Welcome to Solid State Guitar Amp Forum | DIY Guitar Amplifiers. Please login or sign up.

March 28, 2024, 06:24:55 PM

Login with username, password and session length

Recent Posts

 

Newbie question: Why current feedback is not a goal in Mosfet power amps?

Started by Superfuzz, October 07, 2013, 07:16:47 AM

Previous topic - Next topic

Superfuzz

There must be a simple answer for that. Mixing current feedback on typical BJT amps has been a goal for the whole 90's, then mosfets came and no current feedback at all, I was wondering that also for class D amps..

Mosfet trannies are already sensitive to speaker dampening and stuff??

phatt

Hi superfuzz,
I'm speaking as a player here with a bit of tech savvy.

Oh look this stuff has been debated over and over it's up for grabs.

IME, (at least my experience with the Peavey stuff) it's not a mind blowing difference and does little except turn up the volume a tad.

For some that is enough to spin the propeller on there party hat but I've been around Amps and gear for most of my life and it's all ho hum silly maths that may not reap the result claimed by some. Having said that it could help but again I use Peavey as example.

IF as much attention was paid to the preamp setup before adding the fancy power amp bits it would help reap a better result.

If you want the classic Valve sound then do some serious research and you will find they had crap bandwidth, average quality OTransformers and little Feedback or even none. 

Turn them up and the treble suffers due partly to passive Hi Z tone circuits under load. (Hence came lots of treble boost addons)

Now go find a SS amp that does that same trick,, you will find the treble does not suffer at high volumes. Now add far more efficient speakers (Broader bandwidth) and your treble pickup (on a strat at least) will be exceedingly harsh and brittle. now add distortion and the result is a world away from those classic tones you here from yester year gear.

Hartly Peavey likely picked up on the new rage of metal and his preamp designs are mostly aimed at the harsher metal tones.

My guess is that if you built a better/different preamp aimed at the sweeter more mellow tones then the fancy current feedback might be quite helpful but metal is all about preamp contoured distortion and I doubt you would be able to pick the turn of a current feedback dial at a rock concert.

FWIW, Kevin O'Conner (King TUT) in one of his books scoffs at the folly of Current FB in SS Amps.

I've learned to understand that a lot of the really bright technical folk often seem like they all have different opinions and reading several books or even just reading forums they seem to all have different ideas on what is good/best.

Having messed around with heaps of different ideas including the one you mention here I've come to see that there are just so many ways to do things and frankly a lot of them reap same or similar results.

with so much stuff available now in pedal form or dedicated preamp gear you may find a simple basic SS power brick that runs clean is actually more useful than a some of the fancy tricks used to emulate power stage sound.

I remember well when I was given a Book called Art of Electronics (Apparently one of the best on the subject) I read it over and over looking for the holy grail trick (must here somewhere)
Grounded base, bootstrapped, followers current sources. Sadly they all had a BUT at the end of each paragraph.

It finally dawned on me that there is no perfect circuit,, they all have strengths and flaws/limits.

If it helps,,, I use a dead simple power amp,,all my sound/tone/distortion tricks are done in a dedicated preamp section.

I also use darker speakers (lower SPL) as those modern ones gave me way to much headache trying to dial in the mojo.

Higher SPL speakers tend to get peaky as they bump up the spl at only a narrow range of frequencies,, which is where they get the extra loudness from. But if that sounds good to your ears then go for for it I say.

If you want to test how balanced your system is for tone here's my favorite trick.
Stock fender strat, plug in with fairly loud volume and get a tone you like on the Neck pickup.. Now switch straight across to the Bridge PU.
If the treble now rips your ears off,, the system is not going to be useful and you will be forever frustrated.

It's the whole system you have to consider as well as what style of music you are aiming for in the first place.

Also do you mainly use SC pu or HB pu. there are just so many factors I could be here till morning. :lmao:

cheers,, Phil.

teemuk

Current feedback is goal in -some- amps. Having MOSFET or BJT outputs doesn't neccessarily have anything to do with that.

Because current feedback is mostly used to artificially increase output impedance not all designs - like many bass amps, HiFi amps, acoustic amps, etc. - neccessarily benefit from its effects. Whether the designer sees fit to use current feedback also also largely depends on overall design.

