5000uf 50Vs here I come!
 
Jun 15, 2002 at 5:02 AM Post #21 of 32
On that test methodology, Nezer, there are two ways to go about it:

If you build a basic Cmoy type amp, you're listening to the op-amp naked -- nothing else is influencing the sound. But, you're making the op-amp do both current and voltage amplification. This is difficult.

If you add a voltage follower (preferrably in the feedback loop, so you don't add errors), you're allowing the op-amp to stick to voltage amplification, and leaving the current load up to the buffer. This helps the op-amp to sound its best.

Personally, I prefer the first test -- I don't necessarily want all the op-amps I test to sound good. I want distinction, differentiation, so I put them to the nastiest torture test I can devise.

An op-amp that sounds clean and dynamic when driving the load alone will still sound clean and dynamic when buffered -- only more so. And op-amp that sounds good but a little weak alone may fill out with a buffer. An op-amp that sounds like crap alone will continue to sound like crap when buffered. This guides you to the best op-amps, it points out the ones that are merely good but improvable, and it weeds out the truly bad.

Just MHO.
 
Jun 15, 2002 at 5:18 AM Post #22 of 32
What about opamps that have low current output and pretty much need a buffer, such as the AD823? Do you reccommend sticking them in a Hansen board, where they'll buffer themselves? (assuming these are dual opamps we're talking about, some of Nezer's aren't)
 
Jun 15, 2002 at 5:36 AM Post #23 of 32
I recommend testing even the AD823 alone. It's maybe a little weak in the bass line without help, but the overall character is still evident. This op-amp falls into the "merely good but improvable" category.

Also, in this kind of case, you can test with high-impedance headphones, where lots of current is not important.

By the way, the Hansen doesn't fix the AD823's current problem. The "buffers" are just more of the same kind of op-amp, so you end up with AD823s still driving the load. Granted, you split voltage and current gain between the two op-amps, but it's not the same as using a high-current buffer.
 
Jun 15, 2002 at 8:24 AM Post #24 of 32
The different grades of op-amps aren't sorted by anything that human ears can discern. The different grades of OPA627 are sorted by minor variations in:
offset voltage (you can't hear DC)
bias current (you still can't hear DC)
noise (.8uVp-p vs .6uVp-p)
common mode rejection (110dB vs 116dB)
open-loop gain (116db vs 120db)

I have serious doubts anyone can hear the difference between different grades of chips as opposed to chip to chip variations.
 
Jun 15, 2002 at 8:44 AM Post #25 of 32
I've yet to be able to tell the difference audibly between two grades of the same chip. I only suggested getting the high grade when getting samples since a) you're not paying a premium for it; and b) it will be better-behaved. I've run into several different chips where the lower grade behaved badly where the higher grade worked well.
 
Jun 15, 2002 at 2:27 PM Post #26 of 32
I believe higher grade is always measured and specifications verified - hence the price - while lower grade is not, or only a few chips out of the batch are. Regular chips are probably only tested whether they work, and not measured for noise, offset etc. There are exceptions though, depends on the manufacturer. I think Analog tests all of their DAC chips for harmonic distortion. On the flip side, their DAC chips are very expensive. Also, some chips - some voltage references for example - have more than two grades, and those are probably all measured and sold according to precision.

So, with higher grade you're guaranteed to meet specified numbers. With lower grade, you're not but you could still get a chip that's just as good, if you're lucky. Some datasheets give you probabilistic distribution of parameters over the batch of chips, so you can see that there's a 10% chance you'll get a really crappy chip but in most cases they'll be just as good as high grade.
 
Jun 15, 2002 at 10:21 PM Post #27 of 32
Yes, it's true for the higher grades, the limits for some specifications are actually guaranteed. In the case of the OPA627, each chip in the higher grade is tested/rejected for noise while the lower grade is only sampled for noise.

Nonetheless, for some important specifications like open loop gain and input bias current all grades are guaranteed to meet some minimum level of performance.
 
Jun 18, 2002 at 1:03 AM Post #28 of 32
I forgot how fast the samples arrive! They showed-up on my doorstep this morning before I left for work.

I haven't had time to make a test circuit yest but I've decided to go on a variation of the Meta42 design and install a bypass switch for both the class-a current source and the buffer. This will easily allow me to hear each op-amp in all three variations.

Anything I should watch-out for in doing this?

I'm going to post a mini-review of the 637s I just threw into my modded Corda.
 
Jul 21, 2002 at 8:26 PM Post #29 of 32
Came across this thread a little late and thought I'd comment...

I am a big fan of samples. If I want to try a part, I can get it for free. Yes, I am not going to buy 1 million chips like Sony if I like the part. TI doesn't necessarily know this when I order. However I am an engineering student and when some day I am an engineer making these decisions with big money, TI/Analog would probably like me to favor their products.

Besides, he might have had to spend $300 at Digikey to get those parts, but that's at highly inflated <10 quantity prices. The question is what do those parts cost TI? Probably less than $10. And don't comment about how TI had to pay a ton of money for overnight shipping, they didn't, they ship out thousands of orders a day that way and would have a special deal with UPS to get it way cheaper than you could
 
Jul 21, 2002 at 11:32 PM Post #30 of 32
Yeah, but what about when you were trying to e-mail Alps for a sample of their $427 50mm audio potentiometer?
 

Users who are viewing this thread

Back
Top