Wireless World Audio Op-Amp
- Mike H
- Amstrad Tower of Power
- Posts: 20157
- Joined: Sat Oct 04, 2008 5:38 pm
- Location: The Fens
- Contact:
#1 Wireless World Audio Op-Amp
Can you tell I have nothing better to do today
Actually I do but meh
On and off I have been trying to remember the details of a Wireless World article in same magazine circa 1972, which I had eventually got rid of years ago. For no particular reason it all sort of crystallyzed last evening.
The idea of the project was, on the basis that the then available integrated circuits op-amps (still comparatively new back then) weren't really good enough quality for kosher high quality audio, the main issue being noise IIRC.
So this guy's idea was make one out of discretes, whereby you can choose precisely what active and passive components to use.
In addition however, for just audio use you don't need the universal push-pull type output stage, overload clamps, and all the rest of the obscure impedimenta that typically populates an IC op-amp. What you do want though is the basic op-amp function, but again, not necessarily accompanied by massive open-loop gain either (like in a typical IC op-amp). These sorts of considerations can still hold true I think.
Anyroad this is the original circuit as I remember it (the clincher was remembering the use of the 4.7V Zener, after that it all fell into place)
Actually I do but meh
On and off I have been trying to remember the details of a Wireless World article in same magazine circa 1972, which I had eventually got rid of years ago. For no particular reason it all sort of crystallyzed last evening.
The idea of the project was, on the basis that the then available integrated circuits op-amps (still comparatively new back then) weren't really good enough quality for kosher high quality audio, the main issue being noise IIRC.
So this guy's idea was make one out of discretes, whereby you can choose precisely what active and passive components to use.
In addition however, for just audio use you don't need the universal push-pull type output stage, overload clamps, and all the rest of the obscure impedimenta that typically populates an IC op-amp. What you do want though is the basic op-amp function, but again, not necessarily accompanied by massive open-loop gain either (like in a typical IC op-amp). These sorts of considerations can still hold true I think.
Anyroad this is the original circuit as I remember it (the clincher was remembering the use of the 4.7V Zener, after that it all fell into place)
- Attachments
-
- WW discretes op-amp original.gif (6.78 KiB) Viewed 3733 times
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."
#2
(sound of penny dropping). Of course if you have a current source in the tail and one of the arms, you don't need one in the other arm.
Though I think to be a op-amp by definition you do need the high gain.
Though I think to be a op-amp by definition you do need the high gain.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
- Mike H
- Amstrad Tower of Power
- Posts: 20157
- Joined: Sat Oct 04, 2008 5:38 pm
- Location: The Fens
- Contact:
#3
No don't start!
Just that we don't need a gain of 50,000 or whatever it is in this case, 100 or so is quite good enough.
I actually made a pair of these once, however I think there's a mistake in the circuit. After some period of time one of them would go quieter and quieter and stop altogether, maybe when something had got warm
Turns out (LTspice to the rescue) that R2 looks very suspiciously too large in value. To elucidate.....
D2 provides the bias for current sources Q3 & Q5, @ 4.7V. This means that Q5 passes about 4mA, while Q3 400uA.
Therefore each of Q1 & Q2 should pass half, 200uA (or close enough).
200uA x 22k = 4.4V, and considering the DC quiescent bias for Q6 should be exactly the same as for Q5 (4.7V), there is hardly any Voltage drop across Q4. Doesn't take much imagination to realise if it goes just a little bit wonky then the output stage ceases to be biased properly. Memory is extremely hazy but could be that when I measured the output of the non-functioning one (if and when it was non-functioning), it was stuck up near one of the supply rails. That would stop it passing a signal.
This could be a printing / publishing error, so assuming a full-stop is missing from "22k" written on the diagram, perhaps it should be "2.2k" ?
Well try that in sim and of course it makes a whole lot more sense.
The presence of R2 is, if I remember rightly, quote: "to isolate the collector of Q1 from the collector capacitance of Q4." Or something like.
Not sure about R6, here I've got it at 12k to pass about 2mA, whereas originally it might have been 22k for 1mA, but I thought 1mA was a bit low.
In view of this, it needs +/- 15V supply rails (as originally) to maintain this current (it's not too flexible as regards power supply requirements).
An improvement may be made therefore of employing a constant current FET type device here (such gadgets didn't exist in 1972). aka CRD ("current regulating diode", or could be a jFET with an appropriate source bias resistor, but that is getting complicated).
If you can't be bothered then an appropriate value for R6 to produce 2mA with whatever supply Voltages are being used.
Just that we don't need a gain of 50,000 or whatever it is in this case, 100 or so is quite good enough.
I actually made a pair of these once, however I think there's a mistake in the circuit. After some period of time one of them would go quieter and quieter and stop altogether, maybe when something had got warm
Turns out (LTspice to the rescue) that R2 looks very suspiciously too large in value. To elucidate.....
D2 provides the bias for current sources Q3 & Q5, @ 4.7V. This means that Q5 passes about 4mA, while Q3 400uA.
