Ozgur,
Agree with Andrew; use a resistor for load. The collector of the feedback transistor can be directly connected to negative rail.
However, as you set it up, ensure:
1. Resistance from input transistor base to ground equals resistance from feedback transistor base to output.
2. Adjust the input transistor collector resistor so that precisely half the LTP stage current flows through each transistor; this balance is important to low distortion.
3. Match the beta of the LTP transistors.
4. Use a gate stopper resistor of 220R on the output mosfet.
5. Take especial care with heatsinking, as Graham (circlotron) suggested. You might even use two paralleled devices, making four in total, for your output stage.
6. Place a 33pF silver mica lag compensation capacitor between collector and base of the VAS. This is a good start, but may need adjustment empirically for best stability.
Most people find it hard to learn theory unless they have a working, or near working, circuit out front on a workbench. There are exceptions to this of course; EEs are one! Measure everything. Blow up transistors, tiptoe through the silicon crematorium. Investigate the smoke. Curse. Walk away in despair. Drink too much coffee. Ask 'what if....' Ask friends around you, listen. Figure out who is blowing smoke up your @#$ and who is not.
This is how you learn - excitement, frustration, trauma, and money. If you are very young, you can spend nights over text books and take exams. Now you are old, you do not have the privilege, so ENJOY!!
Don't forget to vote tomorrow, either!
Cheers,
Hugh
Agree with Andrew; use a resistor for load. The collector of the feedback transistor can be directly connected to negative rail.
However, as you set it up, ensure:
1. Resistance from input transistor base to ground equals resistance from feedback transistor base to output.
2. Adjust the input transistor collector resistor so that precisely half the LTP stage current flows through each transistor; this balance is important to low distortion.
3. Match the beta of the LTP transistors.
4. Use a gate stopper resistor of 220R on the output mosfet.
5. Take especial care with heatsinking, as Graham (circlotron) suggested. You might even use two paralleled devices, making four in total, for your output stage.
6. Place a 33pF silver mica lag compensation capacitor between collector and base of the VAS. This is a good start, but may need adjustment empirically for best stability.
Most people find it hard to learn theory unless they have a working, or near working, circuit out front on a workbench. There are exceptions to this of course; EEs are one! Measure everything. Blow up transistors, tiptoe through the silicon crematorium. Investigate the smoke. Curse. Walk away in despair. Drink too much coffee. Ask 'what if....' Ask friends around you, listen. Figure out who is blowing smoke up your @#$ and who is not.
This is how you learn - excitement, frustration, trauma, and money. If you are very young, you can spend nights over text books and take exams. Now you are old, you do not have the privilege, so ENJOY!!
Don't forget to vote tomorrow, either!
Cheers,
Hugh
I am surprised too much!
How do you know that we have elections tomorrow?
Anyway. I will do your (and also Andrew's) advices. But;
"Use a gate stopper resistor of 220R on the output mosfet." I already have a gate resistor (270R) do you mean to reduce it 220R or what?
And for heatsinking issue; I have two 0,5 C/W really huge heatsinks. And I supposed to use them... Do you think they will not enough?
How do you know that we have elections tomorrow?
Anyway. I will do your (and also Andrew's) advices. But;
"Use a gate stopper resistor of 220R on the output mosfet." I already have a gate resistor (270R) do you mean to reduce it 220R or what?
And for heatsinking issue; I have two 0,5 C/W really huge heatsinks. And I supposed to use them... Do you think they will not enough?
not huge,
just about normal for a 100W to 150W class AB amplifier on non PA duty AND has a low Ibias.
These may well be far too small for a ClassA amplifier.
Bias up in stages and monitor the temperature of the device cases or the sink very close to the devices.
The hotter you run the devices, the lower the SOAR and the more unreliable the circuit becomes (including the internal temp of the amp case).
What are you talking about?
Theyre 20x10x5 cm and at least 1Kg weight. If you say "theyre not enough" so what kind of heatsink should I use?
Although, I use quarter of it for my tiny headphone amp...;
http://www.diyaudio.com/forums/showthread.php?s=&postid=1260364#post1260364
But if I use a larger one, WAF will be very very low, even impossible to use in living room.
Or should I use active cooling? But this time the fan noise will be a problem...
Hi,
think about it.
Your proposal is dissipating 48W per device, that's 96W per channel. (+-24V, Ib=2A)
Assume external sinks and Ta=30degC.
Rth s-a=0.5C/W
Rth c-s~=0.3C/W.
Delta Ts=96*0.5*correction=48*kCdeg
K~=1.1 when sink is "only" 48Cdeg above ambient.
