Major Issues With Building The BPA200 Reference Servo Circuit!

[Travis Clarke]

Hi guys...I'm having major problems with the BPA200 LM3886-based parallel/bridging amplifier design
which are preventing me from completing my current project: My own chip amp-based multi-channel
hi power car amplifier.

It's almost complete, but I'm having major frustration with this "seemingly simple" servo circuit from AN-1192 Nat. Semi. app. note.

Specifically the servo circuit design example is causing major
headaches with the amplifier modules and power supply (as a result) when connected to the amp modules.

Here's what I have:

1. Noninverting PA100, tested fine alone
2. Inverted-mode PA100, tests fine alone
3. Together these form the BPA200, and work fine.

HOWEVER when I add my SERVO design example, I have terrible noise like "gritty popping" that is periodic and it causes my SMPS to overheat. Also happens when the op-amps are removed from their DIP sockets in the servo board.

I've breadboarded the servo design, which is modular and connects to my amplifier modules.

I've tried everything, checked my work 4 times, but no results.

Here are my questions:

1. Any suggestions or possible starting points to check? Also, does anyone have a schematic of their
own that they have working successfully? I've emailed tlmadsen but no reply.

2. I've noticed the op-amp buffered input circuit causes a "pop" on amplifier power-up. Anyone have a remedy
or suggestions to make that shut up?

Thanks so much an attached are some images of the servo "noise" waveform I see on an oscope, and also the board itself.

Help please! [IMGHTTPDEAD]http://www.gwinnett-mobile.com/photos/100_0507.jpg[/IMGHTTPDEAD]
[IMGHTTPDEAD]http://www.gwinnett-mobile.com/photos/bpa200_servowaveform.jpg[/IMGHTTPDEAD]

 

[gootee]

Have you scoped your +/-15v power supply rails, on that board?

I looks like you have not used decoupling capacitors for each power pin on each opamp. Consider them to be required, unless they later turn out to not be.

Connect a 0.1uF X7R (or probably at least not NPO or C0G) ceramic cap from as close as possible to each supply pin of each opamp, to power ground. Both +'s and -'s power grounds should be connected together, at each opamp, and should run back to the midpoint between the regulators' output caps' grounds, separately from any other grounds.

How are the rest of the grounds connected, exactly? A schematic would be nice, and should show the actual grounding scheme, not like the National schematic where the ground-return paths are not obvious. A component/wiring layout diagram would probably also do. Label all components with type and value or part number. If you can show the amplifiers' setup, too, so much the better.

After installing the decoupling caps, if there is still any problem, then before worrying too much about anything else, I would first change the regulators to higher-capacity ones in TO-220 cases.

What types and sizes of input and output capacitors are you using for the regulators, so far? And what are the regulators' part numbers?

Since this is a separate board, each power rail should probably have an electrolytic cap to ground, where it enters the board, before the regulators. After each regulator, you should probably have a 100uF output capacitor. Those caps should connect as close as possible to the regulator output and ground pins.

You will probably also want to keep all +/gnd/- power and signal wire pairs/triples as close together as possible, probably twisted together.

 

[Travis Clarke]

Hi there. Thanks, I'll give the 0.1uF capacitors a shot, although it would not be my first guess for the root of the problem.

Unlike all my other projects this one is pretty much built literally from the AN-1192 design example, with all grounds connected together.

I'll try putting something useful down on paper. Otherwise yes I do have the 100uF after the regulators.

I'll post first chance I get with more info. Thanks!

 

[gootee]

Hi MartyM,

Thanks for responding. I hope that you'll report your progress.

The decoupling caps are pretty-much a given. i.e. Just because the National example schematic does not explicitly show them does not mean that they did not use them. They almost-certainly did use them. (I wonder if there's an 'evaluation board' available from National, for something similar, with an appnote that would show all of the details.)

The same sort of idea applies to the grounding. Schematics almost never show how it should actually be physically routed. So National's example schematic does not actually show all of the grounds connected together, even though they unfortunately used the same symbol for each ground, which is common practice. For now, we can hope that the grounding will not be too critical, in your servo board's case.

Sorry if you already know all of that. Good luck.

 

[Travis Clarke]

Hi there again. I was going to post the schematic, despite several more futile attempts at diagnosing the culprit.

Same results despite all efforts.

In leiu of posting the schematic (since it's basically literally the design example all the way), and although I hate to give up, I'm moving in another direction.

I found this site by accident:

6 x LM3886 (Parallel) Design Example WITHOUT Using Inverted Mode Amplifier

This guy is using dual non-inverted PA100s (w/ 3 amp ICs each) and instead inverting the signal to the 2nd amp, instead of the amp itself. Basically the same way "bridgeable" car amps work.

