25 watt stomp box amp design

[rring]

Someone might like this.

Its a battery powered 20-30 watt guitar amplifier in a 1590BB enclosure. Its nothing revolutionary, but I was careful with the gain distribution and the tone stack has a nice range of control. The Power amp IC is a differential output car stereo amp. I used a FV-1 for reverb but you could replace that with a belton brick or omit entirely. With my layout there is still room to increase the board size significantly. This is much like other portable amps I have designed, but its louder and has a better tone stack IMHO. I just found a great battery at amazon for $30.. it a replacement battery for a Roomba robot vacuum cleaner. It will run the amp for hours totally cranked and its not big or heavy.

Link to a Video Demo... sorry I'm not much of a rocker

https://www.youtube.com/watch?v=GlwdjTFVmsA

[J0K3RX]

Wow!  That sounds excelent, very natural and warm!! Great playing also! Very interested in this design, would love to see what's under the hood... Schematic etc...

[Kipper4]

Brilliant Thanks Ray
+1 on this ^ too.

[Ice-9]

Sounds great rring, nice little amp. Thanks for sharing. Could also be worth adding the pot controls to the FV-1 and a switch for effect selection as you could have chorus or delay as well.

Wow!  That sounds excelent, very natural and warm!! Great playing also! Very interested in this design, would love to see what's under the hood... Schematic etc...

It says in the you tube vid where the schematic is JOK3RX. I can't remember the url but it's at the start of the vid.

[rring]

I thought I attached the schematic?? here it is. My layout is surface mount but is not that complicated a thru hole version could be done in a couple of hours.

[Kipper4]

I thought I attached the schematic?? here it is. My layout is surface mount but is not that complicated a thru hole version could be done in a couple of hours.

Yes please Suggestions for thru hole chips would be good too.

[rring]

Yes I am  under-utilizing the FV-1 but I was afraid of too many knobs. I put some filtering on the FV-1 to reduce the quantized sound you get with high frequency content with heavy digital reverb in the mix.

[rring]

All those parts can be obtained leaded except the power amp but it is a square inch in area and is easily treated as a leaded part. On my layout I flipped it upside down so that the large ground plane tab (on the bottom) is facing up. I took a piece of 1/2 inch copper pipe an soldered to it directly as a heat sink. This touches the inside of the enclosure - so it stay cool even when you blast it.

[midwayfair]

Jeez, another home run, Ray! A battery-powered amp in a tiny package with built-in effects. 

[rring]

Darn it here is a link to schematic so it can be viewed properly
https://circuitsaladdotcom.files.wordpress.com/2014/09/guitar_porta_rev1-1-6.gif

[aron]

Sounds great!

[Ben N]

Fantastic. Gut shots? Did you have to do much for heat dissipation?

[rring]

The amp is pretty good- the data sheet show 50% efficiency at low levels and gets closer to 70% when driven very hard. That is per the data sheet- I have not verified it. Seems suspect with my understanding of linear push pull amplifiers. Having said that, it does well. With the piece of copper pipe bonded to the IC ground PAD and then touching the inside of the case..it gets warm but only mildly. One big factor is the guitar has a lot of "off time" when playing - as compared to a continuous source like a test  generator or recorded music. So the average power is less than the peak.

[R.G.]

The worst case for an AB amplifier is a square wave at 50% of the power supply. Sine wave testing and ordinary music testing never try this. Fuzz units, however, can supply this waveform, so it's a good practice to allow for it in a guitar amp.

The average power is always less than the peak. Normal recorded music has a "crest factor" of about 20 db - it hits peak powers of 10X the average power out. Most hifi gear takes this into account when specifying transformers and other power supply parts. However, compressed, distorted guitar signals in a band situation can easily have crest factors of only 6-8db.

The big thing that would be an issue here is whether the power amp chip has a thermal shutdown inside the chip. If it does, great, rock on. It will quit when it gets too hot. If it doesn't, you may have to replace it every so often.

