"Small Audio Amplifier" from Ludens.cl

[tubegeek]

https://ludens.cl/Electron/audioamps/AudioAmps.html

I built the 2nd circuit shown at that link, after trying a LM386 chip amp and not really liking it. The application is a bench amp for testing gizmos, so I don't have to walk across the room and fire up the "real" amp every time I want to check something out. I had a small combo cabinet that I had gutted with an 8" speaker and it sounded godawful with the "Smokey" amp I had used briefly - no possibility of a clean listen to what was actually coming into the amp.

Since the whole point of this project was to promote laziness, I didn't want to build anything too extensive or buy any parts. So this circuit - which has intrigued me for a while with its low parts count - called to me.

As presented at the ludens.cl site, this was designed to run from a single supply, but I decided to make a virtual center ground. Since I didn't have any single op-amp chips at hand anyway, there was only the addition of 2 resistors to pop open the piggybank for, and I reckoned I'd swap that for an input coupling cap. (This split-supply approach is suggested in the original article.) I had a wall wart that was giving about 17.5V into a light load, so I thought I'd do it as a nominal +9/0/-9 setup. NE5532 is rated for +/- 20 V so that's OK.

I was stumped when I first turned it on and got an instant of audio and then silence. I did a whole bunch of checking until I discovered that one leg of the IC was crumpled up when I put it into its socket. Straightened out, it works just fine! The circuit's huge headroom comes along with low sensitivity. I'll have to decide what I want to do about that after trying a few pedals with it - the guitar straight in is not particularly loud at all. If I replace the 15K resistor with a 50K pot, that'd work as a gain control, right?

I built it on vero. Here is the schematic as built and the vero layout.

[jonny.reckless]

A slightly different topology will give you much less crossover distortion (esp at high freq), with about the same component count. I used this trick in the Boba FET mini guitar amp output stage.

https://www.diystompboxes.com/smfforum/index.php?topic=118627.msg1104856#msg1104856

[tubegeek]

A slightly different topology will give you much less crossover distortion (esp at high freq), with about the same component count.

Thanks for that - but, look, I've got nearly two dollar's worth of parts and over an hour invested in this one - you think I'm going to throw all that away?

.... let me see what a vero of yours would look like, I'll try laying it out tomorrow....

Am I right that swapping the 15K for a 50K pot would allow me to increase/adjust the gain a bit?

[jonny.reckless]

Yes but beware the NE5534 Is only compensated for closed loop gain of greater than 3. You need to add a 33pF capacitor between pins 5 and 8.

[tubegeek]

Yes but beware the NE5534 Is only compensated for closed loop gain of greater than 3. You need to add a 33pF capacitor between pins 5 and 8.

Thanks for that - but it's actually a NE5532 dual that I'm working with - one section is doing the rail splitting, the other is the input buffer. So I have to take what I get. But I am not going lower than the original gain with the pot I installed: I added the 50K pot in series with the 15K resistor, between the 15K and pin 6. I couldn't foresee needing any LESS gain than I was getting.

I tidied up the build, gave it a box and a power switch, and attached it to the repurposed combo cabinet. It's working out really nicely for my purposes, it'll keep me from getting in trouble playing too loud late at night.

The only misbehavior I've seen so far is at the BOTTOM end - when I ran a ~10 Hz tone from my generator into it, it wouldn't stop oscillating on its own until I turned it off. I guess those 470uF power supply caps aren't big enough to keep it from motorboating. I may put 47R or 100R in series with each power supply rail before the caps just to see if that kills the motorboating. (I wonder if I have a common mode choke somewhere?) With anything in the normal guitar frequency range it's not a problem though. I'll try scoping the HF behavior at some point but I don't really hear any signs of trouble.

I may yet try the low-crossover-distortion circuit yet though Jonny, don't despair! Meanwhile - a project that went together EASILY for once! I've been banging my head against various things lately - this one was relatively simple and easy to get working.

[tubegeek]

A slightly different topology will give you much less crossover distortion (esp at high freq), with about the same component count. I used this trick in the Boba FET mini guitar amp output stage.

https://www.diystompboxes.com/smfforum/index.php?topic=118627.msg1104856#msg1104856

jonny:
Having trouble getting this running (using my own vero layout.) I have a power supply delivering almost 30VDC no-load, I thought it might be good for this project. Getting weird behavior - rail slamming on the virtual ground, overheating transistors. At first I thought that the 2n4001/4003 complimentary pair just wasn't up to the task but I just got in some TIP31C/32C as you spec'ed for the Boba FET and swapped them in (I corrected the different pinout with some leg-bending.) No luck: no output, virtual ground misbehaving (not settling to the center of the two rails.). The 5532 should be OK with ~29V between V+ and V-, although I guess  the spec (+/- 22) could have been exceeded if one rail or the other was shorted to one of the DC inputs and if there resulted in + or - 29 on the chip.

