I did know about that one where he offers just about everything anyone might want which makes it pretty complex if all are built. True some of the facilities can be simplified but I feel that a simple power amp needs a simple pre amp. If OP amps are used the most critical factor really is noise. Distortion depends but the 741 circuit I mentioned can achieve <0.1% thanks to required slew rate which depends on signal levels eg
signal with a peak amplitude of 5 volts at a frequency of 25kHz. An op amp with a slew rate of at least 2 π x 25 000 x 5 = 0.785V/µs would be required.
Would anyone want 5v peak out of a preamp?
There was another design in 83. I've attached it.
What value pot would you suggest bearing in mind it will have ~50k in parallel with the wiper? The poster already appears to have a volume control but can't turn it up much due to the sensitivity of the power amp making setting sound levels difficult.
Just what is wrong with making use of the ~50k by simply using a series resistor to connect his pre amp to the power amp?
Not really. It's just forming a voltage divider with the input resistance of the amp. There will be input capacitance as well. Conceivably this could be an issue but if significant the amp would have problems anyway all on it's own. This is why it's generally best to drive from a low impedance source which you are by the sound of it. Any resistor solution could have the same problem. If you use a pot the wiper will have ~50k in parallel with it's setting. Say you used a 10k and wanted to 1/2 the gain. The centre setting will be a bit out as 5k//50k. Your preamp is likely to be fine driving a 10k. Use 1m and the 50k becomes more significant,The apparent input resistance of the power amp has changed so coupling caps may need looking at again if it's lowered. A series resistor increases the input impedance but means the drive impedance is higher.
You could use either method but a series resistor is what I would try as it's the simplest. Pot a low value and no need to be log. A 10k pot in some respects has least other effects on the circuit but your coupling cap will need changing. You could put it in the preamp before it's coupling cap if it has one but that increases the drive impedance.
Reducing the power amps gain is a bit of an unknown. It may go unstable / wreck the square wave response. Some op amps state stable at unity gain. It's that sort of problem. A fix can be a bit surprising. Speed it up. One way of doing that is add a rc combination across the main feedback resistor to increase gain at higher frequencies. Rather tricky stuff. Phase correction effectively.
The preamp I have attached has a problem - in my view - no trimmers to balance volume from different sources. In my view for a DIY design that's NVG. That may be another solution for you.
Yes. The 2 100k form 50k which will be close to the input impedance of the amp. It has the transistor base loading in parallel with it as well. That relates to it's collector current and HFE or hfe in ac mode. Both vary as supplied and with collector current and Vce. In practical terms this means that the DC current in the 100k's has to be dominant and significantly higher than the load provided by the transistors base current over the voltage levels going into the amp.
The 100k's are in parallel as zero supply rail impedance is assumed. It's an approximation. An idea of the sort of value that the circuit presents. Trying to get accurate figure via calculation is great fun - work through using max and min HFE values, include the the effects of the stage it's driving as well, and the next one.
Similar. Single transistor but easier to explain. Same base circuit but with a collector and emitter resistance. Input resistance will be ~ the base resistances in parallel. The gain of the stage will be very loosely Rc/Re. In an op amps case the gain is so high that the feedback ratio sets the gain very accurately. In the bipolar case the additional load on the input relates to HFE and the collector current. That will be In parallel with the base resistors.
The gain wont be that high.I built one yonks ago to see but that aspect will vary with Vce, voltage drive and current levels. More accurate calc's can be done.These are just techniques for roughing a circuit out. These days the next step would be simulation but maybe more sums. Complex numbers can even be waved about.
There has to be limits though.In that case it is possible the resistors in the feedback network have the incorrect value. These should be 2.7k and 120 ohm it can be difficult sometimes to distinguish color differences between red and orange with some metal film types. I suggest you measure these to see if they are correct. If they are I think likely assuming the CD player is not at fault, the problem is with the light speed attenuator matching. In that case check circuit diagram to see if it has a gain stage i.e. configured like your power amp but using different values. This unit could be intended for use as a line amplifier as a stand alone preamp. With that in series with your normal preamp that may well account for what you are observing.
I hit the wrong keys too at times, 220ohm. That should set the gain of the amp to ~13. You could work that back from the input for a rail to rail output swing. The original will swing very close to that without apparent problems. Not sure on the newer versions.
There is another version not by JLH in this. Also a preamp and a mixer that could be adapted. Also horn loudspeakers and passive cross overs.
https://worldradiohistory.com/UK/Wireless-World/SPECIALS/WW-High-Fidelity-Guide-1977.pdf
How would you place these trimmers?
The actual, measured gain of my JLH Class A is 12.8. So I figure the resistor values is correct.
