Amplifier Design - A Documentation

[Bunkey]

I'm going to try setting these LM386's to a slightly higher gain with a resistor & cap across pins 1 & 8 too, if I can avoid distortion.

Nope.

[Bunkey]

Hmm there's something else going on. A soft fuzz, eminating. Voices. Voices in the distance. Hello?

Hmm.

I've tuned into something, it's all getting a bit... non-linear.. around here.

I think I need another coffee.

[anotherjim]

I have seen an idea to clean the LM386 boost by adding a resistor and cap in series between pins 5 and 8. I have tried it & it kind of works, but only if the cap between pins 1 & 8 has no resistor in series. With resistance in the boost cap, it oscillates big time!

[antonis]

What Jim said.. 
(red resistor & capacitor..)

[iainpunk]

Hmm there's something else going on. A soft fuzz, eminating. Voices. Voices in the distance. Hello?

Hmm.

I've tuned into something, it's all getting a bit... non-linear.. around here.

I think I need another coffee.

i don't know for certain, but it might be corssover distortion. do you have a preamp, what's the schematic and what opamp did you use in the preamp? some opamps like the TL072 don't like capacitive loads, and have some nasty crossover distortion, if you know how it sounds and what to listen for, you can't un-hear it in recordings or even life.
i have done experiments where i have deliberately employed capacitive load to boost this type of crossover distortion, it was meant as the most hideous sounding fuzz ever, i came really close.

cheers, Iain

[Bunkey]

Hmm that's interesting. Looks like a HPF boost? I'm only guessing, as the datasheet shows a bass boost using pins 1-5...

Well I should note that I deviated from the datasheet slightly with the RC values for the zobel network, without understanding what exactly the function of those components is..
I mistook this for a passive filter of some sort and thought I could tame any shrill transients by reducing the value of R - I've done some googling and realised I was wrong, that this is actually in parallel with the impedance of the speaker and is used to flatten the impedance response, so all I was doing was loading down the power amp.
I've been plugged into the 12ohm cabinet for most of this work with just a slight hint of distortion (discounting what's intended) at the standard 20x gain, which cleaned up below 8 on the guitar and wasn't too unpleasant. I then tried a single 8ohm speaker earlier and that's where the issue became apparent; the distortion was worse and wouldn't clean up. This is what led me to believe it might actually be a product of impedance across the output, or a lack thereof. The gain I tried seperately, it was only 30x and just exaserbated this issue.

I'll have to get some more resistors before I can sort this out.

I've purchased one of those cheap LCR T4 meters to get an ESR reading of the various caps I was auditioning on the regulator output too - I could probably use it to measure the inductance of the speaker coils and work out the actual values required for the zobel  Just thinking out loud here really.


The pre-amp is a discreet design that I've been working on in some capacity or another for a long frigging time and I'm actually quite proud of it - which is exactly why I haven't posted a full schematic of the amplifier here Maybe when it's finished and I've got a bigger better amp in the works I'll share it. As I said in the OP it's just a simple design using dual BJT's, diode clipping and a LPF/HPF crossover tone stack. You could build an amplifier just using an off-the-shelf Electra or an LBP1 schematic for the pre-amp if you wanted to, that's kinda the beauty of this project and why I'm trying to keep everything (including the power amp chip selection) inherently simple.


As for the voices possessing the amplifier, well.. I'll light a few candles and say some shit backwards, maybe they'll leave.

[Bunkey]

Hmm again.

Iain you might actually be onto something with crossover distortion. I'll try and record a clip today and see what you make of it..

There seems to be a low threshold below which the distortion suddenly disappears. I have some Ge diodes clipping the signal to ground (vf ~ 300mV) and I'm wondering if there's some audio present on the ground rail which is leaking back through into the pre-amp. It's very likely that I just couldn't discern it as well through the small speakers as I can through the sensitive 8ohm guitar speakers so it kinda got lost in the mix til now.

*Edit: There are a few possibilities including supply instability. I'll demonstrate what it's doing soon - its developed a number of 'features'.


The Zobel theory is a bit of a grey area...

I'm using an 8.2ohm resistor and 47nf capacitor.

