Pimeta slip-up, help needed.
Mar 11, 2006 at 7:48 AM Thread Starter Post #1 of 11

SubRosa

New Head-Fier
Joined
Mar 8, 2006
Posts
47
Likes
0
Hello,

Tonight I began construction on my Pimeta and everything was going fine until I found that I am missing C6 (NP0 Ceramic Capacitor), I know I -had- one but somehow it has disappeared, these things happen. I do not want to place another order for a single part so I have been scrounging through broken (or now broken) electronics for a suitable capacitor. I've found quite a few 0.1" pin pitch monolythic ceramic capacitors but it seems to be impossible to find one of a value between 10-100pf (which I've read are the accepted values). The smallest I've found is blue and numbered 473, I'm not an electronics guru nor complete noob so I've only guessed that this is 47000 -something-, whether it be picofarad or microfarad (though I'm sure it is not uf) this value is much too high for my purposes, or so I've read. I've also found yellow capacitors labelled 104. What possible solution do you see outside of ordering? Will the capacitors I've described be "alright".

I'd also like to ask how important hand matching resistors is, I currently do not have a dmm (broken) and do not want to buy a new one (broke). I assume matching them is to help balance the left and right channels but am I really going to notice if I do not match them? I know I'll need a dmm to check offset and voltages when everything is complete anyways but let's just ignore that fact.

Thank you for your help.
 
Mar 11, 2006 at 5:52 PM Post #3 of 11

tangent

Top Mall-Fi poster. The T in META42.
Formerly with Tangentsoft Parts Store
Joined
Sep 27, 2001
Posts
5,969
Likes
56
Quote:

Originally Posted by SubRosa
I assume matching them is to help balance the left and right channels


No, it's a mixture of one part anal audiophile exercise, and one part good practice to ensure you put the right values in the board. If you're not paying attention to matching, it's easy to put, say, a 1K resistor in where a 10K should be, if you don't have your parts binned neatly. And even if you do have your parts binned, very occasionally you'll come across manufacturing errors, or you'll grab a resistor you used temporarily once and put back in the wrong bin. Double checking the resistor value with a meter helps you avoid these problems.

Quote:

am I really going to notice if I do not match them?


If all of your resistors are correctly labeled and within their rated tolerance, and you use the same nominal value in both channels, then no, you probably won't notice.

With 1% resistors, the worst-case error stack-up (considering just R3 and R4) for a gain of 11 is almost exactly 3 dB. 3 dB is readily noticeable, but in the real world, you will probably end up with a lower gain (which lowers the maximum error spread), and you won't experience the worst case error stack-up.
 
Mar 21, 2006 at 1:09 AM Post #4 of 11

SubRosa

New Head-Fier
Joined
Mar 8, 2006
Posts
47
Likes
0
Hello,

I decided to bite the bullet on C6 and ordered it and some other things, like a serpac h-67, kilo knob, etc. So I've completed my pimeta and miraculously it works! This is my first electronics project in a while. I used OPA551s as buffers instead of BUF634s, Elna Cerafines, OPA227 in ground, tangents suggested input caps (will be using a lot of sources), Wima Caps, Vishay Dales, and RK09 pot. I do, however, have a problem. I bought both OPA2107 and two AD843s and with the 2107 installed the pimeta works beautifully, absolutely silent with no source and music sounds great. With the AD843s on a socketed browndog I get an ungodly squeal/hiss with no source that sort of ramps up as the amp is powered on, I assume this means they are oscillating? I'm powering my pimeta with 2 fresh and tested 9V batteries. All flux has been cleaned off the board and browndog. Any ideas as to what could be wrong and how to fix it?

Also, for those wondering I created a perfect fit for my pimeta with cerafines in the serpac h-67 AC by carefully carving out the rounded corners of the serpac with my dremel, it fits beautifully with no apparent pressure on the caps.

BTW, my amps gain is roughly 2.5

EDIT: I tried raising the gain to 5 and it helped but didn't get rid of the squeal, so I raised it to 11 and the squeal is gone now, the only problem is this will be driving grado phones so...
 
Mar 21, 2006 at 3:23 AM Post #5 of 11

Xakepa

500+ Head-Fier
Joined
Sep 2, 2005
Posts
751
Likes
0
Measure quiescent mAs - 843s sqeeze 13mA each AFAIR, 3-5mA for 227 (maybe less, don't remember), 1.5mA for TLE...if you're more than 2-3mA off 843s are probably oscillating...2.5 gain is lowish.

There's a couple of fast and dirty ways to fix it...like soldering small 0.xx uF caps across the power pins, it might work
 
Mar 21, 2006 at 1:40 PM Post #6 of 11

tangent

Top Mall-Fi poster. The T in META42.
Formerly with Tangentsoft Parts Store
Joined
Sep 27, 2001
Posts
5,969
Likes
56
You really shouldn't draw more than about 30 mA from alkaline batteries. Three 843s will draw nearly 40 mA all by themselves.

