chris ma said:
Hi there,
Sorry to disagree, I like my CLC PSU for CLASS A amp just as much as the regulated ones, in fact the CLC almost generate no heat(in my case). You can save the heat with CLC but also gain in size though
The Butcher
I don't think we are disagreeing. I think we are talking about two different things.
I was addressing still4given's question about heat.
Is it inaccurate to imply that regulated supplies are commonly used or recommended for class-A amps and that they generate additional heat?
EchoWars said:
Which BrianGT power supply? I've noticed a few now. If its the latest Gainclone PSU then I think you can use 1 half of the split supply for each one channel because each rail is completely independant. No CRC filtering though.
I'd add a 0R1 resistor and cap to the ouput, as recommended by Rod Elliott.I've been taking a few measurements of my DoZ during power over the last few days. I'm using 34.5 volts instead of 40 volts and my DoZ is standard without the zener mod. The 1R resistor is still inserted on the power rail for measuring purposes. On initial power up, I notice the current is very close to 3 amp, then drops to 1.2 amps and finally settles at 1.5 amps when everything gets to operating temperature. Initial current could be higher because of the reaction time of the multimeter.
Also, we have to be a little careful here, I don't think reliability has been proven to be an issue (yet?). EchoWars' concerns may well be valid and his mods beneficial but one failure doesn't prove a problem. I'd be very interested in feedback from other DoZ builders. I don't want to discourage others from building a DoZ as it is a very good sounding amp and has been a valuable learning experience for me.
Some of us are running our DoZ at a lower voltage so we may never experience this issue. Would lower voltage mean lower initial current? I dunno.
You have both a smaller transformer and a 1 ohm resistor to limit current. Also, it doesn't take a genius to see that a current surge of well over 5 amps on a transistor with nearly 40 volts across it, lasting for several seconds, is a really really bad thing. And if you don't trust your multimeter, use a scope probe on the 1 ohm resistor. You may see momentary currents higher than you think (or not, that resistor has to help).
There is no voodoo here, and absolutely no speculation. The causes, consequences, and solutions are all well documented in the first post, as well as verification from Rod that he is aware of the large current surge at power up.
There is no voodoo here, and absolutely no speculation. The causes, consequences, and solutions are all well documented in the first post, as well as verification from Rod that he is aware of the large current surge at power up.
Thanks... If you're a member there, fess up..
I checked and rechecked my facts before I posted here, as I know the members at DIY can be a 'tad' critical. As I see it, the DoZ needed that small tweak to allow it to remain in the land of the living for a bit longer than the original design would allow. The changes cannot 'hurt' a thing, and my measurements show decided improvements in the areas that I'm certain led to previous failures. A 'win-win' situation as I see it.
I checked and rechecked my facts before I posted here, as I know the members at DIY can be a 'tad' critical. As I see it, the DoZ needed that small tweak to allow it to remain in the land of the living for a bit longer than the original design would allow. The changes cannot 'hurt' a thing, and my measurements show decided improvements in the areas that I'm certain led to previous failures. A 'win-win' situation as I see it.
There seems (seems??) to be a great deal of comment on the DoZ (most of which I only heard about very recently), and I suppose it's time I got involved.
A couple of points right at the start ...
1) The original DoZ article was simply an exploration into the possibilities - MOSFET vs. bipolar, low power Class-A, etc. The only reason that PCBs became available for it was that people kept annoying me for them
2) Given the original idea, I must confess that it did not get the attention that I normally give a design that is available on a PCB. I had tested it, everything worked, and there were no failures during testing.
Now, on with the explanations. The comments made by ilimzn (I do hope we don't have to pronounce that) are pretty much right on the money. C4 does need to be increased in value, and a 1000uF 10V cap is fine - should fit the PCB without too much trouble.
C6 can be reduced to 100uF, and those two mods will (almost) completely eliminate the big turn-on current peak. This comes at a cost though, and there will be a bigger than normal turn-on thump, because the current source is effectively disabled for around 1.5 seconds after power is applied.
Adding a 0.1 ohm resistor in the emitter of Q3 will help to stabilise the current, but the source is inherently unstable. This is because it is modulated by the signal, so is not a fixed source as is used in many other designs. I got the idea from the original John Linsley Hood Class-A amp, and although it is unstable, it also works well once adjusted. The idea of an unstable current source rather goes against the grain for me, but at the time I never expected anyone to actually build it.
If the source were to be made completely stable the dissipation would have to increase. The reason is that without modulation, you need the quiescent current to be equal to the full peak current needed by the load.
In hindsight, the amp would probably be better with a lower supply voltage. About 35 would be perfect, as this will reduce dissipation in all devices.
It has also been suggested that the zeners need to be bypassed. Not so. The input already has a bypass (C6), and the current source input is bypassed by C4. Although that does connect the zener to the output for AC noise, impedance is very low and noise rejection is high, so that should not be an issue.
