Yes I have and 16 MHz. The beauty of a Darlington is it almost can not give problems. The zener produces some hiss which the 1000 uF takes care of. You could add 0R1 and I uF polyester to the output to give it a boost at the upper frequencies.
Certainly worth looking at. Thanks for doing this. I wonder how it would compare head to head with the LM338 circuit. As you point out, it might deal with the heat dissipation better. It seems cheap and simple. Some people have commented that capacitance multipliers don't sound as good for whatever reason. Are there any disadvantages to this type of circuit compared to voltage regulators apart from subjective measures?
I don't know the LM device. I think getting it wrong is more likely. As I said this is like a big V8 engine. In your version you need a negative regulator also. That's why I did this as they are rare.
In my opinion this suggests that the 1969 version is better If an inexperienced or even expert builder is in a rush goodbye loudspeakers. The 1969 version should be safer and cheaper PSU.
What I like about this design is so few parts used. It should be possible to make this twenty years from now with a few changes of parts. I would bet voltage regulators will change a lot by then.
I think capacitance multipliers are slightly suspect. So many things not quite right. Voltage regulation is important.
In my opinion this suggests that the 1969 version is better If an inexperienced or even expert builder is in a rush goodbye loudspeakers. The 1969 version should be safer and cheaper PSU.
What I like about this design is so few parts used. It should be possible to make this twenty years from now with a few changes of parts. I would bet voltage regulators will change a lot by then.
I think capacitance multipliers are slightly suspect. So many things not quite right. Voltage regulation is important.
Last edited:
One interesting thing this Darlington amplifier shows is how a class AB amplifiers works. I feel this shows if the driver stage or VAS is very quiet then very little colouration of the sound happens. In my opinion most class AB do not get this right. So cheap to solve problems never are. If you like the music is a ripple you want and sawtooth hum you don't. These big Darlington's were used by Bang and Olufsen in very simple highly respected designs. Paul Kemble lists some.
The true voltage regulator should have lower output impedance than mine. That might be irrelevant. Some argue that deliberately using thin cable between PSU and amplifier is a good Idea as long as a 220 uF capacitor is connected between supply rail and 0V. The impedance used as a filter. Whilst I doubt it important the 220 uF is or two off for a balanced supply. That is at the JLH power transistors.
The capacitance multiplier is enhanced by the zener diode as long as the current is sufficient. It takes a miniscule ripple and males it more miniscule. At 50 Hz it's so slow as to make a zener instant. Thus a minor miracle happens.
The true voltage regulator should have lower output impedance than mine. That might be irrelevant. Some argue that deliberately using thin cable between PSU and amplifier is a good Idea as long as a 220 uF capacitor is connected between supply rail and 0V. The impedance used as a filter. Whilst I doubt it important the 220 uF is or two off for a balanced supply. That is at the JLH power transistors.
The capacitance multiplier is enhanced by the zener diode as long as the current is sufficient. It takes a miniscule ripple and males it more miniscule. At 50 Hz it's so slow as to make a zener instant. Thus a minor miracle happens.
Last edited:
Nigel,
thanks for trying that out.
If I interpret the results correctly, worth an extra resistor and capacitor.
Regards
Mike
thanks for trying that out.
If I interpret the results correctly, worth an extra resistor and capacitor.
Regards
Mike
Thank you also Mike. The parts I used were very low cost to make a point. Mostly it's the maths that matters. If you look at it the complexity is slightly less than a LD1084 because no extra protection diodes required.
If you can live with the JLH 1969 version at maximum 1.25 amps LM317 is ideal using 47uF across the gain resistor. Because the device is not connected directly to ground voltages usually are not a problem if not exceeding 40 volts in and out. Probably a 25 vac transformer giving two outputs is ideal. It won't be as forgiving as my design for dropout. I find lm317 very easy to use.
A lady lost her Sony professional Walkman PSU. I recommend her a regulated PSU from Argos a local electrical catalogue store that sells mostly everything. £10. She was astonished at the quality and me too. She let me take it apart ( yikes ). LM317 as used by Naim Audio in the T03 type. The lady was going to buy a new Sony one. She didn't bother. As someone who occasionally fixed Naim equipment PSUs struck me as important. Julian Vereker joked friends were not wrong to say Naim amplifiers were 80% power supply.
Now to the problem of diode noise. In many ways standard diodes are ideal. If an ideal resistor capacitor could be found we could suppress the reverse field collapse in the on junction detectable by am radio. Not easy. In the past I used low cost soft recovery diodes. I don't seem to find them now.
