Sith and Beppe61,
I suppose one can criticise just about every circuit available, even the good ones, even if in the latter case alternative suggestions become with greater frequency opinions or beliefs, the origin of which the commentator is often not too certain of.
Your given circuit is one of many so-called trouble-free/easy/simple options. But quite an improvement can be had with no real increase in cost. To keep this brief I would state a few general pointers to look for in future, rather than another circuit.
But firstly, you mentioned the very valid point of whether 0,01% of distortion is audible in the presence of loudspeaker distortion orders higher. Very low total harmonic distortion (thd) figures in amplifiers are very often of promotional value only. The point to grasp (and this is a whole subject on its own) is that it is the composition of the distortion that is important. High order odd harmonics, especially the 7th, 9th, 11th and 13th can generate such discordant artifacts in the human hearing mechanism, that it can be detected even at levels right on the threshold of hearing. It is thus totally a matter of what that 0,01% consist of. E.g. 0,1% distortion can not be objectionable when consisting mainly of 2nd, 3rd and no higher products. On the other hand amplifiers with thd of 0,01% can generate what is called listener fatigue if containing high order products. This as a summary, but it is this very matter that mostly singles out good semiconductor designs from the rest.
In summary, just the rest:
1. The best output configuration is a full complimentary pair. Dual emitter followers are 2nd, but especially Darlington types as per your circuit are, I fear, low on the list because of limited frequency ability.
2. Somewhat more difficult to see just from a circuit, is the basic requirement that the response of the amp before feedback should cover about the whole audio spectrum, so that feedback stability requirements [compensating capacitor between collector-base of the driver (also called the dominant capacitor)], operates as far up the audio spectrum as possible. Too severe compensation is one of the main causes of high order harmonic generation, etc. The circuit you posted originally does not fulfil this requirement.
3. A driver fed from a current source, an element of which is normally an improved way of temperature sensing over the transistor between the output transistors' bases, is again a better option than doing it with "bootstrapping" as in your posted circuit (that is another harmonic multiplier).
4. Being something of a rebel, I leave maximum currnet limiting out. They do have an effect on distortion, even if marginal, and there are other reasons. Fuses are simple but they can be non-linear if of low enough rating to be effective. So my simple philosophy: Do not short the output terminals! There is no need for that to happen in a domestic amplifier, anyway.
And so on. I may not have done you a favour because I cannot begin to go into any of these details without totally exceeding my welcome. Much of this can be found on web-sites; I am also not unwilling to post basic details to you privately in a very simple document I have prepared. (Only I will have to restore that on my PC; my previous one with a lot of stuff on it was stolen just before Christmas!)
The bottom line is then that your better circuits will contain the above topologies, without necessarily costing more.
Good hunting!
Regards.
I suppose one can criticise just about every circuit available, even the good ones, even if in the latter case alternative suggestions become with greater frequency opinions or beliefs, the origin of which the commentator is often not too certain of.
Your given circuit is one of many so-called trouble-free/easy/simple options. But quite an improvement can be had with no real increase in cost. To keep this brief I would state a few general pointers to look for in future, rather than another circuit.
But firstly, you mentioned the very valid point of whether 0,01% of distortion is audible in the presence of loudspeaker distortion orders higher. Very low total harmonic distortion (thd) figures in amplifiers are very often of promotional value only. The point to grasp (and this is a whole subject on its own) is that it is the composition of the distortion that is important. High order odd harmonics, especially the 7th, 9th, 11th and 13th can generate such discordant artifacts in the human hearing mechanism, that it can be detected even at levels right on the threshold of hearing. It is thus totally a matter of what that 0,01% consist of. E.g. 0,1% distortion can not be objectionable when consisting mainly of 2nd, 3rd and no higher products. On the other hand amplifiers with thd of 0,01% can generate what is called listener fatigue if containing high order products. This as a summary, but it is this very matter that mostly singles out good semiconductor designs from the rest.
In summary, just the rest:
1. The best output configuration is a full complimentary pair. Dual emitter followers are 2nd, but especially Darlington types as per your circuit are, I fear, low on the list because of limited frequency ability.
2. Somewhat more difficult to see just from a circuit, is the basic requirement that the response of the amp before feedback should cover about the whole audio spectrum, so that feedback stability requirements [compensating capacitor between collector-base of the driver (also called the dominant capacitor)], operates as far up the audio spectrum as possible. Too severe compensation is one of the main causes of high order harmonic generation, etc. The circuit you posted originally does not fulfil this requirement.
3. A driver fed from a current source, an element of which is normally an improved way of temperature sensing over the transistor between the output transistors' bases, is again a better option than doing it with "bootstrapping" as in your posted circuit (that is another harmonic multiplier).
4. Being something of a rebel, I leave maximum currnet limiting out. They do have an effect on distortion, even if marginal, and there are other reasons. Fuses are simple but they can be non-linear if of low enough rating to be effective. So my simple philosophy: Do not short the output terminals! There is no need for that to happen in a domestic amplifier, anyway.