A really good tool for power amp experimenters is the remote IR thermometer. These can be bought for under $50, and sometimes under $30. I was just using mine this week in testing the temp rise on the repair boards for Thomas Organ Vox amplifiers I have in test. I was able to find the temperature rises on parts and figure out which ones needed heat sinks and which didn't. I highly recommend the remote thermo. Great tool.

[rring]

yes I was talking a little sloppy... but I was just thinking of a continuous test signal with no dynamic variation vs a guitar which can be idle and or vary significantly with regard to dynamics

the chip has a thermal shut down - which is good, but with my heat sinking it doesn't get very warm - I think its fine. It does get warmer when I run MP3's through it for 30 minutes vs me playing for an hour. But I stop and fiddle and get loud and then quiet...etc.

[PRR]

> data sheet show 50% efficiency at low levels and gets closer to 70% when driven very hard. That is per the data sheet- I have not verified it. Seems suspect with my understanding of linear push pull amplifiers.

TDA7391 data-sheet:
http://www.st.com/web/en/resource/technical/document/datasheet/CD00000162.pdf

I see eff falling under 10% as power falls below a Watt.

The "ideal" efficiency with un-clipped Sine should hit 78% (IIRC, don't make me hit the books; Pi/4 may be the true number). IIRC, then 70% is less than ideal (but better than many-many other audio amplifiers measured honestly).

Efficiency is high because they want the most sound within the car-radio's fixed low supply voltage, so voltage-losses are held to a minimum. They get 11.6V out of a 14.4V supply, 2.8V losses or (because it's a bridge) 0.7V loss in four places (more likely 0.4V up from zero and 1V down from +14.4V).

> worst case for an AB amplifier is .... Sine wave testing and ordinary music testing never try this

Yet perversely, the FTC required long run at 1/3rd power, sine, which is very-near the worst case which we KNOW is never hit in nice reproduction.

Distorty geetar and boom-cars (which FTC never ruled on) may linger near worst-case long enough to hurt. This doomed some of the first transistor car and guitar amps; any sold now have adequate defenses.

The datasheet shows only a tru-hole "TO-220"-11pin (Multiwatt11) package, so it sure can be done hammer-nail instead of watchmaker work.

> I have employed negative feedback from the speaker back to a discrete stage driving the final amplifier. This is a common practice in tube amps to flatten the tonal response of the output transformers, but I find it affects the over all sound in a pleasing manner.

It is often (not always) done in tube amps because a naked pentode's high Z and the wrong speaker can be very nasty screechy, or bass-slap.

However you rarely see a transistor amp without NFB, and never a chip amp. They all have internal NFB (LM386) or NFB in essential external parts (TDA202 et seq). Generally pretty heavy NFB, over 20dB. (Contrast with 5dB-10dB in tube guitar amps, 20dB max in hi-fi bottle amps.)

> negative feedback from the speaker back to a discrete stage

I wondered if your NFB loop did anything at all. The power chip has gain of 32 (one side), the JFET about 4, so gain inside the loop A is about 140. R26 R25 define NFB ratio of 101:1. A*B is 1.38 or about 2.8dB additional NFB on top of the 20+ dB NFB included in the power chip. I assert it makes little difference. However it is a neat clean way to DC bias the JFET.

> The Baxandall

This is a James.
http://home.comcast.net/~stphkeri/James_SimpleToneControl.pdf
http://www.ampbooks.com/home/amp-technology/james-tonestack-analysis/

James requires 10% Audio-taper pots to be "flat" when knobs are centered. Perhaps not a big deal in guitar.... indeed a James mis-built with linear tapers will have a scoped midrange much like the bigger Fenders' nominal response.

Baxandall wrote about a tone network wrapped around a hi-gain stage and being the NFB for that stage. I have not seen anything from him mentioning James-like networks.
http://hifisonix.com/wordpress/wp-content/uploads/2013/05/NegativeFeedbackTone.pdf

I would not even say they are "similar", except that they are both bass/treb 'boost'/cut and do not use inductors. Within those constraints "all" controls are similar; but the Bax works very different to the James.

Today the Bax is commonly done with op-amps. You could do that, and get flatter centered response with more common Linear-taper pots (the reason the Bax became THE hi-fi tone network when Stereo came in).