Delta Tc=48*0.3~=15Cdeg
Tc~=30+48*1.1+15~=98degC.
temperature derating for a plastic packaged device (Tcmax<=150degC) is about 0.42.
A 150W device can only dissipate 150*0.42~63W when Tc=98degC i.e asking a 0.5C/W sink to dissipate 96W is not on.
Try 0.1C/W or if you fancy chancing your arm try 0.2C/W per channel.
Yes, 16W of ClassA runs hot, but not allowed to be that hot.
think about it.
Your proposal is dissipating 48W per device, that's 96W per channel. (+-24V, Ib=2A)
Assume external sinks and Ta=30degC.
Rth s-a=0.5C/W
Rth c-s~=0.3C/W.
Delta Ts=96*0.5*correction=48*kCdeg
K~=1.1 when sink is "only" 48Cdeg above ambient.
Delta Tc=48*0.3~=15Cdeg
Tc~=30+48*1.1+15~=98degC.
temperature derating for a plastic packaged device (Tcmax<=150degC) is about 0.42.
A 150W device can only dissipate 150*0.42~63W when Tc=98degC i.e asking a 0.5C/W sink to dissipate 96W is not on.
Try 0.1C/W or if you fancy chancing your arm try 0.2C/W per channel.
Yes, 16W of ClassA runs hot, but not allowed to be that hot.
yes, that's exactly what I have in stock and have used on 100W Class AB amplifiers capable of going to 200W into 4r. But they don't like higher bias setting. They start to get too warm. Ib=200mA and they tick over at 15Cdeg above ambient. On a very hot UK summer's day the Tc will be about 50degC, before I even ask the amp to pass any signal.
Ozgur,
Andrew is correct; you would need around 0.3C/W for for devices in the output stage - that would be 96W; 96 x 0.3 would be 29C above ambient, so if you were listening on a hot day of 30C, the sink temperature would be almost 60C, almost too hot to touch (65C).
I use a 0.3C/watt heatsink for just 70W of heat, and that gets VERY hot, so yes, you are pushing the limits.
My mistake; 270R gate stopper would be fine.
Ever since Attaturk began the industrial and secular modernisation of Turkey in 1923, given the position of Turkey between the West and the East and the rise of Islam in recent years, your country has become very, very important on the world stage. Of course we know about the elections here in Oz, and the details and debates taking place in your country. The results of this election will be very important, particularly given the situation in the Mid-East, and will have far reaching effects - along with recent developments in Iran.
Mods, please excuse OT......
Back to Electronics. You might find that 2A is not sufficient. If you set the amp up for 48V of total rail, around 44Vpp of signal will be available, generating up to 3.25A into a nominal 8R (actually 6.8R) load.
You should add around 10% more current than this to keep all devices well on and thus avoid current starvation, so you might actually need to run around 3.5A, giving you a high dissipation of 168W total. While fan cooling is noisy, it is very effective, but water cooling is even better......
Options and choices??
Cheers,
Hugh
Andrew is correct; you would need around 0.3C/W for for devices in the output stage - that would be 96W; 96 x 0.3 would be 29C above ambient, so if you were listening on a hot day of 30C, the sink temperature would be almost 60C, almost too hot to touch (65C).
I use a 0.3C/watt heatsink for just 70W of heat, and that gets VERY hot, so yes, you are pushing the limits.
My mistake; 270R gate stopper would be fine.
Ever since Attaturk began the industrial and secular modernisation of Turkey in 1923, given the position of Turkey between the West and the East and the rise of Islam in recent years, your country has become very, very important on the world stage. Of course we know about the elections here in Oz, and the details and debates taking place in your country. The results of this election will be very important, particularly given the situation in the Mid-East, and will have far reaching effects - along with recent developments in Iran.
Mods, please excuse OT......
Back to Electronics. You might find that 2A is not sufficient. If you set the amp up for 48V of total rail, around 44Vpp of signal will be available, generating up to 3.25A into a nominal 8R (actually 6.8R) load.
You should add around 10% more current than this to keep all devices well on and thus avoid current starvation, so you might actually need to run around 3.5A, giving you a high dissipation of 168W total. While fan cooling is noisy, it is very effective, but water cooling is even better......
Options and choices??
Cheers,
Hugh
Options or choices?
Option one; get me more earth for my Class A project...
Option two; find a honest lawyer to me (if its possible of course)..
This is insane. I couldnt realize that much power even in my dreams. Water cooling, 0,2C/W heatsink, for only 20W amplifier. Thats too much for me. I am resigned from that project.
I am happy with my LM4702...
Thx.
