Simple Symetrical Amplifier

[guitar89]

Originally Posted by Esperado

There is something I dont understand: how can your Q3, Q4 bases be at +15V and -15V while their emitters are at 0V ?

[edit] Yes, experimenting in real world and listening seems to me the only way to increase his experience and make-it part of yourself. I'm not sure that all this virtualisation gives no good. It saves time, (money) but kill the soul. (Yes, amps had some kind of soul when they were handcrafted)

For those input transistor, if you notice carefully, there are "A" and "B", which is cross-connection. I only can do vertical or horizontal line, but not slant line, so I use those for connection. Those they are not connected to ground.

I believe so, but my pocket doesn't says that. (i'm still a young student with not much allowance) I personally doesn't particular about time, since these things is our passion (SOUL )

[guitar89]

esperado,
For your information, it works in reality and doesn't even felt different temperature from ambient.
For Transistor to conduct, they must have potential different. Take an example, Q4 have base on 15V, but on emitter end, in order to conduct, it depends on collector of Q1, which is connected through "B". (Q4's emitter to Q1's collector)
Thus I don't think there is possible way for it to burst, melt or 'flame'(burn).

[Esperado]

Originally Posted by guitar89

Thus I don't think there is possible way for it to burst, melt or 'flame'(burn).

Forget, just your schematic was not readable and looked (to me) like the two Q3/Q4 emitters where connected on the same line.

[guitar89]

Originally Posted by Esperado

Forget, just your schematic was not readable and looked (to me) like the two Q3/Q4 emitters where connected on the same line.

I'm agree with you that its not nice ^^ was planning to make a new schematic

[Lazy Cat]

Originally Posted by guitar89

LC, there is a funny thing on my power supply too. Although it is +-40V but it is only 120VA, means for stereo, only 60VA each channel. With load variation from 2ohm - 16ohm, the maximum voltage swing is only +-20V before my trafo gone heaven. (Hope my calculation is correct)
So i guess the 2SA1707/2SC4487 is still suitable for use ? since I have no hope to reach higher....

This transformer is not really a pearl in a necklace it is more like auxiliary spare wheel in a car. For one channel maybe conditionally acceptable but for stereo amp way out of a good choice. Such power supply will worsen stereo separation, cause unstable behaviour, PSRR of this amp demands low rails ripple but yours will be as Himalayas ...

Originally Posted by guitar89

LC, what i'm thinking is such design as attached.

By using those, I get very high open loop gain and low THD.
The VAS are biased as 0.7mA & 10mA
The driver is biased to 1-2mA & 20mA to Output device.

what could be the good number for C13, which is around the OPS, between the Drivers ?

I intend to change those drivers to the 2SA1707/2SC4487 (didn't do in SPICE because no model.....)
and the OPS change from 0281/0302 to NJL4281/4302. (In SPICE, the number is quite outrages......)

Hope it will turn out into a very nice X.x

Your choice of VAS transistors are not optimum, I'm telling you this at least in 10 posts so this will be my final advice regarding them. Please do not use TO-220 BJT's in VAS. Single VAS transistor has to have minimum 100 Vce and beta more than 300. Cascoded VAS: gain transistor's beta more than 500, casode BJT 100 Vce. Double MJE like in your sch is completely wrong choice for the purpose. I become shivered everytime I see those gorillas in the sensiteve VAS position.

[Lazy Cat]

Originally Posted by Sheldon

I've got one channel of ThermalTrak Basic built: http://www.diyaudio.com/forums/attac...rack-basic.jpg

Only significant difference is 2SA1360/2SC3423 for the drivers. Resistor values are as shown, except 2.2R base stoppers for the output devices. Before I fire it up, I could use some guidance on a starting point for the 5K trim pots that parallel the 2.2k resistors into the feedback network. How much current do we want through the zeners, and how much into the feedback network?

Sheldon

Hi Sheldon

Please carefully follow this instructions:
- set 50k MT trimmer in the middle position, this one nulles input offset to zero, so there is no difference in the output DC if you shortened input with wire to GND or if the input is completely opened
- please change both 2,2 k resistors to 5,6 k and start with max resistance of both 5 k MT. These two defines VAS bias current to app. 10-12 mA and at the same time zero output DC offset.
- set 50 ohm MT to zero at start, this one defines output bias current to app. 150 mA

Your choice of VAS (driver) transistors is very good, maybe these two will need small 10-22pF local feedback ceramic caps (Miller compensation) between their collector to base pins (as close as possible) but only in a case if you notice some oscillations at the output.

