LME49720NA for ART Digital MPA?

antichef

pornk rock
I ordered some LME49720NA opamps so I could use one in a little preamp that I got from mrbowes. I had tried the OPA2134 in it and liked it, but I also like the LME49720NA, because it has a "faster" sound to it.

Any reason not to try them out in the ART Digital MPA? There are a *bunch* of OPA2134 in the unit, and I'm sure they're not all directly related to signal amplification, but I guess I could try them one by one and see what happens.
 
Yes, there are a few reasons! First off, unless I have badly erred, the National chip is BJT input, whereas OPA2134 is FET input. The MPA uses only one 2134 per channel, and it's used for the instrument input. That means it needs to be high input impedance, which means you need a FET input opamp. The LME's input impedance is listed as 30K, vs. 10^13 for 2134.

Reason #2, I said that OPA2134 is used for the instrument input, and that's true but not completely true--the mic signal is also routed through the 2134, but only after a discrete stage and a stage of good ol' 5532. Why is this done? Because a mic doesn't need the high input impedance of a FET, and BJTs are generally quieter than FETs, so it's done to get the MPA's nice and low input noise spec. Going straight into OPA2134 generally gets you something on the order of -120dBA, vs. the -134dBA spec (or so, going off memory here) of the MPA.

OK, moral of that story is that before the 2134, there are two other stages beforehand, although I think the 5532 is unity gain. After the 2134 comes the tube, then a couple more 5532s. There is gain at the first discrete stage, the 2134 (responsible for most of the input stage gain), the tube (supposed to be +20dB, but the +20dB switch is actually a pad going into the tube, and there are way too many components around the tube for me to calculate the tube gain), and the final 5532 which adds a hair of gain--10dB I think.

Whew, that's a lot. It's not too likely that swapping anything for 2134 will make much difference, because the signal level is high enough that you won't gain much in noise performance, the slew rate is the same, and of course there's that input impedance problem.

What's more interesting would be to replace the 5532s with the LME chip, especially at the output, where the slew rate might be able to help a bit. Or even at the input, for improved noise performance there (although again, I don't think that will matter too much). But the 5532s are soldered in, so they aren't as easy to swap . . .

#1 on my list of MPA improvements is to rewire the output level knob so it comes right after the tube, rather than after two opamps after the tube. That should allow hitting the tube a lot harder without clipping the following opamps. But it's a complicated circuit, and somewhat involved surgery, so I need to study that a bit more first.
 
interesting -- it was actually a NE5532 that was in the Gaines preamp (that I got from mrbowes) originally, and I replaced it first with a OPA2134, and then with the LME49720NA -- too bad those are soldered in (and presumably SMT? I'll have a look - if not, I can socket'em - I've gotten handy at that.).
 
interesting -- it was actually a NE5532 that was in the Gaines preamp (that I got from mrbowes) originally, and I replaced it first with a OPA2134, and then with the LME49720NA -- too bad those are soldered in (and presumably SMT? I'll have a look - if not, I can socket'em - I've gotten handy at that.).

I don't think there is any SMT on the MPA, except for the digital section of the DMPA. You will have to pull the main board though (except for the input stage 5532, which is on the easily accessed gain pot board), and its screws are glued down :mad:

I never socket anything myself . . .
 
I never socket anything myself . . .
That's because you know what you're doing :D

Rats - bricked it :mad: I destroyed a bunch of 5532s on their way out of the board, and perhaps the high heat of my iron internally damaged even the ones that made it out with all 8 pins. Anyway, the LME49720NA ones didn't work *at all* (I socketed and replaced all 8 5532s) - I just got hiss in one channel and bizarre motorboating in the other when the tube was engaged (and silence when it wasn't). I tried putting back the 5532s without luck - a little mic signal and a lot of noise in one channel, and nothing in the other :( . I'll get some more. Visually anyway, I did a nice job with the sockets, so hopefully all's well except the opamps.

Good thing those group buy pres are working :D -- just at the moment, I'm feeling like I should return to my non-career as a musician :)
 
I guess it's closing the barn door after the cows are gone, but never try to desolder 5532s. I mean, they are what, $0.50 each? Cut them out, then remove the solder from the holes. Much safer than trying to save a bunch of jellybean chips.

You probably destroyed a trace somewhere; you can try to find it, although if these are four layer boards (I don't think the mini-gain boards are; I've never pulled the main board), you could have broken a power or ground plane connection which you can't see.

Start with (power off) measuring continuity between all pins 4 and -V and pins 8 and +V, with the OPA2134 sockets as your reference. If you find no problems there, move on to the signal pins and try to trace where they go. If you find a problem, run a jumper from another point that is still good.

