Whilst its certainly useful to hear the type of noise or nuisance sound in audio circuits and identify it as white, popcorn, 100Hz hum etc, I find it essential to also be able to measure and compare the strength at different points in the circuit with precision. The dynamic range of audio isn't really easy to to come to grips with but an audio millivoltmeter does the trick for the majority of requirements but if you look about, the price of these is often higher than budget 100MHz oscilloscopes!
It's also possible to use a DMM for measuring noise up to about 2kHz with repeatable precision - even higher frequencies too, though typically with a steep roll-off above 4kHz if yours is a typical, low cost meter and RMS converter chip. Obviously, you would need to work around each meter's limitations but I find that I can use even cheapo meters in most situations where I need to track down noise sources, find where the gain went, trouble shoot etc. Most DMMs will respond to sinusoidal AC with reasonable precision up to at least 2kHz and that probably covers most of our practical requirements.
It's also possible to use a DMM for measuring noise up to about 2kHz with repeatable precision - even higher frequencies too, though typically with a steep roll-off above 4kHz if yours is a typical, low cost meter and RMS converter chip. Obviously, you would need to work around each meter's limitations but I find that I can use even cheapo meters in most situations where I need to track down noise sources, find where the gain went, trouble shoot etc. Most DMMs will respond to sinusoidal AC with reasonable precision up to at least 2kHz and that probably covers most of our practical requirements.
Here are the test results. 120R is because I have runout of 100R. Seems ideal. From what I can see the zener helps and helps nicely, none the less it's mostly a capacitance multiplier. The 0.86V is a welcome quirk you won't find in a text book, a primitive Darlington might need 207VAC as it has no feedforward resistors. I would recomend a 2 watt zener, I am using a 1.3 watt.
The output noise is -97dBV and - 62 dBV 100 Hz hum. 18 VDC is 25 dB higher. Thus if a salesman you might say hum > -80 dB ( -87 ) below 18VDC. The amplifier might offer more than 20 dB rejection so this looks to be comfortble.
I got 3 dB better by connecting to 22 000uF centres. That's where to place the speaker return. Some say you shouldn't. It's so easy to be an armchair expert. A year from now tweak that perhaps.
'Certainly a good result but could you clarify or sketch exactly what you meant there, Nigel?
I certainly can do that. I will try 470R. The Idea of 120R was to have enough current for the 207 V requirement. The corner frequency is 1.4 Hz which is useful and explains why the residuals are sinusoidal. 470R might offer lower hum and retain reasonable AC voltage range.
I shall also show how to get a little more voltage.
I must say having a realistic target helps. The actual hum reduction is circa 40 dB or 100 times less. That's peanuts for a voltage regulator. However if a small mistake is made with them it would be about equal.
I had thought of adding a fold back current limiter. The type used on many class AB amplifiers. Better not because it compromises the simplicity.
I suspect TIP142/147 is slightly faster than a voltage regulator. It's more comfortable at high current.
I shall also show how to get a little more voltage.
I must say having a realistic target helps. The actual hum reduction is circa 40 dB or 100 times less. That's peanuts for a voltage regulator. However if a small mistake is made with them it would be about equal.
I had thought of adding a fold back current limiter. The type used on many class AB amplifiers. Better not because it compromises the simplicity.
I suspect TIP142/147 is slightly faster than a voltage regulator. It's more comfortable at high current.
360 R is an ideal value. It actually made for better regulation. It also means a more available 1.3 W zener is fine. BZX C85 20 type used. One of my bargain buys. - 69 dB. an extra 6 dB for zero money. If using more current a 2200 uF could be used. 35 V types from Panasonic FC range. These are all the low cost options. The 22 000 uF is the sensible choice ( 40 V types )
Diagram says better than 760 uV. The spectrum says better still. Eithernet slightly wobbles things. The spectrum is corrent.
I tried getting 1.4 extra volts. - 52 dB noise. That's 17 dB worse and to me out of the question. Not least as 36 V total is a great JLH voltage.
The larger green blob is where I finallly measered from. ( speaker black terminals ) What you see is what it is, thick 2 mm dia wire used. I resisted doing a circuit diagram as it's very unhepful to someone new to this. Anyone who knows it knows it anyway. That is like teaching car mechanics to someone learning to drive. A great idea but not the bigger learning curve. The TIP 142/147 go Base Collector Emitter. Betweeen C and E is a reverse biased protection diode which is a great bonus. I guess it is a 10-amp type.
All in all I think my fuses are sensible. I would syggest at least 225 VA transformer for stereo. 160 VA if 1 amp is enough. In my oppinion it is. I suspect 3 amp fuses will be OK for that. Views on this very welcolm. I have been very lazy and not given it much thought.
The dia should say 760 uV. The spectrum says less. It's at the limit of the scope.
A word of warning. The 1969 JLH protects the speaker due to the output capacitor. For tenetive testing get some 25 V non polar capacitors. 4 x 470 uF will do .These go between the speakers and amplifier. This is the down sidee of a +/- supply.