PS: Ataturk followers (as I am) will have a great victory in that elections. Turkey never will be as Iran.
"Peace at home peace in the World. M.K.Ataturk"
Dxvideo said:
Right. I’m not giving you the meaning of life and you should know by now that your amp’s LTP will work with either a restive load or a resistor-degenerated current mirror.
Under the delusion that it may be appreciated, I just wasted some of my time giving you a very inoffensive and non-argumentative explanation of the benefits of using a current mirror load; benefits which are pretty much technically indisputable (IMO, in correct circumstances, from a subjective point of view also).
That's enough of my time wasted in this thread.
Good bye!
.
from post28 is not quite accurate.
Indeed the mirror forces almost equal balancing of the load currents (close enough for most DC set-ups).
But, that does not help with balancing the collector currents, when current is sourced or sunk from the collector to load connection.
As soon as the VAS, or other connection, extracts a signal from the collector it throws the "balance" out. The current into the base of a BJT VAS can be zeroed out by adjusting the load resistance when not using a mirror, but both BJT and FET VAS will vary the collector current when charging/discharging the VAS Cob etc. and Cdom if fitted. That's why the VAS tends to perform better if it is a low capacitance type and why the Cdom if fitted can help with linearising the capacitance when VAS Vce varies.
I wonder if the mirror being incapable of this adjustment leads to the criticism often thrown at the LTP + Mirror topology for less good sound quality?
But, Juergen,
The LTP cares only about the currents flowing through its transistors; these are responsible for diff action together with the differential input voltage at the two bases.
The input transistor would have around 24V Vce; the feedback transistor around 24V6. This would have no effect on diff action at all.
A current mirror increases OLG very considerably, increasing fb factor, and thus improving THD, bandwidth and Zout. The increased feedback factor focusses attention on amp stability; the OLG must be brought to below unity by the pole frequency, and this too involves some compromises. There is no free lunch, and both approaches can be made to work very well. Andrew's point that the CM is not perfectly balanced because of the base bias on the VAS is absolutely true, too, and is only nulled by using asymmetrical CM degeneration resistors. But that's another story....
Glen,
Calm down. Your advice was certainly sound. Perhaps you could be more sociable and less patronising. People are always free to accept or reject advice.
Cheers,
Hugh
The LTP cares only about the currents flowing through its transistors; these are responsible for diff action together with the differential input voltage at the two bases.
The input transistor would have around 24V Vce; the feedback transistor around 24V6. This would have no effect on diff action at all.
A current mirror increases OLG very considerably, increasing fb factor, and thus improving THD, bandwidth and Zout. The increased feedback factor focusses attention on amp stability; the OLG must be brought to below unity by the pole frequency, and this too involves some compromises. There is no free lunch, and both approaches can be made to work very well. Andrew's point that the CM is not perfectly balanced because of the base bias on the VAS is absolutely true, too, and is only nulled by using asymmetrical CM degeneration resistors. But that's another story....
Glen,
Calm down. Your advice was certainly sound. Perhaps you could be more sociable and less patronising. People are always free to accept or reject advice.
Cheers,
Hugh
Hi,
most say that the collector load on the inverting input is of no consequence to the extent they even omit the collector load.
On symmetry grounds alone I favour equal loads on both sides, but a CCS diode across the inverting load resistor may lead to a fight between the CCS tail current and the two collector load currents.
If the VAS voltage loses out in this fight as it drops out of balance and back in and then dominating the current flows, it may send weird values of current into the VAS and thus the VAS output may not represent a good copy of the non-inverting input.
most say that the collector load on the inverting input is of no consequence to the extent they even omit the collector load.
On symmetry grounds alone I favour equal loads on both sides, but a CCS diode across the inverting load resistor may lead to a fight between the CCS tail current and the two collector load currents.
If the VAS voltage loses out in this fight as it drops out of balance and back in and then dominating the current flows, it may send weird values of current into the VAS and thus the VAS output may not represent a good copy of the non-inverting input.
Juergen,
I'm sorry, I've been sleeping!
If by clamping action you mean some reduction in freedom of current variation, I cannot see that this is a good thing unless it happens at very high frequencies, approaching the pole, where Av must be pulled back to below unity to meet Bode/Nyquist criteria.
Could you explain a little more what you mean?
Cheers,
Hugh
I'm sorry, I've been sleeping!
If by clamping action you mean some reduction in freedom of current variation, I cannot see that this is a good thing unless it happens at very high frequencies, approaching the pole, where Av must be pulled back to below unity to meet Bode/Nyquist criteria.
Could you explain a little more what you mean?
Cheers,
Hugh
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.