[tangent]

In another thread, Kelly said:

Quote:

I challenge any DIYer to build a META42 properly (ie, you have to follow the design and the solder points have to be made properly) using the crappiest component choices you can possibly choose--the cheaper the better. My hypothesis is that this amplifier will still be much better than the Creek.

I decided that I could do worse than the Creek OBH-11 without going to such extremes. Witness the destruction:

Op-amp: TL082 running from 12V. (This is sufficient for the task, according to the datasheet. Real life is a different story....)

Power supply: Build your own from a transformer, a bridge rectifier, and a 100uF capacitor. Given a 20mA load, this will ensure 480mV of ripple on the rails. This is sufficient to avoid frying 2001G, and yet enough that the META42's C2 will still allow plenty of ripple through to the amp stage.

Resistors: No need to go with cheap carbon. Just configure them so that you get a hundred millivolts or so of DC offset on the output of the amp. Not enough to fry the headphones, but enough to cause a loud pop when you turn the amp on, and to make the headphones "swing" improperly.

Power caps: A lone pair of 100 uF Radio Shack 'lytics. If possible get the ones on the back of the peg still in the "Archer" packaging, so you can be sure they'll fail early. Leave out C3, C4 and C5.

C1: X7R Ceramic. 'Nuff said. Make it 100pF, so that, combined with the 10MOhm R2 you need to meet the next requirement, you get a -3dB bass rolloff point of about 160 Hz.

Volume control: Linear taper Radio Shack pot; use a 1Mohm model, if available, and certainly no lower than 250Kohm. Use two mono pots, not a single stereo one. Call it a "feature": integrated balance control. Be sure to get the type with a metal shaft, and then use a metal knob and leave the cases ungrounded so that you get hum in one channel when you touch the knob.

I/O jacks: Cheap 1/8" jacks. Get the kind that don't grip properly, so that you have to keep a constant upward pressure on the output jack to avoid ground hum, scratchiness and channel loss.

Current Source: 220 ohm resistor only. This will bias the op-amp into class A to the tune of about 27mA on average -- within the chip's specs, and way more than necessary. This should serve to completely rob the chip of any dynamic qualities it possesses.

Can one do worse? Hell yes. I'm just getting started.

I invite comments. This thread should be educational -- I've always believed in the instructive value of a bad example.

 

[eric343]

Hey, that's the same config you told me would make a great reference amp

Seriously, though, I think Kelly meant by using the cheapest parts availible, not by mucking up the config... If you wnat to do a horrible sounding config, just leave out a couple parts... Or better yet, swap a few resistors (R2 and R3 come to mind, or R2 and R4... What about R2 and R7?

)

But you could have atleast given the amp a figting chance by using a TL072...

 

[kelly]

Yup, Eric has the spirit of my request down a little better. The basic idea is this: Let's say someone wanted to build and sell META42 amplifiers but wanted to cut every corner possible to get the thing to market and maximize his profit. After having done so, would the amp still sound better than the Creek or not? My guess is that it would.

 

[eric343]

Well, when I get around to building a full-breadboard META42, I'll try out the cheapest components I can find (and there might be a few Vishays or whatever in there, I haven't bought any cheap stuff in ages)

 

[tangent]

Okay, then, here's another try.

But first, on the TL072, I've tested it, and I can't tell it apart from the TL082. Sure, sure, the specs are different, but even from the lowliest TL082 to the highest-spec TL072 I could dig up, the things were indistinguishable to me. Which is to say, barely adequate for testing purposes.

Here's my sane-but-cheesy configuration:

Op-amp: OPA2134PA. This will require 12V to avoid any distortion on tough music with difficult headphones.

Power supply: 12V unregulated wall wart.

Resistors: 5% carbon resistors, unmatched.

Power caps: A lone pair of 220uF lytics -- leave out C3-C5.

C1: Metalized polyester, value chosen so that the -3dB point of the input filter is just barely below 20 Hz.

Volume control: Panasonic EVJ-C. Not only is the part inexpensive, it's board-mounted, which lowers the labor cost.

I/O jacks: The cheapest 1/8" jacks that will still grip securely for at least a year. (Beyond that, the warranty expires, so what the hell...)

Current Source: Leave it out.

Jumper across R1, R7, R8 and R9.

Jumper across 2001G. If you can swing it, try to stuff a resistor divider into the TLE2426 position.

Total parts cost: you could squeeze it down to $30 if you looked hard enough for the parts. ...Which is pretty darn impressive given that the board and buffers alone will cost you $20. More money required for the case and labor, of course.

Would it sound better than the Creek? Hmmmm....hard to say either way.

 

[kelly]

Still seems like an RA-1 beater to me, too. How much does a mahogany box go for?