You can if very willing to experiment fuse the transformer input to protect everything. Tricky but could work. A 1AT fuse plus thermistor could work. Not for class AB. 5R 5A perhaps.This allows tighter fusing. Without the thermistor usually the fuse will blow on start up.
My PC monitor gave up . I am using a tiny TV. Hope the details comes through.
This circuit is great for the 1969 version except the minus side ( TIP 147 ) not required. The hum will be the same as the current is unchanged. It might be the better as the 0V is less complex. People assume the output capacitor is bad news, not really. The Quad 303 has one and sounds far better than many that don't. The 303 with the right speakers is close to perfection. It's a very close performance to class A whilst almost being class B ( 5 mA ). Very like JLH.
Last edited:
I aggree.
In 1973 my teacher Peter Platt ( great teacher with zero confidence , we cured that, he misssed the RAF life ) said this circuit is obsolete. Yes and no. I think it has surprised me. As I said in my missive remove the minus side for 1969 JLH.
The performance is the same as 16 x 22 000 uF caps assuming 6 dB per cap. Without a large copper plate you won't get it and the harmonics will be worse.
This reminds me of big V8 engines that are big enough to work without any extra complexity, not even overhead valves. The reality better than on paper. The old Ford V8 Dillenger liked so muich. Like a JLH fuel was used.
I guess this PSU is OHV ( overhaed valves ). A massive improvement for not much complexity. Then life got harder. Fuel flow being key. .
BTW. The Ford V8 was made to the same tollerences as modern engines more of less. 99% failiures at first. A very primitive engine made to aircraft limits.
In 1973 my teacher Peter Platt ( great teacher with zero confidence , we cured that, he misssed the RAF life ) said this circuit is obsolete. Yes and no. I think it has surprised me. As I said in my missive remove the minus side for 1969 JLH.
The performance is the same as 16 x 22 000 uF caps assuming 6 dB per cap. Without a large copper plate you won't get it and the harmonics will be worse.
This reminds me of big V8 engines that are big enough to work without any extra complexity, not even overhead valves. The reality better than on paper. The old Ford V8 Dillenger liked so muich. Like a JLH fuel was used.
I guess this PSU is OHV ( overhaed valves ). A massive improvement for not much complexity. Then life got harder. Fuel flow being key. .
BTW. The Ford V8 was made to the same tollerences as modern engines more of less. 99% failiures at first. A very primitive engine made to aircraft limits.
Yep, extra loud and clear
I can understand what you meant and thanks for the schematic and extra details, Nigel.Regarding the monitor, I found that local PC repairers (they seem to have greatly increased in number recently) usually have a few lying around the workshop, doing nothing. If they know you are capable of resolving any minor setup problems with them, you might save a bit of cash by at least asking if they are prepared to lend or sell one to you whilst you work out what to do for the long term.
Funny Ian I've just ordered a new one from CPC Farnell. The switch mode fine for once. I will recover the speakers as they look handy. For personal things I have a ten minute to one hour rule. Ten minutes to say I can get inside. One hour to say that's it. In 30 minutes I gave up. CPC very good on price as with the transistors.
It will be a mini Christmas.
My original boss pointed out it took me 47 years to build this PSU. In a way yes. He has all my cast offs like my Magnepans. He has more room. We were Westwood and Mason in Oxford. Him Mason and Leslie Westwood around somewhere. One of us became SSL mixing desks.
It will be a mini Christmas.
My original boss pointed out it took me 47 years to build this PSU. In a way yes. He has all my cast offs like my Magnepans. He has more room. We were Westwood and Mason in Oxford. Him Mason and Leslie Westwood around somewhere. One of us became SSL mixing desks.
Hi Nigel,
just a suggestion: would it be beneficial to split the 360R Zener feed resistor in to 2x 180R and put something like a 470uF from the bottom of each resistor down to ground. In that way, you would put a second order filter on the zener feed and reduce the sawtooth even further?
Kind regards
Mike
just a suggestion: would it be beneficial to split the 360R Zener feed resistor in to 2x 180R and put something like a 470uF from the bottom of each resistor down to ground. In that way, you would put a second order filter on the zener feed and reduce the sawtooth even further?
Kind regards
Mike
Mike I will give that a go as it doesn't make it hard to draw or do. It's looks very possible. Let's speculate at -76 dBV. Sunday morning for that.
Listening to JLH era Sony tuner given to me by my ex Saturday boy James. Today long wave AM 198 kHz. I think a 1922 allocation of frequency. It's almost hi fi ! This isn't the FM which is seriously good.
Listening to JLH era Sony tuner given to me by my ex Saturday boy James. Today long wave AM 198 kHz. I think a 1922 allocation of frequency. It's almost hi fi ! This isn't the FM which is seriously good.