I adopted your feedback values of 100K and 4K7 - the 15K/3K3 having been found to be too low gain previously. The virtual ground is connected to the 4K7 feedback shunt, the (2M) input load resistor, and the speaker (-) terminal only. On the 5532 chip  pins 1 & 2 are connected to the virtual ground, pin 3 is connected to the junction of two 150K resistors from DC input + rail to DC input - rail. The other half the 5532 is the signal buffer/driver as drawn above.

Troubleshooting questions: I'm not used to seeing the chip supply voltages connected to an active element at all, is there any inherent problem with using this as both a rail splitter and a signal buffer in the same chip, powered as per your plan? Also: my standard .1uF ceramic bypass cap - it goes between the chip V= and V-, not the input V+/V-, right? (in between the 100R's on the chip pins as usual, not on the input side?) Final question (I hope) - is the 1K output loading resistor in the Boba FET necessary for this build? I've been testing with an 8 ohm power resistor on the output.

It may well be a bad vero plan or shorts/opens, I will keep at it.

Plan of attack: try it with a lower voltage power supply/get out the jeweler's loupe and scrutinize the board/swap for a different chip/rebuild on a new vero (or perhaps just perf?) layout.

[lietuvis]

Hi Tubegeek if you build the schematic in that little picture above the amp will not work, it's simplified and you need few more components if running from single rail power source. did you look at the actual power amp section schematic of Boba FET provided in here https://drive.google.com/drive/folders/0BzLBW4IzZybIR19RT1QtMVhPdWM?usp=sharing
hope that will help you to get your amp running.

[tubegeek]

Hi Tubegeek if you build the schematic in that little picture above the amp will not work, it's simplified and you need few more components if running from single rail power source. did you look at the actual power amp section schematic of Boba FET provided in here https://drive.google.com/drive/folders/0BzLBW4IzZybIR19RT1QtMVhPdWM?usp=sharing
hope that will help you to get your amp running.

Thank you! - I did take a look at the Boba FET - note that I am not using a single supply, I'm building it with a + and a - dual supply (virtual ground using 1/2 an NE5532 opamp as a rail splitter.) So I omitted the blocking cap at the input and I returned the 4K7 feedback shunt directly to the center virtual ground instead of to ground via a cap as shown in that single-supply version.

Anything else I'm missing? (Maybe I should actually just convert it to the single supply setup, come to think of it. Take the rail splitter out of the equation.)

The fellow at ludens.cl wrote me today (I had emailed him to discuss the other circuit at the top of the thread) - he suggested that the opamp rail splitter may be challenged to source enough current to the virtual ground as I built it.

I suppose I could put an LM386 as the rail splitter instead... that should be able to source a lot of current, right?

I was just trying to use both the halves of the 5532, I don't have any single op amps here. Maybe I should just use it as a buffer with gain for the input - that's probably a more sensible idea.

[lietuvis]

with dual rail power supply you don't need to use any virtual ground. Can you draw what exactly schematic you use and how you connect power supply, speaker etc.
Here is the image for very similar scenario you should be doing just with much more power.
https://www.eleccircuit.com/wp-content/uploads/2015/04/the-75W-75W-stereo-power-amplifier-project-using-TDA2050-and-2N3055-MJ2955.jpg

[jonny.reckless]

The voltage across those 100 ohm resistors in the power supply should be about 400mV at idle. It's only when current is delivered to the load that the current dumping transistors turn on. When running into an open circuit, the transistors should be cold.

Also take a look with a scope and make sure you are not seeing oscillation. Depending on the layout and grounding, parasitic oscillation can occur anywhere from a few kHz to a few hundred kHz. If this is the case, a 100pF cap between base and collector of each output transistor will usually tame it. I've never had much difficulty getting this simple output stage to work on 18V single supply.

Here's the schematic for running with a single supply. It will work for up to 36VDC.


You shouldn't use a virtual ground with this topology since all the speaker current will be referred to it, requiring as much power as the output stage. Instead just bias the output to VDD / 2 and use a large output capacitor to drive the load as shown above.

You can use my sketch if you have a full wave rectified dual power supply, i.e. with 2 large reservoir caps, otherwise use the Boba FET approach.

[Rob Strand]

Also take a look with a scope and make sure you are not seeing oscillation. Depending on the layout and grounding, parasitic oscillation can occur anywhere from a few kHz to a few hundred kHz. If this is the case, a 100pF cap between base and collector of each output transistor will usually tame it. I've never had much difficulty getting this simple output stage to work on 18V single supply.