[guitar89]

Originally Posted by Lazy Cat

This transformer is not really a pearl in a necklace it is more like auxiliary spare wheel in a car. For one channel maybe conditionally acceptable but for stereo amp way out of a good choice. Such power supply will worsen stereo separation, cause unstable behaviour, PSRR of this amp demands low rails ripple but yours will be as Himalayas ...



Your choice of VAS transistors are not optimum, I'm telling you this at least in 10 posts so this will be my final advice regarding them. Please do not use TO-220 BJT's in VAS. Single VAS transistor has to have minimum 100 Vce and beta more than 300. Cascoded VAS: gain transistor's beta more than 500, casode BJT 100 Vce. Double MJE like in your sch is completely wrong choice for the purpose. I become shivered everytime I see those gorillas in the sensiteve VAS position.

wow, mine is himalayas ? X.x i'm going to get another Transformer later I guess..... I will get another transistor as you said to fit.
The MJE in the schematic was substitute X.x it is indeed an gorilla instead of cheetahs.... If in cascode arrangment, I use MJE as cascode BJT, and the 2SA1707/2SC4487 temporarily, is it fine ? (seems the transistor have insufficient beta)

[Lazy Cat]

Originally Posted by Sheldon

How much current do we want through the zeners, and how much into the feedback network?

Sheldon

Hi Sheldon

To rule out any misunderstanding you have to measure three things only:
- output voltage must be measured with oscilloscope to see if you have any oscillation or DC offset. Whatever you do with trimmers, never exceed output DC over +/- 1 V
- voltage drop across 10 ohm VAS emitter resistor must be 100-120 mV optimally, in this case your VAS bias current will be 10-12 mA and input bias will be automatically around 2,3 mA
- voltage drop across 0,22 ohm output emitter resistor must be 33 mV optimally, in this case output bias current will be 150 mA

Regards Andrej

[Sheldon]

Originally Posted by Lazy Cat

- set 50k MT trimmer in the middle position, this one nulles input offset to zero, so there is no difference in the output DC if you shortened input with wire to GND or if the input is completely opened
- please change both 2,2 k resistors to 5,6 k and start with max resistance of both 5 k MT. These two defines VAS bias current to app. 10-12 mA and at the same time zero output DC offset.
- set 50 ohm MT to zero at start, this one defines output bias current to app. 150 mA

Your choice of VAS (driver) transistors is very good, maybe these two will need small 10-22pF local feedback ceramic caps (Miller compensation) between their collector to base pins (as close as possible) but only in a case if you notice some oscillations at the output.

Thanks Andrej,

Check, on the input trimmer.

Will find some 5.6k or close. Setting the trimmers to max resistance makes an easy starting point, with defined minimum. I have been curious though; don't these rail trimmers make the input trimmer redundant? Or is trimming the output offset with these respond differently to changing conditions, than adjusting the input trimmer?

Check on the bias spreader trimmer to zero.

Originally Posted by Lazy Cat

To rule out any misunderstanding you have to measure three things only:
- output voltage must be measured with oscilloscope to see if you have any oscillation or DC offset. Whatever you do with trimmers, never exceed output DC over +/- 1 V
- voltage drop across 10 ohm VAS emitter resistor must be 100-120 mV optimally, in this case your VAS bias current will be 10-12 mA and input bias will be automatically around 2,3 mA
- voltage drop across 0,22 ohm output emitter resistor must be 33 mV optimally, in this case output bias current will be 150 mA

Easy enough, thanks.

Currently I have 0.1R output emitter resistors. I had left those in there from my experiments with Mihai's (Roender) output buffer. He recommended a voltage drop across the emitter resistor of about 18mV for the optimum class B bias. Though he did recommend that if a different value is chosen, it's better to be higher than lower. I had thought that this value is a property of the particular output device. You are recommending about double that. I can do either way.

BTW, my application for this amp with be driving the mid/highs for a 100dB speaker (8 Ohm). I won't need more than 20W for even the highest transients, at live levels. Most listening will be much less. So for normal to somewhat loud, it would be nice to live in class A.

Sheldon

[Lazy Cat]

Hi Sheldon

Base currents of the input transistors are not equal so this difference always causes voltage drop on input 10 k resistor therefore some DC to be present, usually few mV. So if you connect the source which has zero DC on its output (same if you connect input to the ground) the output of the amp will go to DC level of voltage gain multiply few mV, which is undesirable. 50 k MT input offset trimmer is serving to cancel this by supplying (via 1 M series resistor) the current to the bases of input transistors needed to compensate base currents (sum of all three currents is zero - Kirchhoff's current law) thus in effect the voltage drop on 10 k input resistor is zero.