Need ideas for a limiter circuit

[Gaz]

I'm mating a tube preamp to a Class D module (Icepower), and would like to include a limiter circuit (solid state is aok) to keep the power amp from clipping. I'm a bit lost on where to look, but my preliminary research make me think that I need something that is "soft-limiting/clipping." Any basic circuits I can learn from (although I realize most of this circuits are relatively complex). Thanks!

[J M Fahey]

I'm mating a tube preamp to a Class D module (Icepower), and would like to include a limiter circuit (solid state is aok) to keep the power amp from clipping. I'm a bit lost on where to look, but my preliminary research make me think that I need something that is "soft-limiting/clipping." Any basic circuits I can learn from (although I realize most of this circuits are relatively complex). Thanks!

3 possible solutions:

1) if you have no problem with power amp eventually clipping but "want it to clip like a Class AB amplifier and not a Class D one" just add a couple (or a string) of plain diodes across its input, so "they" clip first.
Diode clipping has quite a rounded and smooth edge .

2) if you do not want the power amplifier to ever clip, no matter what, add a non clipping limiter between preamp and power amp.
Scope output so you have as much as possible without squaring peaks.
I suggest the simple and elegant Crate/Ampeg opto limiter, which stops waveform peaks a couple Volts before reaching positive or negative peaks.

3) set preamp to maximum clipped output, balls to the wall, add a screwdriver adjustable "master" at power amp input, scope output and sllloooowwwwllllyyyy rise Master, starting from 0, until you see the wonky tube preamp waveform starts to squarewave on peaks (indicating now the Class D amp has started clipping by itself), then back down a little.
Now nothing the Preamp does will clip the Power amp.
FWIW that is the method I use.

[Mike Sulzer]

I'm mating a tube preamp to a Class D module (Icepower), and would like to include a limiter circuit (solid state is aok) to keep the power amp from clipping. I'm a bit lost on where to look, but my preliminary research make me think that I need something that is "soft-limiting/clipping." Any basic circuits I can learn from (although I realize most of this circuits are relatively complex). Thanks!

The output stage of your tube preamp clips, although not cleanly. You could put an attenuator between it and the amp so that the preamp cannot quite drive the amp into clipping.

[Mike Sulzer]

By the way, I think most of those Icepower modules have a built in soft limiter; who knows how good it is.

[Zozobra]

I've looked into doing something like this for a hybrid amp but I'm yet to put it to the test:
http://sound.whsites.net/articles/soft-clip.htm

[R.G.]

It's good to see this technique being accepted in the larger community. Pre-clipping the preamp before the power amp gets the signal is a neat trick. It can effectively eliminate the razor-edged clipping of high-feedback power amps - the power amps never clip, and always stay linear.

I'm sure it's older than this, but the pre-powe-amp limiter was used to good effect in the Thomas Vox amps of the early-mid 1960s. The "bias" procedure on these amps was actually a pot that adjusted the clipping threshold on the preamp limiter. The procedure was somewhat as JM menioned - turn the limiter to minimum limiting, run the signal into the power amp until the power amp was just hinting at clipping, then turn the mixer/limiter control to make the output noticeably more clipped. This ensured that the limter, not the power amp, was clipping.

As to what clipper to use, that gets tied up in signal level and adjustability. The Thomas Vox limiter had a wide clipping range, from about +/-0.6V to about +/-1.8V. It took two transistors and two diodes plus a handful of R/C to make that work. The Thomas Vox clipping/limiting circuit arranged for a variable amount of DC to run through resistors attached to the clipping diodes, which were ordinary silicon devices. Increasing DC current pulled the clipping points of each diode apart, letting the clip level be adjusted. It's a touchy circuit but sounds good.

There are a variety of clipper/limiters that can be used. About the softest simple limiter is a pair of small signal MOSFET with gates tied to drains. This produces a clipped waveform that never entirely flat tops. It's actually a softer than many tube circuits' clipping. The only real drawback is the clipping voltage is around +/-3V or more depending on the threshold voltage of the MOSFETs. This is a pretty big voltage compared to the input voltage of a solid state power amp (including the IcePower), so you pretty much have to juggle the preamp signal up to clip, then volume control it down to run the power amp inpu at the right level. And it's not all that adjustable as to voltage clipping levels.