Can't really say. Depends on the preamp you are using. If you can post a diagram I may be able to make suggestions. If you are using an op amp for gain a simple solution is 2 gain setting resistor and select as needed with one feedback resistor. A preferred way to do it would be to use a make before break switch to bring another resistor in parallel to get 2 values. That way there is never no resistance connected. Ideally the + and - terminals should see the same resistance connected but that is not that important on ones with a FET input
as mentioned in some TI ap notes. So if you have done that pick a medium value.For lots of inputs have a look at the mixer design in the wireless world I linked to. Probably a better option than the preamp shown in the same magazine where the gain setting isn't so obvious.
The sensitivity mentioned on the other design is close to 0 dBu. A term I hate as have to remember that means 0.775 rms which as usual needed google to find out. So p to p is 0.720 x 2 x 1.414 = just over 2v so swing at the output would be 25.6v. The also use 1.4v rms which is +4 dBu. I'd guess a professional power amp would cope with that. It's hard to find out where these numbers come from.
One thing I didn't mention is gain with feedback. JLH calcs it as sum of R's/ smaller R
ie R1+R2 divided by R2 that being the lower value in his circuit. Same a 1+R1/R2. Higher the open loop gain the closer it will be to the correct value.
Something I am apt to forget on op amps and just use R1/R2. I seem to have picked that up from somewhere. Most likely colleagues at work. Texts will mention the other way.
ie R1+R2 divided by R2 that being the lower value in his circuit. Same a 1+R1/R2. Higher the open loop gain the closer it will be to the correct value.
Something I am apt to forget on op amps and just use R1/R2. I seem to have picked that up from somewhere. Most likely colleagues at work. Texts will mention the other way.
This does seem more like a case of having excessive gain rather than trouble with components like pots. Are your buffers simply buffers or are they actually increasing gain? First up, I would want to measure and compare the signal strengths before and after your buffers to ensure they were doing no more than you suggest.
Incidentally, simply adding a series resistance in the circuit is not the right way to attenuate a signal. Use a pad which approximately matches to the circuit impedances, like one of these:
Incidentally, simply adding a series resistance in the circuit is not the right way to attenuate a signal. Use a pad which approximately matches to the circuit impedances, like one of these:
Yes I did get the earth decoupling resistor value wrong at 120 Ohm that is the value used in the JLH Class AB amplifier that I had been looking at. JLH claimed that could be operated in Class A by increasing the standing current. I won't comment on that here.
Anyway series feedback circuits have a minimum gain of one. At that point the output level is close to the input signal.
What our member reported is a measured gain of 12.8 that is close enough to the mark to say there is no excess gain in the amplifier. From my own experience with the 1996 hardware a CD player is unlikely drive the amplifier to extreme levels of loudness at the low volume setting. The next item on the list is the streamer.
There are two ways one could look at this. First set up a test with a 10 k linear pot on the output and alter this to a point where the signal does not overdrive the power amplifier and base calculations on the split of value from the wiper to each end of that pot.
I looked at the thread for the light speed attenuator but not so far as to find out what the amplification factor is for this. If there is an internal op.amp which is unity gain stable then any feedback resistor network could be altered to reduce the gain of this unit.
I would put more faith in some of the PGA2310 based volume controls on offer online which come complete - rather than the new idea being offered here. One would think if there was commercial merit in this lightspeed form of circuit, such would have already considered by industry professionals.
The PGA2310 is backed by Texas Instruments professional know how.
Anyway series feedback circuits have a minimum gain of one. At that point the output level is close to the input signal.
What our member reported is a measured gain of 12.8 that is close enough to the mark to say there is no excess gain in the amplifier. From my own experience with the 1996 hardware a CD player is unlikely drive the amplifier to extreme levels of loudness at the low volume setting. The next item on the list is the streamer.
There are two ways one could look at this. First set up a test with a 10 k linear pot on the output and alter this to a point where the signal does not overdrive the power amplifier and base calculations on the split of value from the wiper to each end of that pot.
I looked at the thread for the light speed attenuator but not so far as to find out what the amplification factor is for this. If there is an internal op.amp which is unity gain stable then any feedback resistor network could be altered to reduce the gain of this unit.
I would put more faith in some of the PGA2310 based volume controls on offer online which come complete - rather than the new idea being offered here. One would think if there was commercial merit in this lightspeed form of circuit, such would have already considered by industry professionals.
The PGA2310 is backed by Texas Instruments professional know how.
I think there is a chance of going over the top with this particular problem. For instance the amp does not have a balanced input and it's input impedance is there what ever you do about it.
There could also be some confusion. Does the poster want more volume or does he have a volume control that has to cranked right down making levels hard to set.
Fact is that if it's a volume setting issue the simpler resistor suggestions are highly unlikely to cause problems. The reasons for suggesting a lowish value pot for that solution are obvious. Wiper connected to the input cap on the amp.
