[dan1son]

I have an amp I built a little while ago and I just got some parts to try it with some other opamps. It started out with an OPA2132 that has been in it since I built it. I threw in an AD8620 and it worked fine, chip got a little warm as expected but not hot. Then I tried to put an AD8066 in the same circuit and it gets quite hot within 10 seconds or so of the amp being powered on. The DC offset is very small, .3mv and 1.2mv, and it outputs what seems to be clean sound (I'm not going to leave it on very long since it gets so hot).

I'm curious if it is normal for this opamp to oscillate in circuits other opamps don't seem to mind.

 

[PinkFloyd]

I'm trying the 8067 and it really needs a gain greater than 8 to prevent it oscillating.. try upping the gain to around 10 and it should work fine

EDIT: Sorry I misread that...... I thought you were using the 8067

8066 doesn't have a propensity to oscillate or at least the 8065 has never oscillated in my WNA..... as long as the gain is greater than 5 it should be fine.

 

[dan1son]

Darn, I don't have any spare parts here. I just moved and decided against bringing the few parts I actually had. I think my gain is around 5 since I built it primarily for MS-1s.

Do the OPA627, AD843, or AD845 need a larger gain as well? Or should they be ok with 5?

 

[PinkFloyd]

See my edit, I was referring to the wrong opamp

 

[dan1son]

That's strange then... I did notice this opamp is a whole lot faster than any of the other ones, 145mhz. Would that make it more apt to oscillate somehow?

 

[dan1son]

Bump...

I'm still interested in any information as to why this chip won't work in the circuit. Every other chip I currently have in my possession does. NE5532, AD8620, OPA3132... a few more chips are in the mail, OPA627, AD843, and AD845 and now I'm kind of worried some might not work.

The amp is a buffered multi-loop design with dual buf634s per channel. My gain is around 5.

Thanks.

 

[bg4533]

Try putting a decoupling cap from the opamp V+ and V- pins to the power rails if you dont already have one. High speed opamps tend to oscillate more than slower ones. The 8066 has always been stable for me. Are you using an adaptor with the chip or is it mounted directly to a PCB?

 

[dan1son]

It's on a browndog adapter from tangent... and there is no decoupling cap. Not sure what type or what those do.

 

[tangent]

Quote:

Originally Posted by dan1son Not sure what type

Ceramic or film. 0.01uF to 0.1uF are the most widely useful values. Smaller is better, as a rule.

Quote:

or what those do.

It's complex. (Literally! -- the equations involve complex numbers.)

One way to think about bypassing capacitors is that they provide a low-impedance path to ground for noise on the power rails. A capacitor has a falling impedance vs. frequency, so the higher the frequency, the lower the impedance. So for pure DC, the capacitor effectively is not there. As frequency goes up, the tendency for the noise on the rails to be shunted across the cap goes up.

Another way to think about bypassing caps first requires that you ask why the noise is there on the rails in the first place. And that is because the impedance of your power supply is not zero. If it were, no amount of current you could pull from it would cause its voltage to drop; it would be a perfect voltage source. Since that's not practical in reality, the varying load current of your amp translates into varying supply voltage. Putting a capacitor across an impedance lowers the series impedance, because the capacitor acts as a kind of storage reservoir for power.

This is only part of the story. But you begin to get an idea of why bypassing is good, which is the main goal here.

By the way, a better plan than running a single bypass cap from V+ to V- is to use two caps, one from each rail to ground. Put the caps as close to the power pins of the op-amp as you can. Let the path to ground be long and snaky if you must to make this happen.

Using just one cap only shunts noise from one rail to the other. It's not totally useless, but it isn't as good as proper bypassing.

 

[dan1son]

Ok, so do other amps like the pimeta or mint have problems with this chip in the same way? Basically if I used this same power supply on one of those amps I'd be having the same issues with this high speed opamp? Or is something else causing it as well?

I understand adding bypass caps will stop the problems, I'm just curious why there isn't spots for those on any major board I've seen, or why other amps don't have this same problem with this chip.

Complex numbers was definitely NOT one of my favorite courses in college...

 

[tangent]

Quote:

Originally Posted by dan1son Ok, so do other amps like the pimeta or mint have problems with this chip in the same way?

I believe I've seen positive reports of the 8065/66 combo in a PIMETA. I don't believe I've ever seen anyone report on the 8066 in a MINT.

Quote:

if I used this same power supply on one of those amps I'd be having the same issues with this high speed opamp?

I just skimmed the thread, and I can see no other mention of your power supply, so I don't see why you're bringing it up now as the culprit. Perhaps you're thinking my "nonzero PSU impedance" reference was some specific criticism? It wasn't: it's physically impossible to have a power supply with zero output impedance. The problems resulting from that affect all power supplies, to some extent.

Quote:

I'm just curious why there isn't spots for those on any major board I've seen

Speaking only for the PCBs I've been involved with, it's primarily because we tested with a specific set of chips, and none of them required bypass caps. I believe only the PPA was ever tested with the 8065s during development.

Quote:

why other amps don't have this same problem with this chip.

Layout. It's really that simple. A good layout will let a chip work where a poor layout would not. Bypass caps can push a poor layout over into the good layout camp.

 

[dan1son]

Sounds good... I appreciate the information. Somehow I got the PSU being a possible culprit (I wasn't taken it as criticism at all I was just wondering if one PSU would be better in the same spot as another). Do to the fact that I am completely ignorant on PCB layouts I am sure my layout is no where near the requirements needed for a picky opamp.

I really hope this amp works with the 627s I have on the way as well as the 843 and 845s. The 8620 is far too bright with my HF-1s. Sounds great for mellow recordings, but anything else it gets quite exhaustive...

 

[cetoole]

Quote:

Originally Posted by tangent I believe I've seen positive reports of the 8065/66 combo in a PIMETA. I don't believe I've ever seen anyone report on the 8066 in a MINT.

I am currently running a pair of AD8065 on a browndog in my Pimeta, no problems beyond what I have always had, the two day oscillation issue, but I did tack on some small ceramics bypassing the opamps to ground on the underside of the board, soldered directly to the DIP socket before I rolled these opamps in.