millwood said:
Is using a higer power diff. and VAS needed for some odd reason?
It seems to me that power is not really a factor in creating a larger voltage. That is kind of more related to the impeadence that you are putting this voltage accross. If you want to make power, add current to your voltage signal. You don't have to do this in one step, or stage.
Lets say for example you use ECG 26 type transistor as a VAS. This is a good choice for a VAS transistor because its physical construction particulars make it less inheirent to white and other noise. If you operate it at or around 1 mA, then the voltage signal created is very clean; [Vceo=120V, Ic max=50mA, Pd=0.2W max Hfe=250 & f=200MHz case style SP-92]A diode biased class AB emitter follower circuit with constant current sources to bias it, direct coupled will this high impeadence VAS will do the same job as one transistor, but with a lot less effort and noise. The VAS in my Hi Fi amp bias class A at Ico=740uA. Zout =300K Ohms. Zout of first emitter follower circut is ~8K Ohms --- then 120 Ohms ---and finally 4 Ohms
this brings me to the diff. amp. How much current do you think it takes to drive this VAS?
My diff amp circuit uses about 800uA total and each transistor Pd = 14mW. (they must be equal or you will never get DC stability on accout of thetemp. coefficient) and my DC offset is never more than 10-15 mV except for the first 30 sec or so of being turned on and then is never more than 50 mV DC even with a full AC load.
Using low current allows me to use small high gain high freq transistors and with such small current these devices will operate really fast in order to generate inverse odd harmonics for better sound, gives you a higher input impeadence & sensitivity,and a whole lot less
That's a bit like asking how will my chilli taste if I use black pepper rather than white pepper. There are many other things adding to the flavour and texture, and they interact with the pepper in complex ways.
Any input stage will operate more linearly if the bias current is much higher than the signal current. BJTs tend to have exponential gm whereas FETs have square-law gm. You can calculate a BJTs gm by multiplying it's bias current by 40 at room temperature: gm = 40xIc. The gm of a FET is much less predictable and varies considerably from one device to another.
The great thing about FETs are their extremely high dc GS resistance. This means you could run your LTP with 100A of bias and have virtually no input current. This would make an extremely linear differential. The problem with FETs are their unpredictable gm and their higher capacitance (both GS and GD). They need matching very carefully to get similar linearity to BJTs.
As a rule of thumb: for an audio amp (which needs quite high input resistance 30k+) I'd use BJTs when the bias current is <5mA and the signal current is <0.05mA. I'd use JFETs for higher bias currents and/or if the signal current needs to be higher (often the case with low feedback designs). Remember that matching JFETs is a pain in the ****. I am not aware of any "matched pairs" in a single package like you get with BJTs.
Hi, Tradebam,
Thanks for the explenation
I wanted to built this 2nd harmonic preamp, by using single ended design, like a mosfet with only a resistor in its source to -rail, but I wanted the gain to be adjusted, so it will need to put base resistor (about 3k3) and feedback resistor from gate to drain (like 3k3+VR10k to make adjustable gain).
This way the mosfet will form inverting preamp (the output is taken from drain, with about 3k3resistor from +rail)
This will be single ended, but not open loop. Will this design have the needed 2nd harmonic?
Thanks for the explenation
There is a term "the signal current is <0.05mA". What is "the signal current"?
Is the 2nd harmonic produced more by "Single Ended" or "Open loop"?
I wanted to built this 2nd harmonic preamp, by using single ended design, like a mosfet with only a resistor in its source to -rail, but I wanted the gain to be adjusted, so it will need to put base resistor (about 3k3) and feedback resistor from gate to drain (like 3k3+VR10k to make adjustable gain).
This way the mosfet will form inverting preamp (the output is taken from drain, with about 3k3resistor from +rail)
This will be single ended, but not open loop. Will this design have the needed 2nd harmonic?
lumanauw said:
it will have mostly 2nd and 3rd harmonics. I put such a design together a while ago (under the thread "simple pre amp" I think). Nelson has a similar design as well, except that his is a full symmetry.
It sounds very well, to my ears and allows you to use pretty much whatever transistors you want.
lumanauw said:
I'm not to experienced with mosfet power amps, but in the BJT circuit I am making there are two class AB emitter follower stages in succession that drive the output stage, so I can use small current in VAS. Zin is higher than Zout and if you use a constant current source(s) to bias the devices and there can be a big difference in Zin and Zout depending on Beta. Class AB EF stages can have a very small Ic idle. If used with voltage regulated pwr supply, will generate no noise. Smaller low current devices have higher beta and frequency responce(good for third harmonic cancelation in diff. amp.) to limit the symetrical or saturation distortion of SS.
