SMD Discrete Op Amp

[burdij]

I received some prototype boards from the pc fab house last week for the 2520 smd op amp. This is what the project looks like now with the parts stuffed:

Haven't done extensive testing yet, just setup in a preamp. Seems to work well, though. I am running these on +-18V. I may be able to make this available as a "semikit" with all the smd parts installed. The user would need to solder in the through-hole parts and the pins. I just need to construct a "bed of nails" tester to test the partially stuffed boards.

I have also built a rack for housing the 512-style modules. Here are a couple of shots. The rack has a built-in supply that will provide up to 150ma to each of the ten slots. The rack is constructed from a surplus eurocard rack that I had. It required extensive modification.

 

[ask]

Nicy-nice! Keep us posted on the progress. :thumb:

 

[clintrubber]

[quote author="burdij"]The rack is constructed from a surplus eurocard rack that I had. It required extensive modification.[/quote]
Schroff ? Still 3HE ? What did you change ?

Cheers,

Peter

 

[kato]

whoooooooaa, very cool. I will be interested in a few if you make "semikits" available. :thumb:

 

[burdij]

The rack brand was Transrack, a French company, I think. The main problem was with the extrusion between the two end plates. It is way too thick to accommodate the "500" module width of 4.5 inches. I would have had to mill almost an 1/8in. off of each piece in the front in order to fit the modules in vertically. Instead, I used 1/4 x 1in. aluminum bar stock for the front cross pieces and the card connector supports. The only problem with that was I had to do a lot of drilling and tapping.

 

[basscakes]

When do you think you will be offering kits for the DOA? I am interested in getting at least a few...

 

[burdij]

Its going to be at least a couple of weeks, probably. I need to build a fixture and I wanted to torture one of these to see if I can blow it up. A second consideration is that I don't have all the parts available in ROHS certified devices. In order to ship outside the US, I would need to provide a ROHS version.

 

[basscakes]

Well, I am in Chicago, so right down the street. I can buy them in any form I suppose...

 

[Baltimore]

I'm in for a bunch of semikits, depending on price.

kevin

 

[Samuel Groner]

Did you check whether the opamp is short-circuit proof (with full input signal) with the SMD parts?

Samuel

 

[ask]

How close to an API is the sound of the SMD2520? Is it designed to be a perfect clone, or more of a variation on the theme?

 

[burdij]

This implementation is based on the same circuit as below, except with a few minor value changes.

I suspect that the original circuit was not short circuit proof. I don't think the original API part was either. This is one of the things I will be testing, however. Its nice to know how things will self-destruct.

http://www.groupdiy.com/index.php?topic=4796&postdays=0&postorder=asc&highlight=2520&start=165

As far as the sound goes, from reading the extensive comments by others in that thread and a couple of others, the comparison with the original is passable. They certainly have been performing well in the preamp circuits I have tried. I can't give any specific info on distortion numbers yet, though. Certainly there will probably be some slight diffences in noise level and other factors due to the use of SMD parts. I am not matching the transistors in the second differential pair, either.

 

[Samuel Groner]

Do you refer to this schematic: http://web.comhem.se/linusandersson/Micksch2520BC_r1.JPG ?

That looks more or less short-circuit proof--Q9 might have rather high dissipation though, that's why I asked.

Samuel

 

[burdij]

Good point. I was planning on using an MMBT3906 for Q9 but with 18V on the supply and the output shorted to ground, about 300 ma could flow through that transistor. That exceeds the transistors rating by 50%. Instead, I should probably use an MMBT4403 there. Its max current is 600 ma.

 

[Samuel Groner]

300 mA, are you sure? Will need to check the details but for similar designs I had values closer to 30 mA. In any case, even that would be high for an SMD part due to its dissipation limit.

Samuel

 

[burdij]

I'm a little off. I neglected to take into account the gain of the output transistor. I get:

((18V - .35V)/8)/15 = 147 Ma.

as the output transistor base current. This doesn't account for the Q9 base current flowing into the previous stage, either.

That would cause the transistor Q9 to dissipate:

.45V * .147 = .066 W

(assuming that it is also completely saturated)

SOT23 packages are rated at 350 mW so it looks like there would be some margin.

 

[Samuel Groner]

If I understand your writing correctly you're saying that Q7 hase ~150 mA base current? That seems impossible as this would imply a collector current of ~15 A which again is--fortunately--not feasible as it would cause a 84 V voltage drop across R11.

So far I found short-circuit calculations to be too cumbersome to do by hand so I used SPICE with good accuracy. Perhaps I find some time tomorrow to look at it.

Samuel

 

[burdij]

I got the gain of 15 from the datasheet for an MJE243. The gain normally is higher than that but it derates as the current goes up. This was an estimate from a graph. I also have 8 Ohm resistors for R11 as this gives a quiescent current for the entire device more in line with an actual 2520. The consequence of that is that the short circuit current is around 2.2 A. Obviously, in a real world situation, even that would be impossible since the average power supply wouldn't be able to supply current at that level for very long and traces, etc. would start to get melty.

I am interested, though, is seeing your results from a simulation. What I am doing is "seat of the pants" estimates.

 

[Samuel Groner]

For +/-15 V rails we have peak collector currents and dissipations of:

* Q5: 20 mA, 210 mW (on negative swing)
* Q7: 290 mA, 3.8 W (on positive swing)
* Q8: 300 mA, 3.9 W (on negative swing)
* Q9: 18 mA, 210 mW (on positive swing)

The other parts are uncritical. That's a realistic result I'm used to see.

It means that for a continuous short with DC input Q7 or Q8 will probably die (as with most DOAs). With an AC signal (or not complete shorts) there might be a chance of survival as the dissipation is about halfed.

Q5 and Q9 are safe, though the SMD package will get hot.

Samuel

 

[burdij]

Yes, thanks for the analysis. The outputs will get very hot as this will exceed their dissipation rating by a factor of two. The drivers should be okay as they are rated at 350mw. This should at least give users the chance to shut the thing down before the outputs exceed the failure temp of 150 deg. C. in case of a short.