> I previously did it from driver stage to i/p tr emitter but I concluded recently that this is not such a good method.
I would agree with that.
The other way would be to load the output with some Zobel. But compensation cap in the feedback path is easier to handle.
Patrick
I would agree with that.
The other way would be to load the output with some Zobel. But compensation cap in the feedback path is easier to handle.
Patrick
"..................Wirewound has inductance....................."
Would a carbon comp or even a carbon film resistor in parallel with the wire wound be of any use. Say 10 times the value ( 1 ~ 2.2 ohms) of the wire wound ? ( To swamp the inductance ).
Would a carbon comp or even a carbon film resistor in parallel with the wire wound be of any use. Say 10 times the value ( 1 ~ 2.2 ohms) of the wire wound ? ( To swamp the inductance ).
I would suggest to avoid inductance in emitter resistors - I tried in simulation - not good - stored energy like this in FB loop. It causes instability if the value is significant and has no useful effect if the value is insignificant.
Some more clarification asked
You state I can spoil a match with badly matching resistors. It then seems logical that it is also possible to make up for a bad (or not to good) match. The logic (to me) comes from the following resoning: If I can spoil then the resistor does something to vital characteristics. But if vital characteristics are influenced I can also put that to an advantage by influencing it in the right way for non-optimal matching transistors.
Or am I going wrong here?😕
Well, that's interesting. There must be a penalty somewhere
else no one would bother the rather expensive and time consuming matching. What's the penalty?
I am not sure the 2sc3281 will work. I've seen reports of good and bad results. That's why I am building with the trusty old 2n3055's first. If everything is ok, I'll install the 2sc3281's and upgrade to 4 per channel. Then I try and get it working (actually I'll see if it works. If it won't I'm affraid I'll be asking for advice again...).
😀
While I was composing this message, some advice is allready added.
Can I try with the usual startup method of slowly bringing up rail voltage and Iq or will it explode or something
?
Do I understand correctly that putting 10pF cap parallel to the feedback resistor (R6 in my circuit, the one in series with the 470 uF C4) might cure the oscillation problems?
Thankx, Marco
You state I can spoil a match with badly matching resistors. It then seems logical that it is also possible to make up for a bad (or not to good) match. The logic (to me) comes from the following resoning: If I can spoil then the resistor does something to vital characteristics. But if vital characteristics are influenced I can also put that to an advantage by influencing it in the right way for non-optimal matching transistors.
Or am I going wrong here?😕
Well, that's interesting. There must be a penalty somewhere
else no one would bother the rather expensive and time consuming matching. What's the penalty?I am not sure the 2sc3281 will work. I've seen reports of good and bad results. That's why I am building with the trusty old 2n3055's first. If everything is ok, I'll install the 2sc3281's and upgrade to 4 per channel. Then I try and get it working (actually I'll see if it works. If it won't I'm affraid I'll be asking for advice again...).
😀
While I was composing this message, some advice is allready added.
Can I try with the usual startup method of slowly bringing up rail voltage and Iq or will it explode or something
?Do I understand correctly that putting 10pF cap parallel to the feedback resistor (R6 in my circuit, the one in series with the 470 uF C4) might cure the oscillation problems?
Thankx, Marco
Marco,
> Or am I going wrong here?
What you said is, well I have non-linear parts. But I can play around with them such that the non-linearities compensate each other. True. Can't say you are wrong.
I said, I rather have perfectly linear parts all over, and then still try to put them together such that they still compensate for the remaining minute non-linearities. It is like 100-100=0, and 0.1-0.1=0 also. But if I what to be sure, I rather do 0.1-0.1=0.
> What's the penalty?
Damping factor reduces. Less control of the loudspeaker with changing impedance, especially during transients.
> Can I try with the usual startup method of slowly bringing up rail voltage and Iq or will it explode or something ?
I have built many amps that oscillates at the start before tuning. As long as you don't connect an expensive speaker to it, nothing dramatic happens. I have not yet blown one single transistors in 5 years and 8 power amps.
> Do I understand correctly that putting 10pF cap parallel to the feedback resistor (R6 in my circuit, the one in series with the 470 uF C4) might cure the oscillation problems?
A capacitor is one method that will help, but it depends onwhat state you are in without the cap, and what penalty you have top pay. The larger the cap, the more stable, the lower the bandwidth. Exact value you will have to try in the circuit when you HAVE oscillation. 10p is a good start. If 10kHz square waves still looks spiky, then you need to add. If too rounded, then reduce. No shortcut.