 

[eric343]

Now that's better. It's getting into CHA47 territory, now

BTW, if the TL072 is really that bad (it doesn't looks so horrible on the datasheet!) it sounds like I'll have to swap it with an AD8610 module in the output stage of my Arcam... (the cool thing about the Arcam CD72 (and probably the whole DiVA series) is that all the parts are through-hole! talk about easy to mod...)

 

[tangent]

I don't doubt that with careful use the TL072 can sound decent. My test was in a basic CMoy amp, and I only went up to 24V on the power supply. However, you can certainly do better. Even an OPA2132 is a big step up.

 

[aos]

I don't think there'd be that much difference if you just use cheap metal film resistors, cheap caps and cheap potentiometer and keep using EL2001 at least since it's cheap and add another decent opamp. Current source isn't necessary with AD8610 anyway and I'm still satisified with Panasonic pot (maybe because I haven't tried anything else

). If you don't cripple the idea (multiloop) and use proper if not audiophile components, and know where NOT to compromise, my guess is you should be able to easily beat lower grade commercial amps. I've built several different amps in last two years and I'm starting to side with ppl, that this design with proper components can rival a discrete design. I don't think too many amps with 3 digit costs (in Canadian dollars

) stand a chance (although there probably are exceptions). In fact in last few days I'm starting to think that the only reason my DAC sounds so good is this amp, not the DAC chip. Regardless of whether I use internal DAC or external input from line-out of several different CD players and even portables, I can't hear too many differences. Amp makes them all sound great. And that's even without voltage regulators (let alone one for each channel), current sources and other tweaks.

 

[tangent]

It's true that the META42 can make most any chip shine. I accidentally left an OPA2132 in one of my META42s and was happy with it for several hours of listening. I wouldn't have found the problem except that I opened it to show the insides of my newest creation to a friend and noticed a single chip, not the two single-channel-chips-on-adapter I expected to find.

This isn't to say that the amp makes all chips sound alike. Changing chips does make a big difference. But you can indeed be happy with low-end chips in this design. I'd personally draw the line at the OPA2134, though. That's why I used it in my second configuration -- it's the cheapest chip I'd consider "sane" in this amp.

 

[kelly]

tangent
What would the total cost of parts be for your "sane" design and how many hours of labor?

 

[thomas]

I also think these exotic parts are very overrated.

I've built numerous versions of the "meta 42" amp since it was ppl/apheared mentioned it here, ranging from my personal ultraportable version (panasonic evj pot, panasonic caps, miniplug in/out) to a high end version with rel. caps (multicap ppfx) holco resistors, alps black, huge buffered powersupply section, neutrik locking jack, paralleled El2001's, teflon-gold RCAs, point-to-point silver wire etc....

the differences are certainly there, but they all have the same basic sound characteristics. But then again, if you're gonna spend hours building an amp, might as well sepnd a few more dollars for better parts, ie upgrading resistors shouldn't cost more than a dollar or two....

 

[aos]

Exactly, the Vishay-Dales are so cheap that getting them won't make much price difference. And you need only a few (2?) electrolytic caps so why not get Elna Cerafines (even standard Black Gates aren't too expensive). Now, with pot you're getting into high range. High PRICE range that is

.

 

[Nezer]

Quote:

Originally posted by aos Exactly, the Vishay-Dales are so cheap that getting them won't make much price difference. And you need only a few (2?) electrolytic caps so why not get Elna Cerafines (even standard Black Gates aren't too expensive). Now, with pot you're getting into high range. High PRICE range that is .

For the onesy-twosy stuff the DIYer does it's not a big deal but if you're Creek looking to make 10,000 units or more, those pennies add up really fast.

The point I was making on the other thread is that even someone who knows what they are doing could make a META42 sound like crap by cutting corners everywhere (unmatched resistors, no buffers, etc) and using a crap opamp and voltage starving it. Basically you have a crappy CMoy (not that I'm calling CMoy amps crap but in this configuration it is) on a nice PCB.

Of course, someone who knew what they were doing would have to *design* it to sound like crap. There's really no reason an amp like the Creek should ever sound like crap (I'm not saying it doesn't I'm just saying it's sloppy if it does).

 

[wilsonjm]

One of the reasons I haven't been using exotic parts for my amps is because if you want to do this typically you have to order from several web sites many of which have a minimum order, and an expensive shipping/handling charge. this makes it a lot more expensive than normal parts. if a place like digikey stocked them, then i would probably get the expensive parts.

remember that when companies produce amps, they do it a lot cheaper than we do. When I added up the cost for parts for my META42 (ordering from only Digikey and Tangent) the cost came to about $90. If I built 50 amps, the cost would be half that. 1000 amps, about $25-$30 (board cost would go down to almost nothing)