Hello Folks, Happy new Year to you!
Prologue: My Name is Christian and I have recently discovered my interest in electronics and made it my hobby now. Having studied software development and information technologies I must agree that besides some few basics in electronics the whole matter was and is quite new to me. I do this all for fun and to learn something new and exciting.
So I eventually came down to amplifiers. It has been a sincere wish of mine to build an (at least one) amplifier myself. I started to study electronics basics, how transistors and amplifiers work (in general), read a lot of articles in here and other sources, for example pass labs or sound.westhost.com, studied a lot of example circuitries trying to understand what's going on in them. But I never actually built one myself until now. So I finally went on to build a simple headphone amplifier for my all day stuff, including listening to music and watching movies.
Lacking experience in amplifier construction i chose a design that should be very simple and easy to both build and debug. Taking different designs into account my work resulted in:
The Amplifier: It's a single ended Class A amplifier, nothing unusual at all.
The output stage is driven by HEXFETs IRLZ34N - I used them because I have a lot of them in stock, even though I used them only for switching applications in LED drivers so far. Idle current is around 170mA. The 7.33 Ohms are actually made of two triplets of 22Ohm resistors I chose from a stash I found and "matched" them, so that each triplet has indeed 7.33 measured Ohms together. I found the idea of using an LM317 as CCS amazingly simple.
The OPA2134 that I chose for voltage gain was discussed to be "quite good" several times, it's supposed to have sufficiently strong drive capabilities for the used output MOSFET and I must say, from what I hear I tend to agree
With the trimpots in the feedback loop I set the gain to around 10dBu.
R4 is there to provide anti-oscillation resistance in case the volume pot is set to full, given that the OPA2134 has FET inputs. R5 is just for safety either, given the quite high gate capacitance of HEXFETs.
Power supply: Nothing special here either. Transformer, Cap, Regulator.
Not mentioned in the schematic are of course a proper power switch and a 2A fuse in front of the transformer input. The 7812 and 7912 from KEC that I have here are rated at 60dB ripple rejection ... I think that'll suffice even under this load. Furthermore, as stated in the schematic, I have not built the bridge rectifier using discrete diodes but used a DIL7 sized B40D for each polarity.
Construction: The most awesome part of it!
I used the case of a dynavox "hifi amp" that was basically a gainclone amp around a TDA2004R from ST. I scrapped out its contents leaving only plugs and controls I wanted to use and began inserting the power supply. I've built it all on a single prototyping board. Transformer is rated 15VA @ 2x12V and gets quite warm under load.
Well, several hours later it then looked like this:
The heat sinks are connected to each other using heat pipes underneath the board. They do get quite warm, but I can still touch them for several seconds without getting burned
The sound: well, I really cannot tell how it sounds for I simply cannot compare.
It sounds totally amazing! But given from what I was used before, I'd say that's no surprise. My former equipment was a Teac HP10 surround headset with a plastic box that made scratchy noise each time I changed the volume. Now with my new Beyerdynamic phones and this amp I can only say, that I'm experiencing my music collection again .... in a very different way.
I'm really looking forward to your comments, folks
Prologue: My Name is Christian and I have recently discovered my interest in electronics and made it my hobby now. Having studied software development and information technologies I must agree that besides some few basics in electronics the whole matter was and is quite new to me. I do this all for fun and to learn something new and exciting.
So I eventually came down to amplifiers. It has been a sincere wish of mine to build an (at least one) amplifier myself. I started to study electronics basics, how transistors and amplifiers work (in general), read a lot of articles in here and other sources, for example pass labs or sound.westhost.com, studied a lot of example circuitries trying to understand what's going on in them. But I never actually built one myself until now. So I finally went on to build a simple headphone amplifier for my all day stuff, including listening to music and watching movies.
Lacking experience in amplifier construction i chose a design that should be very simple and easy to both build and debug. Taking different designs into account my work resulted in:
The Amplifier: It's a single ended Class A amplifier, nothing unusual at all.
An externally hosted image should be here but it was not working when we last tested it.
The output stage is driven by HEXFETs IRLZ34N - I used them because I have a lot of them in stock, even though I used them only for switching applications in LED drivers so far. Idle current is around 170mA. The 7.33 Ohms are actually made of two triplets of 22Ohm resistors I chose from a stash I found and "matched" them, so that each triplet has indeed 7.33 measured Ohms together. I found the idea of using an LM317 as CCS amazingly simple.