Patrick
> Or am I going wrong here?
What you said is, well I have non-linear parts. But I can play around with them such that the non-linearities compensate each other. True. Can't say you are wrong.
I said, I rather have perfectly linear parts all over, and then still try to put them together such that they still compensate for the remaining minute non-linearities. It is like 100-100=0, and 0.1-0.1=0 also. But if I what to be sure, I rather do 0.1-0.1=0.
> What's the penalty?
Damping factor reduces. Less control of the loudspeaker with changing impedance, especially during transients.
> Can I try with the usual startup method of slowly bringing up rail voltage and Iq or will it explode or something ?
I have built many amps that oscillates at the start before tuning. As long as you don't connect an expensive speaker to it, nothing dramatic happens. I have not yet blown one single transistors in 5 years and 8 power amps.
> Do I understand correctly that putting 10pF cap parallel to the feedback resistor (R6 in my circuit, the one in series with the 470 uF C4) might cure the oscillation problems?
A capacitor is one method that will help, but it depends onwhat state you are in without the cap, and what penalty you have top pay. The larger the cap, the more stable, the lower the bandwidth. Exact value you will have to try in the circuit when you HAVE oscillation. 10p is a good start. If 10kHz square waves still looks spiky, then you need to add. If too rounded, then reduce. No shortcut.
Patrick
Well...EUVL said:
The resistors are inside the feedback loop, so any increase is effectively divided by the feedback before it increases the output impedance. Increasing by 0.1 ohm, say, will be insignificant.
I started by using old 2N3055s and had trouble getting the bias adjustment to stabilize. These were a number of mixed brands of device, different dates. It may have been due to their widely different gains. Keep this in mind if you're starting with surplus 2N3055s.
Re: Some more clarification asked
Add to EUVL's comment:
Increased heat dissipation - reduced efficiency
Lower maximum output voltage swing - reduced output power
Higher wattage resistors will be required - increased cost
deduikertjes said:
Well, that's interesting. There must be a penalty somewhere
Add to EUVL's comment:
Increased heat dissipation - reduced efficiency
Lower maximum output voltage swing - reduced output power
Higher wattage resistors will be required - increased cost
Geoff,
I assumed that efficiency and costs were not factors to be considered, only sound. 🙂
Welcome back.
Patrick
I assumed that efficiency and costs were not factors to be considered, only sound. 🙂
Welcome back.
Patrick
Yes, I agree. But I build my amps mainly for the fun of building (and yes, a large part of the fun comes from building something which is better than what can be bought for a lot of bucks). That said, I like to keep costs down a bit.
That's why I won't be buying a lot of transistors to match and discard the rest. So I was wondering how to get the best from not so well matched devices.
Any hints on pairing resistors with transistors to compensate non-linearities? My guess would be the high gain transistors with the higher value transistors? Yeah, it becomes a bit of a pithole here because : how much resistor variation compensates for how much gain variation., etc. pfff, I guess I'll never figure that one out
Marco
> But I build my amps mainly for the fun of building.
Then why bother with matching, etc.. They will still work unmatched with whatever emitter resistors. And it is still a very nice amplifier, even with 2N3055's.
(Of course I speak from experience.)
Patrick
Then why bother with matching, etc.. They will still work unmatched with whatever emitter resistors. And it is still a very nice amplifier, even with 2N3055's.
(Of course I speak from experience.)
Patrick
Good question 🙄. You got me thinking
Result:
From the original 69 article and other info I gathered that matching was relatively imoprtant. Don't want to spend to much money so I figured: buy a few transistors more, throw out the worst matches, compensate remaining mismatch with resistors and have a very nice amp in the end (and learned a lot on the way which actually gives me the most pleasure).
Good to know though that it is a very nice amp even with unmatched 2n3055's. I hope to hear that in a couple of day's.
Marco
I've a couple of questions regarding mounting of the output devices.
What are the best mounting pads to use for thermal transfer/electrical insulation?
How critical is the length of the leads connecting them to the pcb?
As a general question, how susceptible is this amplifier to RF etc? Hence is layout within the enclosure critical? Has anyone built it with the power supplies in a separate box, or is that not a good idea?
J
What are the best mounting pads to use for thermal transfer/electrical insulation?
How critical is the length of the leads connecting them to the pcb?