The main problem with drawing more than 30 mA from alkalines is that their life starts going down nonlinearly w.r.t. load after that. In this case, the use of alkalines may be creating a secondary problem: the 843s require a pretty stiff power supply in order to come up reliably. The overload on the alkalines may be enough to cause the supply voltage to rise slowly enough to annoy the 843s.

If I'm right about the alkalines being the culprit, you can try two things: 1) swap out just the ground channel op-amp. This won't bring you below 30 mA, but it should at least help. 2) Try NiMH rechargeables. They're capable of several hundred mA of output before their life starts going down nonlinearly.

P.S. Please don't test first with headphones! I'll bet DC offset is high with the 843s in this configuration, too, so you would have found that in a proper test.
 
Mar 21, 2006 at 1:42 PM Post #7 of 11

tangent

Top Mall-Fi poster. The T in META42.
Formerly with Tangentsoft Parts Store
Joined
Sep 27, 2001
Posts
5,969
Likes
56
I just thought up an even more likely explanation: a buffer should be faster than the chip it is buffering, or you risk instability. The OPA551s are only 3 MHz, while the 843s are 34 MHz!
 
Mar 21, 2006 at 4:39 PM Post #8 of 11

Xakepa

500+ Head-Fier
Joined
Sep 2, 2005
Posts
751
Likes
0
Quote:

Originally Posted by tangent
I just thought up an even more likely explanation: a buffer should be faster than the chip it is buffering, or you risk instability. The OPA551s are only 3 MHz, while the 843s are 34 MHz!


Wow...that's a very good point!
 
Mar 21, 2006 at 5:25 PM Post #9 of 11

kwsuh

New Head-Fier
Joined
Jan 15, 2006
Posts
5
Likes
0
The Ground channel is prone to be unstable with unity gain OPA551 buffer and OPA227. I've seen a lot of oscillation with that configuration everytime with my PIMETA. On the other hand, OPA551 with OPA2107 in L/R channel worked very well with the gain of 11.
 
Mar 22, 2006 at 12:09 AM Post #10 of 11

SubRosa

New Head-Fier
Joined
Mar 8, 2006
Posts
47
Likes
0
Hello,

Quote:

Originally Posted by Xakepa
There's a couple of fast and dirty ways to fix it...like soldering small 0.xx uF caps across the power pins, it might work


Is there any negative effect with this solution? Do I need to bypass both opamps separately or just on the browndog adapter?

Quote:

Originally Posted by tangent
If I'm right about the alkalines being the culprit, you can try two things: 1) swap out just the ground channel op-amp. This won't bring you below 30 mA, but it should at least help.

P.S. Please don't test first with headphones! I'll bet DC offset is high with the 843s in this configuration, too, so you would have found that in a proper test.



I'm already using a 227 in the ground and I know I shouldn't test with headphones but I'm using some horrible ear buds anyways, not my main phones.

I'm planning on buying some NiMH rechargeables soon, these alkalines were just handy.

Quote:

Originally Posted by tangent
I just thought up an even more likely explanation: a buffer should be faster than the chip it is buffering, or you risk instability. The OPA551s are only 3 MHz, while the 843s are 34 MHz!


I also thought of this, just as a guess, but I'm no expert so I was unsure. I also have some Intersil HA-5002s in DIP8 that I, with some careful soldering and lead bending, adapted to the BUF634 pin out (without current limiting resistors, +/-V1 and +/-V2 [No, I didn't solder the positive leads to the negatives] are tied together without a resistor, I know this isn't ideal for the sake of the chips but is it acceptable? I'll be careful.) I will try these if all else fails.

Quote:

Originally Posted by kwsuh
The Ground channel is prone to be unstable with unity gain OPA551 buffer and OPA227. I've seen a lot of oscillation with that configuration everytime with my PIMETA. On the other hand, OPA551 with OPA2107 in L/R channel worked very well with the gain of 11.


Thanks for the heads up. I will try OPA277 or another AD843 (I have 4, I just used 227 for ground for lower current draw). With OPA2107 everything sounds perfect at a gain of even 2 with 227 in ground.

Thanks.

EDIT: Well I tried an OPA277 and while it didn't get rid of the squeal completely it really helped with a gain of 2, I think I'll move up to 4/5 and see if it's gone. Putting an AD843 in ground made the squeal about 3X worse.

EDIT 2: I decided to skip trying to up the gain and shoved my frankensil 5002s in and voila, squeal is completely gone at gain of 2 but one of my "creations" is a little fragile so I was wondering if putting an opa551 in the ground buffer along side the OPA277 would be alright?
 
Mar 22, 2006 at 8:41 AM Post #11 of 11

SubRosa

New Head-Fier
Joined
Mar 8, 2006
Posts
47
Likes
0
Sorry to double post, here is a diagram of my 5002 to BUF634 conversion, I hope you can understand what I have done.

attachment.php


All the connections are made under the chip, except 6 to 8, if that wasn't obvious.
 

Users who are viewing this thread

Top