Questions and rumbles about solid state amplifiers

[bhill]

Don't disagree, but you still have the natural response of the speakers as a limitation. I did not say that the frequencies are non-existant, they do exist in the complex waveform output from the pickups, but as harmonics, rather than fundamentals. And we add to those harmonics (or decrease) depending on how we modify the waveform before it is amplified.

But you can be very meh putting that waveform through a hi-fi system into full range speaker cabinets with 50-20khz range. It takes that complex rich waveform and just flattens everything out.

The twin, and other early Fenders, have a hi-fi heritage, but then the circuits were tweaked for instrument usage cutting some, not all, of the higher freqs. 6db/oct first order filters only decrease those freqs not eliminate. I will have to admit that I haven't figured the frequency response for the twin, prob have to sit down with the scheme in the morning and calculate things. My BF Fenders are a pair of Bandmasters, so I guess you could say I have a stereo twin with twice the number of speakers.  It should be the same freqs response, but I will check before I say for sure. I do know that Leo had hearing loss in the higher frequencies so he depended on others he worked with to tell him if the design was getting too ice picky. More later...

[bhill]

In the meantime, we should probably get this back on track

[teemuk]

Pritchard's amps use chip -based solution for power amplification and Pritchard amps are pretty much considered to be among the best solid-state guitar amps ever made, in all kinds of aspects.


It's really all about what you put around those chips.

[bhill]

And if you really want to get into the circuitry of a SS amp, read teemuk's very highly recommended book about them.
http://www.ssguitar.com/index.php?topic=711.0  will get you to the book link.

[Paul Marossy]

And if you really want to get into the circuitry of a SS amp, read teemuk's very highly recommended book about them.
http://www.ssguitar.com/index.php?topic=711.0  will get you to the book link.

Very nice! 

[tca]

I already have Teemu's book. Thank you for remind me of that reading.

[tca]

So I've been reading (Teemu's book, Mullard's, etc) and looking for schematics of SS amps. One thing that I've noticed is that, although there is "not" (or is there?) a amplifier specific topology for guitar, there is a type of configuration that statistically pops out: 2 BJTs with a voltage feedback biasing (yes, the FF type configuration). This voltage feedback biasing appears in a simple arrangement, as in the case of a FF, but also using the class-ab to feedback the driver circuit. Two examples:
- Peavey standard, and the PA standard
- Heathkit TA-16

I've listen to the Peavey standard and it sounds great. I'm curious about that Heathkit TA-16? How does it sound? (I've goolgled that?!?)

[R.G.]

although there is "not" (or is there?) a amplifier specific topology for guitar

There really isn't an amplifier topology for guitar. I believe all the ones I've ever seen are reflections of hifi practice, perhaps modified for lower cost and more ruggedness.

It's worth noting that some of the earlier amplifier setups before the diffamp input style had prominent (by hifi standards) even-order distortion, so may have been subtly more preferred by guitarists. That is definitely something the guitar ear could dig out.

[tca]

although there is "not" (or is there?) a amplifier specific topology for guitar

There really isn't an amplifier topology for guitar.

I got that

Notes on the previous schematic:
- very low input impedance
- to much gain +-50dB!
- and of course to much fuzz (obviously), that would make the amp with a very specific sound, only fuzz type sounds comes out it.

Meanwhile for those following my rumbles, here is what is on my breadboard (as the previous schematic, you may recognize the driver part):



Cheers.

P.S. (edit)

I always end up using a booster to further drive the amp in to distortion (in both schematics) so I guess I should try and make a 5 transistor amplifier to get the level I want, instead of the 4 transistor classical class-ab.

[Thecomedian]

I've noticed that the discussion of frequency response often confuses the reduction of certain frequencies with the elimination of those frequencies. The treble frequencies of a guitar pickup (and most guitar amps & speakers) are definitely reduced. That doesn't mean they're not there, or not important. Half power is a lot different than no power.

Don't disagree, but you still have the natural response of the speakers as a limitation. I did not say that the frequencies are non-existant, they do exist in the complex waveform output from the pickups, but as harmonics, rather than fundamentals. And we add to those harmonics (or decrease) depending on how we modify the waveform before it is amplified.

