*Project* - CMOY Headphone amp

Quite possibly.

The OPA2134 only has a bandwidth of 8MHz! -3dB at about 100KHz is probably good to aim for. With your current resistor values this would be around 160pF. That sounds like it'd be a non standard value, so try for something close to that I guess.

Lol, you can tell I'm pretty new to this audio stuff...:p

So having a cut-off frequency at 100kHz should filter out anything higher than audio frequency then? If the bandwidth is only 8MHz then why is it even necessary to filter out radio frequencies which start at 100MHz or so?

Thanks again!


Anyway, I've been busy today with uni work so no real progress was made on the amp itself. However, I've now drawn up what I think is a final design for which I'll need a few more capacitors:

design1kp9.png


please ignore the poor quality.:o

I've also been looking at how to wire up the front panel. I can hopefully fit the power switch, power LED, volume knob and two audio jacks all on one end of the container.:)
 
Lol, you can tell I'm pretty new to this audio stuff...:p

No problem mate. I still don't know all that much myself to be honest. Lots of things still confuse me. A good place for DIY audio stuff is... www.diyaudio.com (hope i'm allowed to link there?). Lots of smart people over!

So having a cut-off frequency at 100kHz should filter out anything higher than audio frequency then? If the bandwidth is only 8MHz then why is it even necessary to filter out radio frequencies which start at 100MHz or so?

Well, apparently (wikipedia, who knows how accurate? lol):

Radio frequency (RF) is a frequency or rate of oscillation within the range of about 3 kHz to 300 GHz.

But yes, it creates a gradual roll off with the -3dB point at about 100KHz. The roll off will start slightly earlier though, which is why you need to start slightly higher than 20KHz. -0.5dB at 20KHz is unlikely to be audible to most :)

Anyway, I've been busy today with uni work so no real progress was made on the amp itself. However, I've now drawn up what I think is a final design for which I'll need a few more capacitors:

design1kp9.png


please ignore the poor quality.:o

10nF from each rail to ground too ;) I think they're more important than the rail to rail cap. I seem to recall the rail to rail cap is useful in bridged designs.


Oh, and apparently some headphones like to be driven with an output impedance of 100R. Perhaps a future tweak could be to add some kind of switch on the back to toggle a 100R resistor, in series with the output, in and out of the circuit. I think that resistor was shown in the original schematic, but it's gone now :)
 
Oh, and apparently some headphones like to be driven with an output impedance of 100R. Perhaps a future tweak could be to add some kind of switch on the back to toggle a 100R resistor, in series with the output, in and out of the circuit. I think that resistor was shown in the original schematic, but it's gone now :)

I suspect this is something to do with power matching, although I'd have thought simply adding a resistor in series with the ouput would not simply set the output resistance to that value...

That tactic was mentioned in the tutorial I was orininally following as a method for reducing noise. I tried it yesterday and it didn't make much difference. However I just tried it again now and it reduced the slight background hiss by loads! Using a 220R has almost completely eradicated the hiss. I don't know why it made a difference now and not yesteray, but one of the other modifications I made must have made a difference. Either way, the background hiss on the Vaughan Williams .ogg I'm listening to at the moment is louder than the amp's hiss, which is awesome!

Oh, and thanks for that link.:)
 
I suspect this is something to do with power matching, although I'd have thought simply adding a resistor in series with the ouput would not simply set the output resistance to that value...

It... pretty much does as far as I am aware :) Though it does also waste power, which is dissipated in the resistor. Not so much of a problem with headphone amps.

edit - Although, maybe if it is inside the feedback loop this wouldn't be the case. If it is, then what I have said below would only apply if this resistor is outside of the feedback loop I think.

That tactic was mentioned in the tutorial I was orininally following as a method for reducing noise. I tried it yesterday and it didn't make much difference. However I just tried it again now and it reduced the slight background hiss by loads! Using a 220R has almost completely eradicated the hiss. I don't know why it made a difference now and not yesteray, but one of the other modifications I made must have made a difference. Either way, the background hiss on the Vaughan Williams .ogg I'm listening to at the moment is louder than the amp's hiss, which is awesome!