@Gaz: I'd scope out the amount of signal needed to clip the IcePower module, then design a clipper that clipped at just less than that, and check this out by switching the clipper in/out while watching the output waveform. The IcePower module will have a relatively fixed input clipping point, so once you get the input clipping point set, you won't need to adjust it. Fixed is always better if you don't really need adjustable.

[Gaz]

3 possible solutions:

1) if you have no problem with power amp eventually clipping but "want it to clip like a Class AB amplifier and not a Class D one" just add a couple (or a string) of plain diodes across its input, so "they" clip first.
Diode clipping has quite a rounded and smooth edge .

2) if you do not want the power amplifier to ever clip, no matter what, add a non clipping limiter between preamp and power amp.
Scope output so you have as much as possible without squaring peaks.
I suggest the simple and elegant Crate/Ampeg opto limiter, which stops waveform peaks a couple Volts before reaching positive or negative peaks.

3) set preamp to maximum clipped output, balls to the wall, add a screwdriver adjustable "master" at power amp input, scope output and sllloooowwwwllllyyyy rise Master, starting from 0, until you see the wonky tube preamp waveform starts to squarewave on peaks (indicating now the Class D amp has started clipping by itself), then back down a little.
Now nothing the Preamp does will clip the Power amp.
FWIW that is the method I use.

Thanks, Juan!


#1, Do you think an arrangement like @zozobra posted would be worthwhile? http://sound.whsites.net/articles/soft-clip.htm

#2 - Which Ampeg/Crate design would be a good one to study? (i.e. the simplest one for me to distill the limiter circuit from).

#3 Having no knowledge of limiting circuits, my first inclination was to do #3, and to be honest I'm not sure why not all preamps aren't just attenuated to that point before power amp clipping Is it really to just make the amp feel more tube-like?

[Gaz]

The output stage of your tube preamp clips, although not cleanly. You could put an attenuator between it and the amp so that the preamp cannot quite drive the amp into clipping.

Thanks, I may just do this, and if I understand correctly, you mean the same as JM did in his #3 suggestion.

By the way, I think most of those Icepower modules have a built in soft limiter; who knows how good it is.

I'm not sure. I emailed them, and they did not mention it, just that input sensitivity was 2.3Vrms, and no need to go past 2.5Vrms (limit before damage is 12Vp).

[Gaz]

It's good to see this technique being accepted in the larger community. Pre-clipping the preamp before the power amp gets the signal is a neat trick. It can effectively eliminate the razor-edged clipping of high-feedback power amps - the power amps never clip, and always stay linear.

I'm sure it's older than this, but the pre-powe-amp limiter was used to good effect in the Thomas Vox amps of the early-mid 1960s. The "bias" procedure on these amps was actually a pot that adjusted the clipping threshold on the preamp limiter. The procedure was somewhat as JM menioned - turn the limiter to minimum limiting, run the signal into the power amp until the power amp was just hinting at clipping, then turn the mixer/limiter control to make the output noticeably more clipped. This ensured that the limter, not the power amp, was clipping.

As to what clipper to use, that gets tied up in signal level and adjustability. The Thomas Vox limiter had a wide clipping range, from about +/-0.6V to about +/-1.8V. It took two transistors and two diodes plus a handful of R/C to make that work. The Thomas Vox clipping/limiting circuit arranged for a variable amount of DC to run through resistors attached to the clipping diodes, which were ordinary silicon devices. Increasing DC current pulled the clipping points of each diode apart, letting the clip level be adjusted. It's a touchy circuit but sounds good.

There are a variety of clipper/limiters that can be used. About the softest simple limiter is a pair of small signal MOSFET with gates tied to drains. This produces a clipped waveform that never entirely flat tops. It's actually a softer than many tube circuits' clipping. The only real drawback is the clipping voltage is around +/-3V or more depending on the threshold voltage of the MOSFETs. This is a pretty big voltage compared to the input voltage of a solid state power amp (including the IcePower), so you pretty much have to juggle the preamp signal up to clip, then volume control it down to run the power amp inpu at the right level. And it's not all that adjustable as to voltage clipping levels.

@Gaz: I'd scope out the amount of signal needed to clip the IcePower module, then design a clipper that clipped at just less than that, and check this out by switching the clipper in/out while watching the output waveform. The IcePower module will have a relatively fixed input clipping point, so once you get the input clipping point set, you won't need to adjust it. Fixed is always better if you don't really need adjustable.