And so on. I may not have done you a favour because I cannot begin to go into any of these details without totally exceeding my welcome. Much of this can be found on web-sites; I am also not unwilling to post basic details to you privately in a very simple document I have prepared. (Only I will have to restore that on my PC; my previous one with a lot of stuff on it was stolen just before Christmas!)
The bottom line is then that your better circuits will contain the above topologies, without necessarily costing more.
Good hunting!
Regards.
consort_ee_um said:
Dear Sir,
thank you sincerely for your very kind and much valuable support.
I hoped very much in a your positive reply because before starting any search for a new power amp solution I would very much like to check if with a little mod on the amp I have at hand I could get some nice improvements, as I said.
1) Very very interesting indeed.
The fact that I could get a sound as nice as the one of the Albarrys from a very simple DIY kit is even amazing for me.
Now I understand that kits can be so good to be on a par with very fine commercial products.
2) Thank you very much again for your very kind support in my experiments.
I will scan the circuit as soon as possible and I will send it to you for comments.
As I said apart from the weak bass I found the amp quite good sounding after all.
But the bass response is so weak that it seems that a low filter is engaged.
Thank you so much again for your kindness.
Kind regards,
beppe
richie00boy said:
Dear Friend,
you are right, I am using a old GFA 545, but I have also a Samson Servo 260 that I bought on the basis of very favourable reviews.
After that, looking for a stronger bass, I bought the Adcom.
A little better but still not up to a enjoyable performance.
The music is without the necessary body.
"I still haven't found what I am looking for".
Well I have (the Albarrys) but they are so difficult to find used.
I would like a similar performance.
Thank you.
Kind regards,
beppe
I've built this kit to use as a sub amp and have had nothing but trouble with it blowing up. It's fine up to a certain point but once I crank up the volume it suddenly makes a very loud hum from the speaker and then the output pair die. I'm using the exact components from the schematic. I'm interested in the suggested removal of the current limiting components, but am concerned that will make it even more keen to blow.
Any advice welcomed.
If you couldn't already tell, I'm a complete novice to this whole subject.
Many thanks,
Dan
Any advice welcomed.
If you couldn't already tell, I'm a complete novice to this whole subject.
Many thanks,
Dan
Have you actually bought and built the kit, or have you just cloned it? What voltage rails are you using?
There is an issue with the current limiting circuitry that could potentially cause destruction of the output devices, but I do not recommend removing the limiter, just adding flyback protection diodes from the output to each rail.
There is an issue with the current limiting circuitry that could potentially cause destruction of the output devices, but I do not recommend removing the limiter, just adding flyback protection diodes from the output to each rail.
Thanks for the reply!
I'm using a 2x30v torroidal transformer. Gives 33v off load which is rectifies to about 45v DC. 2x30 is what is recommended in the literature:
velleman k8060 manual
I am using the kit as supplied (except for all the bits I've blown up which have been replaced with parts from farnell).
What about the manufacturer of output transistor - how relevant is this? It didn't come with anything fancy as I remember.
Can you elucidate a bit more on the flyback circuit please?
I'm using a 2x30v torroidal transformer. Gives 33v off load which is rectifies to about 45v DC. 2x30 is what is recommended in the literature:
velleman k8060 manual
I am using the kit as supplied (except for all the bits I've blown up which have been replaced with parts from farnell).
What about the manufacturer of output transistor - how relevant is this? It didn't come with anything fancy as I remember.
Can you elucidate a bit more on the flyback circuit please?
The schematic has a flaw in it's output protection, in two ways:
1) there are no flyback diodes (from E to C of transistors, they are normally reverse polarized in circuit, so go from output to + rail, and from - rail to output). Some darlington pairs have these integrated inside, though. If they are absent, the amp will self destruct every time the protection engages and anything but a pure resistive load is used. This is because any inductance in the load will provide an induced voltage pulse over that of the supply rails, which will not be damped because once the protection engages, the output becomes a current source, i.e. very high impedance. This spike will reverse-polarize the B-E junction way over it's maximum and destroy the transistor. The diodes are there to 'clamp' any inductve kickback ('flyback') to the power rails.
2) This is a simple current limiting protection, that makes no alowance for the actual SOA of the output transistors. At least a single slope protection could have ben implemented with the same parts count.
Regarding output transistor manufacturer, it may be relevant. There are many cheap fakes around, even for already cheap devices. It is more than likely these will find themselves in a kit due to the low price.
1) there are no flyback diodes (from E to C of transistors, they are normally reverse polarized in circuit, so go from output to + rail, and from - rail to output). Some darlington pairs have these integrated inside, though. If they are absent, the amp will self destruct every time the protection engages and anything but a pure resistive load is used. This is because any inductance in the load will provide an induced voltage pulse over that of the supply rails, which will not be damped because once the protection engages, the output becomes a current source, i.e. very high impedance. This spike will reverse-polarize the B-E junction way over it's maximum and destroy the transistor. The diodes are there to 'clamp' any inductve kickback ('flyback') to the power rails.