It can also do this too as it goes. It can attenuate RF from entering via the output leads. In the case of speaker cables, these act as huge antennas :)

It also isolates the amplifier from inductance and capacitance at the output, both of which I believe can actually lead to instability, in the form of 'oscillations', where the amp sits there generating a sine wave on it's own, with no actual user input signal. Generally regarded as a very bad thing with audio amplifiers. It heats up the output stage, especially with very high frequency oscillations, and can cause the transistors to blow in some cases. It will also increase distortion, both thermal distortion and, I guess, the fact that it will have to 'work harder' to generate the actual signal, and then the oscillation riding on top of that.

If it is indeed unstable and you don't want that high of an output impedance, you can put an inductor in parallel with this resistor. Lower frequencies will pass straight through this inductor, but as the frequency rises, the inductor will block it and the resistor will attenuate this signal.

Did you make absolutely no changes at all to the circuit, or anything else? Hiss can be caused by oscillation or general instability. It may be that something, RF?, is actually interfering!

If you've changed the circuit, it may be that the change has made the amp unstable. Though if you have changed it, I guess it would be adding the things I suggested to prevent RF lol

If you know someone with an oscilloscope you could borrow, these things can be useful to show you what's going on :) Fortunately for me, my uncle is an EE, so I scrounged one of his older cheapy ones off of him. It does the job :)

Oh, and thanks for that link.:)

No problem!
 
Another thing, I would edit it in but it could turn out a lengthy reply and I wouldn't want you to miss it in the edit.

Grounding. It can actually make a difference to audible noise. In the audio world, star grounding is a good way to achieve good grounding. Star grounding is where you take each ground to one point, the 'star' ground point, and then tie that point to ground.

Mainly you will want to "separate" the grounds of small, sensitive signals, like the input and feedback, from that of the noisy power supply. The signal grounds are more of a reference point really, so these need to be as clean as possible. The grounds still have to be electrically connected in some way.

I'm still not anywhere near an expert on this, but I have had good luck with the following:

- Create a main star point. This could be an unused trace on your prototype board.
- Create a point at the power supply, just after the main filter capacitors, and take this to the main star point with a nice thick wire.
- Create a star point for input signals. As it's a dual opamp, you may choose to have two of these. One for each channel. This could be another one or two unused strips on your board.
- Take the input returns, including the input socket ground tabs, to the input star point. If per channel, take it to the input star for that channel.
- Take all opamp input grounds, volume control ground and feedback network ground, to the input star point.
- Take the output signal return tab to the main star point.
- Take any decoupling capacitor returns to the main star point. Ideally I would think you actually want the main star point quite close to these capacitors if you can. Use thicker wire.
- Connect input star point(s) to the main star point with a low valued resistor, say 1 to 10 ohms, to keep it clean.

I think that covers it in this case :)
 
Disaster! I think I've destroyed the Op Amp. I don't know what's wrong with it, but no matter how many times I rebuild the amp, I get very little gain and a stupid amount of distortion.:(

I'm going to keep trying to fix it for a little longer, but I'm afraid I might have to buy another one...:(
 
Disaster! I think I've destroyed the Op Amp. I don't know what's wrong with it, but no matter how many times I rebuild the amp, I get very little gain and a stupid amount of distortion.:(

I'm going to keep trying to fix it for a little longer, but I'm afraid I might have to buy another one...:(

Did it just die all of a sudden? Or did you change something and it died?
 
Phew. Another rebuild seemed to fix it. There must have been a mistake somewhere. Perhaps I'm just getting tired, but I seem to be making more mistakes now...!


Anyway, I tried bigger coupling capacitors but they made little difference to the sound at all, and were physically bigger too. So I'm back with the 0.1uF ones.

I also put those 10nFs in the power supply, no problems there.:)

Tomorrow I start designing the final board for use in the amp its self.
 
A short update:

I've now got the amp fully working again (still in the prototype stage) with a couple of modifications. I implemented the star-grounding system that markie recommended, and also added a few capacitors to improve the high frequency response and to hopefully reduce RF interference. I also got hold of a 10k logarithmic potentiometer so the volume control now accurately reflects the dB scale.