Therefore each of Q1 & Q2 should pass half, 200uA (or close enough).
200uA x 22k = 4.4V, and considering the DC quiescent bias for Q6 should be exactly the same as for Q5 (4.7V), there is hardly any Voltage drop across Q4. Doesn't take much imagination to realise if it goes just a little bit wonky then the output stage ceases to be biased properly. Memory is extremely hazy but could be that when I measured the output of the non-functioning one (if and when it was non-functioning), it was stuck up near one of the supply rails. That would stop it passing a signal.
This could be a printing / publishing error, so assuming a full-stop is missing from "22k" written on the diagram, perhaps it should be "2.2k" ?
Well try that in sim and of course it makes a whole lot more sense.
The presence of R2 is, if I remember rightly, quote: "to isolate the collector of Q1 from the collector capacitance of Q4." Or something like.
Not sure about R6, here I've got it at 12k to pass about 2mA, whereas originally it might have been 22k for 1mA, but I thought 1mA was a bit low.
In view of this, it needs +/- 15V supply rails (as originally) to maintain this current (it's not too flexible as regards power supply requirements).
An improvement may be made therefore of employing a constant current FET type device here (such gadgets didn't exist in 1972). aka CRD ("current regulating diode", or could be a jFET with an appropriate source bias resistor, but that is getting complicated).
If you can't be bothered then an appropriate value for R6 to produce 2mA with whatever supply Voltages are being used.
- Attachments
-
- WW discretes op-amp improved.gif (6.86 KiB) Viewed 3729 times
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."
- Mike H
- Amstrad Tower of Power
- Posts: 20157
- Joined: Sat Oct 04, 2008 5:38 pm
- Location: The Fens
- Contact:
#4
Cited application was for one pair as phono EQ pre-amps followed by a second pair as volume / tone controls, the outputs going to your power amps.
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."
#5
It would be simple to translate the circuit into using valves as well.
ECC83 long tailed pair, CCS in one arm and the tail, feeding 5687 cathode follower with another CCS as its load. Throw in three pentodes it could all be done with valves .
ECC83 long tailed pair, CCS in one arm and the tail, feeding 5687 cathode follower with another CCS as its load. Throw in three pentodes it could all be done with valves .
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
- Mike H
- Amstrad Tower of Power
- Posts: 20157
- Joined: Sat Oct 04, 2008 5:38 pm
- Location: The Fens
- Contact:
#6
Go on then!
Wow it will do 10V peak into 10k, gain set to X100, and with a flat response to 100 kHz.
For practical purposes (limited to audio band) might be prudent to put a 100pF cap between Q6 collector and base.
'Tis true though the first op-amps were valves. Problee needs dual split supply rails
Wow it will do 10V peak into 10k, gain set to X100, and with a flat response to 100 kHz.
For practical purposes (limited to audio band) might be prudent to put a 100pF cap between Q6 collector and base.
'Tis true though the first op-amps were valves. Problee needs dual split supply rails
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."
- Mike H
- Amstrad Tower of Power
- Posts: 20157
- Joined: Sat Oct 04, 2008 5:38 pm
- Location: The Fens
- Contact:
#9
Just for Nick
Very interesting ~ R3 is necessary to equalize U1 U2 anode currents, without it U1 anode tries to equal B+. In fact R3 controls the DC anode Voltage. Here 100V, if = 150k, becomes 150V (but gain is reduced).
U2 anode could be held at a fixed 100V from some source, but this way it's allowed to find its own level.
Open loop gain = X95
Very interesting ~ R3 is necessary to equalize U1 U2 anode currents, without it U1 anode tries to equal B+. In fact R3 controls the DC anode Voltage. Here 100V, if = 150k, becomes 150V (but gain is reduced).
U2 anode could be held at a fixed 100V from some source, but this way it's allowed to find its own level.
Open loop gain = X95
- Attachments
-
- valve op-amp.gif (7.49 KiB) Viewed 3706 times
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."
#10
Yep, thats what I was thinking of. And if you connect out to inv-in you get very close to a augmented cathode follower
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
#11
Nick wrote:Yep, thats what I was thinking of. And if you connect out to inv-in you get very close to a augmented cathode follower
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
- Mike H
- Amstrad Tower of Power
- Posts: 20157
- Joined: Sat Oct 04, 2008 5:38 pm
- Location: The Fens
- Contact:
#14
Just so's both grids got the same grid leak resistor.
Although that idea is scuppered somewhat by the DC connected signal source, but means can easily switch it between the 2 inputs.
Re R3, see post 4:53pm above.
@ Nick yes that would be a "unity gain buffer", like wot you can do wiv real op-amps innit
Although that idea is scuppered somewhat by the DC connected signal source, but means can easily switch it between the 2 inputs.
Re R3, see post 4:53pm above.
@ Nick yes that would be a "unity gain buffer", like wot you can do wiv real op-amps innit
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."
#15
@ Nick yes that would be a "unity gain buffer", like wot you can do wiv real op-amps innit
That can sink a lot of current, so a possible A2 driver stage.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.