For example,

- TubeWorks MOSvalve amps with MOSFET outputs are not configured as source followers (as usual) but common source amps. They also use very little negative feedback. In such setup the output impedance is quite high and there is very little need for additional current feedback.

- Many TDA7xxx series chips (that have MOSFET outputs, now as source followers with plenty of closed loop negative feedback) are often used in current feedback configuration ...at least in guitar amps. ...not so often in bass amps.

- Amp like Ampeg SVT3-PRO again uses MOSFET outputs, open loop. The output impedance is moderately high already.

- Randall's new G3 series power amps use MOSFET outputs, they are open loop, and the amps also fake the characteristic response of high output impedance amp with an additional fixed EQ circuit.

- Traynor DG amps use MOSFETs in class G switching. Other output devices are BJT. They employ current feedback.

- The new Quilter amps have class D outputs and they aqlso employ current feedback (at least accoring to patent).

...I think you get the point. It varies. Anyway, I'm too busy right now to think any more examples.


...So, I can't quite agree that current feedback is "not a goal" as it in many cases in fact is. However, in certain cases the designer(s) have not seen current feedback as important part of the design.

I don't really find that having anything to do with using MOSFETs, though.

Roly

The "goal" really is to try and emulate a valve/tube amp sound, and one means that some designers use is to increase the output impedance to valve amp levels using current feedback.

Current feedback, a.k.a. variable damping, was a fad in the Hi-Fi world c1950's, but was out of favour by the time stereo came along.

These days a high output impedance might be in favour for tenor guitar, but you certainly wouldn't use it in the other cases @teemuk mentions, bass guitar, Hi-Fi, or acoustic guitar amps, which want fidelity and tighter speaker cone control.
If you say theory and practice don't agree you haven't applied enough theory.

Superfuzz

Ok, thank you all for all the explainations..

Since I' spent a lot of time experimenting with stuff, I'll try to tell you something more of what I feel/need when playing since my experience as a player is a lot greater than my electronic skills (I'm only an hobbist there).
Just to find a point of what could be the direction of my research:

I've built my sound around SS amps, I play fast harsh stuff (when I say harsh I speak about playing distortions thru custom cabs with tweeters inside, it's like the end of the world) so transistors have always been perfect.
My eq. tastes are also mid-less so, tubes are useless also there..

THEN (lately) I started missing that kind of hump that comes with palm muting and stuff like that, so I first started messing with pedals (I've built a lot of them in years) and eqs but, I found that adding let's say, a peak around 100/150hz (where speakers usually have their inpedance peak) sounds close BUT then everything gets muddy and lifeless..especially when I hit my deepest distortion, it gets confused...not that kind of "breathe" that you can have with tube amps..

So, bigger experiments came:

Lately I evolved into this new setup, all my pedalboard works as a preamp, and then I go straight to power amp input of my SS amps (usually a '77 H|H ic100L). I like it a lot, very "clear" and it's also louder (to the point I can saturate my power amp input LOL)

Then while recording, I find it lacking a bit of that Hump anyway, so I first tried my old Carlsbro slave amp (wich I converted to current feedback) but for some reason I get oscillations, so I tried an Orange OR120 (power amp only also here) and, BLING. A bit softer, but so much richness, it's perfect in pair with my HH.


NOW: wich kind of power amp (SS obviously) may be the best halfway solution for my needs?

P.S.
I know I'm a lot OT, but I was considering the whole current feedback thing as a keystone for my needs before..

P.P.S.
If you're have read everything I wrote above, and you're still reading thing, it means a lot to me, thank you SO MUCH, really!  <3)



Roly

Quote from: Superfuzzmy old Carlsbro slave amp (wich I converted to current feedback) but for some reason I get oscillations

Okay, let us dip a toe in a little bit of theory here.

Solid state amps typically have quite a lot of voltage feedback.  This is simple feedback derived from the output node, the voltage across the speaker.  While a speaker in a cab represents a "complex" load, that is it looks electrically like one or more resonant circuits, but because solid state amps have such low output impedances this complex load can't have much effect on the output voltage.

In a simple resistive load the current is always in phase with the applied voltage, according to Ohm's Law, but with a complex or reactive load such as a driver in a cab, the current through the load may be flowing a long way out of phase with the applied voltage.

