[SDA]

Searches here and on Headwize revealed nothing relevant, so I decided to start a thread to ask for help with a problem I'm having with a MINT. If I just suck at using the search function, feel free to tell me so.

Now then, the problem. Everything sounds good (or as good as things get with my $2 test headphones) for a few seconds, then all of a sudden things get nasty, crunchy, and distorted. At first I thought that the NiMH battery I was using for tests was just dying and caused the opamp to clip, but when I left my multimeter's leads on the voltage test points the voltage stayed at around 9.5v (it was just topped off) through the Point of Ugliness. OK, so it dropped by a couple millivolts, but I don't think that would be an issue. Besides, when I tried it with an alkaline I had on hand it did the exact same thing.

Then I decided to test DC offset. I didn't know how to test much else, and I figured I was bound to happen on a useful measurement eventually. I did: DC offset on the left channel stayed at 0-1mV for a few seconds, then suddenly started to rise (quickly, this can't have been happening over more than half a second), then went all over the map (28mV to negative 10mV, moved around too quickly for my DMM to give me more than readings here and there) at the exact same time that things got ugly. The same thing happened with the right channel.

I've cleaned the board, checked my connections, checked voltages... what could be the problem here?

 

[Pars]

We probably need a little more info to help you out (pics would be great also).

What amp is this? A MINT built from the PCB available from Tangent? A CMoy (some people call them mints cuz theyre usually put into mint tins)?

What opamps, etc.? It kind of sounds like it is going into oscillation. You might switch your DMM to mA mode, and connect it in series with one of the battery leads. Does the current draw go up alot when it goes into its "nastiness"? Anything getting hot (opamp(s))? Also check the DC volts (to ground) for the V+ and V- pins on the opamp(s). Are they about even (and 1/2 the supply, so for example, if you are running a single 9V, both at 4.5V, one + the other -?

Chris

 

[SDA]

Sure, sorry for not providing enough to begin with. I used to work as a tech support, yet I still forget the importance of information

It's a MINT built on the PCB from tangent, yeah. It's using an AD8620 and mostly unremarkable parts (C1: .27uF BC MKP-416s, C2: 470uF 16V Panasonic FCs, etc.).

Testing current consumption does reveal something interesting/weird. For the five or ten seconds that all is well, current consumption stays around 14-15mA (which strikes me as high by a couple mA, especially since it displays -1mA when the leads aren't connected to anything), but when things go bad current consumption drops into the single-digit range pretty quickly. (Slow enough that I can read it go down almost digit by digit, quick enough that it's over in a second or two.) Nothing seems to be getting hot, though.

Is signal ground (I think it's that, it's the pad labelled "G" just to the left of input ground in this) the same thing as power supply ground? If so, I think you might have gotten it: voltage between + and ground starts a little below 5V as soon as power is connected and keeps rising until it hits 8V, voltage between - and ground starts around the same place and keeps falling until it hits 1V. I have no idea what this means, but it doesn't seem like it should be happening at all.

 

[MisterX]

It sounds like you have one the PSU caps in backwards....can you post a pict?

 

[Syzygies]

I'm a newbie myself, but in the spirit of tech support, gotta check: Do you know what "virtual ground" is? Op amps would love a true dual supply with V+, V- balanced around a zero volts ground. Can't be done with batteries, but op amps just care about differences, so we make V+ 9 volts, V- 0 volts, and ground halfway in between at 4.5 volts. The CMoy uses a pair of resistors as a voltage divider to accomplish this, the MINT uses a "rail splitter" designed for this exact, very common purpose.

If you imagined that the op amp ground was like any other ground, this could explain your issues. But a better question: You'd have to go off the ranch to get into this kind of trouble, what did you do that was unusual, not in Tangent's directions? Like finding a computer bug, you actually already know what you did, it struck you as odd when you did it, you just have to remember. Stare at the swinging watch, you're getting very sleepy...

Edit: I like MisterX's explanation better, it fits the timing you're seeing.

 

[SDA]

No pictures until I get a functional camera, sorry

The PSU caps seem to be in the right way.. the "negative" stripes match with the negative poles, and the caps look exactly as they do in this picture. (They're FCs too, so that reference should at least be worth something.)

I know virtual grounds aren't perfect, but surely a TLE2426-based virtual ground circuit should be at least pretty close to half-and-half, right? Closer than having one rail just 1V from ground with a 9.5V supply, anyway.

There is one thing that I did differently from the docs on tangent's site: I messed with the Jung multiloop resistor values a little, because I didn't have a couple of the resistor values in the MINT schematic on hand. The values I used, which read pretty much like those in the PIMETA schematic, were R3=1K, R4=3.32K (wanted a pretty low gain), R5=3.32K, R6=1M. Could those values cause problems with the MINT?

I'm going to play with the amp a little more and see what else I can dig up... thanks to everyone who's helped so far!

 

[Earwax]

Quote:

Originally Posted by SDA No pictures until I get a functional camera, sorry The PSU caps seem to be in the right way.. the "negative" stripes match with the negative poles, and the caps look exactly as they do in this picture. (They're FCs too, so that reference should at least be worth something.) I know virtual grounds aren't perfect, but surely a TLE2426-based virtual ground circuit should be at least pretty close to half-and-half, right? Closer than having one rail just 1V from ground, anyway. There is one thing that I did differently from the docs on tangent's site: I messed with the Jung multiloop resistor values a little, because I didn't have a couple of the resistor values in the MINT schematic on hand. The values I used, which read pretty much like those in the PIMETA schematic, were R3=1K, R4=3.32K (wanted a pretty low gain), R5=3.32K, R6=1M. Could those values cause problems with the MINT? I'm going to play with the amp a little more and see what else I can dig up... thanks to everyone who's helped so far!

I don't think changing resistor values can cause the type of problem you're seeing. Something is out of whack in the power supply area. If the caps aren't reversed it's possible that the TLE is busted. But, this symptom seems familar. I think someone recently posted about an amp with widely unbalanced power rails and it turned out that the power supply was shorted to one of the signal lines somewhere.

What type of case are you using? Battery power or wall wart?

 

[tangent]

Sounds like something's connected to something it oughtn't be, through a somewhat high resistance. That's why it works for a while, and then fails as more current goes through the fault path. I had similar things happen to me on a CMoy when using vulcanized rubber washers to try and isolate the I/O jacks from ground. (I realized later that vucanized rubber has carbon in it, which is...you guessed it...conductive.)

Measure resistance between virtual ground and all of the op-amps' pins. Do all of the readings make sense? Do the same between the buffer outputs and vground.