However if you did that, it is some redesign and considerable re-layout (and you already got yours done). And the James is widely used in guitar amps.

Also you would have to find some place else for the R14 R13 C11 network which is what *makes* this a singing guitar amp instead of a dull hi-fi amp (or car radio).

The R7 C3 time constant looks a little brutal to me--- rolls 2 octaves off the guitar's bottom. Choice of C3 value should be "to taste", depending on speaker size, desired tone, and if playing with a bassist or not. A 1u/2u/5u switch could be another knob to play with.

[PRR]

The TDA7391 includes clipping detection, which is easily implemented.

In stage-amps, if you are not clipping, you coulda brought a smaller amp.

Refer to datasheet Figure 3. Ignore "TO uC". Replace the 4.7K resistor with a 3.3K resistor and LED in series.

This detector *seems* to be "slow". It does not say "about to clip" (as good studio gear does). It says "clipping 10% of the time!". So lack of blink does not mean dead-clean. However occasional blinks on transients are probably not "dirty".

On hi-fi this simple blink is pretty useless. In mixed music the peaks are quite narrow (even on loudness-war mixes). You will miss the brief blinks. However a nice simple overdrive guitar will probably give visible blinks at the edge of overdrive. Sustained head-basher music can keep the LED lit much or all the time.

[bool]

Clipping indicator LEDs are easily and very cheap-ly driven with a 555 monostable.

One caveat: a regular non-cmos 555 has a tendency to introduce audible "clicks" into whatever circuit it's built into if not generously bypassed.

[rring]

PRR - good comments

I used a different package for this part then the data sheet you posted shows

The TO-220 package may a good alternate for other builders

Just to clarify this is the data sheet for the amp - I used (the only difference is the package) which is a surface mount version of the same part:

http://www.st.com/web/en/resource/technical/document/datasheet/CD00001662.pdf

the efficiency is zero... at no output because the  bias current is about 60mA! - so I was just throwing out 50% because most people will use it in that range.

It wasn't clear to me that the how much if any negative feedback was employed with this IC. I could not tell from the block diagram - so that is interesting information(regarding the amount of negative feedback typically employed).

My added negative feedback is not much - it does have an effect... I adjusted the feedback level until I could hear the difference and I liked the way it sounded. I was just curious to try it and I intentionally DC coupled it to affect the dc bias of the jfet - as you mentioned.

The other tonal shaping aspects (the small caps in the op amp feedback..etc) were all chosen to sound good to my ear- by all means adjust them as desired -  I started out with 10uf for C3 and C11.

you pointing this out is good - so other builders can see where they might want to make changes.

My bad on the nomenclature - I think of the James and Baxandall as the same because they behave the same- in that they leave the mid constant. I am going to be more accurate in the future with my nomenclature.

I don't care about flat center(linear pots) so much... as long as all of the desirable tone control resides within the extremes(which it does) besides with all of the high end peaking - flat is not flat anyway.

I don't get how the taper will create a mid scoop because only the rotational position of the shaft changes for a given wiper value -if the taper is changed - in any case I just used linear taper pots and they worked well and I moved on.

I intentionally used the passive topology, instead of in the feedback loop of and op amp because I wanted the attenuation. The FV-1 has little head room and I wanted to manage gain so I could have gain up front to improve noise figure of the entire circuit

[PRR]

> how the taper will create a mid scoop

Imagine the James. Resistors and caps are in 10:1 top:bottom ratios. This gives 1:10 midrange loss. To get 1:10 bass or treble loss we use 10% audio taper pots.

Now instead use linear (50%) pots.

Set the knobs at "5" on a 0-10 dial.

Midband loss is 10:1 (20dB) and bass or treble loss is 2:1 (6dB). The extremes of the audio band are 10dB-14dB louder than the middle of the audio band. "Mid-scoop".

This is solely "user interface". You could instead mark the knobs at 10% and get "flat" at that mark. However then you have 20dB of boost spread over 240 degree rotation and 20dB cut in only 27 deg rotation. Perhaps if you only want fine degrees of boost or major cut this would be best. However it is intuitive to have roughly similar dB/degree both ways.