Question on amplification/ rail voltage/ operating current?
https://sound-au.com/tcaas/jlh1996.pdf
“For a 15W (sinusoidal) output into an 8Ω load, an 11VRMS drive voltage is required..With the circuit shown, using the improved current control layout – which is rather less efficient than the boot-strapped load for Tr3 which I originally proposed – the rail voltage needed is +/- 22V.. ..this means that the output transistor operating current must be at least 2A to allow this”
This will lead to a dissipation, in each output transistor, of 44W.” ..So far things are clear
Built the 2003 Geoff Moss, but I think the same rules should apply as to the ‘96 version!?
What I don’t get, if I change voltage of the supply (I use a lab supply) volume doesn’t change? If I use 10 or 20 V doesn’t make a difference in volume. Until on low voltage end lets say 3-5V the sound starts to distort. Must say I used the headphone out of my laptop which has a lower P-P voltage than a line out. But anyways:
Does that mean, that amplification level = volume is merely dependent on output voltage of the preamp?
Does that also mean, that exaggerating rail voltage and operating current wouldn’t make any effekt other than creating heat?
So this would mean for example that the output swing of my ext. Soundcard with lets say +-2V and the impedance of the speaker are determining the max. Volume?
So the strategy would be to measure distortion and lower voltage and operating current to reach an optimum? And live with amplification level?
Thank you for your advices/ estimations/ evaluations. Also environment will thank!
https://sound-au.com/tcaas/jlh1996.pdf
“For a 15W (sinusoidal) output into an 8Ω load, an 11VRMS drive voltage is required..With the circuit shown, using the improved current control layout – which is rather less efficient than the boot-strapped load for Tr3 which I originally proposed – the rail voltage needed is +/- 22V.. ..this means that the output transistor operating current must be at least 2A to allow this”
This will lead to a dissipation, in each output transistor, of 44W.” ..So far things are clear
Built the 2003 Geoff Moss, but I think the same rules should apply as to the ‘96 version!?
What I don’t get, if I change voltage of the supply (I use a lab supply) volume doesn’t change? If I use 10 or 20 V doesn’t make a difference in volume. Until on low voltage end lets say 3-5V the sound starts to distort. Must say I used the headphone out of my laptop which has a lower P-P voltage than a line out. But anyways:
Does that mean, that amplification level = volume is merely dependent on output voltage of the preamp?
Does that also mean, that exaggerating rail voltage and operating current wouldn’t make any effekt other than creating heat?
So this would mean for example that the output swing of my ext. Soundcard with lets say +-2V and the impedance of the speaker are determining the max. Volume?
So the strategy would be to measure distortion and lower voltage and operating current to reach an optimum? And live with amplification level?
Thank you for your advices/ estimations/ evaluations. Also environment will thank!
..ok will place a poti there to be able to try! ..i have read in the original article: https://sound-au.com/tcaas/jlh1969.pdf Output Voltage is proportional to input voltage. In his Table for 0.9V Vin the circuit will need 38 V Rails supply. ..normally for the 1996 and 2003 2005 versions +-22 V Rails are proposed (44 Volt total). How is it if your Line/DAC delivers 4V p2p. Would it make sense to further rise rails voltage? Are these assumptions of the original article still valid? Thanks for the advice olddiy! ����
Just a few facts. This is not a simulation and is shown exactly as built. It is not 100 % the best it can be. It is easy and cheap to do and should not put off a newbe. Even at 1 amp things are very warm . If we use 4 amps total the deign ripple of - 60 dBV should be insight. ( - 63 dBV ? ).
What has surprised me is we are close to a LD1084 regualtor in everything. I have manaaged -103 dB with that after many troubles with one. Realistically without a spectrum analyser you won't get that. I totally doubt LD1084 could give 5 amps day in day out.T0220 verse T0147 case here.
I didn't give an output ripple as it is a single line now. Instead 22 000 uF ripple displaced to read the peak to peak ( very important it should be low ). I would say a capaacitor ripple rating of 10 amps at 85C good guidance. A 105C type better. These are Samwha computer grade with screw terminals 40 V. .
Still usuing the tiny TV as a monitor so I exspect typos.
https://cpc.farnell.com/multicomp/lpr35v229m35x45/capacitor-22000uf-35v/dp/CA04832?st=22 000 uf
I notice 10 amps is hard to get to and keep the price OK. 4 of these should be ideal. I think similar exist everywhere. 5.7 A at 85C. Ignore the 2000 Hr rating. Mind you switch off when not in the house.
f you look at the drawing the 1.3V Darlington voltage is met. ( zener to Vout ). Below 200 VAC it still just about works ! I had doubts about a ready made Darlingtons. Not now. Also the feedforward resistors get the best speed from the device ( discharging junction ca[pacitance ) . It likely isn't a fator either way.
I notice 10 amps is hard to get to and keep the price OK. 4 of these should be ideal. I think similar exist everywhere. 5.7 A at 85C. Ignore the 2000 Hr rating. Mind you switch off when not in the house.
f you look at the drawing the 1.3V Darlington voltage is met. ( zener to Vout ). Below 200 VAC it still just about works ! I had doubts about a ready made Darlingtons. Not now. Also the feedforward resistors get the best speed from the device ( discharging junction ca[pacitance ) . It likely isn't a fator either way.