The motorboating could be caused by oscillation, even if you cut an input trace somewhere so there is no signal. It's not a huge surprise with a faster chip. You should add bypass capacitors to all chips you are replacing.
 
I'll give it a shot - I suppose I could also test that way with my new sockets?

I also contacted ART about getting a replacement board or even :o sending it to them.

Multi layer boards -- the bane of newbs!! :D
 
answer #1 from ART - no replacement boards! I'll find out about service next.

I'm starting to get my courage back up regarding trying to fix it, and I got a bunch of crisp new 5532s, so I can restore the stock circuit -- I have one of those "pens" that is full of conductive material - are those worth considering?

I recognize that if it's a multi-layer board, and the trace I messed up is on the inside, that I have some challenges.
 
What's more interesting would be to replace the 5532s with the LME chip, especially at the output, where the slew rate might be able to help a bit. Or even at the input, for improved noise performance there (although again, I don't think that will matter too much). But the 5532s are soldered in, so they aren't as easy to swap . .

yes thats best, but there are more stable chips. The motorboating is oscilations, if you socket chips you really need to use machined sockets the flat pin ones are bad for oscilations and moaterboating and usually need decoupling caps across power an ground

Machined-contact sockets provide lower contact resistance and greater
retention.

fa93_1.JPG
 
I did use machined sockets (not because I knew what I was doing, but because that's what they were selling at the electronics store)

anyway -- wish me luck :D
 
And the other -- whew!!! I just learned some stuff about ghetto circuit board repair that I wish I didn't have to know. I think at least one of the analog outs is still hosed, but I don't use those :)

Anyway - my findings... all references to circuit board locations are for the Digital MPA, and may not translate to the MPA Gold.

The short story / low hanging fruit: replacing the already socketed OPA2134 chips in U202 and U203 with LME49720NA increased gain some and made the unit sound a little "faster" with more noticeable detail. That could be good or bad (more gain probably isn't ever bad), depending on what you like about the sound.

The long story: Thanks to my stupidity, I'm able to swap out any opamp in the unit - there are a total of 8 5532s and 4 OPA2134s (half of each on each side). I'm sure the circuits are identical, but the board layout is not exactly symmetrical from side to side - no doubt because circuitry that is common to both channels gets in the way - no biggie, it's close enough to follow (heh, as if I could follow a complicated circuit like that :) ).

There are 3 5532s for each channel on the main board, and 1 on each of the smaller boards to which the gain and impedance pots are attached. These last two are U200 and U201 -- I didn't notice any change when replacing those two - maybe they're related to analog out. I left them as 5532.

I replaced all 6 total 5532s on the main board for LME49720NA, and did notice a small improvement - a little less noise and maybe some more headroom -- but the improvement is DEFINITELY NOT WORTH THE RISK OF BRICKING THE UNIT -- I found out that ART does not sell replacement circuit boards - you bust it, you're on your own.

There are two remaining OP2134s - U420 and U421 - I tried replacing U421, and got no signal through the channel when I did -- I don't really understand this -- maybe a bad opamp? I don't think so, because I think I used it at another position. I put back the OPA2134 and everything was happy again. I didn't try to do the same thing with U420.

One last tip - when you're putting it back together, there's a long ribbon cable that goes from the left channel side of the main board way over to the output board on the back -- make sure you run this cable around the front side of the tube, instead of the back side, where it gets too close to the power transformer and picks up some hum.

edit: an old quote from Jim Williams on another board on this very subject:

"Audio band parts like OPA2134, AD 8599 are usually safe, the LM4562/LME49720 is a 55 mhz bandwidth part, watch out." -- jeez, it really sucks being ingorant -- anyway, I'm liking the change for now, and if I notice anything wacky, it's easy enough to switch back.
 
Last edited:
The short story / low hanging fruit: replacing the already socketed OPA2134 chips in U202 and U203 with LME49720NA increased gain some and made the unit sound a little "faster" with more noticeable detail. That could be good or bad (more gain probably isn't ever bad), depending on what you like about the sound.

I might try this at some point if I manage to pick up some freebie LME49720NA opamps. Did you test for oscillation with a scope?
 
I might try this at some point if I manage to pick up some freebie LME49720NA opamps. Did you test for oscillation with a scope?
I did not -- I have no scope, except for what I may be able to pull off with my computer/interface (which will chop off the really HF anyway).

I dunno - I left it on all night, and this morning, it still sounds good :D

But that's the right approach (not to stress on it, and just to try it if convenient) -- the real action with these things is changing the tubes.
 
that's pretty funny :D It would be even funnier if it didn't hit so close to home :D
 
Last edited:
Back
Top