But you can be very meh putting that waveform through a hi-fi system into full range speaker cabinets with 50-20khz range. It takes that complex rich waveform and just flattens everything out.

The twin, and other early Fenders, have a hi-fi heritage, but then the circuits were tweaked for instrument usage cutting some, not all, of the higher freqs. 6db/oct first order filters only decrease those freqs not eliminate. I will have to admit that I haven't figured the frequency response for the twin, prob have to sit down with the scheme in the morning and calculate things. My BF Fenders are a pair of Bandmasters, so I guess you could say I have a stereo twin with twice the number of speakers.   It should be the same freqs response, but I will check before I say for sure. I do know that Leo had hearing loss in the higher frequencies so he depended on others he worked with to tell him if the design was getting too ice picky. More later...

since humans usually cant hear above 16khz-20khz, and since even if you could it would sound incredibly soft and weak, we can throw those frequencies out completely.

an ideal Cutoff frequency high or low pass or band pass filter would have all values within a certain range at full voltage, and immediately drop to zero voltage on all undesired values. The pass filters dont behave like this, and so attentuation occurs along frequency ranges outside the cutoff point(s). Both systems (if the former were real), have their benefits and drawbacks. If you wanted to have 500 hz at 10 volts and 1000 hz at 5 volts with a perfect filter, you would have to find a way to pass only 1000 hz+ through another network to drop it down, and so on, unless you want to lose 1000 hz altogether. With the latter method, you need to produce a circuit that just "sounds right", without actually cutting out frenquencies perfectly, which allows for a little more fudge factor, but can still be built into a complicated (or op-amp) design that can give you exactly the voltage properties per frequency you want. It just depends how much time you feel like spending making it

This is more or less what frequency filters and analog circuit design tutorials seem to say.

it is my personal opinion that if we didn't have harmonics in our voice boxes, we couldn't have language. Im trying to imagine a single 1000 hz tone that sounds like A in english....

If you lost harmonics in an output system, im assuming you'd just hear beeps.

[R.G.]

it is my personal opinion that if we didn't have harmonics in our voice boxes, we couldn't have language. Im trying to imagine a single 1000 hz tone that sounds like A in english....

The human voice apparatus generates essentially sawtooth waveforms at the vocal cords, and the harmonics of these are shaped by resonances in the vocal tract. This is the reason that we can hear speech at different frequencies and sing. The fundamental does not carry the intelligence, nor do the harmonics, nor do the absolute positions of the resonances. What carries information in human speech is the relationship/ratio in frequency of the resonances.

[tca]

Hi again, a few more notes on increasing the output impedance, or decreasing the damping factor of a SS amp using a mixed mode feedback. I've simulated both schematics with a non-ideal speaker sim and by controlling the feedback network I could get the typical non-linear response characteristic. What I also get is a 15-20dB, or bigger, reduction in the total gain of the amplifier to get this response. This of course opens another problem, that in order to get the same initial power, before the mixed feedback, the circuit needs a pre-amp and also the input of the amplifier must not clip to early... need do redesign the driver part. I do see a light in the end of the tunnel but I was thinking in a simple way of increasing the output impedance without killing "almost" all the gain! Is there a clever way of doing this without building a 60dB driver and then killing 20dB? Probably not, "there ain't no such thing as a free lunch".

I'm really enjoying Teemu's book.

Cheers.

[KMG]

Some words about transformer coupled SS guitar amps.
Look at my page there you can find wide range of this type amps (250mW-90W)
http://milas.spb.ru/~kmg/fetpa_en.html

[tca]

Some words about transformer coupled SS guitar amps.
Look at my page there you can find wide range of this type amps (250mW-90W)
http://milas.spb.ru/~kmg/fetpa_en.html

Nice amps, thanks. But I was looking for a transformer-less amp.