(This can be a bit of a hard concept to grasp, but just think of energy being stored in the speaker system somewhat like winding up a a watch spring, and "winding up" and returning energy by "unwinding" can take place at times different to the applied voltage drive.)

Voltage feedback, strictly negative feedback, is out of phase and always opposed the driving signal, but it is possible for feedback of any sort to be applied in phase with the driving signal and a self-sustaining oscillation will occur.  This is the basis of all oscillators, signal generators, and the like.

Now with current feedback the phase of the feedback can be a very sensitive function of the driver and cab resonances, and it is quite possible for feedback that is negative or out of phase in the mid band to rotate sufficiently at the extremes of frequency response to become in phase and cause transient instability, ringing, our outright oscillation, either low frequency putt-putt "motorboating", or high frequency shreaks or whistles.

For this reason current feedback is typically limited to around %10 of the level of the voltage feedback in the same amplifier, and attempts to apply more current feedback may result in various forms of amplifier instability.

HTH
If you say theory and practice don't agree you haven't applied enough theory.

Kaz Kylheku

Quote from: Superfuzz on October 08, 2013, 03:46:16 PM
THEN (lately) I started missing that kind of hump that comes with palm muting and stuff like that, so I first started messing with pedals (I've built a lot of them in years) and eqs but, I found that adding let's say, a peak around 100/150hz (where speakers usually have their inpedance peak) sounds close BUT then everything gets muddy and lifeless..especially when I hit my deepest distortion, it gets confused...not that kind of "breathe" that you can have with tube amps..

I get that lifeless sound with poor "hump" when I turn down the current feedback on my amp, and the "breathe" when I crank up the current feedback. It's a huge difference.


   
   
ADA MP-1 Mailing ListMusic DIY Mailing List
http://www.kylheku.com/mp1http://www.kylheku.com/diy

Superfuzz

Quote from: Roly on October 08, 2013, 06:02:55 PM
Quote from: Superfuzzmy old Carlsbro slave amp (wich I converted to current feedback) but for some reason I get oscillations

Now with current feedback the phase of the feedback can be a very sensitive function of the driver and cab resonances, and it is quite possible for feedback that is negative or out of phase in the mid band to rotate sufficiently at the extremes of frequency response to become in phase and cause transient instability, ringing, our outright oscillation, either low frequency putt-putt "motorboating", or high frequency shreaks or whistles.

For this reason current feedback is typically limited to around %10 of the level of the voltage feedback in the same amplifier, and attempts to apply more current feedback may result in various forms of amplifier instability.

HTH

That's a big help! Is there any formula to calculate the ratio between voltage feedback resistors and current feedback ones? Now is too late here but tomorrow I'll upload the simple schem of the mod I did to my carlsbro slave 1000, I must have messed with values too much!

Roly

errruuumm ... there are quite a few circuits around, their signature is that they have a very small resistance in series with the speaker load to sense the current (as a voltage).

Now we have two voltage signals, the voltage feedback from the output, and a voltage proportional to the current (with a phase component depending on how the speaker+cab is reacting at that moment), and we sum these voltages with a summing resistor network, which is then applied to a stage so the overall effect is negative feedback (this summed feedback signal is subtracted from the input at that point).

This is roughly;

Vf = -(Ve+(Vi/10))

where Vi = k Vmax Sin(mag + j phase)

And you have an overall feedback network that is now phase sensitive.

"An amplifier is an oscillator going somewhere to happen."

http://en.wikipedia.org/wiki/Phase_margin

https://www.google.com.au/search?q=amplifier+gain+and+phase+margin

For an amplifier to be unconditionally stable it must not have gain where it has high phase rotation, and it must not have high phase rotation where it has high gain.

The conditions for oscillation, being unconditionally unstable, are pretty much the opposite (but other considerations like starting, distortion, frequency stability come into consideration, e.g. Fx LFO's, trem/vib/chorus/&c)

Attach: how many feedback paths can people see?

If you say theory and practice don't agree you haven't applied enough theory.

Superfuzz

I can see clearly the Current Feedback path, wich actually is DC coupled, then I got some more troubles identifing the Voltage Feedback network, but it must be that 680k res parallaled with the two caps and resistors..

J M Fahey

Quote from: Superfuzz on October 07, 2013, 07:16:47 AM
There must be a simple answer for that. Mixing current feedback on typical BJT amps has been a goal for the whole 90's, then mosfets came and no current feedback at all, I was wondering that also for class D amps..

