Been a member so many years is there a bit of thread jacking allowed ???
I will presume yes just for a sec😀
I will still place my question under investigation about the stability issues of the alchemist taking also in mind the experience coming from the P3A
To refresh the alchemist is a P3A only in the output stage we have 4 drivers and 4 outputs
Tip41 for driver and B1047 for outputs are a moderate combo but my gut say that the stability issues is coming from the pcb design and the wide spread of drivers on the pcb ( expect each and every one of them to behave thermally on its own )
so one that redesign the pcb locate the drivers close and in a common heatsink probably together with the Vbe multiplier ( than now is housed in the output heatsink ) might end up with far less compensation ,and in total with a very fast amplifier sounding very nice and benefit of much faster outputs and the additional kick provided by the 4 independent drivers ...
My question and thoughts has been placed before but only if constructed one may tell the results ..in between a few theoretical answers might become handy ..
Thanks for your time and sorry for the brake
Kind regards
Sakis
I will presume yes just for a sec😀
I will still place my question under investigation about the stability issues of the alchemist taking also in mind the experience coming from the P3A
To refresh the alchemist is a P3A only in the output stage we have 4 drivers and 4 outputs
Tip41 for driver and B1047 for outputs are a moderate combo but my gut say that the stability issues is coming from the pcb design and the wide spread of drivers on the pcb ( expect each and every one of them to behave thermally on its own )
so one that redesign the pcb locate the drivers close and in a common heatsink probably together with the Vbe multiplier ( than now is housed in the output heatsink ) might end up with far less compensation ,and in total with a very fast amplifier sounding very nice and benefit of much faster outputs and the additional kick provided by the 4 independent drivers ...
My question and thoughts has been placed before but only if constructed one may tell the results ..in between a few theoretical answers might become handy ..
Thanks for your time and sorry for the brake
Kind regards
Sakis
fig3 shows the device currents but omits what happens >1.5Apk at the output.
Extend that using a simulator to predict the driver and pre-driver currents when the output goes to +-15Apk (i.e. extend the graph by a factor of 10 vertically).
My apologies as well for slightly moving away from the topic - see what I've done in the middle of December - >CFA-CFPx2 amplifier<. OPS is also a double-CFP, but in this particular case having BGT drivers and HEX-FET output devices. All 4 drivers are set together, along with Vbe spreader, tracking their temperature. Thermal stability is excellent. Running it in my test setup since then.
P.S. This is to reply a question from Sakis. BTW, we can continue this discussion in my thread, if you like.
P.S. This is to reply a question from Sakis. BTW, we can continue this discussion in my thread, if you like.
Last edited:
drivers biased to 100mA are not common, but very possible.
I think your mention of pre-driver is a typo !
Pre-driver biased to 5mA to 10mA is common. I can't recall any that do 50mA.
Imagine the dissipation stress of a To92 with 40 or 50V supply rails biased to 50mA !!!
I already discovered the problem of biasing so high. In order to bias to 100mA, the resistor across the NPN and PNP pre-drivers has to be 15ohm because there is only 2Vbe drop across which is 1.4V. 15ohm at the emitter reflects only 1.5K impedance at the base of the predriver even if you count on beta=100. Same applies to the first EF stage of the 3EF. The input impedance will be too low and load down the VAS ( which is current drive).
Unless using CCS to pull the current for the pre-drivers, it will defeat the use of 3EF to raise the input impedance of the OPS that is so important to get higher open loop gain. In fact, if you follow my thread of the schematic of the OPS, you'll see I reduced the current down all the way.
Ha ha, just by talking out loud, I discover a lot of my own mistakes. I no longer interested in darlington also.
Last edited:
Member
Joined 2009
Paid Member
yes, after all this is diy-audio, where members build amps, not talk-audio....
even the thread title is misleading, it is not true that people do not use darlingtons in their amps.....
I disagree. You want to look at as much option as possible. You want to study the strong and weak points of each device, topology and circuit design as much as possible. It is much faster and cost saving than just go with the first one and commit to the pcb and buy all the parts, then find out two months later that you can do better IF you would just spend the time to do the research ahead of time.
You want to make sure you choose the correct transistor, transformer, capacitors, topology the best you can. Not blindly spending money and commit to something that you likely trash and have to start all over again just for the sake of "just do it" and "gain experience". In real job, you'll get fired if you fail once, try again, then try again!!!!
Last edited:
In a real job you have to do the research yourself though. 🙂 Right now you are substituting our time and effort for your own.
In my day job I frequently build prototypes on breadboard and measure their performance in the lab, before committing to a PCB spin. I would certainly recommend that for a high performance audio output stage. I can build a prototype output stage and measure it quicker than I could simulate all of the thermal stability and parasitic oscillation issues.