One thing I forgot to do was get a bigger cap to replace the 220uF one in power supply. Does anyone know if simply using two 220uF caps in parallel will have the same effect as adding a single bigger one, in terms of aiding the "big hit" response? I only ask as I have two 220uFs from the original rudimentary power supply I built.

I've also designed the final layout for the strip-board that's going in the enclosure, and made a 'mock up' of it in the actual stripboard I'll be using:

dsc06074mediumdu4.jpg


dsc06073mediumxb0.jpg


It looks extremely cramped: that's because it is. The finished product will look a lot neater: all component legs will be insulated and the wires will be the minimum length possible to reduce clutter. I'm confident that I can make this work...:o
 
Last edited:
A short update:

I've now got the amp fully working again (still in the prototype stage) with a couple of modifications. I implemented the star-grounding system that markie recommended, and also added a few capacitors to improve the high frequency response and to hopefully reduce RF interference. I also got hold of a 10k logarithmic potentiometer so the volume control now accurately reflects the dB scale.

Out of curiosity, did it make any difference to the hiss?

One thing I forgot to do was get a bigger cap to replace the 220uF one in power supply. Does anyone know if simply using two 220uF caps in parallel will have the same effect as adding a single bigger one, in terms of aiding the "big hit" response? I only ask as I have two 220uFs from the original rudimentary power supply I built.

With capacitors in parallel you add the uF, but the voltage stays the same. You will get lower ESR too. So yeah, two 220uF per rail = 440uF per rail.

I've also designed the final layout for the strip-board that's going in the enclosure, and made a 'mock up' of it in the actual stripboard I'll be using:

dsc06074mediumdu4.jpg


dsc06073mediumxb0.jpg


It looks extremely cramped: that's because it is. The finished product will look a lot neater: all component legs will be insulated and the wires will be the minimum length possible to reduce clutter. I'm confident that I can make this work...:o

Neat! Can't easily make it out myself, but still.... the planning is immense! :cool:
 
First time I've seen a "Suntan" cap. :D

It's quite satisfying when you build something and it works, and annoying when it doesn't...

FWIW, I managed to finish of my little tripath amp today, I've a got few extra caps to remove and replace, but even so it works which is always nice.

Pic:
imgp1181pe6hk6.jpg

I didn't build it from scratch (SMD parts are a PITA), it came as a module and it's been modified a little. :)
 
Agree, the planning is really good! My planning is usually not even on paper, I tend to do it and hope for the best, but I know this is unwise :p . Suntan caps? I got lots of them, they are really cheap and seem decent enough. Heres a bank of them with slow charging circuit built for a class-A amp which never worked (despite its incredible simplicity);


Your amp will look much tidier when all the parts are fixed on and the leads cropped. As a precaution you might consider using a small resistor in series with the power supply (100ohm-ish) when first powering it up, and if you note any unusual behaviour you can switch it off quickly and the potential damage should be quite limited. I always do this when first powering up speaker output amplifiers since if something is wrong here and you connect straight to the main supply the result can be literally explosive :eek:! Assuming you have a multimeter you should measure only a small voltage loss accross this safety resistor.

I got too many things to build, I'm on the second of 3 power amplifiers I'm constructing, though it is nearly finished.
 
^ Good call with the resistor, sounds similar to the lightblub trick when testing mains powered amps.

I'm not knocking the caps, if they work then it's all good - looks like you have plenty od capacitance there. It's good to see that there is a few DIY minded people around here, I like seeing these projects. :)
 
Theres 48,000uF on that board at 63v (though designed for 30v use). I dare anyone to lick the back of that board when its charged :p:eek:. I see your on DIYAudio too Mike, my amp(s) are on there but not yet in the main solid state gellery since I still haven't fitted any front panel. I also like looking through other people builds, can give you new construction techniques too.
 
