Chip amp limiter diodes, how do they work? Specific example

[Morocotopo]

Hi guys, I´ve been reading about limiting the signal just before it enters a chip amp (LM3886, LM1875, etc.), I mean, between the preamp out and the power amp in, to avoid ugly distortion and/or internal thermal protection (that sounds AWFUL!). The simplest way is a couple of antiparallel diodes to ground, a simple clipper. But someone in a thread mentioned the Vox Busker regarding this subject, so I looked into that. This is the corresponding schem part:



I don´t quite understand how this works... more complicated than a simple pair of diodes to ground. The 680 ohm R I guess  makes the forward voltage higher, but I don´t get why the diode pair has a different gain set R... Could someone care to explain a bit?

Also R.G. mentioned somewhere that the Vox Royal Guardsman also has a diode limiter pair, looked into that, but that is even more difficult for me to understand because it involves DC bias, so one step at a time!

Thanks

[brett]

Hi
I'm no expert, so I'll probably miss some of the detail (and maybe make a mistake or two). 

But some of the obvious things I see here include a 680/15k voltage divider in the diode feedback loop.  This reduces the voltage difference across the diodes, and therefore increase the voltage "threshhold" for conduction via the diodes (to Vf(tiny current) x 15680 / 680). 

Also, the maxiumum rate of change in conduction through the diodes is very low compared with circuits like the tubescreamer, due to both the effect of the voltage divider, and also the significant resistance in series with the diodes (15k).  This means that when the input resistance is low (master volume turned up), the clipping should be quite soft, as there is still quite a bit of gain even though the diodes are conducting.

It is certainly an interesting circuit.
cheers

[R.G.]

I personally don't like to put any funny stuff in the feedback path of a power amp chip. I know the theory says that it's fine, but I've seen lots of fried output transistors on power amps too.

I prefer to compute what the largest input voltage is which will still keep the power amp from clipping, then limit before the power amp to that level. There is then no question about whether the limiting will affect the power amp's stability.

Some designers are not so conservative - obviously.

[Morocotopo]

Thanks for the answers. R.G., how would you approach limiting before the power amp? Simplest is antiparallel diodes... how to make the onset of limiting gradual? Maybe a string of antiparallel Ge diodes, since I think those have the smoothest transition into conduction... or was it LED´s? Anyway, the idea is to keep it simple. I guess the next step would be active limiting...

[brett]

Hi

Maybe a string of antiparallel Ge diodes, since I think those have the smoothest transition into conduction

Yes, Ge diodes have about twice the "knee" of Si diodes.  You could add resistance to a single diode so that diode conduction is limited. e.g. for an amp with 10k input resistance, you might use 1 or 2 k in series with the diodes.

Even better would be a full compressor, like the Dyna.
cheers

[Morocotopo]

Thanks Brett for the clarification. Yes, another option would be a compressor like the Dyna, but there you have to contend with attack, noise, recovery speed... another set of variables, and a diferent "feel". More parts.

[Morocotopo]

By the way: what do you think about breadboarding chip amps? Not practical? I mean, because of possibility of oscillations, high currents, parasitic capacitance, layout... ´cause I´m thinking about trying some of the ideas presented here...

[R.G.]

By the way: what do you think about breadboarding chip amps? Not practical? I mean, because of possibility of oscillations, high currents, parasitic capacitance, layout... ´cause I´m thinking about trying some of the ideas presented here...

The smaller the chip amp, the easier it is to get away with less-than-perfect setups, as a rule. Many of the effects that bite you in prototypeing are current related, and the lower the power, the less current involved. At least that's my personal rule of thumb.

Introspectively, it may be that I like pre-power-amp mods better than in-the-power-amp mods because I can breadboard without mucking with the high current ends. Don't know.

In any case, I wouldn't feel too bad about messing with limiters in the feedback of chip amps up to about LM1875 (very good chip amp!) or TDA2030 size. These are about 20W, but they are *very* difficult to insert into a breadboard because the leads are too big. You'll have to solder on dummy leads of smaller stuff or ruin some contacts in the breadboard. They also *need* heat sinks or they'll either self destruct or self limit in the case of the LM1875 and make it very confusing to work with.

It's easier to take a working perf layout and modify only the parts that are involved with the feedback network or input network, which is where most of this happens anyway. That leaves the power connections and bypassing alone.

[petemoore]

  The power supply is a little more involved, but you could just use 2 x 9V batteries for a while.
  BBDing isn't that much different than any other method of making connections, except it tends to be fidgetty and prone to noise.
  The other main difference and reason why BBDing isn't as common with amps is because of the increased potential, as always, make certain you and your components are rated to withstand the voltage applied.

[teemuk]

The other thing with breadboards is that you cannot trust that the connections are always perfect. The interconnections between components are based on fairly loose sockets that in many cases are also worn badly. If a connection fails and things open circuit it usually isn't that dramatic when you're dealing with battery-powered circuits or small signal, low current circuits. BUT it's a whole different issue when you bring power amps that are able to pass several amperes of current into game. In those cases you don't want that feedback loop to open circuit or an other half of the PNP/NPN bridge to loose it's connection to power supply or load, as the effects will likely toast your speaker, a bunch of components as well as the breadboard. If you're lucky you get away with some arcing that ruins a bit of the board. With power amps it's usually worth the trouble to etch a board for prototyping. I'm not even sure how much current those breadboards are rated to handle.

[Minion]

I suggest you just try finding out what the Max voltage output of your preamp stage is then set the gain of the power amp stage accordingly as to avoid clipping the power amp ...

Ive built several LM3886 amps and Have never had a problem with feeding them very high signal voltages or with the thermal protection kicking in, My first LM3886 guitar amp had 4 cascadeing gain stages in the preamp and probably had an output signal over 5v and the LM3886 had no problem with it and there was no heat problem .....


CHeers

[Morocotopo]

Thanks Minion for the data. What voltage are you powering your 3886 amps with? Because that would change/set  the chip´s headroom, together with the gain as set by the gain setting resistors, right?

[Minion]

well that particular amp used a 2x18v Transformer to get about +/-26v into a 8 ohm load which is an easy load for those voltages and I used a pretty huge heatsink but I was also useing the Insulated version of the chip which needs a big heatsink ..... I persdonally like the sound of the TDA7293 (mosfet chipamp) better for guitar amps but it is harder to use but also more powerfull .....

Cheers