The OPA2134 that I chose for voltage gain was discussed to be "quite good" several times, it's supposed to have sufficiently strong drive capabilities for the used output MOSFET and I must say, from what I hear I tend to agree
With the trimpots in the feedback loop I set the gain to around 10dBu.
R4 is there to provide anti-oscillation resistance in case the volume pot is set to full, given that the OPA2134 has FET inputs. R5 is just for safety either, given the quite high gate capacitance of HEXFETs.
Power supply: Nothing special here either. Transformer, Cap, Regulator.
An externally hosted image should be here but it was not working when we last tested it.
Not mentioned in the schematic are of course a proper power switch and a 2A fuse in front of the transformer input. The 7812 and 7912 from KEC that I have here are rated at 60dB ripple rejection ... I think that'll suffice even under this load. Furthermore, as stated in the schematic, I have not built the bridge rectifier using discrete diodes but used a DIL7 sized B40D for each polarity.
Construction: The most awesome part of it!
An externally hosted image should be here but it was not working when we last tested it.
I used the case of a dynavox "hifi amp" that was basically a gainclone amp around a TDA2004R from ST. I scrapped out its contents leaving only plugs and controls I wanted to use and began inserting the power supply. I've built it all on a single prototyping board. Transformer is rated 15VA @ 2x12V and gets quite warm under load.
An externally hosted image should be here but it was not working when we last tested it.
Well, several hours later it then looked like this:
An externally hosted image should be here but it was not working when we last tested it.
The heat sinks are connected to each other using heat pipes underneath the board. They do get quite warm, but I can still touch them for several seconds without getting burned
The sound: well, I really cannot tell how it sounds for I simply cannot compare.
It sounds totally amazing! But given from what I was used before, I'd say that's no surprise. My former equipment was a Teac HP10 surround headset with a plastic box that made scratchy noise each time I changed the volume. Now with my new Beyerdynamic phones and this amp I can only say, that I'm experiencing my music collection again .... in a very different way.
I'm really looking forward to your comments, folks
I wouldn't be surprised if there weren't any open-loop, let alone with the significant amount of feedback provided by the opamp. After all, the LM317 is a current sink (high impedance) looking into the low-impedance FET source-follower output. Walt Jung examined the performance of LM317 current sources in this article, finding an output impedance of 100 kOhms or greater across the entire audio band, and >=300 kOhms at <=10 kHz.
So let's see how much noise the '317 would be producing to begin with. IMO there would be two components:
1. Vref / input noise at unity gain, converted to current noise by Radj
2. Radj shot noise as a result of the current flowing through it
(1) is about 20 µV across the audio BW --> 20 µV / 7.33 ohms = 2.7 µA or 19 nA/sqrt(Hz) in this circuit
(2) computes to sqrt(2qI)/sqrt(Hz) ~= 0.23 nA/sqrt(Hz) --> negligible
The circuit shown has about 9 nV/sqrt(Hz) and a gain of 4, which is 36 nV/sqrt(Hz) (or 5 µV) output wise. It would take about 2 ohms of source follower output resistance to increase minimum noise floor by 3 dB. The simulation for a Szekeres at 170 mA with an IRF510 gives me about 1.5 ohms. So the effect would just be detectable - assuming good cooling of the LM317 and not taking FET noise contribution into account. With the buffer outside the feedback loop.
Inside the feedback loop, even a fairly small amount of feedback would make the output stage contribution entirely negligible, and we have plenty of that even at, say, only 1 MHz of loop GBW.
Noise (and distortion) wise, I'd be more worried about the 100k pot in this circuit (R_source,max = 25k).
Exactly. And what's wrong in the schematic is that the actual pot from that casing IS 25k and not 100k as in the simulation. I never noticed that
Anyway, reflecting onto this after over 1.5 years i can only say:
I love this amp! And it's still in use together with my Beyerdynamic DT-990 headphones. As easy as it is, as awesome is it - as long as you don't mind its energy efficiency, of course. I have tried out some other designs but never achieved such good results. It's practically the best sounding amplifier I've built so far - and one of the hottest...
Cheers, Christian
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.