It's always wise to use a Zobel network on the output when driving speakers.

[tubegeek]

Great advice folks, much appreciated. I'm going to revamp this project, it's so simple a rebuild is easy enough. FYI I had a nice conversation with the ludens.cl designer over email and he's giving me pretty much the same advice I'm getting here. My next move is just to do a single supply as per jonny's Boba FET schematic, unless I can scrounge up an appropriate AC transformer and do a conventional rectified split supply. If I absolutely HAVE to use the other half of the 5532 I'll use it for input gain/buffer.

I confess I have a distaste for input caps. Probably a stupid plan to follow given that this amp is for testing and I can run into DC leakage on the input from unknown equipment.

Some of the requested info:
Schematic as built:


Vero build top side:


The vero layout as built. Note that when I replaced these transistors with the TO-220 parts, I did account for the different transistor pinout by crossing the legs as needed:

[Kipper4]

Can't wait for the outcome and final schematic, I've been unhappy with the quality of my noisy cricket even for testing, looking to upgrade and add a switch for speaker or headphone. Very handy when late or all night testing.
Well done keep going.

[tubegeek]

looking to upgrade and add a switch for speaker or headphone.

Reminds me to ask: does a headphone send connection profit from a zobel network?

[jonny.reckless]

It's always wise to use a Zobel network on the output when driving speakers.

True for a Hi-Fi amp, I think it's less important for a guitar amp since the speaker is basically inductive at high frequencies (there is no crossover) so improves rather than compromises the phase margin.

[jonny.reckless]

Great advice folks, much appreciated. I'm going to revamp this project, it's so simple a rebuild is easy enough. FYI I had a nice conversation with the ludens.cl designer over email and he's giving me pretty much the same advice I'm getting here. My next move is just to do a single supply as per jonny's Boba FET schematic, unless I can scrounge up an appropriate AC transformer and do a conventional rectified split supply. If I absolutely HAVE to use the other half of the 5532 I'll use it for input gain/buffer.

I confess I have a distaste for input caps. Probably a stupid plan to follow given that this amp is for testing and I can run into DC leakage on the input from unknown equipment.

Some of the requested info:
Schematic as built:


Vero build top side:


The vero layout as built. Note that when I replaced these transistors with the TO-220 parts, I did account for the different transistor pinout by crossing the legs as needed:

The way you've built that won't work because the little op-amp driving your virtual ground can't source or sink anywhere near enough current to drive the speaker. You would be better off doing a bridge tied load with 2 output stages, or just using the Boba FET approach and something like 470uF to 1000uF output capacitor.

[tubegeek]

The way you've built that won't work because the little op-amp driving your virtual ground can't source or sink anywhere near enough current to drive the speaker. You would be better off doing a bridge tied load with 2 output stages, or just using the Boba FET approach and something like 470uF to 1000uF output capacitor.

Live and learn. BTW I stuffed the redux board today, just need to do the enclosure & pots & sockets still. Thanks jonny!

[Steben]

Can't wait for the outcome and final schematic, I've been unhappy with the quality of my noisy cricket even for testing, looking to upgrade and add a switch for speaker or headphone. Very handy when late or all night testing.
Well done keep going.

Did you use the noisy cricket with big cabs?

[Kipper4]

Absolutely not. It would eat my speakers, the thing pops and cracks like a street dancer on a trip.
I've already replaced the circuit once in 7 yrs and since then the input and output caps.
Probably some of it is down to using it a lot for breadboarding experiments. Lose connections and one the fly live componant changes add to the excitement of the poor Cricket.

It's installed in an old speaker from a reel to reel. The tolex still bears the scars from the hinges that made the speakers swing out to the side. In case of an impromptu 70s type party.

You know the kind of thing. Your friend popped to the off license for a party seven of double diamond, two packs of ciggies a bottle of wine and a family sized bag of crisps.
As soon as they hit the door you know it's on, the speakers are being swung out.
It's Saturday night and the cones are pushing air.
I guess this little speakers seen some sights. It must be older than me.
Still going strong. Looking forward to an upgrade. 1 > 2 Watt would be enough me thinks.
Thinking of something  a little more hifi maybe.

[tubegeek]

In case of an impromptu 70s type party.
You know the kind of thing.

I love that description, I feel like I was there!

Thinking of something  a little more hifi maybe.

If you want my opinion, no way. You've already got a cheap alnico fullranger in there, it's gonna be really hard to beat that for guitar fun. Perfect thing for a small amp.