Adding a resistor from the wiper of VR1 to ground isn't a bad idea, but if you use a good quality multiturn pot the wiper will never go open circuit. If it makes anyone feel better, then by all means add it.
As for reliability ... anything that runs as hot as the DoZ is bound to be less reliable than cool running gear, but I have had very few negative comments over the years, and many of those were because of fake transistors. I do accept that the switch-on current peak is easily capable of exceeding the SOA of Q3. When the amp is cold, it will be fine, but a 150W (or more) peak dissipation with hot transistors is likely to cause failure.
I hope this helps, and that everyone can get their DoZ up and running without blowing up too many transistors. I will also update the construction page to highlight the change of value for C4 and C6. As for the output coupling cap, I do suggest that it is 4,700uF in the bill of materials, so the comment about the -3dB frequency being 26Hz is based on the original article, rather than the construction notes.
Cheers, Rod
A couple of points right at the start ...
1) The original DoZ article was simply an exploration into the possibilities - MOSFET vs. bipolar, low power Class-A, etc. The only reason that PCBs became available for it was that people kept annoying me for them
2) Given the original idea, I must confess that it did not get the attention that I normally give a design that is available on a PCB. I had tested it, everything worked, and there were no failures during testing.
Now, on with the explanations. The comments made by ilimzn (I do hope we don't have to pronounce that) are pretty much right on the money. C4 does need to be increased in value, and a 1000uF 10V cap is fine - should fit the PCB without too much trouble.
C6 can be reduced to 100uF, and those two mods will (almost) completely eliminate the big turn-on current peak. This comes at a cost though, and there will be a bigger than normal turn-on thump, because the current source is effectively disabled for around 1.5 seconds after power is applied.
Adding a 0.1 ohm resistor in the emitter of Q3 will help to stabilise the current, but the source is inherently unstable. This is because it is modulated by the signal, so is not a fixed source as is used in many other designs. I got the idea from the original John Linsley Hood Class-A amp, and although it is unstable, it also works well once adjusted. The idea of an unstable current source rather goes against the grain for me, but at the time I never expected anyone to actually build it.
If the source were to be made completely stable the dissipation would have to increase. The reason is that without modulation, you need the quiescent current to be equal to the full peak current needed by the load.
In hindsight, the amp would probably be better with a lower supply voltage. About 35 would be perfect, as this will reduce dissipation in all devices.
It has also been suggested that the zeners need to be bypassed. Not so. The input already has a bypass (C6), and the current source input is bypassed by C4. Although that does connect the zener to the output for AC noise, impedance is very low and noise rejection is high, so that should not be an issue.
Adding a resistor from the wiper of VR1 to ground isn't a bad idea, but if you use a good quality multiturn pot the wiper will never go open circuit. If it makes anyone feel better, then by all means add it.
As for reliability ... anything that runs as hot as the DoZ is bound to be less reliable than cool running gear, but I have had very few negative comments over the years, and many of those were because of fake transistors. I do accept that the switch-on current peak is easily capable of exceeding the SOA of Q3. When the amp is cold, it will be fine, but a 150W (or more) peak dissipation with hot transistors is likely to cause failure.
I hope this helps, and that everyone can get their DoZ up and running without blowing up too many transistors. I will also update the construction page to highlight the change of value for C4 and C6. As for the output coupling cap, I do suggest that it is 4,700uF in the bill of materials, so the comment about the -3dB frequency being 26Hz is based on the original article, rather than the construction notes.
Cheers, Rod
I added the resistor from wiper to ground because the first pots I used were junk...and the input cap would try to charge as I adjusted the volume. I eventually replaced the pot with a decent one (why is a simple, halfway decent pot SOOO damn expensive??) and simply left the resistor there. Seems cheap insurance.
Indeed, if C6 is reduced to 100µf, then the value of C4 can be made as low as 1000µf, but you are correct that turn on thump might be an issue. I'd suggest that, with the zener modification, a compromise of 220µf would be appropriate.
Large value 10V caps for C4 are available in a 10mm diameter, which fits fine...that's what the average computer motherboard uses for filter caps. If C6 was reduced to 220µf, I would still use a 2200µf 10V cap, simply because it is common size for the above mentioned application.
I know the zener would be fine without bypassing. But it soothes my sense of continuity to include it.
Thank you Rod...you are a gentleman and a scholar.
Indeed, if C6 is reduced to 100µf, then the value of C4 can be made as low as 1000µf, but you are correct that turn on thump might be an issue. I'd suggest that, with the zener modification, a compromise of 220µf would be appropriate.
Large value 10V caps for C4 are available in a 10mm diameter, which fits fine...that's what the average computer motherboard uses for filter caps. If C6 was reduced to 220µf, I would still use a 2200µf 10V cap, simply because it is common size for the above mentioned application.
I know the zener would be fine without bypassing. But it soothes my sense of continuity to include it.