You build up the current with each stage until you get to your speaker. I basically made a diff. and VAS to drive about 22V RMS accross 300K Ohms and used AB current amp stages(3) biased with current sources to give enough current to drive 4 Ohms sure it takes more parts, but they are small and the result is good.
Literature said differential cancels 2nd harmonic, but they do not mentioned cancelation of 3rd harmonic.
What can cancel 3rd harmonic?
Cunningham,
I have made experience with bias in VAS (bipolar VAS). I get impression, that smaller bias gives more detailed sound, up to a limit (the needed minimal current to drive the next stage). But this always drag to use of darlington or 3-4 darlington output stage (because we wanted small bias in VAS).
What do you think of CFP differential in post #17? The input transistor only works in 127uA (isn't that very low?)
But Doug Self book mentioned about noise in this CFP differential. He wrote that the compromising value of the R is 2k2 (in #17 it is 4k7). He wrote that bigger than that, the system will generate noise (Johnson Noise?)
What noise is he talking about? Is it audible, like hiss in the speaker, if in CFP differential the R is very big (like 10k) to get much smaller bias?
Furthermore, what is the advantage in that CFP differential, to have that low bias in the first transistor? Bipolar tends to work better in lower bias? How low can we go?
I'm talking about the change of current due to the music signal. Eg: the bias thorugh one transistor may be 500uA. When the amp is playing music this current might vary by +/- 50uA. The latter is what I mean by signal current. I'm simply saying that the change of current must be small compared with the bias current.
Lumanauw,
I'm not sure, but it looks like the diff circuit in post #17 that the two transistors on the bottom are the current sources and the two in the middle facing each other (NPN) are voltage followers to lower the Z ; increase the current. This would allow the Diff pair(PNP) to operate with only 127uA and still funtion properly, also raising the input impeadence. This is somewhat clever looking and I would be interested in how it performs, like two stages cramed into one. I believe I read someware on this thread that the signal current with relation to the bias current should be low and do agree with that. In this case it would take a coulple of micro-amps of signal current to drive this diff. circuit; higher input Z.
This circuit sort of has the properties of a feedback pair follower circuit. I see the feedback pair technique used on some amp designs from this forum. You can get very high gain this way, but the feedback pair is rarely used without neg. feedback. Makes a good switch though.
Noise can be audio, subaudio, DC, or even RF. On the circuit I am making right now, I had a problem with RF occilation until I worked out a miller compensation or bypass shceme. Occilation was at about 100 KHz and roughly 5 W of power, not sure excactly, but it is a good thing that radio stations don't use this freq. or I would have been causing a lot of interference!!
Any noise that goes into and is in the bandwidth responce of the amp circuit is going to be bigger noise on the other end. That is determained usually by the signal to noise ratio. RF noise can usually be filtered out, Zobel networks and such; or by just mounting your circuit in a metal enclosure.
I'm not sure, but it looks like the diff circuit in post #17 that the two transistors on the bottom are the current sources and the two in the middle facing each other (NPN) are voltage followers to lower the Z ; increase the current. This would allow the Diff pair(PNP) to operate with only 127uA and still funtion properly, also raising the input impeadence. This is somewhat clever looking and I would be interested in how it performs, like two stages cramed into one. I believe I read someware on this thread that the signal current with relation to the bias current should be low and do agree with that. In this case it would take a coulple of micro-amps of signal current to drive this diff. circuit; higher input Z.
This circuit sort of has the properties of a feedback pair follower circuit. I see the feedback pair technique used on some amp designs from this forum. You can get very high gain this way, but the feedback pair is rarely used without neg. feedback. Makes a good switch though.
Noise can be audio, subaudio, DC, or even RF. On the circuit I am making right now, I had a problem with RF occilation until I worked out a miller compensation or bypass shceme. Occilation was at about 100 KHz and roughly 5 W of power, not sure excactly, but it is a good thing that radio stations don't use this freq. or I would have been causing a lot of interference!!
Any noise that goes into and is in the bandwidth responce of the amp circuit is going to be bigger noise on the other end. That is determained usually by the signal to noise ratio. RF noise can usually be filtered out, Zobel networks and such; or by just mounting your circuit in a metal enclosure.