As a general question, how susceptible is this amplifier to RF etc? Hence is layout within the enclosure critical? Has anyone built it with the power supplies in a separate box, or is that not a good idea?
J
> What are the best mounting pads to use for thermal transfer/electrical insulation?
Beryllium Oxide TO3 insulation washers. Banned in the EU (toxic).
Next best thing readily available Aluminium Oxide 0.5mm thick.
I would just use mica. Not a hugh difference.
> How critical is the length of the leads connecting them to the pcb?
Depends on layout, but as short as possible.
> As a general question, how susceptible is this amplifier to RF etc?
Please read Geoff's wonderful website.
> Has anyone built it with the power supplies in a separate box, or is that not a good idea?
Good idea. Last bank of caps after cap multiplier in the amp housing, IMHO.
Patrick
Beryllium Oxide TO3 insulation washers. Banned in the EU (toxic).
Next best thing readily available Aluminium Oxide 0.5mm thick.
I would just use mica. Not a hugh difference.
> How critical is the length of the leads connecting them to the pcb?
Depends on layout, but as short as possible.
> As a general question, how susceptible is this amplifier to RF etc?
Please read Geoff's wonderful website.
> Has anyone built it with the power supplies in a separate box, or is that not a good idea?
Good idea. Last bank of caps after cap multiplier in the amp housing, IMHO.
Patrick
It would also help if you use HF-Litze.
Like 120x 0.1mm enamel wire in parallel, twisted together.
Patrick
Like 120x 0.1mm enamel wire in parallel, twisted together.
Patrick
Am I right in concluding that it is better to have the main PSU in an enclosure, including the capacitance multiplier, then to have a bank of capacitors with the amp modules in a second box? Or would the cap multiplier have to be in the same box as the amp?
> Am I right in concluding that ......
Concluding is the wrong word. Design (including electronic or eletro-mechanical) is always a matter of personal choices. There is no absolute truth, just personal preferances.
I personally would keep the capacitor multiplier in the PSU box. But some people might choose otherwise. There is no right or wrong. And there are arguments both ways.
Patrick
Concluding is the wrong word. Design (including electronic or eletro-mechanical) is always a matter of personal choices. There is no absolute truth, just personal preferances.
I personally would keep the capacitor multiplier in the PSU box. But some people might choose otherwise. There is no right or wrong. And there are arguments both ways.
Patrick
Thats actually a very "post modernist" philosophical approach to design. I'm not wishing to dispute a person's right to have a preference for a certain type of sound but I still think that when all the facts are in we should be able to say objectively that some circuits reproduce a waveform more faithfully than others.....whether the result appeals to a person or not. (I also am aware that this is a deeply devisive area both in philosophy and audio. And I'm not up for a fight at the moment!)
Jonathan,
In most subjective areas - and audio is subjective - post modernism is arguably no less appropriate than other philosophies.......
You can be subjective about the objective, and in matters of perception this best describes what we do in audio. But a subjective view of the objective remains subjective - very much a matter of taste, and that is the issue.
Some like tubes, some like SS. You can't get away from it. People hear things differently - same in the visual arts.
Cheers,
Hugh
In most subjective areas - and audio is subjective - post modernism is arguably no less appropriate than other philosophies.......
You can be subjective about the objective, and in matters of perception this best describes what we do in audio. But a subjective view of the objective remains subjective - very much a matter of taste, and that is the issue.
Some like tubes, some like SS. You can't get away from it. People hear things differently - same in the visual arts.
Cheers,
Hugh
Thanks Hugh, It's nearly 11pm here so God knows what the time is in Melb'! No, seriously, I appreciate your response. (I was hoping for none!) I'll have to think about what you said......especially the middle bit. Just off the top of my head, without thinking about it too deeply (and I've just got back from a boring day at the Gabba) I may still end up where I was in the earlier post. I'm happy for personal preferences but I still think that we can conclude, objectively, that one sq/wave or sine wave is more nearly related to an original than another. I am thinking I may end up with a sort of pair of "parrallel universes" on this topic but I will stop now as (A) I don't want to ramble on without a bit more thought and (B) the Moderator will probably pull the pin on this subject.......
My daughter says your overnight minimums are higher than ours on occasions recently.....amazing.....regards to Melb' ....I've got fond memories of MAC from the 1970's.
Jonathan
My daughter says your overnight minimums are higher than ours on occasions recently.....amazing.....regards to Melb' ....I've got fond memories of MAC from the 1970's.
Jonathan