Here's why I decide to pick this method and stop trying to use the Nat. Semi. servo example circuit:

-Both amplifier modules are identical, no inverted mode configuration needed
-Lower circuit complexity & roughly the same cost
-My DC offset of my non-inverted mode amp was actually lower than that of the inverted mode! (??)
-Parts are in stock and Mouser & Digi-Key, some on eBay!

All in all, not my first choice but given the circumstances more practical solution for me....still wish I could find the problem!

Thanks for trying to help and I hope this will be useful to someone.

 

[AndrewT]

Hi Marty,
that link does use inverted mode.

 

[peranders]

Marty, hate to say it but compactness and a good pcb layout is a virtue. If you have a ratsnets, it can be virtually impossible to make it work properly. You must have a good ground and a good scheme of feeding the circuits with supply voltage.

 

[Travis Clarke]

You're right, they're inverted mode. I must have been insane to write that they're not. Arrgggh well I was really tired yesterday.

At least the principal operation doesn't change (two identical units, one with inverted signal).

But circuit layout and construction was not the issue with the servo circuit. It was obviously a functional issue.

Oh well.

 

[gootee]

But, as one of our more-knowledgable members, Peranders, was so-politely trying to convey to you, circuit layout and construction ARE functional issues. They are one of the predominant causes of problems such as the one you had.

Conductors are quite imperfect and must be considered to be RLC components. Please do some searches for 'star ground'. The problems are very real.

There are also real potential problems with radiated and conducted electromagnetic fields, which can be avoided with proper layout and construction techniques. e.g. Keep all enclosed loop areas as small as possible. Keep all conductors as short as possible. Keep low-level conductors away from higher-power stuff. Etc etc.

You should probably try re-wiring your project, using a star ground and good layout and construction techniques. Then, at least, any problems could be assumed to be 'real' problems, and attempts to diagnose or to help diagnose them could be made with more hope of finding definitive answers.

-----

BY the way, you also did not address some of the questions and suggestions that someone already spent their time to provide for you: Did you not scope the power supply rails on that board? Did you add decoupling caps to each power pin? Did you run the decoupling grounds separately back to the regulator caps' ground midpoint? Did you try TO-220-cased regulators? What are the part numbers of the regulators you are using? Did you try it with 100uF BEFORE each regulator, too? (It might also be helpful if you showed the rest of the project, i.e. the complete setup instead of just a part of it.)

Unless you report back, or post images of, exactly what has been tried, and what the result was for each suggestion, it will be much more difficult to receive effective help via any discussion group.

On the other hand, if you want to first try using further discussion to determine whether or not what has been suggested is worth your time and expense to try it, that is almost always fine, too. i.e. Asking "Why?" is good. And saying "I am not sure." is good.

BUT, go one way or the other. i.e. Either do it or else ask or argue about it. It's probably best to not make assumptions which simply discount or ignore the suggestions and questions which others have invested their time and effort to formulate, to try to help you. Of course, I realize that, given limited time, that might be 'problematical', especially if you are not yet familiar with those who are providing suggestions etc. I do not consider myself to be an expert at all of this, yet. But peranders and AndrewT should both be considered to be extremely experienced and knowledgable experts. You should consider yourself to be lucky if they take an interest in helping you, and their posts should always be carefully read. And I, at least, typically don't suggest or ask something unless I think I know that there's a pretty-good reason for it. But your asking questions instead of making assumptions will usually take care of any unreasonable suggestions or questions, in any case. That's one of the beauties of a discussion forum: There's usually someone 'looking over our shoulders'. It's like a near-real-time 'peer review' process. You just need to try to guide the discussion until you are satisifed that the process has worked.

 

[Travis Clarke]

Well I appreciate your response althought the servo circuit is a bit of a mute point now (will be using the alternative solution, see prior post).

To answer you question(s):

"BY the way, you also did not address some of the questions and suggestions that someone already spent their time to provide for you: Did you not scope the power supply rails on that board? Did you add decoupling caps to each power pin? Did you run the decoupling grounds separately back to the regulator caps' ground midpoint? Did you try TO-220-cased regulators? What are the part numbers of the regulators you are using? Did you try it with 100uF BEFORE each regulator, too? (It might also be helpful if you showed the rest of the project, i.e. the complete setup instead of just a part of it.)"

Yes, yes and yes, etc. Those are simple issues but again, it's a functional issue somewhere (because it's a cyclic problem, a clue to something making the circuit oscillate or making the op amps hit rail voltages then back again, etc.)

I believe I have a "suspect" in the problem...it's possible but not practical to spend the time to find the culprit.

I'll take some of your comments to heart and provide more background info on my next post; much like yourself I normally only post when I have something useful to contribute. It seems awful easy to accidentally leave out info/context when posting!

Thanks & have a good weekend all! I'll update when I build/try the alternative circuit (using noninverted & inverted signals to identical PA100s).