Calculating for the 8ohm Jensen speaker using its datasheet would suggest a 7.5ohm resistor with a 3.3uF cap, which means either the LM386 datasheet is acounting for a speaker with a very different inductance, or the zobel network isn't intended to flatten the impedance response at all...

[iainpunk]

the candles and backwards speaking might help, i guess shielding the circuit is better tho. its probably picking up some AM radio.

the zobel really needs more than 10 ohm, and less than 22, if the resistance is to low, the Capacitance will take over and cause more crossover distortion. also, its quite normal to expect distortion on the 8 ohm, since its bridged, the amp can only "see" 4 ohm, which is to low for the lm386, so distortion is guaranteed! in bridged mode, you NEED atleast 16 ohm.

also, the inductance you measure with and inductance meter isn't there anymore when the coil is actually moving due to the current. it turns in to resistance.

cheers, Iain

[Bunkey]

It is magic. I conjure the spirits. I have bridged dimensions.



PS. That confirms a lot about the impedance. Thank you.

[Bunkey]

We'll just gloss over how flawed my understanding of fundamental concepts is huh...


*The inductance, and the parallelled amplifiers.

The shielding I have a handle on. Let me just claw back a modicum of respectability there 

[Bunkey]

Another day, another awesome resource: http://philsal.co.uk/wp-content/uploads/2018/12/Analogue-Guitar-Pedal-Design.pdf

I spend so much time at the deep end trawling through numerical equations and scientific publications just trying to understand the simple things I need to know to get the job done. I've said it before but I find the process exhausting, absorbing that sort of material and hoping the answers might appear between the lines. I can do numbers but it's not an intuitive process for me.
That's why I find it so refreshing, like the valvewizard site, when I come across something such as this which explains circuit design concepts in a practical sense relative to what it is I'm actually trying to do. It's easy to get lost on a forum where everyone seems to have a given understanding of electronic theory, it's almost novel to think that we're basically all guitarists here; I mean, how many guitarists do you actually know irl? Yet here we all are in one place, presenting ourselves as engineers and mathematicians instead. You'd be forgiven for forgetting, I find that quite funny.


Anyway, I took a break from building amplifiers to work a little bit on building an amplifier..

This one's my 1982 A&R Cambridge A60 stereo that I found in a skip (along with a Linn turntable - not even kidding) and lovingly rebuilt last year. No thread of mine passes without its mention.
I received a delivery of some nice 20awg cotton/PVC solid core wire for the guitar amp yesterday, so what better way to show it off...

[amptramp]

Cotton-covered wire is used a lot for the restoration of antique radios.  A number of places cater to those of us who restore radios and this is one of the products they carry.

[Bunkey]

The boot laces get me 

I had some all-cloth 22awg in the past that was a pain to strip without fraying. This pvc type with the cloth braid though is the single best wire I've ever worked with.

I'm guessing you're not in the UK, but if you are Home Bargains has a 2 pack of tiny Rolson folding knives in the DIY aisle just now and it's a real dream combination.


There was a good wee place like that in Hexham actually but it seems to have been swallowed up by the rising tide of late.

[Bunkey]

I had made a rather large oversight on the original supply which was helped picked up by Rob on another thread..

The 100r resistors I had in line with the supply for noise suppression were, at a theoretical 500mA load from the bridged LM386's, trying to drop 50v across themselves and thus limiting the supply to 180mA at 18v; preventing the op amps drawing the current they need and causing all sorts of mess above 1/3rd or so volume.

I've removed them and replaced with 1ohm 2W resistors I had spare (also tried without). I've also swapped the zobel resistors for 15ohm which had much less to do with anything.
We are on the right track but at full whack I'm getting some crazy distortion which also sounds supply related...

It's prominent when there's a tiny signal input (almost like a gated sputteriness), playing loud is raucous but more like the real tone when the amp is quiet, likely the electronic sounding distortion is just getting swallowed up in the mix.
It is loud now.


So, Pop Quiz time:

What exactly is causing this crazy distortion when the volume is way up near the top?

Makes for some cool AF phasery vibes but heck it's not intentional.
Bridged LM386's. Wall supply is rated 600mA @18v. Everything as above. Wild plectrum scrape to keep you on your toes.

