I did basically the same thing with a 2SA1302 and it actually sounded great. It made a ton of heat running in class A of course. I even did 4 transistors in parallel with matched Hfe that worked as well. That was 25+ years ago, I’m getting old! If my memory serves me, I used Dale RN-25 emitter resistors also on heatsinks. It was a great learning experiment.
Again, voltage gain is not the same thing as beta. At 1MHz, output is still a copy of input, no sign of phase delay. From post #1 :
https://www.diyaudio.com/community/attachments/fig10_1mhz_i_o-jpg.1245879/
When I use my fast impulse generator with 10ns rise time, there is still not much difference between input and output rising edge.
This reminds of my first amplifier I made when I was at school. It was a darlington triple running of an old filament transformer, AC coupled of course. I think it gave about 250mW. Surprisingly adequate for casual listening, and definitely a step up from what ever 80's ss amp I was running at the time SQ wise. Eventually ended up running off of a quad 33. I still have it in a box somewhere.
fT is defined as the current gain - bandwidth product (Technically where |h21| falls to unity, but close enough for government work). Divide by beta, and it roughly gives the bandwidth (where roll off starts as a unity gain follower). Where you are taking full advantage of the transistor’s gain, that’s the achievable BW. If fT is 30 MHz, and beta 150 you can expect 200 kHz BW. The real bandwidth is more determined by the forced beta of the circuit - if full current gain is not needed, it simply goes out further before rolling off. Driving off a 32-ohm capable headphone out with an 8 ohm load is only a forced beta of 4, or from a 50 ohm output of a sig-gen would be about 6. You can easily see a 5 MHz roll off point. Pretty damn flat. Drive it from a higher impedance source and you’ll see the breakpoint creeping in on you.
So, the stereo version has been completed, in the Al metal case, with all wiring, power supply, capacitance multiplier and connector.
It is now happily playing in my workshop.
Almost everything has been already said in post #1 together with necessary measurements, so one more that shows 250kHz square response, both input (blue) and output (yellow), and we can read some small delay in the middle of the rising and falling edges, remember the time base division is 500ns/div, or 0.5us/div. So definitely negligible for any kind of audio or bat frequency signal. The load was 8ohm.
The last one is a view to my test bench, with the amp running, on the scope screen we can see 100kHz square response, I/O
Have fun,
Pavel M.
https://pmacura.cz/audiopage.html
It is now happily playing in my workshop.
Almost everything has been already said in post #1 together with necessary measurements, so one more that shows 250kHz square response, both input (blue) and output (yellow), and we can read some small delay in the middle of the rising and falling edges, remember the time base division is 500ns/div, or 0.5us/div. So definitely negligible for any kind of audio or bat frequency signal. The load was 8ohm.
The last one is a view to my test bench, with the amp running, on the scope screen we can see 100kHz square response, I/O
Have fun,
Pavel M.
https://pmacura.cz/audiopage.html
Last edited:
Hi Hugo,
Thank you, I have been testing this little amp also with my main 3-way floor standing speakers and as long it stays below clipping, the sound is very good as can be expected from a solid state amplifier with very wide frequency response and output impedance low enough. With a usual speaker sensitivity of 86-88dB/2.83V/m, I have measured max. undistorted usable SPL of some 76dB(A), measured with a SPL meter set at A curve. This is fine for normal, not very loud listening.
I would not call the sound sweet. It is rather detailed, dynamic, with no intrinsic background noise. When staying below clipping, the harmonic (and intermodulation, which is only different way to show the same non-linearity) distortion of this amplifier is inaudible with any music signal. The proof is to record output of the amp and compare it, level and time matched, with original sound data, doing the DBT controlled test. This is the only way to tell audibility of distortion.
The nonlinearity of this circuit is a textbook, frequency independent (in audio band) transfer function. Below please see THD vs. frequency plots of total harmonic distortion and H2 - H5 harmonics.
Another way to show that distortion is independent of frequency is THD vs. output voltage plot, for 1kHz and 10kHz frequencies, you can see that they do overlap. This is quite unusual, right.
Both measurements were done on a completed amplifier in the box, idle current is 500mA and transistor beta is 151 at 150mA. Load was 10ohm this time. With 8ohm, we get exactly THD=1%/0.3W into 8ohm with 500mA idle current, as seen in the next plot.
I wonder if this would be a good amp to power compression drivers on horns (for home use). I think it would be at least worth trying. From all the amplifiers I testes, all produced detectable noise when very close to the mouth. I can pad the compression drivers down, but having a dedicated amp without padding sounds better to me.
I have tested it with Monacor HT-88 horn tweeter with 99dB/W/m sensitivity.
https://www.monacor.com/products/components/speaker-technology/hi-fi-tweeters-/ht-88/
And yes, it is the only amplifier (from those I have) that has no audible noise with this driver. There are two necessary conditions to get it with this follower amplifier:
1) you have to use battery power, or, better option, you MUST use capacitance multiplier behind the usual transformer/bridge/capacitor power supply, or a good regulated DC power supply,
2) the output noise is then completely defined by the sound source. I use Topping D10s DAC which has output noise of 2uV (measured over 20Hz-2kHz flat) and with this DAC there is zero audible noise. But, with a mediocre sound card that has usually 10 - 20uV output noise, you might hear something from that driver. My Hypex NC252MP has audible noise when connected to HT-88, though very low.
You can easily derive it from amplitude response
. Flat, of course . But, remember, depending on signal source output impedance. Keep it as low as possible, always below 100ohm. With 50ohm source impedance, F(-3dB) is about 7MHz and phase would start to roll at 700kHz. If you need even more flat phase below 100Hz, then simply increase input coupling cap value, which is now 220uF,I am using a DSP with 2 V outputs, that should be good enough to drive the 113 db/W sensitive DCX464 to loud enough levels. I am definitely putting this on my list of low power amplifiers to try when I actually have some time to build them, I am busy with the speakers so far. I can imagine 4 channels of these in one box - or maybe even better, two stereo units, one for each side.
Please just take into account that you will need some vent drills or slots in the metal case. Idle current per module needs to be 500mA, so it dissipates 6W when supplied from 12Vdc. Both Q1 heatsink and 10R/5W resistor run quite hot (like 60-70°C the heatsink and 100°C the resistor). It is still manageable, but completely closed box is impossible. The transformer runs at about 70°C in my case.
Anything is possible, but why do it? Transformers are not miracle workers; the only way the input impedance will be increased is by using a step down transformer which means you need an even higher voltage output from the source.
Not sure what is the practical utility of the circuit in this thread anyway. NP has pretty much exhausted the studies of simple circuits and this one does not bring anything new. A reasonably usable single device amp can only be built using fets or tubes, or perhaps by cheating and using a darlington transistor.