Thanks, R.G. I've actually read your article on the Vox limiter (https://www.premierguitar.com/articl...-in-vox-amps-1)

I did email Icepower a while back about input sensitivity, and this was their answer:

"Yes the differential input voltage needed for 720W in 4 ohm is 2.3Vrms. If you input is 2.46Vrms differential you will get 836W at 4 ohm with 5% THD+N, so more than 2.5Vrms input does not make much sense. Keeping the max input level at ~2.5Vrms (3.6Vpeak) also gives better/lower idle noise than with larger gain in your preamp."

The actual input max to the module is 12Vp.


I will have to audition some clippers, but if like the effect, what would be reasonable threshold to set? 2.5Vrms according to their comments? I'm thinking it may be cool to add a clip light for myself, so I can have a visual for when the threshold is being crossed.

[R.G.]

Whenever I see "clipping" I always shift over to peak/instantaneous voltages, not rms. 2.3Vrms is 3.25V peak, 2.5V rms is 3.535V peak. So you're talking about clippers that limit voltage to between 3.25V and 3.54V.

I'm interpreting the comments from IcePower as saying in effect "... the input is not distorted (much, anyway) at input voltage peaks of 3.25V, but another quarter of a volt starts pushing the amp into the beginnings of clipping", which is how I interpret any amplifier maker saying that it's XXXX W at Y% distortion.

In my mind, the difference in real sound level between 720W and 836W is not enough to pursue, even given that you do retain some hearing after using those kinds of levels So for me, I'd try to set the limiter to go max outputs of 3.3 or 3.4V peak and call it a day.

As an architectural side note, the power amp is running linearly at all points under these 3.3-3.4V peaks, so you can in fact put a "master volume" after your clipper, and give yourself the ability to turn down the blasted thing and retain exactly the same kind of overload/overdrive from the power amp at lower power levels. I would be horribly tempted to design the clipper to run the IcePower at its full 720W or so with the "master volume" full up, and let it be turned down for lower effective powers.

But back to the main thread. So you want to design a clipper that never lets its output signal go over +3.3V (for instance...) nor under -3.3V. We as a group believe that softer clipping is better on this point, so the trick becomes designing a clipper that starts softly clipping somewhat below that point, maybe 2.6, 2.8, 3.0V or so, and softly increases clipping until it won't do any more output above 3.3. One way is to hand-pick small signal MOSFETs as I mentioned to easter-egg the right clipping points. Two will be needed, and they probably won't match, but a little asymmetry in this clipping just might sound good. Matter of taste.

MOSFET clippers are so soft that it might not be clipp-y enough. In that case, either a string of diode pairs, diodes tied to clipping voltage references (which is what the Thomas Vox thing really did, they just varied the clipping references), or some such. This will take some auditioning.

I'd like to say I have a canned design all ready for you, but I don't. My last foray into this was for a +/- 1.4V clipper at the input of a more standard SS amp.

Another issue lurking here is what AC voltages your preamp puts out. If the tone you like from your preamp that you just want to be made [MUCH] louder, you'll need to need to figure out the happiest marriage of padding down or amping up the sweet spot for tone from the preamp output and then massaging that down to the right place for how much you want to overdrive the clipper circuit, if any. This is going to be hugely subjective, I think. And should be.

It's an interesting design question. I'd be happy to add any design help I can.

[Gaz]

R.G., thanks so much for your thoughts, and I'll take any help I can get!

Whenever I see "clipping" I always shift over to peak/instantaneous voltages, not rms. 2.3Vrms is 3.25V peak, 2.5V rms is 3.535V peak. So you're talking about clippers that limit voltage to between 3.25V and 3.54V.

Note taken - how I will sPEAK about voltages from now on regarding clipping issue

As an architectural side note, the power amp is running linearly at all points under these 3.3-3.4V peaks, so you can in fact put a "master volume" after your clipper, and give yourself the ability to turn down the blasted thing and retain exactly the same kind of overload/overdrive from the power amp at lower power levels. I would be horribly tempted to design the clipper to run the IcePower at its full 720W or so with the "master volume" full up, and let it be turned down for lower effective powers.