Thank you Rod...you are a gentleman and a scholar.
SOA protection
1. Build a speaker protection circuit with time delay.
2. Current limiter for Q3. See attached circuit. The collector of the BC548 might well be connect to base of Q2. Could someone comment on this suggestion.
The 1K ohm resistor is for charging up the output capacitor in case the loudspeaker is not connected and essential for the speaker protection circuit which should be connected right after.
1. Build a speaker protection circuit with time delay.
2. Current limiter for Q3. See attached circuit. The collector of the BC548 might well be connect to base of Q2. Could someone comment on this suggestion.
The 1K ohm resistor is for charging up the output capacitor in case the loudspeaker is not connected and essential for the speaker protection circuit which should be connected right after.
Attachments
DC speaker protection is really not a problem considering the output is capacitor coupled, unless you are using a very small and sensitive speaker - otherwise, most speakers are capable of withstanding the 50-something ms pop if the output suddenly fails, as long as you do not repeatedly try it. That being said, a delay + relay may be a good idea, however, if you are using one, you might as well use a DPDT relay and while the speaker is disconnected, connect an 8 ohm resistor instead. That way only a very small delay is sufficient.
As for your current limiter - it is another form of current source, so if you are inserting a resistor and another transistor in there, you may as well limit the current at it's proper value, in which case you actually only need to connect the existing transistor differently. You should keep in mind that the limit operates only when there is ~~0.5V across the resistor so this is how much of the potential output voltage swing you lose. On the plus side, the current source would be very stable. On the minus side, limiting the current at 5A already violates the SOA for the transistors Echo used. Slower transistors are much more fogiving in this regard, but also offer less performance.
As for your current limiter - it is another form of current source, so if you are inserting a resistor and another transistor in there, you may as well limit the current at it's proper value, in which case you actually only need to connect the existing transistor differently. You should keep in mind that the limit operates only when there is ~~0.5V across the resistor so this is how much of the potential output voltage swing you lose. On the plus side, the current source would be very stable. On the minus side, limiting the current at 5A already violates the SOA for the transistors Echo used. Slower transistors are much more fogiving in this regard, but also offer less performance.
My idea was to avoid the Surge current at switching on to destroy TR2 ( see the first post of this topic). Tr2 should seldom goes beyond 2Amp (??) in my own steady state setting. The current limit transistor would not work normally but only at switching on.
Correct me if I am wrong that TR2 seldom goes beyond 2 or 3 amps.
Correct me if I am wrong that TR2 seldom goes beyond 2 or 3 amps.
Sorry folks, I mixed up the transistor designation in JLH with Doz.
I meant the current limiter (as shown on the attached diagram is connected to base and emitter of Q3) is for protecting Q3. If there is a surge current larger than 5 to 6 amp flowing through Q3, then the current limiter will bypass the Q3 base current limiting Q3 collector(emiter) current to 6 amp utmost.
Of course, the suggestion by Echoware are fine too.
I meant the current limiter (as shown on the attached diagram is connected to base and emitter of Q3) is for protecting Q3. If there is a surge current larger than 5 to 6 amp flowing through Q3, then the current limiter will bypass the Q3 base current limiting Q3 collector(emiter) current to 6 amp utmost.
Of course, the suggestion by Echoware are fine too.
Hm, I believe you are missing the point of the whole circuit.
Q3 _IS_ supposed to be a current source which is NEVER supposed to source more than the idle current. The only reason it does is because with the original parts the reference current for it becomes excessive. But there are many ways to make a current source - or, to be more precise, a current limiter. So, in essence you are proposing a current limiter for a current limiter. Why not just fix the first one?
That being said, the current limiter you are proposing has some virtues compared to the one in the DOZ - and one problem, up to 1V loss of headroom. There are also other topologies that avoid this.
Q3 _IS_ supposed to be a current source which is NEVER supposed to source more than the idle current. The only reason it does is because with the original parts the reference current for it becomes excessive. But there are many ways to make a current source - or, to be more precise, a current limiter. So, in essence you are proposing a current limiter for a current limiter. Why not just fix the first one?
That being said, the current limiter you are proposing has some virtues compared to the one in the DOZ - and one problem, up to 1V loss of headroom. There are also other topologies that avoid this.
ilimzn ,
thanks, understand.
Can you comment on the JLH Iq control circuit (1996 sep issue Fig 3) ?? It used a MJE371 with emitter connected to a 0R33 which is in series with Q3 collector (Tr2 in JLH) and its base to a reference supply....seems like a control loop by itself. Would it smooth out the start-up surge ??
thanks, understand.
Can you comment on the JLH Iq control circuit (1996 sep issue Fig 3) ?? It used a MJE371 with emitter connected to a 0R33 which is in series with Q3 collector (Tr2 in JLH) and its base to a reference supply....seems like a control loop by itself. Would it smooth out the start-up surge ??
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.