If you can live with the JLH 1969 version at maximum 1.25 amps LM317 is ideal using 47uF across the gain resistor. Because the device is not connected directly to ground voltages usually are not a problem if not exceeding 40 volts in and out. Probably a 25 vac transformer giving two outputs is ideal. It won't be as forgiving as my design for dropout. I find lm317 very easy to use.
A lady lost her Sony professional Walkman PSU. I recommend her a regulated PSU from Argos a local electrical catalogue store that sells mostly everything. £10. She was astonished at the quality and me too. She let me take it apart ( yikes ). LM317 as used by Naim Audio in the T03 type. The lady was going to buy a new Sony one. She didn't bother. As someone who occasionally fixed Naim equipment PSUs struck me as important. Julian Vereker joked friends were not wrong to say Naim amplifiers were 80% power supply.
Now to the problem of diode noise. In many ways standard diodes are ideal. If an ideal resistor capacitor could be found we could suppress the reverse field collapse in the on junction detectable by am radio. Not easy. In the past I used low cost soft recovery diodes. I don't seem to find them now.
SBYV 28- 100 or 200 at 3A would be do for 1.25 A version. 8 diodes required. Mouser have them cheap enough. Soft recovery diodes. If you just add let's say 10nF to standard diodes they give a harsh sound worse than no capacitor. Watch for that as some say you should. The soft recovery diodes are a excellent compromise.. Larger soft recovery diodes exist at a price.
Been looking at your project nigel, it looks very promising, I would like to build one of my own, I don't have any tip transistors, but I do have some darlington mj11032/mj11033, with a hfe of at least 1000, don't you think they could work nicely ? I'm not very skilled with soa charts
I will research them. The beauty of this Idea is mostly anything will work as long as voltage and current levels meet basic specs. For a T0220 device I would limit to 2 amps. Also try to make sure excess voltage isn't too high. 5 volts about right.
If you don't have an oscilloscope but think it all is working well enough the best 0V point can be found with a crocodile clip to earth if a fat wire used. Ear to the speaker for least hum. Switch off between tests.
Darlington's with BD139 and TIP35 could be best of all. BC337 40 2N3055H also.
I will look in the morning.
If you don't have an oscilloscope but think it all is working well enough the best 0V point can be found with a crocodile clip to earth if a fat wire used. Ear to the speaker for least hum. Switch off between tests.
Darlington's with BD139 and TIP35 could be best of all. BC337 40 2N3055H also.
I will look in the morning.
Having bought a job lot of MJ15003 for my Geoff Moss version, I also use the BD139/MJ15003 (rejects from selection for the amp) as a Darlington with great success in my Cap multiplier. Essentially the same circuit as Nigel's but without the zener.
Mike
50 amps is a nice idea and gain is ideal ( > 1000 < 25 amps). I still call them 5 amp devices which could be 10 amps if the excess voltage is minimal ( 4 volta ? ). FET company Exicon said something that seems true. If a FET says 10 amps then with reasonable care they are 10 amps. If a bipolar transistor says 10 amps safe to say 5 amps under realistic conditions. The problem is getting the heat away. The T03 can helps greatly. If you read carefully the silicon might just be OK with 175C. The can top 125 C is likely. The heatsink 60C. If you put your fingers on the can and find it too hot it's likely it's too hot. T0247 looks on paper to be as good. Not really. Good enough. T0220 is some kind of a joke above 2 amps. If only the companies that make fakes made good fakes with real numbers. TIP35/36 in T03 would be nice.
Why 50 amps? Transients.
Gogny managed to get 50 watts 1 ohm from 2N3055 using 3055 as a driver ( yes, Darlington ) . This is because the gain of a 3055 is 5 when like that. 7 amps peak or 3.5 amps constant. That's about righ. 1/4 the max current. The 3055 can take masssive base currrent.
A friend of mine is a vastly experianced engineer and was designing a Helmholtz coil for caliibratring magntometers. I was very dubious about it working. He knew the maths and it looked he was right. The machine never worked for long. He always knew the reason and it wasn't the transistors ( really). I designed a simple version. That's no good he said, we can't have the boss understanding it. A paper engiineer came in. He was very arrogant. The staff found a draw full of PDFs on thee subject with circled imformation. He hadn't a clue. He had a PHD! My friend now works at Harwell and has retained his job. He had only just moved there beforre Lockdown. BTH he built most of the mixing desks sold in the 1980s. He is the World expert on NE5534/32. Very senstive to PSUs. SSL desks had very big PSUs.