That's interesting, and I always imagined the master volume (there's actually two switchable as it's two channels) before the clipper, so any clipping from the 'soft-limiter' would be obscured by the sheer volume, and perhaps even feel more like a tube power amp starting to clip. I will surely have to try both.

But back to the main thread. So you want to design a clipper that never lets its output signal go over +3.3V (for instance...) nor under -3.3V. We as a group believe that softer clipping is better on this point, so the trick becomes designing a clipper that starts softly clipping somewhat below that point, maybe 2.6, 2.8, 3.0V or so, and softly increases clipping until it won't do any more output above 3.3. One way is to hand-pick small signal MOSFETs as I mentioned to easter-egg the right clipping points. Two will be needed, and they probably won't match, but a little asymmetry in this clipping just might sound good. Matter of taste.

MOSFET clippers are so soft that it might not be clipp-y enough. In that case, either a string of diode pairs, diodes tied to clipping voltage references (which is what the Thomas Vox thing really did, they just varied the clipping references), or some such. This will take some auditioning.

Ok, thanks, I can envision now how the clipping will function in relation to the input sensitivity of the module. What are some small signal MOSFETs to look at? I am FET illiterate (besides the high voltage ones I've used for a tube-voltage FX loop). I'm also going to have a go at the VOX style. The Icepower module has a handy .5A 15V supply on it I could use for the reference.[/QUOTE]

Another issue lurking here is what AC voltages your preamp puts out. If the tone you like from your preamp that you just want to be made [MUCH] louder, you'll need to need to figure out the happiest marriage of padding down or amping up the sweet spot for tone from the preamp output and then massaging that down to the right place for how much you want to overdrive the clipper circuit, if any. This is going to be hugely subjective, I think. And should be.

Yes, I'm pleased with the preamp tone, and am starting to realize there's no real reason I can't just passively attenuate the preamp signal so that it never exceeds 3.4Vp, but now am set on trying the clipper I measured my preamp, and it puts out about 7Vp before clipping, and 20Vp maximum on the Overdrive channel. The clean channel is a bit less at max.


Also, I was able to find one schematic that is a hybrid preamp, and uses an Ice module - The Fender Bassman TV series. I'm having a bit of trouble understanding exactly how they are doing it. Are D1 and D2 zeners being used as a limiter (their voltage rating seems to high), and then the following opamps being used as buffers (the Icepower modules have a 7K input impedance)?

Fender Bassman TV Series Schematics.pdf

Thanks again, R.G.!

[J M Fahey]

Are we talking Guitar amplifier or Bass amplifier here?

If Bass, you need tons of clean pkower, period, and preamp is used basically clean.

FWIW the Fender "Bassman TV series" you linked to is a fraud, pure and simple, the only "DNA" they show is *cosmetics*.
Oh, and "do nothing Tube preamp, since used that way itīs always clean, does not colour sound and might as well have been made with Fets or even Op Amps.

Tube amps (and preamps) "do their thing" when clipping or at least being driven real hard into non linear areas which does not happen here.

ICE power module is an excellent Hi Fi/PA type power amp, VERY SS, sharp edge squarewave clipping, high damping, zero compression ... the absolute opposite of any Tube Bassman power amp.

Diodes D1/D2 are just a fail safe protection, keep signal below 5V peak or 3.5V RMS ... what Dr Icepower recommends

Same as D4/5/6/7, they protect U1a and U2a .
In theory there is 0Vac there, those are virtual ground points ... emphasis on *virtual*.

If circuit malfunctions or simply loses 1 or 2 15V rails you do NOT have the self cancelling action any more and Tube signal out , if untamed, can and will destroy delicate Op Amp inputs,so these diodes keep peaks below 700mV peak.

Now for a Guitar preamp:

Tubes are not magic or fairy dust or Talismans, acting by mere presence (contrary to 99.99% Internet disinformation being spread out there), they do some very definite things when driven hard, which we happen to like (me included )

IF you get a nice sounding tube waveform, reamplify it faithfully and drive guitar speakers, your ears will have a hard time finding it is not "100% tube".

Just imagine a Champ driven full blast, going through a Jensen speaker, being picked up by an SM57 and being reamplified by tens of kilowatts into a Stadium audience, through a *clean* power amp bank (to boot, triamplified, limited, etc.) into a mountain of high performance PA speakers, including horn drivers, bullet tweeters, etc. ... sound will still be very recognizably TUBE type.