2) This is a simple current limiting protection, that makes no alowance for the actual SOA of the output transistors. At least a single slope protection could have ben implemented with the same parts count.
Regarding output transistor manufacturer, it may be relevant. There are many cheap fakes around, even for already cheap devices. It is more than likely these will find themselves in a kit due to the low price.
OK that all sounds good then.
When the limiting kicks in suddenly the back EMF from the loudspeaker (as it is an inductive load) has nowhere to go because the output stage becomes high impedance, so it blows up the output transistors due to exceeding their Vce. The diodes 'catch' the flyback and allow it a route back into the rails, limited to a diode drop above the rail voltage hence not exceeding the Vce of the output devices.
One diode goes from the output to each rail. The positive rail has the anode of one diode, and the negative rail has the cathode of the other diode. They join in the middle at the output node.
I suspect the limiting is kicking in too early to allow you to use the full potential of 4 ohm loads though.
When the limiting kicks in suddenly the back EMF from the loudspeaker (as it is an inductive load) has nowhere to go because the output stage becomes high impedance, so it blows up the output transistors due to exceeding their Vce. The diodes 'catch' the flyback and allow it a route back into the rails, limited to a diode drop above the rail voltage hence not exceeding the Vce of the output devices.
One diode goes from the output to each rail. The positive rail has the anode of one diode, and the negative rail has the cathode of the other diode. They join in the middle at the output node.
I suspect the limiting is kicking in too early to allow you to use the full potential of 4 ohm loads though.
Thanks once again. That makes sense I think although that means this kit must have blown up with every person who's ever tried to use it to its maximum!?
Regarding Richie00Boy's comment
Finally, I have noticed that TIP142/147 can be purchased with the e/c diode built in. Does this make those a better choice or is modifying the circuit with external diodes just as good or even better.
Many thanks (again) - your help has been excellent.
Regarding Richie00Boy's comment
does this mean that the current limiting can/should be removed once the flyback diodes are fitted to utilize the full power of the amp, or will it be even more unstable?
Finally, I have noticed that TIP142/147 can be purchased with the e/c diode built in. Does this make those a better choice or is modifying the circuit with external diodes just as good or even better.
Many thanks (again) - your help has been excellent.
If you remove the limiter there is no longer any need for the flyback diodes as the output stage will not be suddenly switching off. Of course it won't hurt to leave them in.
They probably rely on the built-in diodes of the TIP transistors, but this is such a 'universal' part that there's probably makers out there who leave the diode out... hence the blow-ups. Velleman probably just source who's cheapest at the time. Even if you got parts with diodes in for sure, it wouldn't hurt to add additional ones.
The amp is IMO a little ambitious to expect the claimed power into 4 ohm loads, I think the limiter probably goes some way to them fiddling the power figures. If you remove the limiter be sure your heatsink is massive.
They probably rely on the built-in diodes of the TIP transistors, but this is such a 'universal' part that there's probably makers out there who leave the diode out... hence the blow-ups. Velleman probably just source who's cheapest at the time. Even if you got parts with diodes in for sure, it wouldn't hurt to add additional ones.
The amp is IMO a little ambitious to expect the claimed power into 4 ohm loads, I think the limiter probably goes some way to them fiddling the power figures. If you remove the limiter be sure your heatsink is massive.
It turns out that the output pairs I have been using are indeed without diode protection across collector and emitter. This goes a long way to explain the burn-ups. I now have some more on order and a bagfull of suitable diodes to play with.
One final thing I'd like to clarify with Richie00boy:
Is this in addition to the collector and emitter diode, ie are you assuming that the darlingtons I have already had the protection in? If you are not assuming this, why place the diodes where you suggest rather than across collector and emitter, effectively mimicking the protected versions?
Sorry if I'm being stupid but I'd like to be clear(er) what I'm doing.
Many thanks again.
Dan
One final thing I'd like to clarify with Richie00boy:
Is this in addition to the collector and emitter diode, ie are you assuming that the darlingtons I have already had the protection in? If you are not assuming this, why place the diodes where you suggest rather than across collector and emitter, effectively mimicking the protected versions?
Sorry if I'm being stupid but I'd like to be clear(er) what I'm doing.
Many thanks again.
Dan
sith said:
1. Should be noticed!
This amplifier has got very low Input impedance,
set by 3K3 input bias resistor (and also feedback resistor is equally 3K3 to avoid offsets.)
( R3, R8 )
You need a preamp or transistor for buffering.
At least for many CD-player outputs. And of course for most other signal sources.
2. How about balance in the input pair.
Looks like they share 1 mA.
But what I can see FAR from 50% + 50%.
Either the zener in long tail should be changed to increase current (now it is ~ 1 mA).
Or the 680 ohm resistor ( R7 ), at one transistor input,
that sets this balance, should be adjusted to achieve
at least 48% + 52% balance, or better, of current sharing in input pair.
lineup
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.