The noise that I'm asking is the noise generated if the 4k7 resistor in the schematic is swapped by bigger value, like 10k or more. Doug Self stated this on his book (section about CFP differential), he wrote that the optimal value for this R is 2k2. Bigger than that it will produce noise.
What noise occurs if we put big resistance in the place of 4k7 resistor?
Is that noise generated because the I bias at first transistor is so low (lower than 100uA=making noise?)
What noise occurs if we put big resistance in the place of 4k7 resistor?
Is that noise generated because the I bias at first transistor is so low (lower than 100uA=making noise?)
Mikeks! I'm so tired. It is hard to understand all post .Which post can i read to achieve my dream?
"Why?" , What do you want to ask?
Can opamp Ha-5033 cancel offset voltage? http://www.stanford.edu/class/ee214/Handouts/ho15opamp.pdf
.Which post can i read to achieve my dream?"Why?" , What do you want to ask?
Can opamp Ha-5033 cancel offset voltage? http://www.stanford.edu/class/ee214/Handouts/ho15opamp.pdf
lumanauw said:
I am not familiar with Douglas Self, I do most of my own circuit designing. Anyway, 4.7K is Rc for the diff. pair. I would think that raising the impeadence of the diff. pair would make it more succeptable to noise, maybe RF? It would certainly increase the voltage gain. Perhaps this is what he is referring to. Everything has noise and this would increase the Voltage of the noise. I'm not certain what kind of noise Mr. Self is referring to, but I suspect it is Radio Frequency noise.
Keep in mind that the AC load line for each diff. transistor doesn't have to be equal, but may require rebiasing.
If your circuit is direct coupled to the speaker, the power of each device must be equal or you will get DC drift on the output. It helps if your bias current is low, that you mount your circuit in a metal box in order to shield RF from the very sensitive input.
In the circuit that I am making, the diff. pair bias at idle is: Q1=450uA & Q2=330uA but Q1-Rc=33kOhms and Q2-Rc=15kOhms
The output current to the VAS transistor is only 3uA.
If I even touch the base of VAS transistor, it makes all kinds of really ugly noise. Some audiable and some RF and changes the DC offset of my entire circuit.
. I found that a lot of my problems due to high freq. and noise were controlled by simply mounting the circuit within a metal enclosure.
thanh said:
thanh, mikeks is right: those fast amps using ltp are current feedbck amps (mikeks prefers to decribe them with a different name tho.,
).you can take a look at some current feedback opamps and realize them with discrete parts. Go to TI.com and check out an application note named "Voltage Feedback vs. Current Feedback Op Amps".
You can also check out some of John Curl's designs. They tend to be fully symmetrical and extremely fast too. the schematic for JC1 i believe was extensively discussed here, with John himself participating (lots of insights from a real expert, not those wannabes here).
Good luck.
millwood said:
thanh, mikeks is right: those fast amps using ltp are current feedbck amps (mikeks prefers to decribe them with a different name tho.,
There is no such thing as a 'current' feedback amp. with a diff. input stage...
http://www.anasoft.co.uk/EE/currentfeedbackmyth/currentfeedbackmyth.html
I have just simulated Slone's circuit with 50V supply.,gain =10K/330 =33 . When output voltage is 40, it tend to convert a sine wave into a square wave
I also tested with single ended input, the result is the same...
Mikeks!http://www.anasoft.co.uk/EE/current...edbackmyth.html
Bjt + fully symmetry = odd harmonics can be increase 2 times
VFB : voltage feedback -voltage subtraction
CFB : voltage feedback- current subtraction
I think current feedback-voltage subtraction and current feedback-current subtraction are not suite for audio amp
In my country there is a litle audio book . I have read a vietnamese audio book which I think it is translated from "Randy Slone,Slone, High-Power Audio Amplifier Guidebook with Projects: 50 to 500 Watts for the Audio Perfectionist " ,a slone's books. I know CFB amp by searching in internet.
I also tested with single ended input, the result is the same...
Mikeks!http://www.anasoft.co.uk/EE/current...edbackmyth.html
Bjt + fully symmetry = odd harmonics can be increase 2 times
VFB : voltage feedback -voltage subtraction
CFB : voltage feedback- current subtraction
I think current feedback-voltage subtraction and current feedback-current subtraction are not suite for audio amp
In my country there is a litle audio book . I have read a vietnamese audio book which I think it is translated from "Randy Slone,Slone, High-Power Audio Amplifier Guidebook with Projects: 50 to 500 Watts for the Audio Perfectionist " ,a slone's books. I know CFB amp by searching in internet.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.