Am I just trying to draw too much current from the wall or is this what op-amp clipping sounds like when there's no current limiting on the supply? It sounds like a flat battery to me

[Bunkey]

Onset is much sooner into an 8ohm load than it was with the 12ohm.

I guess that's to be expected with a supply induced 'feature'.

[Bunkey]

*Creative Possibility*


This one is called: 'Even a Stopped Clock Tells the Right Time Twice a Day'

Seriously though I need to fix this.

Onset @ 8ohm (Bridged = 4ohm) is about 9am (shown here)
Onset @ 12ohm (Bridged = 6ohm) was about 3pm (shown previously)

...I think I'm literally just below the threshold of what what of impedance the bridged LM386's can handle at full volume (though the datasheet says 4ohm).

I also think a little bit of current limiting on the supply was no bad thing so that might be something to experiment with...
5.6ohm is as big as I can go without dropping more than the required headroom for the regulator.
30ohm would limit an 18v supply to the rated 600mA of the wall transformer but it would also drop 18v across itself at full load so I'm not sure how the finer points of that work... 

[iainpunk]

that phasery clipping reminds me of moving bias. if you add an LFO to a guitar signal, which is then clipped, you have some moving asymmetry, which gives such a phaser-like sound, but its generally less pronounced, and overwhelming.

i'm not saying its that, but it might be.

cheers, Iain

edit:
the LM386 can handle 4 ohm at 4v Vcc, and only the version with the N-4 suffix, check your chips for the suffix!
if it doesn't have that suffix, it can't handle the 18v, just 12... and 8 ohm, not 4
also, it can only reach the full 1W of power at 32ohm!!!

i think another power chip would give you more of everything, power, headroom, less distortion, better sound quality etc..
i recommend the TDA2040, it can run from 9v to 40v

you make me want to give my old DIY amp another shot, inductor loaded class A amp.

cheers, again

[Bunkey]

Cheers Iain, yeah these are the N4's.

I have asymmetrical clipping in the preamp so based on what you're saying it could just be that running into the available headroom of the chips.

I've tried a single LM380 and it is better quality and little more well-behaved but it was a very hard sound into the tiny cab.. The LM386's are a little more compressed and rounded.
I'll have to audition that again now that I've lifted the current limiting off the supply, maybe it'll sound nicer.

Will try pairing a couple speakers for 16ohm too and see what happens.

The bridged configuration seems to impart more of an interesting and open character than a single chip, which is very much focused and dare I say boring to listen to..
I'm inclined to make this work with a 16ohm load but hmm we'll see.

Where are you getting the minimum impedance for a given Vcc btw?
The TI sheet seems to imply that anything from 4ohm is acceptable @ any supply 5-18v.
Not saying that's wrong - I'm just curious to save myself a bit of trial-and-error.

[iainpunk]

Where are you getting the minimum impedance for a given Vcc btw?
The TI sheet seems to imply that anything from 4ohm is acceptable @ any supply 5-18v.
Not saying that's wrong - I'm just curious to save myself a bit of trial-and-error.

the data sheet tells you that the minimum impedance should be 4 ohms, but not what the power is which it can actually supply to that load.

when there is a Sinus wave of 2v RMS (2.83v 0-to-peak 5.66 peak-to-peak), there will be 1W on a 4 ohm load, that means that everything more than 5.66v peak-to-peak will distort and heat the chip and cause trouble. to keep the trouble from happening, you should use higher impedance, like 8 ohm, which will reach 1 watt with a sinus wave of 2.8v rms, (4v pk or 8v pk-pk)

cheers

[Bunkey]

Woohoo! A slightly clearer shade of mud

So into my bridged config with the 12ohm load, where each chip sees 6ohm:,
The limit for one chip would be 6.9Vpp @ 1W and looking at the datasheet the Vcc to achieve that would be.. gawd the best I can do into a 6ohm load is about 5Vpp with a 9V supply for 0.5W power, after which there's no increase in output no matter what the Vcc.

Right I'm starting to see the problem here. Starting to realise that 'overvolting the chip' as Paul said ageees ago doesn't actually mean exceeding the maximum rating either but rather there's no frigging point to it unless the load is adequate, it's just generating heat and causing problems like you say.

ffs


*Closes LM386 datasheet*
*Opens LM380 datasheet...*

Enter stage left: Another coffee.