No PCB licking for me, I've already got a few black marks on my desk, and don't want one on my tounge! :eek: ;)

DIYAudio is addictive, I spend far too much time on there. :o Still, at least I now have better carpentry and electronics skills since I first registered due to building various projects... :)

So far the only amp I've built from scratch is the 41hz.com amp6basic, but I'm planning to build the amp8 into the chassis of a broken amp. There is one small problem with this plan tho, as all the amps I've found locally have only had small problems, so I fixed them instead, as it seems to shame to use a chassis of a perfectly good amp!

I'll keep an eye on the solid state gallery for your amps, I find once you have a working amp finishing the case always takes longer than you think...
 
Out of curiosity, did it make any difference to the hiss?

With capacitors in parallel you add the uF, but the voltage stays the same. You will get lower ESR too. So yeah, two 220uF per rail = 440uF per rail.

Neat! Can't easily make it out myself, but still.... the planning is immense! :cool:

I didn't notice any difference to the hiss to be honest, but I'm not a true audiophile really!

Re the caps: that's what I thought.:)

The planning was essential, it only just fits on the board at all!



Mike - That's a nice looking amp! I've no idea what tripath means, but I'm sure it was pretty satisfying to make!:p



Dr. Em:
That Class A looks awesome! I'd be scared to even go near it, never mind lick it!
 
Okay, I finally finished soldering it up. First things first, a few comments on the whole process:

-It was incredibly difficult. I knew it was going to be fiddly, but there was the odd moment when I could have done with an extra pair of hands to hold everything. Even pinning things down with blu-tac wasn't enough when I made a mistake and had to remove a couple of components.

-I've realised that my Antex 21/25W soldering iron is pretty shoddy. The end is too fat and I think there is some sort of residue on the tip so only a small part gets hot enough.

-Stripboard is horrible to work with. The copper strips tend to peel off if a component is soldered then removed. It was also very weak where I had to cut the strips, and frequently started to peel away.

-I tried extremely hard to insulate all bare metal legs/connections, but found that the heatshrink shrank too much axially and ended up leaving the odd part exposed. Even when using insulated wires, the plastic insulation would shrink when the wire was being soldered in.



Despite all the problems I encountered I manage to fit everything in, with only one minor deviation from the hand-drawn plan on the previous page. I was really pleased with this, my hard-work in the planning stage appears to have paid off!:)

Having soldered it all up I was certain it wouldn't work first time - it was just too much to ask for such a complex circuit on such a small board to work properly first time around. I was right. The LED didn't light up on the first try.

A bit of fiddling with the multimeter showed that there was 9V across the main star ground point and the battery's 9V rail which was obviously not correct - it should have been closer to 4.5V. The LED was not lighting because it was connected over the main ground rail and the 9V rail, thus no voltage across it. A bit of knife action on the back of the board revealed that there was a short from one rail to the other. This was easily cleared.

The LED then lit up, and I was greeted with fully amplified sound!:p Sheer delight such as this I had never known before! I was expecting hours of laborious debugging work before it even approached a working state!


Anyway, enough geeky gabbling, here are some more photos to keep you all happy:

Stripboard prepped for soldering. Notice the knife cuts to isolate different parts of the circuit:

dsc06076mediumkv5.jpg



The finished circuit! Took me about 3 hours to get to here:

dsc06077mediumcm8.jpg



Check out how cramped it is! In particular, note the densely populated top-end, between the 2 coupling capacitors. All the resistors are vertical to save on board real-estate!

dsc06081largekk1.jpg



The next big challenge is fitting the damn thing into that little aluminium box. I've drawn up some plans for the front panel, and it's very, very tight.


Finally: another warning not to short-circuit 9V batteries. I don't know how this one happened, but it melted through 3 little plastic bags.:o
dsc06082mediumyh4.jpg
 
Wow, you certainly stuffed it in on the stripboard! I'd have been tempted to make it a little wider, so you could have the caps a little closer to the board. Still, if it works then it's all good. How does it sound?

EDIT: FWIW, Tripath is the name of a company who makes class D amp chips. They are known to have a very clean sound, and are pretty simple to put together.
 
Back
Top Bottom