Mosfet trannies are already sensitive to speaker dampening and stuff??

Who says so?  ;)

FWIW *my* amplifiers, Bipolar and MosFet, *all* use current feedback, prominently higher over a little voltage feedback, since 1972, when I moved from tubes (think Fender Blackface clones) to SS and instantly noticed SS was "dry" in the low range and "flat/boring/uninteresting/dull" on the high end.
We are talking flat uncorrected amps here.

The end result?:
my guitar amps bite and thump, a cutting and punchy sound onstage.
Bass amps slap and thump.
So much so that there is a dedicated "Slap" button on all of them.

Of course, we are talking closed loop amps here, where 99.99% of sound characteristics come from what you use as a feedback loop.

In some modern cases Teemuk mentions, they use tubes as drivers and open loop MosFets as current boosters, which has, say, a couple ohms output impedance which "kills" high damping.

Personally I prefer not to rely on part to part randomness and set amp parameters from the beginning, in a repeatable way.

As of your conversion oscillating, well, yes, it can happen sometimes, *voltage* phase (which is what most amps care about and compensate for) and *current* (through the speaker) phase, which is what you sample for current feedback amps are not the same, speaker phase and impedance are *weird* and all over the map.

Personally I have had little or no problem with that because I use very simple "classroom example" designs, not over the top open loop gain to begin with, nor killer HF response (who needs perfect 100KHz squarewaves in a *guitar* amp anyway?) :loco

As of the latest (last 10 years or so) offerings by Randall and others, they are clearly designed by excellent but *not* guitar oriented engineers.
A Randall *without* current feedback?
A Marshall getting 350W out of 4 paralleled/bridged chipamps ??? ?Which by the way, die like flies when you spray some Raid.
So much so that it started a small Industry offering aftermarket plug-in replacement modules (besides those offered by Marshall themselves).
C'mon, gimme a break !!!!

phatt

@ Jaun Fahey,
                     Would it be fair to say that all this would do little to a badly designed preamp section?
My point I guess is that in my (obvious) *limited experience* I've found that I can achieve far more by taking very careful note of preamplification.

I have messed around with this trick a bit but it never did enough to make it worth the effort.

My thoughts are that at high SPL rock concert levels the idea may reap better results but at the levels I play (small venues) it brings little to the sound/tone.
interested in what the teck folk think. 


Oh yeah I just had a run in with a Skyteck biamped 4 input PA.
3 x TDA7294 chip amps,, one for the satilite speakers and 2 bridged to run the built in sub woofer.

fixed the cracked solder join and worked perfect for all of a minute until I turned up the sub and bang ,, blew the sub chips.
It claimed 300 watts mosfet in big print.

Phil.

J M Fahey

*maybe* it pulled 300W (DC) from the PSU while dying.
Hey!! , 300W is 300W, isn't it  :lmao:

Superfuzz

Ok, here's the mod I did to my Carlsbro SLAVE 1000 power amp.. you can see I used a rotary switch (2X4) to vary the amount of CFB, from stock (nothing) to "max", as the CFB increases the overall gain increase..maybe it does too much gain per step...this may be the problem.. the original gain is 10k/220r (there's also the real schematic attached below)


J M Fahey

Interesting.
My standard mixed feedback which I have been using since 1972 adds 2 voltages in series, one a sample of the output voltage (10K/220 ohms for a roughly 40:1 gain) and the 4 ohms speaker in series with a 0.1 ohms resistor, also roughly 40:1 , which combine for a roughly 20:1 gain.

Why those values?

I had been building Fender clones up to that point, when I switched to SS sound *was* different.

Instead of magic, I started measuring, scoping, etc. , to find *why* and what could I do to get a closer sound.

One thing I instantly noticed was that the tube amp signal at the output jack doubled (+ 6 dB) when unplugging the 4 ohms speaker.
My NFB network was designed to do the same.

Maybe not a *big* step, as Phatt says, but definitely a step in the right direction.

And instead of adding external EQ (which also works, of course), it cost me less than 10 cents, how to beat that?

Specially because I make my own precision 0.1 ohms resistors: 4 cm of 0.50mm diameter Constantan wire.

Still use that, 41 years later.