[tca]

There is in fact a simple idea that enables to control the damping factor in a very simple way. This can be seen on the Mullard's book "Transistor audio and radio circuits" pg. 39. There are various examples of this, namely 1W, 3W (>= 20Ohm output impedance) and 10W amps (4 transistor amplifiers). The idea consists in putting the feedback not from the collector of the first transistor but from its emitter. Here is what I've cooked up:



The speaker qualities does shows up in this way. Resistor R3 can be varied to fit several output impedances without perturbing to much the working point.

Still have to adjust the frequency response to my taste but so far so good.

Cheers.

P.S. (edit)

R3 also controls gain. And BTW it does sound good for listen to music!

P.P.S.

I guess this is just a singleton topology that is why it works as it does and works great with a guitar, nice 2nd harmonics!

P.P.P.S.

Using this http://sound.westhost.com/project56.htm and the simulator I get roughly Zo=8Ohm, unity damping factor.

[PRR]

> a simple way of increasing the output impedance without killing "almost" all the gain!

You generally need significant voltage feedback at DC to set the DC operating point.

There's no God-given reason to have voltage feedback at audio.

Your NFB network tends to be a resistor from output to "-" input, then a resistor and capacitor to ground. Make that to-ground resistor zero, the cap quite large. Perhaps 68K, 470u, and <10r.

That sets the DC gain to unity, so the output DC voltage can be set near half-supply.

That sets the AC gain over 6,000, your forward path probably does not have that much gain, so the "voltage feedback" is ineffective for audio, for anything over about 35Hz.

Your plan in #34 has a "horrible" flaw (which may be a happy discovery). The output gain and NFB is very different pull-up and pull-down. Wave is lopsided.

> The speaker qualities does shows up

What, the 2dB bump at 74Hz? A classic Fender Twin shows nearer 6dB bump, because it has barely-hardly-any NFB. Some Ampegs have no NFB and 10+dB bump at bass resonance. This works best with stiff drivers in open-back cabinets. The slump from insufficient baffling is countered by the rise from low damping. (When Marshall used softer drivers that favored closed-back boxes they used more NFB than the Fender they were copying.)

[KMG]

tca, you use not exact emulation of speaker impedance.
This version is closer to real speaker


I use this loadbox for silent testing amps

[tca]

> a simple way of increasing the output impedance without killing "almost" all the gain!

You generally need significant voltage feedback at DC to set the DC operating point.

There's no God-given reason to have voltage feedback at audio.

Your NFB network tends to be a resistor from output to "-" input, then a resistor and capacitor to ground. Make that to-ground resistor zero, the cap quite large. Perhaps 68K, 470u, and <10r.

That sets the DC gain to unity, so the output DC voltage can be set near half-supply.

That sets the AC gain over 6,000, your forward path probably does not have that much gain, so the "voltage feedback" is ineffective for audio, for anything over about 35Hz.

Your plan in #34 has a "horrible" flaw (which may be a happy discovery). The output gain and NFB is very different pull-up and pull-down. Wave is lopsided.

> The speaker qualities does shows up

What, the 2dB bump at 74Hz? A classic Fender Twin shows nearer 6dB bump, because it has barely-hardly-any NFB. Some Ampegs have no NFB and 10+dB bump at bass resonance. This works best with stiff drivers in open-back cabinets. The slump from insufficient baffling is countered by the rise from low damping. (When Marshall used softer drivers that favored closed-back boxes they used more NFB than the Fender they were copying.)

I could set the DC gain to 1 and increase the gain by decreasing the value of R3 but that also makes the amp to oscillate when  the guitar is connected. By doing that the 74Hz bump can be as much as 10dB.

tca, you use not exact emulation of speaker impedance.

I'll check my numbers, thanks.

[tca]

tca, you use not exact emulation of speaker impedance.

Just calculated both versions, blue my sim (Thiele/Small), red yours.



Your version seems much more harsh/drastic approach... I really don't know what is the best approximation to a real speaker.

Cheers.

[KMG]

I have different simulation results, solid lines - amplitude, dotted - phase.
Your resonance is about 70Hz, my - 120Hz (close to real guitar speakers).

In earlier post I show real measurements of guitar cabinet & load box with guitar amplifier having output impedance about 15 Ohm.