In my day job I frequently build prototypes on breadboard and measure their performance in the lab, before committing to a PCB spin. I would certainly recommend that for a high performance audio output stage. I can build a prototype output stage and measure it quicker than I could simulate all of the thermal stability and parasitic oscillation issues.
and you need to open a thread for all this? surely the things you ask about have been discussed here many times over...if only you used the search button....https://www.google.com.ph/#q=diyaudio+darlington+output+stages+site:www.diyaudio.com
Member
Joined 2009
Paid Member
Nonesense, modern power Darlington devices can be easily used above 30W with good results; probably twice as much.
I wonder what the real cost benefits of this 'lower component count' really is.
Remembering that using Darlingtons will lock you into a much lesser range of alternatives too (especially if you go to higher end commodity devices), meaning increased supply chain risk, and elevated procurement costs. Enough to offset the three or four lesser components? I don't know.
Then we have to consider, if we're going to reduce component count at the expense of quality (or should I say purity) then why not go all the way to say a chipamp - which come in reasonable power these days and achieve quite good performance for it too?
The death of the Darlington is because it is in a limbo area between products that address specific applications better than it can. Not to mention that other purposes they traditionally may have been used for, such as motor drivers and servo amplifiers, have moved onto more effective devices and packaging for their applications too.
Yes, agree with this approach. I like the reassurance you get when a circuit works on strip-board. Reduces the chance of wasting money on PCBs... 🙂
Just one small issue here.
Keep in mind that beta/Ie graphs for darlingtons require some interpretation. They can't be simply compared with singletons. The reason are the internal resistors. As base current increases, most of it actually pass through the resistor before sufficient Vbe is reached, which shows up on the graph as abysmal low beta for very low currents. The graph also has a lesser inflection point which is usually well hidden inside the beta droop curve of the second transistor where the same occurs with the Vbe of the second transistor, where it's really only the 'driver' doing the amplification.
There is also the mixup between classical darlingtons which are made for (SLOW!) switching and perhaps some voltage regulation, and newer darlingtons made using tripple diffusion or even LAPT processes specially for audio (mostly Sanken). They only have a resistor between the B-E of the output transistor. This is still a problem as in the actual circuit there will be an external emitter resistor, so there is still a point of hand-over where the output turns off (lack of Vbe) and the driver still contributes output current.
This is not ideal but hardly insurmountable.
Also... use class A? 😛 and most of the problems will dissapear, leaving the usual ones pertaining to heat and SOA.
Keep in mind that beta/Ie graphs for darlingtons require some interpretation. They can't be simply compared with singletons. The reason are the internal resistors. As base current increases, most of it actually pass through the resistor before sufficient Vbe is reached, which shows up on the graph as abysmal low beta for very low currents. The graph also has a lesser inflection point which is usually well hidden inside the beta droop curve of the second transistor where the same occurs with the Vbe of the second transistor, where it's really only the 'driver' doing the amplification.
There is also the mixup between classical darlingtons which are made for (SLOW!) switching and perhaps some voltage regulation, and newer darlingtons made using tripple diffusion or even LAPT processes specially for audio (mostly Sanken). They only have a resistor between the B-E of the output transistor. This is still a problem as in the actual circuit there will be an external emitter resistor, so there is still a point of hand-over where the output turns off (lack of Vbe) and the driver still contributes output current.
This is not ideal but hardly insurmountable.
Also... use class A? 😛 and most of the problems will dissapear, leaving the usual ones pertaining to heat and SOA.
Last edited:
Yeah, you can build a 60 watter with a MJ11015/6 pair. But you can build a better one with discretes. And unless you want to use some crazy expensive Sanken devices forget about 100+ watts without some twisted topology that uses more in support components. That "150 watt' amp using the TIP142/7 pair is nonsense. Maybe 50W IYL and they like to blow up. Back it off to 30W by reducing the rails and it will hold up fine. The TO-3's would be more robust but they ain't cheap unless you find some NOS in a back room somewhere.
If I was only running 25V rails, didn't need the last word in distortion or thermal stability, wanted to keep board space to a minimum, and didn't want to put together a special order I'd use a pair of TIP122/7's. That's where they really come in handy. Because they're there and you don't want to use up stuff that's suitable for something bigger.
Guys, I think this is the major problem using darlington as power transistors in parallel. You have two individual transistors as 2EF in every parallel darlington. You have to match both thermally and electrically of 2Vbe for each darlington in parallel. That's where you run into problem and that's where discrete 3EF has the big edge. Forget the variation of beta and fT.
Yes, today's darlington has much higher fT, that's not even a problem. It's the thermal matching. I totally gave up the idea of using darlington as the power output transistors.
Discrete 3EF using single pre-driver and driver in the first two stages. They are by default matched. You only worry about the power BJT alone.
Last edited:
Sure - however, under normal circumstances that sort of current only happens during crossover. Be that as it may, you are right, the resistor is typically not chosen with an audio output stage in mind. The one in the Sanken darlingtons is only on the emitter of the driver (no base bleeder) and the value of the resistor is more what one would expect in an actual amp.
- Status
- Not open for further replies.