Chip amps and clipping/protection behavior

Status
Not open for further replies.
BWRX said:
I don't know how many times this has to be stressed:

The LM1875 exhibits HARD CLIPPING.

My apologies. I should have said that its clipping style is softer than the other LM chipamps because LM1875 doesn't contain the louder clipping Spike protective system. So that was really meant to be in a comparison, but I sure goofed because it wasn't stated in such a comparison.

I believe that the applied test of whether an amplifier soft clips or not, is if it could be used for musical effect. In this case, LM1875 isn't at all suited to such an effect. In this view, it is a hard clipping amplifier.

Any way I look at it, you're entirely right.

However, the LM1875 has a crazy little audio trick. If either the average 4 amper limit is reached or the support circuit is too weak, the amplifier will hard clip as expected.
However, if demands of transients go up to 12 ampers without running the average demand up to 4 ampers, then the LM1875 exibits a "soft clip like" effect, because it doesn't brutalize the transients like other hard clip amplifiers--when used within those conditions.
That part is somewhat unusual, and it confuses me a bit.
 
danielwritesbac said:
I should have said that its clipping style is softer than the other LM chipamps because LM1875 doesn't contain the louder clipping Spike protective system.

The Spike protection in the other LM chips has no effect on the clipping behavior during normal operation! If you overdrive the input of the amp the output will clip, simple as that. If the protection is active clipping should be the least of your worries.


danielwritesbac said:
However, the LM1875 has a crazy little audio trick. If either the average 4 amper limit is reached or the support circuit is too weak, the amplifier will hard clip as expected.
However, if demands of transients go up to 12 ampers without running the average demand up to 4 ampers, then the LM1875 exibits a "soft clip like" effect, because it doesn't brutalize the transients like other hard clip amplifiers--when used within those conditions.

Do you ANY have evidence of this? If not, please stop just presenting your "ideas" as if they were facts.
 
audfile74 said:
Andrew T. , bwrx. the best way for you both to handle this is to start your own topic instead of log jamming mine. if you don't agree with daniel that's fine. but please do not turn my topic into a slapboxing session. ,do it in your own topic!

Its okay. Those are good guys. Sometimes I need a notice to stop and think and get back to the main point. My writing skills do falter. Their posts are intended to inspire clarification and/or correct errors.

BWRX,
The LM1875.pdf mentions the 4 amper limiter and that transients may slip past the limiter for more. The limit on transients is 12 ampers and I forgot where that is documented.
EDIT: The only specifics I have for you at this time is that 100uF onboard smoothing caps (or smaller) do help increase the success of an extra power exploitation--in one example where LM1875 replaced LM3886 in the role of very loud techno playback (intense dynamics). I will attempt to verify that capacitance figure's effects. /edit
I mentioned LM1875 for a "qualifier question" and since it didn't inspire any interest due to the power output, then the kit with significantly more (considering +3db per each double in output)--more power is Peter's LM4780, which can be used like LM3886's or it can be used in parallel for extra power.
Well, I thought it made sense.

Andrew,
I think we're going to be just fine on a discussion with the safer isolated chips since there's no terrible caveat if the screw gets loose, and I do want to see amplifiers safety earth grounded whenever possible.
On that note, why don't you author a permanent thread with illustrations on safety earth grounding?

Please do.
On the recent thread where we got into a disagreement, it was because the amplifier chronically slipped off the insulator. I'd sure like to see some additional safeties (extra ground strap on the heatsink?) for the non-isolated chips (LM1875, TDA7294, LM3875T, etc. . .) to facilitate possibly working the mains fuse or circuit breaker if the chip slips off the insulator pad. Well, those do need some additional safey measures. Do you think that the insulator could be thermal-glued to the chip?
And, I never again want to step onto/into your pet peeve again. So, please do consider authoring a permanent thread so that everyone can have a safely grounded reference. 😉

Audfile74,
Its fantastic that the owner of AudioSector, Peter Daniels, has arrived to answer your questions. He's very polite, so when he makes a mild suggestion, do pay close attention because his suggestions are usually "known good baseline," well tested, and important for success.

Most kits, including Audiosector, are "power amplifiers" so I think that the next step is to get Peter's recommendations on the active preamplifier subject, and then purchase your supplies.
 
Daniel, if you look at the schematic diagram on page 5 of the LM1875 you'll maybe recognize the dual slope SOA protection circuitry. Assuming that the datasheet drawing is correct it can clearly be derived that the max output current is just about 5A, even for short peak.

Say we were to draw 5 amps at a high output voltage. Then the zeners won't conduct, thus Q43 is off. The output's dual collector current ratios are given with 3:500, the protection circuit runs off the x3 leg. So at 5 amps we have 5*3/503=30mA. These drop 0.6V across R32B and cause Q36 and Q44 to turn on, shunting away drive current from the output transistor. If the current draw would occur during a phase of low output voltage (even below GND) due to reactive load / back-EMF, then the threshold gets gradually lower, because Q43 turns on, lifting the base of Q34 a little via R22 which then lowers the amount of voltage drop across R32B needed to reach 0.6V Vbe at Q34 accordingly. This dual slope behaviour is perfectly reflected in the SOA graph on page 3

12 Amps would mean 1.44V drop and at that point Q36 has long started to conduct. 12 Amps just won't happen, ever. And besides, I have dealt with this chip and recall never getting any more than ~5 amps, as measured in the supply lines. Because the thesholds depend somewhat on temperature (Vbe drops with increasing temp), the max. current will be lower when the chip is hot. If you manage to keep the junction a freezing temps, you might be able to pull some higher levels, say 6A. But 12A is totally unrealistic. BTW that's the peak rating for its big brother, the LM3886.

- Klaus
 
KSTR said:
. . . the max. current will be lower when the chip is hot. If you manage to keep the junction a freezing temps, you might be able to pull some higher levels, say 6A. But 12A is totally unrealistic. BTW that's the peak rating for its big brother, the LM3886.
- Klaus

Working with it again today to confirm. . . you're bound to be correct on those figures.
I should have checked more thoroughly. I had measured with 4 ohm load, and used an output measurement to make an estimate, not the actual current draw. This was some time back, so I had forgotton.
Efficiency, not amperage--ah, sorry about the blunder.

The rest of the information seems to be holding on, in that, without the Spike system, when LM1875 just starts to clip, it is only the loudest pitches affected--somewhat different than the Spike equipped LM chips.
Depending on configuration, some of my LM1875 builds have done the "fuzzy buzzy" almost tube style clipping while other LM1875 builds have done the more raspy transistor sound. Those that did a "nearly" soft clip, did somewhat softer than a Tripath, although not the same as my 1962 era tube amp. I simply must remember to ask ZDR about his LM1875 with 12v rails (battery) application.
What could cause the difference in the "style" of clipping behaviors, between two amplifiers, both LM1875?

Today, I checked heat output with a power circuit specifically designed to help temperature. The LM1875 (brewed up especially for testing) had 100uF caps onboard in addition to a pair of 100nF econo ceramic. Several similar configurations were tested. This ran considerably cooler than the usual approach of a pair of large caps. However, there's several different methods (on various threads) that allow these results, one of which does use the "pair of large caps." Puzzling. The lower temperatures are really useful in allowing a bit more voltage transformer (more headroom), or just saving some cash on heatsink sizes.
Is there some documentation on this particular topic?

Thanks for your help!
 
Status
Not open for further replies.