Products - ON Semiconductor
http://douglas-self.com/ampins/wwarchive/More transformerless amplifiers Mar58 p1.jpg
http://douglas-self.com/ampins/wwarchive/More transformerless amplifiers Mar58 p2.jpg
This is a valve JLH in a way. A very famous circuit that almost is a good idea. Take a pentode valve to be very like a transistor. One critical advantage a valve has is it's a depletion device as are JFETs ( mostly, silicon carbide even ) . This means a single resistor can set the current. Sadly the JLH is not as easy to describe. Valves actually auto rest using this single resistor when they age. That's magic! The reasons it works is exactly like a JLH in most ways.
http://douglas-self.com/ampins/wwarchive/More transformerless amplifiers Mar58 p2.jpg
This is a valve JLH in a way. A very famous circuit that almost is a good idea. Take a pentode valve to be very like a transistor. One critical advantage a valve has is it's a depletion device as are JFETs ( mostly, silicon carbide even ) . This means a single resistor can set the current. Sadly the JLH is not as easy to describe. Valves actually auto rest using this single resistor when they age. That's magic! The reasons it works is exactly like a JLH in most ways.
Thanks Nigel, that sounds promising, still even with 10 amps, more than adequate.
Can I use your design still? Or does it need a more powerfull zener, lower base resistor
Can I use your design still? Or does it need a more powerfull zener, lower base resistor
It will be fine exactly as it is. 360 R seemes ideal (180 + 180 ). I wouldn't go lower than that. The beauty of the circuit is you can listen to ideas. Try 470 + 470 perhaps? Try 2200 uF + 2200 uF although it might take a few minutes to get the last 1/2 volt when that. The critical part is to make the 0V centre line of thick metal. 2.5 mm^2 mains cable is OK ( fatter if you can ). Listen for lowest hum where that fixes. A PSU is very like a musical instrument how you tune it. It's very easy if you know how to think about it.
If you use different vollages just calculate that you don't go above 0.8 watts at the zener. Use the zener I use if possible. 6/360 = 17 mA. 0.017 x 20 = 0.34 watts typical. The Darlington slightly changes that. I used 120R at fiirst. That was risky at 253 VAC. It was to be certain that the diode working point was fully tested. In theory it's better, it wasn't and spoilt somethings.
I think many class AB amplifiers could use this. If so I would construct a Darlington using fast devices. 6 amp reverse biased diode collector to emitter and copy the feedforward resistors from a typical ready made ( MJ3001 ? ). I would expect a more class A sound. It might seem less powerful when it isn't really. Less grunge. The design seems to tollerate voltage variations very well so no great changes needed. Off the shelf types are more fussy.
The heat sink always is twice as big as you think ! That's a joke that isn't.
If you use different vollages just calculate that you don't go above 0.8 watts at the zener. Use the zener I use if possible. 6/360 = 17 mA. 0.017 x 20 = 0.34 watts typical. The Darlington slightly changes that. I used 120R at fiirst. That was risky at 253 VAC. It was to be certain that the diode working point was fully tested. In theory it's better, it wasn't and spoilt somethings.
I think many class AB amplifiers could use this. If so I would construct a Darlington using fast devices. 6 amp reverse biased diode collector to emitter and copy the feedforward resistors from a typical ready made ( MJ3001 ? ). I would expect a more class A sound. It might seem less powerful when it isn't really. Less grunge. The design seems to tollerate voltage variations very well so no great changes needed. Off the shelf types are more fussy.
The heat sink always is twice as big as you think ! That's a joke that isn't.
I am just about to strip down the PSU. I tried a BD135 ( 39 ) 2SC5200 primitive Darlington. The dream team as many might think. C5200 is a very large, very fast and cheap package. How could it possile be wrong? Well it is. The TIP142 is better and that's all there is to it. Just like typical DIY Audio we could make it work and it will. By them it's only for the Grand Masters of DIY.
One good thing I did prove is the feedforward resistors make it miildy worse on ripple.
Ready made Darlingtons like TIP142/147 most likely are faster than LM/LD series reguators. LM/LD possibly have traits of instability which are never completely hidden. The Darlington is the most stable thing I know of. Well almost al I will try that now.
One good thing I did prove is the feedforward resistors make it miildy worse on ripple.
Ready made Darlingtons like TIP142/147 most likely are faster than LM/LD series reguators. LM/LD possibly have traits of instability which are never completely hidden. The Darlington is the most stable thing I know of. Well almost al I will try that now.