Not much point guessing what voltage a cd player puts out. Best measure it. The web suggests typically -10dBu which is ~0.25v rms. In that case it would need an amp adding with a gain of just under 3. The problem with dBu is that it's vu meter related so in practice higher may come out. Hard to say without knowledge of how this spec is derived. There will be a reason for the use of +10dBu / -10dBu leaves plenty of scope for that if an amp is designed for 0dBu input. Likely to be too much,
What does a streamer put out - pass. It may be in it's spec. Probably wont be.
The simplest way of adding gain would be a decent op amp. For mixed gains an inverter could be used. Resistor values can simply be switched using make to break switches to ensure the amp see resistance all of the time. Same with a none inverted set up. The feedback resistor could be switched but for various reasons most would choose the inputs. The none inverted has an advantage because the none inverted input shouldn't be left floating. A resistor to ground is needed which means always the same input impedance. All simple with dual rail supplies but not so simple with single. There maybe some ideas around but on the face of it the + and - inputs would be connected to dividers. Some sums might cause a switch to using a separate opamp for each input. Hopefully people can now see how their equivalent resistances are worked out compared with the usual circuit. Maybe they can't. Tough.
There could also be some confusion. Does the poster want more volume or does he have a volume control that has to cranked right down making levels hard to set.
Fact is that if it's a volume setting issue the simpler resistor suggestions are highly unlikely to cause problems. The reasons for suggesting a lowish value pot for that solution are obvious. Wiper connected to the input cap on the amp.
Not much point guessing what voltage a cd player puts out. Best measure it. The web suggests typically -10dBu which is ~0.25v rms. In that case it would need an amp adding with a gain of just under 3. The problem with dBu is that it's vu meter related so in practice higher may come out. Hard to say without knowledge of how this spec is derived. There will be a reason for the use of +10dBu / -10dBu leaves plenty of scope for that if an amp is designed for 0dBu input. Likely to be too much,
What does a streamer put out - pass. It may be in it's spec. Probably wont be.
The simplest way of adding gain would be a decent op amp. For mixed gains an inverter could be used. Resistor values can simply be switched using make to break switches to ensure the amp see resistance all of the time. Same with a none inverted set up. The feedback resistor could be switched but for various reasons most would choose the inputs. The none inverted has an advantage because the none inverted input shouldn't be left floating. A resistor to ground is needed which means always the same input impedance. All simple with dual rail supplies but not so simple with single. There maybe some ideas around but on the face of it the + and - inputs would be connected to dividers. Some sums might cause a switch to using a separate opamp for each input. Hopefully people can now see how their equivalent resistances are worked out compared with the usual circuit. Maybe they can't. Tough.
I do wonder about the high value resistor on the input to TR1. JLH will have had reasons for using the values he did. If built bear that in mind.
Where a diode is used as it is some circuits use the BE junction of a small transistor. Not sure if there is any advantage really but may be to remain at the same temp as the transistor. That's the aspect to think about. One circuit used a ZTX part and from memory their cases are thinner than others.
Some of the TR's are available. The odd numbered PNP isn't. Maybe the pnp version of the lower npn would suite. Check the specs. Power tasnsistors usual problem, not available. Spec's can be found.
The most interesting aspect of this circuit is how close it compares to the JLH 1969 with the alternative quiescent current control circuit published in Wireless World 1970 fig 2 on page 608.
The main difference in the cicuit submitted by A.H.Cavert is the omission of a 500uF base modulation cap between Tr2 emitter and base. The omission converts a push pull system into one that is single ended, one less efficient, and devoid of the distortion cancellation properties of push pull operation.
Chronologically the JLH circuit predated the Calvert circuit idea also published in Wireless World and unlike JLH and Nelson Jones earlier, whose works warranted full articles the privilege of publication of a full article was not accorded to Mr Calvert.
The power transistors he chose came in TO36 packages which are round. Thus the heat sinks would need isolating from the chassis and the devices have far worse frequency characteristics than ones chosen byJLH even 2N3055H are better.
As JLH has a large following on balance I doubt anyone apart from you would consider building Calvert's circuit.
Can we now close this discussion and move on.
Last edited:
AjohnL - you seem to be confused between op-amp gains. In the non-inverting configuration the gain is set by (Rf+Rg)/Rg where Rf is the feedback resistor and Rg the one which goes to ground. In the inverting configuration the gain is (-)Rf/Rin.
As regarding potentiometers for the JLH, you seem to be stuck on the 100k input bias impedances. Using high value pots (50k or 100k) will increase noise. There is no requirement for high value pots, and 10k will be adequate. Assuming any preamp or CD player is able to drive it.
As regarding potentiometers for the JLH, you seem to be stuck on the 100k input bias impedances. Using high value pots (50k or 100k) will increase noise. There is no requirement for high value pots, and 10k will be adequate. Assuming any preamp or CD player is able to drive it.