Now if you feed a preamp straight into a loud clean SS power amp you start by losing 50% of the Tube sound, since there is neither a clipping (or crunching) Tube power amp and damping is "all wrong".

And if you nicely symmetrically re-clip that wonky Tube waveform you took so much trouble to get, with any kind of SS clipper, pick any of them: Diode/Zener/diode connected FET/LED/biased Diode (VOX)/you-name-it ... then whatīs the point of using Tubes in the first place?

Beyond cosmetics or Mojo that is.

Thatīs why for me the only "honest" way to do it (which is also the simplest one) is to clip preamp **even if it will later be used "clean"**, what we are doing is finding its limits, then feed that into any loud clean Power amp you like, and carefully adjust it (a screwdriver adjustable Master Volume is as good as it gets) until it *just* clips and leave it there.

Which I can predict (not surprisingly ) will let clipped Tube waveform peaks NOT surpass 3.5V peaks or whatever ICE Power specifies

In a nutshell: the adjustment suggested is *visual* , by clipping preamp out, monitoring speaker out, and setting non-user-accesible Master Volume trimmer to just clipping.

Which will also be stupid loud and destroy any sensible Guitar cabinet you can build .. unless you find a 24x12" cabinet acceptable, that is.

Being realistic, exact same adjustment procedure applies to a Tube preamp driving a 15 to 25W chipamp usable in a bedroom, or an LM3886/TDA7294 used in a Club.

FWIW my own "100W Marshall/Mesa head killer" uses a 300W RMS Class AB amplifier padded down to reproduce faithfully *any* waveform and level those heads can send to a speaker.

And speaker does not know the difference, all it sees is a certain voltage and current waveform driving its voice coil, has no clue where that came from and reproduces it as air pressure variations.

Ear canīt tell the difference either.

[Mike Sulzer]

"...Tube preamp, since used that way itīs always clean, does not colour sound and might as well have been made with Fets or even Op Amps."

What am I missing? It looks to me as though a humbucker could easily overdrive the second tube with the gain up. Even a strat ought to do it. But also tubes do not have to be dried hard into clipping to color the sound somewhat.

[R.G.]

That's exactly right, J.M. - if you can hear tube tone on a PA or a recording, that means that you can reproduce tube tone live through a non-tube power amp on stage. There is one caveat on that, and I'll go there in a moment.

I once had a neighbor who wanted to perfect his lawn. (there is a point here, and I'll get to it in a second )
We live in a very rocky area, and so when he found a small rock sticking up through his grass, he got his shovel and started to dig it up. This is very common here. However, after about an hour of digging and prying, attempting to lift the small rock, he had enlarged the hole to about 1.5m wide. At this point, he realized that the "rock" was not a chip. It was, instead the PLANET protruding through his grass, and he began filling the hole back up. We sometimes find what looks like a small problem that gets bigger the more we dig into it.

Guitar amp clipping is one of those. There are a lot of special cases, caveats, and other things to think about.

What helps me is to try to split problems into smaller pieces and examine the smaller pieces. Here, as J.M. clearly explains, you can hear tube sound from any amplifier that clearly reproduces the originally-produced tube sound. The only question is where do you stop -producing- and start -re-producing?

There are some other sub-cases. Gaz, if your preamp produces 7Vpk on the normal preamp output, and 21Vpk on the overdrive, you'd see two different kinds of clipping - (probably) the preamp tube soft distortion on the normal channel, and (probably) the power amp input clipping on the distortion channel. The bigger signal (probably) causes a different kind of distortion when driving the power amp input, and (probably) different yet when the PI is a LTP versus a concertina. The details start multiplying.

[Mike Sulzer]

Thanks, I may just do this, and if I understand correctly, you mean the same as JM did in his #3 suggestion.



I'm not sure. I emailed them, and they did not mention it, just that input sensitivity was 2.3Vrms, and no need to go past 2.5Vrms (limit before damage is 12Vp).

Go here: https://icepower.dk/products/amplifi...mparison-chart

Scroll down and click on "see comparison chart" . Look down for the soft limiter entry.

Yes, I did not mean to double post Juan's idea. I had not seen his post yet.